Oxford University Press is a department of the University of Oxford.

It furthers the University’s objective of excellence in research, scholarship

and education by publishing worldwide. Oxford is a registered trade mark of
Oxford University Press in the UK and in certain other countries.
Published in South Africa by
Oxford University Press Southern Africa (Pty) Limited
Vasco Boulevard, Goodwood, N1 City, Cape Town, South Africa, 7460
P O Box 12119, N1 City, Cape Town, South Africa, 7463
© Oxford University Press Southern Africa (Pty) Ltd 2015
The moral rights of the author have been asserted.
First published 1999
Sixth Edition published in 2015
Seventh Edition published in 2019
All rights reserved. No part of this publication may be reproduced, stored in
a retrieval system or transmitted, in any form or by any means, without the
prior permission in writing of Oxford University Press Southern Africa (Pty) Ltd,
or as expressly permitted by law, by licence or under terms agreed
with the appropriate reprographic rights organisation, DALRO, The Dramatic, Artistic
and Literary Rights Organisation at dalro@dalro.co.za. Enquiries concerning
reproduction outside the scope of the above should be sent to the Rights Department,
Oxford University Press Southern Africa (Pty) Ltd, at the above address.
You must not circulate this work in any other form
and you must impose this same condition on any acquirer.
Public Economics Seventh edition
Print ISBN: 978-0-190723-18-7
ePUB ISBN: 978-0-190449-18-6
Typeset in Utopia Std 10pt on 12pt
Acknowledgements
Publisher: Janine Loedolff
Editor: Language Mechanics
Managing designer: Judith Cross
Typesetter: Aptara
Cover reproduction by: xxxxx
The authors and publisher gratefully acknowledge permission to reproduce copyright material in this book. Every effort has been made to trace
copyright holders, but if any copyright infringements have been made, the publisher would be grateful for information that would enable any
omissions or errors to be corrected in subsequent impressions.
Links to third party websites are provided by Oxford in good faith and for information only. Oxford disclaims any responsibility for the
materials contained in any third party website referenced in this work.
Table of Contents
Contributors
Preface

PART 1 Perspectives on the role of government in the economy

The public sector in the economy

1.1 Introduction
1.2 Legacy of the past
1.3 Fiscal challenges
1.4 e study eld of public economics
1.5 e public sector in South Africa
1.5.1 Composition of the public sector
1.5.2 Size of the public sector
1.5.3 e relationship between the public and the private sectors
Key concepts
Summary
Multiple-choice questions
Short-answer questions
Essay questions

Benchmark model of the economy: Positive and normative approaches

2.1 Basic assumptions of the benchmark model
2.2 e benchmark model and allocative efficiency
2.2.1 Condition 1
2.2.2 Condition 2
2.2.3 Condition 3
2.3 Efficiency and economic growth
2.4 Market failure: An overview
2.4.1 Lack of information
2.4.2 Friction and lags in adjustments
2.4.3 Incomplete markets
2.4.4 Non-competitive markets
2.4.5 Macroeconomic instability
2.4.6 Distribution of income
2.4.7 Further notes on market failures
2.5 Enter the public sector: General approaches
2.5.1 Allocative function
2.5.2 Distributive function
 2.5.3 Stabilisation function
2.6 Direct versus indirect government intervention
2.7 Note on government failure
Key concepts
Summary
Multiple-choice questions
Short-answer questions
Essay questions

Public goods and externalities

3.1 Private goods and the benchmark model
3.2 Pure public goods: De nition
3.3 e market for public goods
3.4 Who should supply public goods?
3.5 Mixed and merit goods
3.6 Externalities
3.6.1 Negative production externality
3.6.2 Positive consumption externality
3.6.3 Concluding note
3.7 Possible solutions to the externality problem
3.7.1 Pigouvian taxes and subsidies
3.7.2 Direct regulation
3.7.3 Creation of regulated markets
3.7.4 Support for alternative markets
3.7.5 Legal solution: Property rights
3.8 Global public and merit goods
Key concepts
Summary
Multiple-choice questions
Short-answer questions
Essay questions

Allocative efficiency, imperfect competition and regulation

4.1 e social costs of statutory monopolies

4.2 Natural monopolies
4.2.1 Characteristics and effects
4.2.2 An overview of policy options
4.3 Competitive restructuring
4.4 Privatisation
4.5 Regulation
4.5.1 Rate-of-return regulation
4.5.2 Price-cap regulation
4.5.3 Sliding-scale regulation
4.5.4 Regulation in developing countries
Key concepts
Summary
Multiple-choice questions
Short-answer questions
Essay questions

Equity and social welfare

5.1 Introduction
5.2 Nozick’s entitlement theory
5.3 Other Pareto criteria
5.4 Bergson criterion
5.5 Efficiency considerations
Key concepts
Summary
Multiple-choice questions
Self-assessment exercises
Essay question

Public choice theory

6.1 Introduction
6.2 e unanimity rule and the Rawlsian experiment
6.3 Majority voting and the median voter
6.4 e impossibility theorem
6.5 Majority voting and preference intensities
6.6 Optimal voting rules
6.7 Government failure
6.7.1 Politicians
6.7.2 Bureaucratic failure
6.7.3 Rent-seeking and corruption
6.8 Concluding note
Key concepts
Summary
Multiple-choice questions
Short-answer questions
Essay questions

PART 2 Public expenditure

Public expenditure and growth

7.1 e constitutional framework
7.1.1 e Constitution and public goods
7.1.2 Constitutional entitlements
7.2 e size, growth and composition of public expenditure
7.2.1 Size and growth of public expenditure
7.2.2 e changing economic composition of public expenditure
 7.2.3 Functional shifts in public expenditure
7.3 Comparisons with other countries
7.4 Reasons for the growth of government: Macro models
7.4.1 Wagner and the stages-of-development approach
7.4.2 Peacock and Wiseman’s displacement effect
7.4.3 e Meltzer-Richard hypothesis
7.5 Micro models of expenditure growth
7.5.1 Baumol’s unbalanced productivity growth
7.5.2 Brown and Jackson’s microeconomic model
7.5.3 Role of politicians, bureaucrats and interest groups
7.6 Government and the economy: Long-term effects
Key concepts
Summary
Multiple-choice questions
Short-answer questions
Essay questions

Government intervention to reduce inequality and poverty

8.1 Inequality and poverty in SADC countries
8.2 e budget, redistribution and poverty alleviation
8.2.1 Changing the distribution of primary income
8.2.2 Changing the distribution of secondary income
8.2.3 Broad considerations for the design and assessment of policies
8.3 Targeting of government spending programmes
8.3.1 e bene ts and limits of targeting
8.3.2 e costs of targeting
8.3.3 e effectiveness of targeting
8.4 Fiscal incidence
Key concepts
Summary
Multiple-choice questions
Short-answer questions
Essay questions

Social insurance and social assistance

9.1 e two components of income security systems
9.2 Social insurance
9.2.1 e income protection role of insurance
9.2.2 e rationale for social insurance
9.3 Social assistance
9.3.1 e choice between cash and in-kind transfers
9.3.2 Conditional cash transfer programmes
9.4 Incentive effects of income protection systems
9.5 e South African income security system
9.5.1 Coverage
9.5.2 Scope and adequacy
 9.5.3 e effectiveness of South Africa’s social assistance programmes
Key concepts
Summary
Multiple-choice questions
Short-answer questions
Essay questions

Social services
10.1 Arguments for government intervention in education and healthcare
10.1.1 Education
10.1.2 Healthcare
10.2 e case for in-kind subsidies
10.3 Social service delivery
10.4 Service delivery in education in South Africa
10.4.1 Resources and outputs in the South African basic education system
10.4.2 Educational outcomes in South Africa
10.4.3 Service delivery issues in basic education in South Africa
10.5 Service delivery in healthcare in South Africa
10.5.1 e South African health system
10.5.2 Levels, growth and composition of government spending on health in South Africa
10.5.3 Access and quality of health services
Key concepts
Summary
Multiple-choice questions
Short-answer questions
Essay questions

PART 3 Taxation

Introduction to taxation and tax equity

11.1 Sources of nance
11.2 De nition and classi cation of taxes
11.2.1 Tax base and rates of taxation
11.2.2 General and selective taxes
11.2.3 Speci c and ad valorem taxes
11.2.4 Direct and indirect taxes
11.3 Properties of a ‘good’ tax
11.4 Taxation and equity: Concepts of fairness
11.4.1 Bene t principle
11.4.2 Ability-to-pay principle
11.5 Tax incidence: Partial equilibrium analysis
11.5.1 Concepts of incidence
11.5.2 Partial equilibrium analysis of tax incidence
11.6 General equilibrium analysis of tax incidence
11.6.1 A selective tax on commodities
 11.6.2 A general tax on commodities
11.7 Tax incidence and tax equity revisited
Key concepts
Summary
Multiple-choice questions
Short-answer questions
Essay questions

Tax efficiency, administrative efficiency and flexibility

12.1 Excess burden of taxation: Indifference curve analysis
12.1.1 Lump-sum taxes and general taxes
12.1.2 Selective taxes
12.1.3 Tax neutrality
12.2 Excess burden: Consumer surplus approach
12.2.1 e magnitude of excess burden
12.2.2 Price elasticities
12.2.3 e tax rate
12.3 Administrative efficiency
12.4 Flexibility
Key concepts
Summary
Multiple-choice questions
Short-answer questions
Essay questions

Personal income taxation

13.1 e comprehensive income tax base

13.1.1 International taxation of income: e residence principle versus the source principle
13.2 e personal income tax base
13.2.1 Exclusions
13.2.2 Exemptions
13.2.3 Deductions
13.2.4 Rebates
13.3 e personal income tax rate structure: Progressiveness in personal income taxation
13.4 Economic effects of personal income tax
13.4.1 Economic efficiency
13.4.2 Equity
13.4.3 Administrative efficiency and tax revenue
13.4.4 Flexibility
Key concepts
Summary
Multiple-choice questions
Short-answer questions
Essay questions
 Company income tax, capital gains tax and income tax reform
14.1 Why company taxation?
14.2 e company tax structure
14.2.1 e company income tax base
14.2.2 Types of company tax
14.2.3 Taxation of small and medium-sized enterprises
14.3 e economic effects of company tax
14.3.1 Economic efficiency
14.3.2 Company taxation and investment
14.3.3 Fairness
14.4 Capital gains tax
14.5 Income tax reform
14.5.1 e at rate income tax
14.5.2 e dual income tax
Key concepts
Summary
Multiple-choice questions
Short-answer questions
Essay questions

Taxation of wealth
15.1 Wealth and types of wealth taxes
15.2 Why tax wealth?
15.2.1 Equity considerations
15.2.2 Efficiency considerations
15.2.3 Revenue and administrative considerations
15.3 Property taxation
15.3.1 e tax base
15.3.2 Tax rates
15.3.3 Assessment
15.3.4 Equity effects
15.3.5 Efficiency effects
15.3.6 e unpopularity of property taxation
15.4 Capital transfer taxes
15.4.1 Economic effects of capital transfer taxes
15.4.2 Capital transfer taxation in South Africa
Key concepts
Summary
Multiple-choice questions
Short-answer questions
Essay questions

Taxes on goods and services, and tax reform

16.1 Types of indirect taxes
16.2 Indirect taxes: A general critical assessment
16.3 Value-added tax
16.3.1 e economic effects of VAT
16.4 Personal consumption tax
16.4.1 e rationale for a personal consumption tax
16.4.2 e disadvantages of a personal consumption tax
16.5 Tax reform: International experience
16.5.1 Patterns of taxation in industrialised and developing countries
16.5.2 International tax reform
16.5.3 Globalisation and tax reform
16.5.4 International tax competition and tax harmonisation
16.6 Tax reform in South Africa
Key concepts
Summary
Multiple-choice questions
Short-answer questions
Essay questions

PART 4 Fiscal policy

Public debt and debt management

17.1 e concept of public debt
17.2 Public debt sustainability
17.3 Size and composition of the public debt
17.4 eory of public debt
17.4.1 Introduction
17.4.2 Internal versus external debt and the burden on future generations
17.4.3 Inter-temporal burden
17.5 Should the government tax or borrow?
17.5.1 Allocation (efficiency)
17.5.2 Distribution (equity)
17.5.3 Macroeconomic stability
17.5.4 Summary of views on the impact of debt
17.6 Public debt management
17.6.1 Introduction
17.6.2 Bonds and the cost of borrowing
17.6.3 Foreign borrowing
17.6.4 Public debt management objectives I and II: Minimisation of state debt cost versus
macroeconomic stabilisation
17.6.5 Public debt management objective III: Development of domestic nancial markets
17.6.6 Public debt management objectives IV: Financial credibility
17.6.7 Contingent liabilities
Key concepts
Summary
Multiple-choice questions
Short-answer questions
Essay questions

Fiscal policy
18.1 Introduction
18.2 e nature of scal policy
18.2.1 De nition
18.2.2 Fiscal policy and the budget constraint
18.2.3 Goals of scal policy
18.2.4 Instruments of scal policy
18.2.5 e scal authorities in South Africa
18.3 e macroeconomic role of scal policy
18.3.1 e Keynesian approach
18.3.2 Shortcomings of anti-cyclical scal policy
18.3.3 e structural approach to scal policy
18.3.4 Renewed interest in active scal policy
18.4 A brief re ection on the scal consequences of the Great Recession
18.5 Fiscal policymaking frameworks
18.5.1 Rules versus discretion in scal policy
18.5.2 Fiscal policymaking frameworks and scal outcomes
18.5.3 Concluding comment
18.6 Fiscal policy in South Africa
18.6.1 Macroeconomic aspects of scal policy in South Africa
18.6.2 e scal policymaking framework in South Africa
18.7 Fiscal reforms in sub-Saharan Africa
Key concepts
Summary
Multiple-choice questions
Short-answer questions
Essay questions

PART 5 Inter-governmental fiscal relations

Fiscal federalism
19.1 e economic rationale for scal decentralisation
19.1.1 e Tiebout model
19.1.2 Public choice perspective on scal federalism
19.1.3 Other reasons for scal decentralisation
19.2 Reasons for scal centralisation
19.3 Taxing and spending at sub-national level: e assignment issue
19.3.1 Stabilisation function
19.3.2 Distribution function
19.3.3 Allocation function
19.3.4 Tax assignment
19.4 Tax competition versus tax harmonisation
19.5 Borrowing powers and debt management at sub-national level
19.6 Inter-governmental grants
19.6.1 Unconditional non-matching grants
19.6.2 Conditional non-matching grants
19.6.3 Conditional matching grants (open-ended)
19.6.4 Conditional matching grants (closed-ended)
19.6.5 e rationale for inter-governmental grants
19.7 Inter-governmental issues in South Africa
19.7.1 Constitutional issues
19.7.2 Inter-governmental transfers and the Financial and Fiscal Commission (FFC)
19.7.3 Provincial nancing issues
19.7.4 Urban economics
Key concepts
Summary
Multiple-choice questions
Short-answer questions
Essay questions

References
Answers to multiple-choice questions
Index
Contributors
Tania Ajam is Associate Professor in Public Policy, Finance and Economics at the School of Public
Leadership at Stellenbosch University. She is a public policy analyst and an economist with broad
experience in the design, analysis and implementation of scal policy, intergovernmental scal relations
and government-wide monitoring and evaluation systems. She has served on the board of the South African
Reserve Bank, the Financial and Fiscal Commission and on the Davis Tax Review Committee. She was
appointed to the President’s Economic Advisory Council in 2019.

Estian Calitz is Emeritus Professor and a former Executive Director: Finance and Dean of Economic and
Management Sciences at the University of Stellenbosch. During South Africa’s period of political transition,
he was the Director of Finance of the national government (1993–1996). He is a former president of the
Economic Society of South Africa.

eo van der Merwe is Professor of Economics at the University of South Africa. He was chair of the
Department of Economics for more than seven years. He presently teaches Public Economics at
postgraduate level. His research interests include social security issues and the economics of language.

Krige Siebrits is a Senior Lecturer in the Department of Economics at the University of Stellenbosch. His
research interests include Public Economics and Institutional Economics.

Tjaart Steenekamp is retired Professor of Economics at the University of South Africa where he taught
Public Economics. His research interest includes the economics of taxation.

Ada Jansen is an Associate Professor in Economics at Stellenbosch University. She has a PhD in Economics
from Stellenbosch University, and lectures Public Economics to undergraduate and postgraduate students.
Her main research area is Public Finance, with a specialisation in taxation.

Ian Stuart is the acting head of the Budget Office at the National Treasury of South Africa. e division is
responsible for running the budget process, and preparing all budget documents. Before this, he headed the
Treasury’s scal policy unit. Ian is also a member of the IMF’s technical advisory panel on public nance
management, and a research fellow of the Stellenbosch University economics department. He holds a
MCom Economics from Stellenbosch University.

Philip Black was Extraordinary Professor of Economics at the University of Stellenbosch, a past President of
the Economic Society of South Africa and a previous Managing Editor of the South African Journal of
Economics. He taught Microeconomics and Public Economics at post-graduate level.
Preface
Public Economics provides a comprehensive introduction to the study of public economics within a South
African and a southern African context. In this book, theory is explained with reference to southern African
institutions, practices and examples. e aim of the book is to equip the student with basic analytical skills
and to demonstrate the application of these skills to practical issues.
Our emphasis is on developing the student’s understanding of the theoretical issues pertaining to the
role of government in a mixed economy as re ected in expenditure on government functions and the
nancing of such spending by means of various taxes and loans. e conceptual framework for the book is
that of a medium-sized, open developing economy, exposed to the forces of democratisation and
globalisation. Many textbooks offer public economics as a study in applied microeconomics. Although this
book largely follows the same approach, it also discusses the macro- and microeconomic dimensions of
scal policy.
In this seventh edition, we retain the approach followed in previous editions by including several cross-
references to and examples from countries making up the southern African region. In view of the
interrelatedness of the economies of southern African countries as well as cultural and historical links
between these countries, students from neighbouring countries should therefore also nd the contents of
the book accessible and relevant. We hope that the book will impart to southern African students greater
appreciation for the broader economic environment in which they operate.
As the book deals with constantly evolving policy and analytical issues, we have used the latest relevant
gures and information provided in recent issues of the Quarterly Bulletin of the South African Reserve Bank
and the 2019/20 Budget of the South African Government to update the material in this edition.
is edition retains several pedagogical features, including the following:
• Important concepts that are speci c to the discipline are reviewed and explicitly de ned in the text. For
easy reference, these concepts are marked in bold where they are rst mentioned. At the end of each
chapter they are listed as important concepts, with the relevant page number in parenthesis.
• A brief summary appears at the end of each chapter.
• Examples of multiple-choice questions are included at the end of each chapter. e correct option or
options for these questions can be found just before the Index at the end of the book.
• Each chapter also contains separate lists of examples of short-answer questions and essay questions.

Key changes
Part 1 provides perspectives on the role of government in the economy. e most important changes in
this edition are highlighted below.
• Various sections of Chapter 2 were partly or entirely rewritten, notably Sections 2.4.1, 2.4.4 and 2.5.1. In
addition, Figures 2.1 and 2.2 were replaced. Chapter 2 is now co-authored by Ada Jansen.
• Chapter 3 re ects substantial editorial revision. Section 3.4 now provides an important perspective on
the production of public goods and examples of such goods.
• Chapter 4 is now entitled ‘Allocative efficiency, imperfect competition and regulation’. e material on
competition policy in the sixth edition has been replaced by three new sections on competitive
restructuring, privatisation, and regulation. e chapter also contains two new boxes. Box 4.1 discusses
the experience of the South African electricity utility Eskom, which is experiencing serious nancial
difficulties. Box 4.2 provides a schematic illustration of the determination of electricity tariffs in South
Africa. Figure 4.3 replaces the one in the sixth edition. is chapter is now co-authored by Estian Calitz.
• In Chapter 5, Figure 5.5 has been changed to show the outward shift of the production possibilities curve
(PPC).
• In Chapter 6, Section 6.4 on the impossibility theorem was substantially rewritten and a new Figure 6.1
was added. Parts of Section 6.7.3 have also been rewritten to provide more clarity on rent-seeking and
Figure 6.4 is new. is chapter is now co-authored by Estian Calitz.

Part 2, which addresses public expenditure, has also seen some major changes.
• Feedback indicated that lecturers generally did not include the chapter on cost-bene t analysis in the
sixth edition in their curricula. Hence, this chapter is omitted from the seventh edition. is change
made it possible to discuss other topics in more depth.
• Chapter 7 (on public expenditure and growth) was updated and slightly revised.
• Chapter 8 is now the rst of three chapters on governments’ role in the reduction of poverty and
inequality. It retains material from Chapters 8 and 9 of the sixth edition, but the sections on targeting
(Section 8.3) and scal incidence (Section 8.4) were rewritten and expanded. Krige Siebrits wrote this
chapter.
• Chapter 9, also written by Krige Siebrits, remains focused on the social insurance and social assistance
components of social security systems. e material in the sixth edition on the characteristics and effects
of income security programmes in South Africa were retained and updated. is chapter also contains
new material on the income protection role of insurance (Section 9.2.1), the economic rationale for
social insurance programmes (Section 9.2.2), and the incentive effects of means tests (Section 9.4). e
sections on the pros and cons of cash and in-kind transfers (Section 9.3.1), conditional cash transfer
programmes (Section 9.3.2) and the incentive effects of income protection systems (Section 9.4) were
revised and expanded. Section 9.3.2 now includes Box 9.1, which provides an overview of Tanzania’s
Community-Based Conditional Cash Transfer Programme.
• Chapter 10 represents an expansion of the discussion of service delivery in education and health in
Chapter 8 of the sixth edition. e material on the delivery of social services in South Africa (Section 8.4
in the sixth edition) has been revised and updated. is chapter now also contains a new discussion of
the economic arguments for government intervention in the markets for education and healthcare. Krige
Siebrits wrote this chapter.

Part 3, which focuses on taxation, has also been updated and now incorporates the recommendations by
the Davis Tax Committee, inter alia. All the tax chapters are now co-authored by Ada Jansen. e most
salient changes in Part 3 are outlined below.
• Behavioural economics is growing in importance in public economics and elsewhere. Hence, Chapter 12
now contains a box (Box 12.1) that provides insights from behavioural economics into the causes of tax
evasion.
• Previous editions of this book discussed personal and corporate income tax in the same chapter. is
edition splits these two topics into separate chapters. Chapter 13 explores personal income tax, while the
new Chapter 14 is titled ‘Company income tax, capital gains tax and income tax reform’.
• Wealth tax is now discussed in Chapter 15, while Chapter 16 deals with taxes on goods and services as
well as tax reform in South Africa and further a eld.

Part 4 covers scal policy and public debt.

• e sequence of chapters in this part of the book was changed: the discussion of public debt (Chapter 17)
now precedes that of scal policy (chapter 18). is change made it possible to incorporate views on
public debt more clearly and fruitfully in the discussion of scal sustainability in Chapter 18. Ian Stuart
now co-authors Chapter 17.
• In the sixth edition, there was some duplication in the material about scal activism in the chapters on
scal policy and public debt. is has now been largely eliminated.
• A subsection on the public choice view in the public debt chapter of the sixth edition was moved to
become Section 18.3.2.2 in the scal policy chapter of this edition. e numbering of the subsequent
subsections of Section 18.3.2 changed accordingly.
• Section 18.4, which was Section 16.4 in the sixth edition, now has the title ‘A brief re ection on the scal
consequences of the Great Recession’”. is section was shortened and the somewhat complicated Figure
 16.4 in the sixth edition was omitted. A new Box 18.3 discusses scal consolidation approaches and the
expansionary scal contraction hypothesis.
• Section 16.5 in the sixth edition was rewritten as Section 18.5 to present the rules-versus-discretion
debate in the wider context of scal policymaking frameworks and to incorporate empirical ndings
about the connections between such frameworks and scal outcomes.

Part 5, which focuses on inter-governmental scal relations (Chapter 19), has not undergone major
changes, but has been updated.

is edition of Public Economics is once again the product of a long process of discussions and deliberations
among the authors, as well as discussions with colleagues and students who have used the earlier editions.
We are grateful to them. A special word of thanks is due to our colleagues at Oxford University Press
Southern Africa, Janine Loedolff (Publisher), Nicola van Rhyn (Senior Project Manager), and Liezl Roux
(Development Editor) for their strategic contributions, technical pro ciency and, above all, patience! We
also thank Language Mechanics for careful and professional language editing of the manuscript. Krige
Siebrits has replaced the late Philip Black as co-editor of the book.

Plan and outline of the book

We have three objectives with this book:
• To provide students with an opportunity to study public economics with reference to South African and
southern African institutions and practices
• To teach public economics as a eld of study characterised by the dynamic interaction and tension
between macro- and microeconomic considerations
• To explain public economics with reference to the enormous challenges posed by economic and political
reforms in the region and the growing integration of individual countries into the global and regional
economies.

e book is divided into ve main parts.

Part 1 outlines theoretical perspectives on the role of government in the economy. Chapter 1 begins
with an introduction to the study eld of public economics and a discussion of the nature of the public
sector in South Africa. It includes a discussion of the political and institutional context within which the
public sector functions as well as a brief overview of the major views on the role of government in the
economy. Chapter 2 explains the rationale for the role of government in the economy in terms of market
failure and distinguishes between the allocative, distributive and stabilisation roles of government. e next
two chapters examine the different dimensions of the government’s allocative role, focusing on public
goods, externalities and natural monopolies. Chapter 3 provides more detailed discussions of two important
sources of market failure, namely public goods and externalities. ese two types of market failures re ect
the incompleteness of many real-world markets. In addition to theoretical perspectives on public goods and
externalities, this chapter also outlines policy implications and options. ese include the regional and
global dimensions of some public goods and the international spillover effects of some externalities. Chapter
4 outlines the economic effects of monopolies and discusses policy interventions to rectify one
manifestation of this phenomenon, namely electricity supply. It highlights the social costs of imperfect
competition by contrasting outcomes under perfectly competitive and monopoly conditions and discusses
the nature and economic effects of natural monopolies. Both the traditional policy approach and the more
recent ‘new model’ approach towards decreasing-cost industries are explained. e discussion of the new
model covers its three elements (competitive restructuring, privatisation and regulation). Inequality as a
type of market failure is the focus of Chapter 5, which also examines the criteria for government
intervention. Experimental evidence that questions the validity of the self-interest hypothesis is also
discussed. Chapter 6 discusses the institutions and mechanisms of public (or social) choice, and examines
their efficiency and equity properties. It also explains the phenomenon of government failure.
Part 2 deals with public expenditure. Chapter 7 outlines theories of government expenditure growth and
comments on their explanatory value for South Africa. It also considers the long-run economic effects of
changes in government expenditure. is chapter includes a discussion of the role of infrastructure
investment in lowering production costs and ‘crowding in’ private investment. e important topics of
poverty and inequality in South Africa are discussed in Chapter 8. is chapter also discusses broad
conceptual and programme-design aspects of the redistributive and poverty-reducing roles of government,
targeting of government spending programmes, and scal incidence analysis. Chapter 9 focuses on social
insurance and social assistance. It provides an overview of the economic theory of insurance and explains
the need for government intervention to overcome the inability of private markets to provide insurance
against the effects of certain income losses. In addition, it discusses the choice between providing such
assistance as cash or goods and services, the pros and cons of adding conditions to cash transfer
programmes, and the incentive problems associated with income security programmes. e chapter
concludes with comments on various aspects of the South African income security system. Chapter 10
emphasises the importance of the effective provision of social services. It deals with the market failures that
justify government involvement in the provision of education and healthcare; the case for in-kind subsidies
(which are important policy instruments in the provision of social services), and effectiveness in the
provision of social services. e chapter also discusses aspects of public provision of education and
healthcare in South Africa.
A major part of this book is devoted to taxation (Part 3). In Chapter 11, the focus is on the principles of
taxation (the properties of a good tax, equity and tax incidence). Tax efficiency and the excess burden of a
tax, administrative efficiency and tax exibility are the main topics of Chapter 12. is chapter also includes
a discussion of tax evasion and measures to reduce it. e next four chapters analyse different types of taxes,
with the emphasis placed on theory as well as the South African and southern African experiences. Chapter
13 covers personal income tax (PIT). e impact of PIT on work effort and private savings is analysed.
Chapter 14 now focusses exclusively on company income tax, capital gains tax and income tax reform. It
explores the economic effects of company tax and explains two recent income tax reforms, namely the at
tax and the dual tax. e taxation of wealth (including a more detailed treatment of property taxation) is the
subject of Chapter 15. Chapter 16 focuses on taxes on goods and services, including the personal
consumption tax. is chapter concludes with a section on international tax reform and tax reforms in South
Africa since the late 1960s.
Two seemingly opposing forces have strongly in uenced debates about scal policy and the economic
role of government in South Africa. On the one hand, there are heavy demands on the scus to deal with
poverty, unemployment and the high degree of inequality. On the other hand, the imperative to maintain
macroeconomic stability and simultaneously promote sustained economic growth within a very competitive
global economy imposes severe constraints on the scus. is debate is captured in Part 4 (public debt and
scal policy). Chapter 17 (public debt and debt management) and Chapter 18 ( scal policy) are particularly
pertinent to the second objective of this book, that is, to highlight the dynamic interaction and tension
between macro- and microeconomic considerations in public economics. Chapter 17 deals with theories of
public debt and includes several sections on public debt management. Chapter 18 includes a discussion of
the evolution of views on the macroeconomic role of scal policy, focusing on the distinction between the
Keynesian and the structural approaches, and the choice between discretionary and rules-based scal
regimes. e chapter also provides an overview of the scal policy experiences of South Africa and other
countries in the southern African region.
e last part of the book (Part 5) deals with inter-governmental scal relations. In the sole chapter of
this part (Chapter 19), the theoretical issues pertaining to scal federalism are discussed. e structure of
government as embodied in the Constitution of South Africa and the implications for inter-governmental
scal relations in South Africa receive speci c attention.

How to use this book

As lecturers with experience at both residential and distance-learning universities, we (the authors of this
book) share a great deal of research and practical experience in different aspects of public economics and
scal policy. Hence, we designed this book with the following different kinds of students in mind:
• Students at distance-learning universities who will have a fairly self-contained text that the lecturer can
supplement with further explanations and tutorial exercises
• Residential students who will be able to study on their own, leaving lecturers time to analyse and debate
topical issues, discuss the self-assessment exercises and explain the more difficult material
• Postgraduate students who are completing a course in public economics for the rst time and need to do
self-study in order to digest more advanced material
• Postgraduate working students who did not complete an undergraduate course in public economics, but
who need to acquire this knowledge for non-degree purposes in order to advance their careers.

e material contained in this book may be more than can be digested in a semester course at
undergraduate level. If public economics is presented as a theoretical course in applied microeconomics,
the focus could be on Parts 1, 2, 3 and 5. If the course designer requires a good mix of applied
microeconomic and macroeconomic theory as well as practice in public economics, Part 4 is a must. A short
course in public nance could include Parts 1, 2, 3, 4 and 5, but the volume could be reduced by excluding
Chapters 4, 13, 14, 15, 16 (Sections 16.4–16.6). e material can therefore be adapted to suit speci c types of
courses.
Our educational approach is fairly straightforward. We explain public economics to students with a basic
understanding of macroeconomics and microeconomics at the undergraduate level. We make extensive use
of diagrams and some basic algebra. Each chapter begins with an introduction and a number of study
objectives. e student should use these objectives to maintain his or her focus on what is essential in the
chapter. We highlight important concepts in bold in the text and list them (together with page references) at
the end of each chapter. Each chapter also contains a summary, as well as examples of multiple-choice
questions and lists of short-answer questions and essay questions that the student can answer to test his or
her understanding of the material. A comprehensive literature list containing consulted sources is included
at the end of the book.
roughout the book, we have tried to present factual information about public economics in southern
Africa in relation to theory and international experience, rather than as separate and dull descriptions of
statistical trends and super cial features. We trust that this will enhance the relevance of both empirical
observation and theoretical understanding.

Estian Calitz
Tjaart Steenekamp
Krige Siebrits
The public sector in the economy

Estian Calitz

e aim of this chapter is to demarcate the study eld of public economics with reference to key issues and
scal challenges facing the South African economy and the Southern African region in general. e role of
government is fairly similar across the region, and in the sections and chapters that follow, we describe the
nature of the South African public sector against the backdrop of the theory of public economics and
contemporary views on the role of government in the economy. We also consider the interaction between
the public, household, and formal and informal private (business) sectors of the economy.

Once you have studied this chapter, you should be able to:
give a brief overview of different views of the role of government in the economy
list key issues confronting Southern African governments regarding their role in the economy
distinguish between the main institutional categories of the public sector
discuss the salient features of and trends in the size and composition of the South African public sector
discuss various aspects of the relationship between the public sector and the rest of the economy.

1.1 Introduction
In economics, we study the way in which society chooses to allocate its resources in order to satisfy a
multitude of needs and wants. As these resources are both scarce and have alternative uses, it is necessary
for society to prioritise its needs and ensure that they are met in a declining order of importance. e income
or budget constraint necessitates these choices. In the process, needs are converted into effective demand,
and resources are allocated and used accordingly. We are therefore interested in the allocation of resources
and the distribution of the bene ts derived from resource use.
In public economics, we study the impact of the public (government) sector on resource allocation and
distribution. In a mixed economy, the balance between the supply of and the demand for resources is
pursued through either the market system or the political system. In the market system, prices are the
equilibrating mechanism in the interplay between supply and demand, which, in turn, are determined by
factors such as the preferences and income of consumers, the costs of production factors and the prevailing
technology. Needs that cannot be or are not satis ed via the market system are channelled through the
political process. e equilibrating mechanism between supply and demand in a political system based on
democratic principles is the ballot box, and the ‘price’ is the tax that people pay.
Most Southern African countries have a parliamentary democracy with an executive president elected by
parliament. In South Africa, a constitutional change requires a two-thirds parliamentary majority (or a 75%
majority in respect of articles pertaining to the supremacy of the Constitution and the rule of law). Key
features of the Constitution of the South African democratic state, as established on 27 April 1994, are the
Constitutional Court, the Bill of Human Rights and the Human Rights Commission. Other important
features of the country’s democracy and of good governance are an independent judiciary, an independent
Auditor-General reporting to parliament and an independent central bank (the South African Reserve
Bank). e seat of parliament is in Cape Town (the provincial capital of the Western Cape), the Appeal Court
is located in Bloemfontein (the capital of the Free State province) and the Constitutional Court is in
Johannesburg (the mining and industrial ‘capital’ of the country and the capital of the Gauteng province).
ere are three tiers of government:
• e executive (the national public service) functions at the national level and is situated in Pretoria.
• At provincial level, there are nine provincial governments.
• At local government level, there are 278 municipalities.

We will discuss the nancial interaction between the different tiers of government in Chapter 19.
A large portion of Southern Africa’s total resource use is channelled through the political process.
Resource use in the public sector differs from pro t-driven resource use in the private sector, which is
nonetheless indirectly in uenced by the nature of the political environment and the functioning of the
political system, including a variety of economic policies and regulatory measures. e efficiency and equity
with which resources are allocated in the public sector as well as the impact of political decisions on private
economic behaviour are therefore of paramount importance to the economic performance of a country. is
point is best illustrated by South Africa’s recent economic history and the major rethink internationally of
the appropriate role of government in market-based economies.

1.2 Legacy of the past

During the ten to fteen years prior to the constitutional change and the rst few years of the new state, the
South African economy performed poorly. Real economic growth did not keep up with population growth:
in 1994, real per capita gross domestic product (GDP) was 16% below its previous peak (recorded in 1981). It
was only in 2006 (25 years later) that the 1981 level was surpassed. Broadly de ned, about one third of the
labour force was unemployed and South Africa had one of the most uneven distributions of income ever
measured internationally.
At the time of the constitutional change, poverty was indeed – and still is today – the most discomforting
feature of the South African economic landscape. e lack of job opportunities among millions of previously
disenfranchised citizens of the new South Africa is partly the result of statutory and other regulatory
measures that inhibited occupational and geographical labour mobility in the past. More recently, the job
losses due to the international nancial crisis of 2007–2009 aggravated the unemployment problem in South
Africa. Furthermore, the allocation of government resources on a per capita basis has historically been much
skewed, resulting in many of the poor having only limited access to basic social services. A large proportion
of the population has had experience of undernourishment, inadequate housing and poor education, and
has only had limited access to basic public services such as primary healthcare. Generally speaking, millions
of people were and still are ill-equipped to participate meaningfully in the modern (formal) sector of the
economy. Under these circumstances, the need for appropriate policies is evidently urgent. Faced with the
high expectations of the citizenry, scal measures that promote economic growth and improve people’s
ability to participate gainfully in the market economy constitute important policy instruments in the hands
of the government.
Government’s share in the economy increased steadily in the period between 1960 and 1990 (see Table
1.1 on page 13, which contains different nancial indicators of the size of government). When the rst full
democratic elections took place in 1994, many experts felt that the total tax burden, the national budget
de cit (or borrowing needs) and the public debt were too high to achieve sustainable economic growth and
development. Concerns for government nances at the time included the increasing arrears and leakage in
tax collection at the national level as well as the tendency for government expenditure to exceed budgeted
gures regularly. e growth of the government’s current expenditure was such that it crowded out public
investment; that is, ever-decreasing funds were available for government investment in social and physical
infrastructure such as schools, hospitals and roads.
e growth in public employment, which constitutes the main component of current government
expenditure, was a source of alarm. Between 1990 and 1999, for example, the number of employment
opportunities in the general government sector (that is, excluding public sector business enterprises and
corporations) was on average 16% higher than in the previous ten years. Between 1989 and 1999, job
opportunities in the formal private sector dropped by 19%. ese were hardly the characteristics of a thriving
economy.

1.3 Fiscal challenges

During the three or four decades before and up to the 2007 international nancial crisis, a strong consensus
emerged that the efficient management of developed and developing economies required smaller rather
than bigger budget de cits as well as lower rather than higher levels of public debt. ere was also general
agreement that scal prudence required a thorough revision of the basic functions for which government
should be responsible in order to contain the growth of government expenditure in a way that also ensures
effective support for the real needy in society. Along with the international shift towards market-based
economies that followed the demise of communism and the command economies of Eastern Europe during
the late 1980s and early 1990s, key ingredients of economic restructuring included the privatisation of
various activities hitherto undertaken by the state, a total revision of the role and functions of the state, and a
concomitant reprioritisation of government expenditure. It is important to note that the emphasis was on
how the public sector could be restructured to free up more resources for the development function of
government and to target the destitute effectively, without jeopardising macroeconomic stability or
increasing the share of government in the economy (preferably even reducing it).
e case for a smaller and more efficient government was the result of a growing dissatisfaction with the
effectiveness of government. Contributing factors included the failure of the command economies of
Eastern Europe, the increasing unaffordability of the welfare state in Western Europe, the poor economic
performance of (especially) developing countries with increasingly ineffective and ballooning governments,
the negative impact on economic freedom and initiative of excessive and poorly functioning government
regulations, the alarmingly high prevalence of corruption and rent-seeking activities, the poor service
delivery of government agencies in developing countries, which corresponded with an inability to attract
and/or retain expertise, and – generally – the failure of governments to generate wealth in cases where they
wanted to engage in or prescribe on production activities better left to the private sector. All of these also
went hand in hand with calls on the private sector to assume its role in an efficient and fair manner.
Countries seldom have the opportunity to thoroughly revamp all their institutions and policies at once.
However, this has been one of the main features of the South African experience since the 1990s and it
occurred against the backdrop of the above-mentioned changing views on the role of government in the
economy. Similar institutional changes are also taking place or have been effected in Angola, Namibia,
Mozambique, Zambia and many other African countries. e restructuring of the public sector in South
Africa after 1994 affected all groups in society, including the way in which business is conducted in the
domestic private sector as well as the nature of the country’s trade and investment relations with the rest of
the world.
It also affected the nature of public goods and services to be provided by the different tiers of
government. In addition, it affected and changed the individuals and groups of individuals who bene t from
public goods and services as well as the way in which the tax or nancing burden is distributed among
individuals and rms.
It changed the way in which consultation takes place between politicians and the electorate, and
between politicians and various sectional interest groups in society, be it interest groups in business,
organised labour or civil society.
Finally, it affected the basis on which employees are appointed and managed in the public sector as well
as the nature of interaction between politicians and bureaucrats, and between bureaucrats and the clients of
government.
 What has been remarkable about the development of scal policy in South Africa since 1994 is the ability
of the government to strike a good balance between the dictates of the market (requiring scal prudence)
and the scal pressures for redistribution, poverty reduction and socio-economic development.
Unfortunately, some of the good policy decisions were undermined by poor service delivery, resource waste
and the increasing occurrence of rent-seeking activities, fraudulent behaviour and nepotism. We will return
to some of these themes in later chapters of the book.
e international nancial crisis of 2007 and beyond was a serious wake-up call. Con dence in
mainstream (consensus) views on the respective roles of government and the private sector in the mixed
economy were shaken. e sub-prime nancial crisis in the United States resulted in the bankruptcy of huge
and prominent nancial institutions, thus posing a systemic risk to its nancial system. e subsequent
nancial contagion rapidly spread far and wide. e ensuing economic recession, which became known as
the Great Recession, was the biggest since the Great Depression. It jeopardised the solvency of companies in
the real and nancial sectors of developed economies all over the industrial world, and resulted in bailout
and rescue plans by governments of unprecedented proportions. More than that, the scal sustainability of
entire countries (such as Greece, Spain and Ireland) was threatened and required multilateral rescues. e
recession in developed countries had a major adverse effect on developing countries such as South Africa,
although the impact was generally less severe than in the industrial world.
e Great Recession gave rise to a major review of the appropriate role of government in the economy
and the required nature of economic policies. e cost of bailouts of nancial institutions and scal
measures to counter the recessionary impact led to dramatic increases in the ratio to GDP of budget de cits
and public debt in many industrial countries. Countries that had joined the European Union in terms of the
Maastricht prescriptions of a budget de cit–GDP ratio of no more than 3% and a debt–GDP ratio of no more
than 60% quickly found themselves in huge transgression of these scal criteria. For example, in 2010, the
United Kingdom recorded a budget de cit–GDP ratio of 11,4% and a general government debt–GDP ratio of
79,4%.1 Proponents of scal activism argue that their views had been vindicated: contra-cyclical scal policy
is justi ed and nancial markets need more regulation. But the huge de cits and debt levels clearly
threatened scal sustainability, and the appropriate way towards sustainability was and still is heavily
debated.
Subsequent to the establishment of full democracy in 1994, the South African government carved out a
particularly good scal track record, at least at the macro level. In tandem with monetary policy,
macroeconomic stability and scal sustainability were achieved up to the time of the 2007–2009
international nancial crisis. Although not spectacular by international standards, the economy had
recorded sixteen consecutive years of positive economic growth, averaging about 3,5% per year. e
economic upswing that started in September 1999 and ended after 99 months in November 2007 was the
longest by far since World War II. After introducing in ation targeting in 2002, the monetary authorities
succeeded in containing in ation within the target range of 3% to 6% or bringing it back into the target range
fairly quickly. e country’s performance with regard to economic growth, job creation and the combating of
poverty improved on the performance during the fteen years before the constitutional change, but still fell
far short of the country’s economic needs. Moreover, the Great Recession was a major setback, causing
about one million job losses. Many of the scal reforms that have formed part of the economic policy
package since 1994 are highlighted in this book, often with reference to theoretical issues. Attention is also
given to new challenges to scal policy on account of the increase in unemployment that accompanied the
economic downswing that has taken place since November 2007 and the continued high incidence of
poverty and economic inequality.

1.4 The study field of public economics

Public economics is the study of the nature, principles and economic consequences of expenditure,
taxation and nancing, and the regulatory actions undertaken in the non-pro t-making government sector
of the economy.
In order to understand this de nition, we rst note that economic policy entails the application by
government institutions of measures (instruments) that are designed to in uence economic behaviour in
order to achieve certain outcomes. Besides various macroeconomic goals, such as economic growth,
balance-of-payments stability or income redistribution, many sectoral and microeconomic goals are also
pursued, with the ultimate objective of improving the material well-being of people.
Turning now to the elements of our de nition of public economics, note rstly that the main areas of
decision-making are expenditure, taxation, nancing and regulation. ese elements are also called the
instruments of scal policy. In public economics, we study the nature and impact of these instruments.
e rst three instruments (expenditure, taxation and nancing) entail the procurement by the state of
private funds (at the cost of either private expenditure or saving) and the spending of these funds. In
economic terms, the use of these instruments constitutes the direct mobilisation and allocation of scarce
resources. Examples include the spending of tax income on primary healthcare and the borrowing of funds
to build an irrigation dam or a highway.
Regulation, by contrast, entails enacting a law or administratively proclaiming an enforceable
instruction that leads to a different allocation of private resources from that which would apply in the
absence of such government intervention. e allocation of resources is now in uenced indirectly.
Examples are the regulation by government that forces the manufacturers of motor vehicles to install
platinum catalysts in the exhaust pipes of vehicles to reduce the emission of carbon monoxide and the
regulations that prescribe the behaviour of natural monopolies such as electricity supply.
Different types of expenditure, taxation, government borrowing and government debt can be
distinguished. In taxation, for example, we distinguish between taxes on income, wealth, and goods and
services. e various categories of taxation or expenditure have different economic consequences. For
instance, a tax on wealth (for example, a property tax) will affect different people in society from those who
would be affected by a value-added tax on red meat. In other words, the distributional effects differ. e
choice of a particular tax therefore depends on how the government wants to change the distribution of
income or wealth in the economy, or on how an efficient allocation of resources can be pursued.
Economists have developed several important scal criteria on which economic decisions in the public
sector are or should be based and that may be applied when recommendations on taxes or expenditure
allocations are formulated. ese governing criteria are derived from the two concepts of efficiency and
equity, which are paramount and distinguishing features of economics.
e study of the nature and economic consequences of decisions falls in the realm of positive
economics, posing questions such as the following: If I take step a (for example, raising income tax), what
will happen to b (for example, the supply of labour in the economy)? e development of criteria, on the
other hand, has to do with normative economics, focusing on what ought to be. For example, if I want d (for
instance, a more even distribution of income), what should step c be (in other words, what type and level of
taxation should be introduced)?
Public economics enters the realm of normative economics when we consider diverse questions such as
the rationale for government involvement in the economy and how political decisions should be taken (in
other words, what kind of voting system should be used) to ensure efficiency and equity in the allocation of
resources.
Generally, two broad views of the role of government in the economy are encountered.
1. e individualistic view of government (sometimes also called the mechanistic view) recognises the
supremacy of the individual and his or her freedom of choice. e main point of departure and the test
for whether or not government intervention is called for is the maximisation of individual welfare.
Government action is seen as a re ection of individual preferences and government institutions have no
role independent of these preferences. e role of government is limited to correcting for market failure,
a topic that we explore in various chapters of Part 1. is view is closely associated with the proponents of
the free-market economy and gained widespread international support towards the end of the twentieth
century, notably after the collapse of the command economies of Eastern Europe. e international
nancial crisis referred to above embodied elements of market failure, as well as of government failure
that we also explore.
2. e public interest or collectivist view of the role of the state is sometimes also referred to as the
organic view. Collective choice and preferences are thought to exist independently of individual
preferences and the goals of society as a separate organism differ from the goals of the constituent
individuals. Instead of individual welfare, it is social welfare or the national interest that should be
 maximised. e bene t to the individual, consequently, is derived from the bene ts accorded to or
achieved by the group.

From a policy perspective, the individualistic approach focuses relatively more on the efficiency of resource
allocation and economic growth, whereas the collectivist approach is relatively more concerned with
combating poverty and equity issues, notably the distribution of income as justi cation for government
intervention. In practice, government policies often re ect a combination of the two approaches.
One is indeed likely to nd a combination of policies emanating from both views in the model of the
developmental state, a term coined in 1982 by Chalmers Johnson, an American political scientist. e ANC
government propagates this model, although many features of South African society arguably do not t this
mould (see Burger, 2014). Leftwich (1995: 401) describes developmental states as those states whose politics
have concentrated sufficient power, autonomy and capacity at the centre to shape, pursue and encourage
the achievement of explicit development objectives, whether by establishing and promoting conditions and
the direction of economic growth or by organising it directly, or a varying combination of both. e
developmental state has been characterised as embodying inter alia a determined development elite, a
powerful, competent and insulated economic bureaucracy, a weak and subordinate civil society and the
effective management of non-state economic interests (Leftwich, 1995: 405–420). e focus appears to be
more on the way in which political and economic control is obtained and exercised, and seems to allow for
much variety as regards the nature and principles of the economy and economic policy. Selective
government intervention that distorts relative prices rather than a neoclassical minimisation of price
distortions by government has nonetheless been identi ed as a feature of the successful developmental state
(Grabowski, 1994: 413). Countries regarded as examples of the developmental state2 as well as their
economic policies are too diverse to allow for the de ning of a common economic policy of success. As
always, many other factors besides economic policies determine economic success.
Returning to the elements of our de nition of public economics, the term ‘non-pro t-making’ signi es
the absence of pro t maximisation as the leading motive, or one of the leading motives, in decision-making
on the mobilisation and allocation of resources. e absence of the pro t motive means that other criteria
for decision-making have to be employed. We will see that the nature of public goods is such that their
supply does not allow for decentralised price determination in a competitive market economy. Note that the
government is not the only non-pro t-making sector in the economy. Many welfare, church and service
organisations exist as non-pro t organisations. ese institutions are often referred to as non-governmental
organisations (NGOs). When they receive the status of public bene t organisations under tax law in South
Africa, donors to these entities are entitled to special tax bene ts and the entities themselves also enjoy
special tax treatment.
Does our de nition of public economics include a study of public corporations (alternatively referred to
as public enterprises or state-owned companies) that operate in the hybrid area between the government
and the private sector, such as Eskom in South Africa? If these entities were driven strictly by the pro t
motive, they would not t our de nition. However, as long as political appointees serve on or control the
boards of these entities, as long as they render certain socio-economic services on behalf of the government
and rely on government nancial support, and/or as long as they behave in a monopolistic manner, they are
not pure private institutions. In countries that have embarked on privatisation, such as South Africa, public
enterprises often nd themselves in transition between being a public entity and a private company.
Consequently, it is not easy to pinpoint their exact position on the spectrum between public and private
entities, and the criteria in terms of which to study their behaviour are not that clear. In our study of public
economics, we do not include a separate section on these kinds of activities. We do, however, analyse
aspects of their functioning when we discuss topics such as externalities, imperfect competition, user
charges, privatisation, public-private partnerships and macroeconomic stabilisation.

1.5 The public sector in South Africa

From our discussion in the previous section, it is clear that a wide range of diverse activities is studied in
public economics. In order to structure our thoughts, we will now examine the composition and size of the
public sector in South Africa, which is fairly typical of public sectors elsewhere in Africa, before we brie y
review the relationship between the public and private sectors.

1.5.1 Composition of the public sector

What are the constituent institutional components of the public sector? We present them in Figure 1.1 as a
set of rectangles within rectangles. e South African Constitution speci es three levels or spheres of
government. e central (or national) government (see the inner rectangle in Figure 1.1) consists of all the
national government departments. When the various extra-budgetary institutions are added (such as the
National Research Foundation, the National Health Laboratory Service and the Urban Transport Fund, a
number of government business enterprises that sell most of their output of goods and services to
government institutions or departments at regulated prices, social security funds and universities), we refer
to the combination as the consolidated national government. (Note that in the national accounts, the
universities are in fact classi ed as part of the public sector.) e aforementioned entities have access to
funds that are additional to budgetary allocations, such as user charges, levies and other non-tax income.

Figure 1.1 The composition of the public sector

As pointed out earlier, the second and third tiers of government in South Africa constitute nine provincial
governments and 278 local authorities (or municipalities) respectively; these are shown in the second
rectangle in Figure 1.1. Together with the consolidated central government, the general departments of
provinces and local authorities and certain business enterprises such as the trade departments (for
electricity, water, transport and so on) of local governments are constituent components of the general
government. For the most part, general government thus represents the non-pro t activities of the public
sector. e allocation of resources is determined by political considerations, and is nanced through the tax
system and user charges (or loans that have to be repaid out of taxes at a later stage).
e next category of public entities (see the outermost rectangle in Figure 1.1) consists of nancial and
non- nancial public enterprises (also referred to as state-owned enterprises or companies) such as Eskom,
PetroSA, the South African Broadcasting Corporation (SABC), Telkom, Transnet, the Land Bank and the
Public Investment Corporation. ese activities are managed much more along business lines and, in the
case of corporations such as Eskom and Telkom, decisions are often taken on the same basis as in the private
sector. For as long as these corporations are subject to government control, either in the form of
shareholding or the appointment of directors, they are classi ed as part of the public sector. Conversely,
should any public sector activity or body (or a part of it) be privatised, it will thereafter be reclassi ed as part
of the private sector.
To summarise: We refer to the three tiers of government (in other words, the general services and certain
business enterprises of national, provincial and local government) as the general government, and to the
combination of general government and public corporations as the public sector.

1.5.2 Size of the public sector

e size of the public sector differs according to the indicator used. If we are interested in the size of the
burden that the government imposes on current taxpayers, we may use the total tax income of the general
government as an indicator and express it as a percentage of the gross domestic product (or national
income). By this criterion, the government’s average share in the South African economy during the period
2008–2017 was 25,4%. We know, however, that government expenditure is not only nanced through tax
revenue, but also by means of non-tax income (such as dividend and property income, mining leases and
administrative fees) as well as borrowing (loans). We will thus obtain a more accurate picture by looking at
government expenditure. We can identify two aggregate indicators.

e rst of these aggregate indicators is the total amount of resource use by government (in other words, the
nal demand by or the exhaustive expenditure of government) in any year. From Table 1.1, we note that the
total resource use by the public sector (that is, consumption and investment spending of national,
provincial and local government as well as of public enterprises, valued at market prices) increased from an
average of 20,2% of the gross domestic product over the period 1960–1969 to an average of 27,5% during the
1980s. It then decreased to an average of 23,3% during 2000–2007, that is, before the international nancial
crisis impacted on South Africa. It then rose sharply to average 27,4% for the period 2008–2017. Even this is
not the complete picture.
Not all government expenditure is in the form of nal demand for goods and services (that is, exhaustive
expenditure). e government also makes transfer payments (subsidies, current transfers and interest on
public debt) to targeted bene ciaries or entities outside the public sector. ese are called non-exhaustive
government expenditure. e government mobilises the resources, but they are used by the recipients, who
exercise the nal demand in accordance with their preferences. (Note that the national government also
makes transfer payments to provinces and local governments, as discussed in Chapter 19. ese are
regarded as internal ows within the public sector and are not counted as payments from the government
sector to other sectors of the economy.) If we add interest payments and transfers to the household, business
and foreign sector, we can obtain an accurate picture of the extent of resource mobilisation by the
government, our second aggregate indicator of the size of government. During 2008–2017, the South African
public sector was instrumental in mobilising 39,8% of the national resources, a signi cant increase on the
average ratio of 34,5% of the preceding eight years. is gure is also higher than the average gures for all of
the other previous periods shown in Table 1.1. e highest resource mobilisation for any year since 1960 was
in 2015, namely 41,0%.
Table 1.1 Average size of the South African public sector by different indicators, selected periods (current prices; %
of GDP)

Notes: a In order to link data before and after 1994, data before 1994 were downscaled by an adjustmet factor of
.899537403 so as to remove FISIM (Financial Intermediation Services Indirectly Measured) from series
before 1994. e factor represents the ratio between two sets of gures for 1994–1996, namely interest
inclusive and exclusive of FISIM.

b Include social bene ts to households and production subsidies as well as miscellaneous transfers and
transfers to international organisations; for latter two, average ratio to GDP between 1994 and 2009 used as
estimate for data before 1994. Data before and after 1994 for the subcomponents were not strictly
comparable and assumptions were made in order to ensure reasonable comparability of the aggregate. No
seperable data for capital transfers before 1995 were available but the amounts from 1995 onwards were
relatively small.
Source: SARB (various years) Quarterly bulletin (various issues); SARB (various years) Electronic data

As a result of the diverse nature of government activities and the corresponding differences in the factors
that determine the allocation and distribution processes in the public sector, we are not only interested in
the aggregate size of the public sector, but also in its constituent components. Note in particular the opposite
trends of general government consumption expenditure and public investment as well as the rising share of
transfer payments. Interest on public debt as a percentage of GDP increased substantially up to the second
half of the 1990s and again since 2014. Commensurate with government’s priority to ameliorate the
consequences of poverty, transfer payments to households, as a percentage of GDP, likewise rose
signi cantly during the past twenty years. All of the above trends are recurrent themes in this book (see, for
example, Chapters 7–10).

1.5.3 The relationship between the public and the private sectors
What is the relationship between the public sector and the rest of the economy? A number of important
aspects of this relationship may be identi ed with reference to the familiar circular ow of income,
expenditure, and goods and services (see Figure 1.2).
• Government is a supplier of public goods and services. Households and businesses pay for these goods
and services through taxes (and user charges). Government then uses this revenue to acquire factors of
production as well as to purchase private goods and services (as intermediate inputs), all of which are
used in order to produce public goods and services. Government departments, of course, use outputs of
other departments and state-owned enterprises as intermediate inputs as well. Government activities
are relatively labour-intensive and, as a result, compensation of employees constitute the largest input
cost (averaging 54,9% of consolidated general government output during 1995–2017).3

Figure 1.2 The government in the circular flow of income, expenditure, and goods and services
• e size of government in the mixed economy is such that its purchases of goods and services exert
important in uences on the economy. At the sectoral level, for instance, government spending is often
decisive for the construction and engineering sectors. Privatisation entails goods or services formerly
supplied by government as part of the ow of public goods and services to be sold or transferred to and
rede ned as goods and services supplied by private rms. In the case of public-private partnerships
(PPPs), the private sector produces and often delivers the goods and services, but government
determines (through the PPP contract) the quantity and quality of the goods and services. Often private
investment cannot be undertaken unless the necessary public infrastructure (for example, roads and
electricity networks) is in place. At the macroeconomic level, changes in the aggregate level and
composition of government expenditure are important factors in determining economic stability and
growth. Excessive expenditure growth can, for example, be in ationary or may crowd out private
investment, thus retarding economic growth.
• e way in which government nances its expenditure also has important economic consequences. e
kind of taxes used and the rates levied in uence the after-tax distribution of income and thus the well-
being (welfare) of individuals as well as the decisions by private businesses regarding the allocation of
resources in the private sector. e tax system can promote or obstruct efficiency and equity.
• If there is a budget imbalance (surplus or de cit), the government exercises an in uence on the
balance between saving (S) and investment (I) or on the balance of payments (in other words, the
balance between exports (X) and imports (M)). is is shown in Figure 1.3. In national accounting terms,
a budget imbalance (that is, G ≠ T) is re ected in either an imbalance between private investments (I)
and savings (S), that is, S ≠ I (internal imbalance) or an imbalance between imports (M) and exports (X),
that is, M ≠ X (external or balance-of-payments disequilibrium), or both.
• If there is a budget de cit, for example, tax revenue (T) is less than government expenditure (G), or T < G,
the government has to borrow money. e size of its de cit and the way in which it is nanced is very
important for macroeconomic stability, depending on one’s view of the impact of budget de cits on the
economy. Government borrowing occurs via the nancial markets and represents a use of either
domestic (Sd) or foreign savings (Sf), as shown in Figure 1.3. (In Figure 1.3, T, S and M represent leakages
from the income-expenditure circular ow, while G, I and X are additions to or injections into the
circular ow.) Part of the savings may nd their way into nancing government investment, which is
included in government expenditure (G). In the case of a budget surplus, the government supplements
the supply of savings in the economy.
• While the government can in uence the course of the economy, it is also extensively affected by what
happens in the economy. In an economic recession, for instance, government revenue falls, or grows at a
slower rate. is may impair its ability to provide public services, especially if its debt or budget de cit is
already relatively high. Government also bears the brunt of its own decisions via their adverse effect on
the economy, such as when high budget de cits result in higher interest rates, thereby increasing the
government’s interest bill.
Figure 1.3 The impact of a budget imbalance on the financial sector in the context of the circular flow of income and
expenditure

Key concepts
• budget imbalance (surplus or deficit) (page 15)
• central (or national) government (page 11)
• consolidated national government (page 11)
• developmental state (page 9)
• exhaustive expenditure (page 12)
• extra-budgetary institutions (page 11)
• general government (page 11)
• individualistic (or mechanistic) view of government (page 9)
• instruments of fiscal policy (page 8)
• local authorities (or municipalities) (page 11)
• non-exhaustive government expenditure (page 12)
• non-governmental organisations (NGOs) (page 10)
• non-profit-making (page 10)
• privatisation (page 15)
• provincial government (page 11)
• public corporations (page 10)
• public economics (page 8)
• public enterprises (page 10)
• public infrastructure (page 15)
• public interest or collectivist view of government (page 9)
• public-private partnerships (PPPs) (page 15)
• public sector (page 11)
• regulation (page 8)
• resource mobilisation (by the public sector) (page 12)
• resource use (by the public sector) (page 12)
• state-owned enterprises or companies (page 11)
• transfer payments (page 12)

SUMMARY
• In public economics, we study the impact of the public (government) sector on resource allocation and
distribution. is requires a mechanism whereby demand for and supply of public goods and services
can be equated in the non-market sector of the economy. In a political system based on democratic
principles, this is the ballot box and the ‘price’ is the tax that people pay.
• Resource use in the public sector differs from pro t-driven resource use in the private sector, which is
nonetheless indirectly in uenced by the nature of the political environment and the functioning of the
political system, including a variety of economic policies and regulatory measures. e efficiency and
equity with which resources are allocated in the public sector as well as the impact of political decisions
on private economic behaviour are of paramount importance to the economic performance of a country.
• Since 27 April 1994, the new South African Constitution has provided for important features of
democracy and good governance, such as an independent judiciary, a Constitutional Court, a Bill of
Human Rights, a Human Rights Commission, an independent Auditor-General reporting to parliament
and an independent central bank (the South African Reserve Bank).
• Fiscal measures that promote economic growth and improve people’s ability to participate gainfully in
the market economy constitute important policy instruments in the hands of the government.
• In 1994, there were concerns about scal sustainability because of increasing arrears and leakage in tax
collection at the national level, the tendency for government expenditure to exceed budgeted gures
regularly, and high and rising public debt. e growth of the government’s current expenditure was such
that it crowded out public investment: ever-decreasing funds, as a percentage of GDP, were made
available for government investment in social and physical infrastructure such as schools, hospitals and
roads.
• During the three or four decades before and up to the 2007–2009 international nancial crisis, a wide
consensus developed about scal prudence and the restructuring of the public sector (inter alia through
privatisation and deregulation) to free up more resources for the development function of government
and to target the destitute effectively, without jeopardising macroeconomic stability. e case for smaller
and more efficient government was the result of a growing dissatisfaction with the effectiveness of
government in developed and developing countries.
• Against the backdrop of international events and shifting views on the role of government in the
economy, the constitutional change in 1994 provided South Africa with the unique opportunity to
restructure or revamp all of its institutions and policies thoroughly. e new government generally
achieved a good balance between the dictates of the market (requiring scal prudence) and the scal
pressures for redistribution, poverty reduction and socio-economic development. Unfortunately, in later
years some of the good policy decisions were undermined by poor service delivery and resource waste as
well as increasing occurrences of rent-seeking activities, fraudulent behaviour and nepotism.
• e international nancial crisis of 2007–2009 and beyond was a serious wake-up call. Con dence in
mainstream (consensus) views on the respective roles of government and the private sector in the mixed
economy was shaken. e recession in developed countries had a major adverse effect on developing
countries such as South Africa and other countries on the African continent, although the impact was
generally less severe than in the industrial world.
• Since 1994, the South African government has carved out a particularly good scal track record, at least
at the macro level. e economic upswing that started in September 1999 and ended after 99 months in
November 2007 was the longest by far since World War II. e Great Recession was a major setback,
causing about one million job losses.
• Public economics is the study of the nature, principles and economic consequences of expenditure,
taxation, nancing and the regulatory actions undertaken by the non-pro t-making government sector
of the economy.
• ree approaches to the role of government in the economy can be distinguished: the individualistic,
collectivist and developmental state views. e ANC government propagates the latter model, although it
can be argued that many features of South African society do not t this mould.
• e ‘public sector’ consists of the following:
- Central (or national) government (in other words, all of the national government departments) and
various extra-budgetary institutions, which together are known as the consolidated national
government
- Provincial and local government institutions, which, together with national government, constitute
general government
- Public corporations, also called public or state-owned enterprises (or companies).
• ere are different indicators of the size of the public sector and its share in the economy, for example,
tax revenue, resource use (or exhaustive expenditure) and resource mobilisation (that is, resource use
plus transfer payments), all of which are normally expressed as a ratio of gross domestic product (GDP).
• e relationship between the public sector and the rest of the economy can be analysed with reference to
the four-sector circular ow of income, expenditure, and goods and services. While the government can
in uence the course of the economy, it is also extensively affected by what happens in the economy.
• In the macroeconomic identity, the budget balance equals the sum of the balance between domestic
savings and investment (S + I) and the balance between imports and exports (M + X).

MULTIPLE-CHOICE QUESTIONS
1.1 Which of the following statements applies or apply to the individualistic view of government?
a. e main point of departure is the maximisation of individual welfare.
b. Government institutions have no role independent of individual preferences.
c. e role of government is limited to correcting for market failure.
d. is view is also referred to as the mechanistic view of the role of government.
1.2 Which of the following statements is or are wrong?
a. e individualistic view of government best describes the role of government in South Africa.
b. e collectivist view of government best describes the role of government in South Africa.
c. e developmental state view of government best describes the role of government in South Africa.
d. e developmental state view is a special variant of the individualistic view of the role of
government.
1.3 Complete the sentence that follows by choosing the correct alternative from the options below.
When the government has a budget de cit, …
a. the country exports more than it imports
b. private savings exceed private investments.
c. the budgets of local governments as a group are in surplus.
d. the economy will show either an internal or an external imbalance, or both.
1.4 General government …
a. consists of all the general departments of the government.
b. refers to the activities of national, provincial and local government.
c. refers to the general divisions of public corporations.
d. authorities in South Africa undertake no capital expenditure.
1.5 Which of the following statements about the public sector in South Africa is or are correct?
a. After 1994, the biggest increase in the general government tax burden on average occurred
between 2000 and 2007.
b. e share of public sector resource mobilisation has increased consistently from its average during
1995–1999.
c. e growth of the government’s current expenditure was such that it crowded out public
investment from time to time.
 d. e ratio of current expenditure to GDP provides the best measure of the relative size of
government in the economy.

SHORT-ANSWER QUESTIONS
1.1 Distinguish between the following:
a. positive and normative economics
b. general government and public sector
c. resource use and resource mobilisation.
1.2 Explain the difference between the individualistic and the public interest view of government.
1.3 What are the features of the developmental state?
1.4 List the components of exhaustive government expenditure.

ESSAY QUESTIONS
1.1 Brie y review the salient changes in the size and composition of the South African public sector during
the past few decades. Which of the changes, in your opinion, is or are incompatible with the
requirements of a thriving economy?
1.2 Give an outline of the various dimensions of the relationship between the public sector and the rest of
the economy.

1 By comparison, South Africa’s national government debt–GDP ratio of 34,7% in 2010 was relatively low (see Section 17.3 of Chapter
17 for further discussion). Even if sub-national government debt is added, the gure remains comparatively low.
2 For example, Singapore, Botswana, Malaysia, South Korea, Indonesia and, more recently, China.
3 During and after the international nancial crisis the trend was strongly upwards, that is, from 50,9% in 2007 to 57,1% in 2017.
Benchmark model of the economy: Positive and normative
approaches

Philip Black and Ada Jansen

is part of the book starts with a brief review of the neoclassical theory of general equilibrium, which has
become something of a benchmark model over the years. It is a benchmark model precisely because it does
not presume to provide an accurate description of the real world. Rather, it should be seen as a frame of
reference or a starting point in the sense that it describes a stable economy in which all resources are
optimally utilised. As such, it may help us to understand and appreciate real-world problems better. As shall
become apparent in the chapters that follow, the model can in fact accommodate a large variety of
alternative assumptions. is built-in exibility enables it to yield alternative predictions that bring us closer
to the real world.
Section 2.1 of this chapter provides a brief description of the basic assumptions of our benchmark model,
while Sections 2.2 and 2.3 discuss its equilibrium properties. ese three sections thus provide a vision of
how the world ought to work and as such represent a good example of what some commentators refer to as
‘normative’ economics. Section 2.4 begins to enter the domain of ‘positive’ economics by comparing the
benchmark model with the real world. is is done by introducing the concept of market failure, a generic
term describing broad categories of human behaviour that deviate from the ideal assumptions of the
benchmark model. Sections 2.5 and 2.6 brie y deal with the role of the public sector in coming to grips with
these market failures and related real-world problems (Chapters 3–6 will deal with these issues in more
detail). e chapter concludes with Section 2.7, which comments on government failure.

Once you have studied this chapter, you should be able to:
identify the critical assumptions of the two-sector model
define what is meant by a Pareto-optimal allocation of resources
articulate the three conditions for a general equilibrium
distinguish between allocative efficiency, X-efficiency and ‘dynamic’ efficiency (or economic growth)
discuss the broad categories of market failure
explain the allocative, distributive and stabilisation functions of government
distinguish between direct and indirect forms of government intervention.

2.1 Basic assumptions of the benchmark model

e benchmark model is based on a host of patently unrealistic assumptions that are set out in Box 2.1. It
assumes that the individual consumer or producer is fully informed about the economy, unaffected by the
actions of other consumers or producers, completely mobile in the occupational and spatial sense of the
word, and always striving to maximise his or her own utility or pro t within perfectly competitive markets.
Any exogenous disturbance will merely set in motion a series of more or less instantaneous adjustments that
will automatically return the system to a stable equilibrium. Indeed, there would be few economic problems
to speak of in such a blissful state, and little need to write this chapter and many of the subsequent ones.

Box 2.1 Assumptions of the two-sector model

• There are two individuals, A and B, who are the suppliers of two factors of production, the producers of two
commodities and the consumers of both of these commodities all at the same time. Each individual is
initially endowed with fixed quantities of the two factors, capital (K) and labour (L), and uses these
endowments to produce two commodities, X and Y. Each individual consumes both commodities.
• There are no external effects associated with consumption and both individuals have fixed tastes, as
reflected in the existence of smooth and ‘well-behaved’ individual indifference curves. These curves (which
show combinations of goods that yield the same levels of utility) are convex with respect to the origin, cannot
intersect and exhibit diminishing marginal rates of substitution.
• The two production processes are both characterised by unlimited factor substitutability, diminishing marginal
productivities and constant returns to scale. The latter assumption rules out internal (dis)economies of
scale, while there are also no external costs or benefits in production. These assumptions together ensure
the existence of smooth and ‘well-behaved’ isoquants (recall that an isoquant is a curve that shows
combinations of inputs that yield the same level of output).
• As consumers, A and B maximise utility, and as producers, they maximise profit. Both are perfectly informed
about their respective environments, and are perfectly mobile in the occupational and spatial sense of the
word.
• The commodity and factor markets are all perfectly competitive, which implies that each market behaves ‘as
if’ there were a large number of individual demanders and suppliers involved, none of whom can influence
price.
• These assumptions together ensure the existence, uniqueness and stability of a general equilibrium.

One does not have to be an economist to realise that real-world economies do not behave in the way in
which the neoclassical model predicts. Perhaps the only real market that comes close to this is an auction,
where suppliers and demanders all come together to reveal their preferences to an independent and
knowledgeable auctioneer who establishes the equilibrium price before any trading takes place. In the
neoclassical model, we simply assume the existence of this knowledgeable and omnipresent auctioneer –
also called the ‘Walrasian auctioneer’– and give him or her the task of establishing equilibrium prices in all
markets, more or less simultaneously.
Before delving into the model itself in Sections 2.2 and 2.3 below, it is worth noting that a theory does not
necessarily stand or fall by its assumptions. To be sure, assumptions are important if one wanted to explain
and predict some real-world phenomenon, as are the functional relations used to make predictions and test
the validity of the theory. But if one’s aim is merely to develop a normative theory – such as a benchmark
model of resource allocation – it is hardly appropriate to judge it only in terms of the realism of its
assumptions.

2.2 The benchmark model and allocative efficiency

Economic efficiency is conventionally de ned in terms of both allocative efficiency and technical efficiency
(or X-efficiency), and can also refer to a country’s ability to achieve economic growth. is section focuses
on allocative efficiency (Section 2.3 discusses X-efficiency and economic growth).
 Allocative efficiency refers to a situation in which the limited resources of a country are allocated in
accordance with the wishes of its consumers. An allocatively efficient economy produces an ‘optimal mix’ of
commodities. Under conditions of perfect competition, the mix of outputs will be optimal if consumers
maximise utility in response to prices that re ect the true costs of production. It is therefore evident that
allocative efficiency involves an interaction between the consumption activities of individual consumers
and the production activities of producers. In an economy with no public sector and no consumption or
production externalities, allocative efficiency in the general equilibrium context requires the simultaneous
concurrence of three familiar conditions. Brief discussions of these conditions follow.

2.2.1 Condition 1
e rst condition is Pareto optimality in consumption. Consider a two-person, two-commodity exchange
economy in which the total supply of the two commodities is xed. e two consumers (A and B) each has
an initial allocation of the two goods (X and Y), and exchange these commodities as long as it is mutually
bene cial for them to do so. e rate at which they are willing to exchange one good for the other is given by
the marginal rate of substitution of good X for good Y (MRSxy). As long as the MRSxy differs between A and B,
there is room for bene cial exchange and trade continues. Once the two individuals have exhausted all
mutually bene cial exchange opportunities, the MRSxy will be the same for both. is implies that a Pareto
efficient allocation has been reached. A movement from the initial allocation to this efficient allocation is
called a Pareto improvement (the exchange has increased the utility of at least one of the consumers,
without decreasing the utility of anyone). Economic efficiency in consumption occurs if no interpersonal
reallocation of commodities can increase the utility of either of the two consumers, A or B, without thereby
decreasing the utility of the other. All efficient allocations occur where MRSAxy = MRSBxy. e exchange
contract curve is obtained by connecting all these allocations.
Figure 2.1 uses the familiar Edgeworth-Bowley box diagram to illustrate the condition of efficiency in
consumption. e xed total supplies of the two commodities give the dimensions of the box: either of the
two vertical axes, 0A J or 0BK, gives the total amount of commodity Y, while either of the horizontal axes, 0AK or
0B J, gives the total amount of commodity X. Individual A’s indifference curve map originates in the
southwestern corner, 0A, and that of individual B in the northeastern corner, 0B. Only three out of the large
number of indifference curves are shown for each of the two individuals, that is, UA1, UA2 and UA3 for
individual A, and UB1, UB2 and UB3 for individual B. Starting from an initial allocation at point P (where the
indifference curves UA2 and UB1 intersect), consumers exchange commodities (given that their MRSxy are not
the same). When they reach point T, all bene cial exchange opportunities have been exhausted. e
efficient allocations in this exchange economy are the points where the indifference curves of the two
individuals are tangent, for example, points S, T and R. e exchange contract curve in Figure 2.1, 0A0B, can
be derived by repeating this exercise many times from different initial allocations.
In a competitive market setting, each individual will maximise utility subject to his or her own budget
constraint. is implies choosing the commodity mix for which the MRS equals the corresponding
commodity price ratio. at is:

for all consumers in the market. is condition holds under perfect competition as all consumers are price
takers and face the same prices.

2.2.2 Condition 2
We now relax the assumption of xed output supplies and make production activities variable. is brings us
to the second condition for allocative efficiency: Pareto optimality in production. is means that it should
not be possible to increase the output of any commodity without decreasing the output of at least one other
commodity. Put differently, in a non-optimal situation, it is always possible to increase the output of one
commodity without thereby decreasing the output of other commodities.
e model assumes that each of the production sectors commence with an initial amount of the ( xed)
factors of production in the economy, labour and capital. e initial allocation of inputs between sectors X
and Y may not be the most efficient option. A reallocation of the inputs L and K can bene t both sectors and
move the economy from an inefficient allocation to a Pareto efficient allocation, as shown in Figure 2.2. A
move from point C (the initial allocation) to point E is a Pareto improvement, as it increases the output of Y
while maintaining the output of X. If such gains are no longer possible, a Pareto efficient production point
has been reached. Figure 2.2 illustrates this. e xed total supplies of the two factors of production
determine the dimensions of the box: either of the two vertical axes, 0xV or 0yW, gives the total supply of
capital, while either of the two horizontal axes, 0xW or 0yV, gives the total supply of labour.

Figure 2.1 Efficiency in exchange

e isoquant map of Sector X (that is, its production function) originates in the southwestern corner, 0x,
and that of sector Y in the northeastern corner, 0y. Only three out of a large number of isoquants are shown
for each of the two sectors, that is, X1, X2 and X3 for sector X, and Y1, Y2 and Y3 for sector Y. e next step is to
nd all of the points where the two sectors’ respective isoquants are tangent, for example, points E, F and G.
At such points of tangency of the isoquants, the rates at which one input is substituted for another to
produce constant output levels (the marginal rate of technical substitution, MRTSlk) are the same. e two
sectors use inputs efficiently at such production points, in the sense that it is not possible to reallocate inputs
between sectors to increase the output of one sector, without thereby decreasing the output of the other
sector.
If this exercise is repeated many times, the contract curve for production is derived, shown as 0x0y in
Figure 2.2. Note that each point on the contract curve represents a Pareto-optimal allocation of the two
resources, K and L, between the two sectors, X and Y: at point E (or F or G), it is not possible for either sector
to increase its output without the other sector having to cut back its own output. is is clearly not true of a
point such as C where either of the two sectors can increase its output without causing a reduction in the
output of the other. Point C is not on the contract curve; as discussed in Section 2.3, it represents an X-
inefficient outcome.
 In terms of the familiar two-sector model, this condition requires that each of the two sectors should
maximise output subject to its own cost constraint. Hence, for each sector the MRTSlk = where the latter
is the market-determined equilibrium input price ratio. e assumption of perfect competition ensures that
the two sectors operate at the same point on the contract curve (for example, E, F or G in Figure 2.2), and
that the economy nds itself at a point on the production possibility curve (PPC). e latter is simply the ip
side of the above contract curve: it brings together all of the output combinations along the contract curve
within a more conventional diagram and is shown in Figure 2.3. ere the PPC is depicted as the curve MN,
on which the points E, F and G represent the same output combinations as their equivalents on the contract
curve in Figure 2.2.

Figure 2.2 Efficiency in production

e slope (or rate of change) of the PPC in Figure 2.3 is given by and is known as the marginal rate of
product transformation of X for Y, or MRPTxy. e latter, in turn, equals the corresponding marginal cost
ratio, which is easily proved with the aid of Figure 2.3.
Consider a small movement from point F to point h in Figure 2.3 such that the resources gained by sector
X equal the resources lost by sector Y. With factor prices assumed unchanged, this means that the increase
in the total cost of sector X will equal the decrease in the total cost of sector Y, that is, ∆TCx = ∆TCy. Now,
since

or

therefore
Figure 2.3 Production possibility curve

MCx is the marginal cost of production in sector X. Since under perfect competition each sector will ensure
that its own marginal cost equals the corresponding market price, that is, MCx = Px and MCy = Py, the
following holds:

is is given by the slope of a tangent drawn to a point on the PPC, for example, the slope of line tt’ at point F
in Figure 2.3.
e second condition thus implies a point on the PPC at which it is impossible to increase the output of
either of the two sectors without thereby decreasing that of the other. Under perfect competition, price will
equal marginal cost in each sector, so that the MRPT, which equals the marginal cost ratio for the two
sectors, also equals the corresponding price ratio.

2.2.3 Condition 3
e third condition for allocative efficiency requires that producers and consumers achieve equilibrium
simultaneously. Given that the slope of the PPC (that is, MRPTxy) equals the corresponding ratio of marginal
costs, and thus also the corresponding equilibrium price ratio, the third condition can be written as
follows:

is indicates equality between the (marginal) rate at which each consumer is willing to substitute one
commodity for the other and the rate at which it is technically possible to do so. Put differently, the economy
is producing the combination of commodities X and Y at minimum cost and in compliance with the
preferences of the consumers. If MRPT does not equal MRS, it is possible to alter production and
consumption to achieve a Pareto improvement. is can be illustrated as follows. If the but the
MRPTxy = it means that consumer A is willing to sacri ce 2 units of X for 1 unit of Y, while at this
production point 2 units of Y will be gained if 2 units of X are sacri ced. A Pareto improvement is possible if
more of Y is produced.
Simultaneous compliance with these three conditions will ensure production of the optimal output mix,
which is shown by the parallel lines tt’ at point F and vv’ at point F’ in Figure 2.4. e slope of the line tt’
equals the MRPTxy and the corresponding marginal cost ratio, while the slope of vv’ equals the marginal
rates of substitution for the two consumers. Points F and F’ in Figure 2.4 are thus consistent with Equation
2.6 above: they represent the third or top-level condition, and hence also the rst and second conditions for
a general equilibrium.
Point F is a Pareto-optimal top-level equilibrium in the sense that it is not possible to increase the output
of either of the two sectors or the utility of either of the two consumers without thereby reducing that of the
other.
Note that at this point of production (F), total output is X2 and Y2. ese quantities determine the
dimensions of the Edgeworth Box in Figure 2.4. However, the precise location of the top-level point on the
PPC will depend on the underlying assumptions of the model, particularly the initial distribution of
resources between the two individuals, A and B. If one of the two individuals owns most of the initial capital
and labour resources, and has a particularly strong relative preference for commodity Y, it stands to reason
that our model will generate a top-level equilibrium lying on the PPC and close to the y-axis in Figure 2.4
(and Figure 2.3). Chapter 5 revisits this important issue.
Figure 2.4 Consumption and overall equilibria

2.3 Efficiency and economic growth

Technical efficiency or X-efficiency refers to a situation in which the maximum possible output is obtained
from a given set of resources (put differently, a situation of technically efficient production). is necessarily
implies a position on the PPC, such as points C0 and S in Figure 2.5.
Points inside the PPC, such as R, indicate the presence of X-inefficiency. X-inefficiency (sometimes also
termed organisational slack) can be caused by a number of factors ranging from a lack of motivation by
production agents to a lack of information about market conditions, incomplete knowledge of production
functions, and incomplete speci cation of labour contracts.
Figure 2.5 X-inefficiency and economic growth

Clearly, X-efficiency alone is an insufficient measure of economic efficiency since technically efficient
production of goods as such does not necessarily re ect the needs of consumers. In common sense terms, it
is pointless to produce goods efficiently if people would rather consume some other combinations of those
same goods. Put differently, X-efficiency ensures that society is on its PPC, but cannot determine where
society should be on this curve. As shown in Section 2.2.3 above, the latter point is given by the top-level
equilibrium, and represents an economy that is X-efficient as well as allocatively efficient.
It is also possible to de ne economic efficiency in dynamic terms. Dynamic efficiency has to do with
increases in the quantity and/or productivity of the factors of production (that is, with economic growth).
e sources of growth are conventionally de ned to include savings, investment in the form of both physical
and human capital formation, technological inventions and innovations, and increases in the availability of
labour with different skills. From a general equilibrium perspective, the net effect of sustained economic
growth can be shown in Figure 2.5 as an outward shift in the PPC, for example, from M0N0 to M1N1, and by a
concomitant change in the competitive equilibrium, for example, from point C0 to C1.

2.4 Market failure: An overview

e perfectly competitive model is nothing more than a theoretical nicety, a normative ideal against which
real-world conditions can be judged. is section brie y distinguishes between several instances of market
failure. e most important of these are discussed in more detail in the remaining chapters of Part 1.

2.4.1 Lack of information

Producers and consumers sometimes lack information that is essential for making rational decisions.
Producers may be unaware of the existence of certain resources they require or of the latest technologies
available in their own lines of business. Similarly, consumers may be ignorant of potentially harmful
properties of some goods and services they consume or unaware that certain goods are available at lower
prices than what they are currently paying (Bohm, 1978). Information problems also arise in labour markets:
unemployed workers are often unaware of job vacancies, while employers are often unaware of available job
seekers who can ll their vacancies. In addition, many transactions are characterised by asymmetric
information, that is, situations in which either the buyer or the seller has more or better information than
the other party has about the terms and implications of the exchange.
Information-related problems of this nature feature prominently in the rest of this book. Regulatory
authorities are often ignorant about the pricing of local and global public goods or the size of and solutions
to externality problems (Chapter 3). Governments are also unsure about the most efficient way of regulating
natural monopolies (Chapter 4), dealing with poverty, inequality and service delivery (Chapters 5, 8, 9 and
10), avoiding or minimising principal–agent problems between and among politicians, bureaucrats and
voters (Chapters 6, 7 and 19), and about the economic impact of various taxes (Chapters 11–16). In scal
policy-making, the possibility of unforeseen GDP changes means that governments can never be sure if
actual tax revenue will reach the budgeted level (Chapter 18) and neither do they know precisely what the
cost of their debt will be in the future (Chapter 17).
Very few individuals and institutions have all the information they need when they have to take decisions
that would affect their own well-being. Even if consumers, producers and job seekers were prepared to pay
for the required information, this would not always help. It would clearly be impossible for them to calculate
costs of this nature without knowing exactly what information they will obtain. But if they knew the latter,
there would be no need to acquire the information in the rst place, let alone calculate the cost of acquiring
it. ey thus have no choice but to accept their lack of information as a binding constraint.
is raises the question whether governments in free societies should bear the responsibility of providing
individual citizens with information for private decision-making. Governments sometimes do perform this
function. In South Africa, for example, virtually all consumer goods must be cleared by the South African
Bureau of Standards. ey must also have labels displaying their weight, mass, volume and, more recently,
their ingredients, including chemical additives. ese standards have become increasingly important for
many African countries, partly because of the minimum standards imposed by regions and countries with
which they are trading. On balance, however, pro t-driven private institutions tend to acquire and
disseminate information more efficiently than government agencies do.

2.4.2 Friction and lags in adjustments

Most markets do not adjust very rapidly to changes in supply and demand. While this may partly be the
result of a lack of information, it is also true that resources are not very mobile at the best of times, that is,
even when the necessary information is at hand. is has given rise to the development of search markets in
which agents spend time and resources searching for information and, in the process, incur so-called
friction costs. Labour may take time to move from one job to another or from one place to another, while
physical capital can only move from one location to another at very irregular intervals. It takes time for an
entrepreneur to move out of textile production into computer software production or for a civil engineer to
become retrained as an electrical engineer, even when they are fully aware of the new opportunities
available in the market place. Acquiring and adapting the latest technologies may also take both time and
money, especially when the requisite skills are not readily available. e discussion of the effectiveness of
scal policy in Chapter 18 will revisit the issue of the effects of lags.

2.4.3 Incomplete markets

Markets are often incomplete in the sense that they cannot meet the demand for certain public goods (such
as street lighting, defence or neighbourhood security). Neither do they fully account for the external costs
and bene ts associated with individual actions. When a company pollutes the air or river water, damaging
the health of consumers using that air or water, it is not usually held responsible for its actions, and neither
are the consumers compensated for the additional health expenses that they might incur. Chapter 3
discusses these issues in more detail.
2.4.4 Non-competitive markets
Non-competitive markets are the rule rather than the exception. Monopoly often results from government
intervention in markets, using policy instruments such as licenses and regulations. Such monopolies are
called arti cial or statutory monopolies. Statutory monopolies cause social costs (a loss in consumer
surplus). ese social costs can be reduced by deregulating the industry, that is, by removing the arti cial
barriers to entry imposed by government. e natural monopoly is another market failure associated with
non-competitive markets. is market form results from industries exhibiting decreasing average costs over
the entire production range for which there is market demand. Public utilities such as bulk water providers
and rail transport are examples of natural monopolies. If these industries were to produce at the perfectly
competitive market equilibrium, they would be making losses. Government therefore has an important role
to play in ensuring that production and pricing are at a point that is close to the social optimum level.
Chapter 4 considers these issues in more detail.

2.4.5 Macroeconomic instability

At the macroeconomic level, markets may be slow to react to sudden exogenous shocks to the system. Left to
themselves, markets may take too long to adjust to changing external conditions and it is often necessary for
domestic policy-makers to take appropriate actions aimed at stabilising their currencies. e important role
now played by monetary and exchange rate policies can be viewed as an attempt on the part of governments
to deal with the problem of market failure at the macroeconomic level.
A more recent example of a serious exogenous shock is the 2007–2009 global nancial crisis, which has
even been compared with the 1929 Great Depression. Swift and appropriate actions were required from
policy-makers the world over to mitigate its effects. It was prompted by a liquidity shortage in the American
banking system, and has led to the downfall of large nancial institutions such as Lehman Brothers, bailouts
of banks by domestic governments and downswings in stock markets throughout the world. e effects of
the crisis were more severe in developed countries than in the developing world, with the International
Monetary Fund (IMF) and the European Union (EU) having to bail out several countries whose public debt
burdens had become unmanageably large (including Ireland, Greece and Portugal). is issue is taken up
again in Chapters 17 and 18.

2.4.6 Distribution of income

Perhaps the most important shortcoming of the neoclassical model of general equilibrium is its neutrality in
respect of the distributions of wealth and income and the related poverty problem. e model operates like a
‘black box’: what you get out of it is what you put into it. For all practical purposes, the distributional
outcome – as re ected by the precise top-level equilibrium on the PPC – is determined by the initial
distribution of capital and labour between the two individuals. If the initial distribution is highly unequal,
then so too will be the nal distribution. We take up this issue again in Chapter 5 and spell out some of the
relevant policy implications in the discussions of poverty and redistribution-related issues in Chapters 8, 9
and 10.

2.4.7 Further notes on market failures

ese examples of market failure are but a small sample of the many things that can go wrong in a modern
capitalist economy. It is clear that real-world economies are not very efficient in the conventional economic
sense of the word. ey do not, and probably never will, achieve a Pareto-optimal allocation of resources.
Neither will they produce an equitable distribution of income on their own, or even a distribution that is
acceptable to the broad community or its representative government.
But this does not mean that any other form of economic organisation, such as a socialist or centrally
planned economy, will be any better at allocating scarce resources in accordance with the true wishes of the
community. ere is now enough evidence to suggest that countries that have abided by the principles of the
market have generally grown more rapidly and have spread the bene ts of that growth more effectively than
countries that have opted for overly interventionist or centrally planned systems. Indeed, not all economists
have bought into the notion of market failure; as Box 2.2 shows, they have viewed the phenomena outlined
in this section as manifestations of the normal adjustments made by free agents in the face of changing
circumstances. e key issue here is whether these adjustments are being made in response to negative or
positive externalities imposed by others and whether they can be made in good time with little or no cost to
the broader society, or, conversely, whether government intervention can speed up the process and close
some of the gaps discussed above without undermining economic efficiency.
What cannot be denied is that the public sectors of most market-based economies have played a
signi cant role in creating wealth for their growing populations. In these economies, the public sector has
been charged with the task of providing certain public goods (such as defence and law and order), dealing
with the problem of externalities, protecting consumer interests through health and other standards,
regulating natural monopolies, and preventing abusive behaviour by monopolies and oligopolies. e
sections and chapters that follow will deal with these issues.

Box 2.2 Market failure: Alternative view

The Austrian school believes in the superiority of a free-market system in which government’s role is limited to
that of protecting the freedom of individuals and communities. To Austrian economists, the notion of market
failure, as conventionally defined, is something of a misnomer.
According to them, it should rather be viewed as an aspect of the dynamic process by which markets
continually adjust to changing tastes and technologies, with individuals and enterprises constantly searching for
and filtering information in an attempt to stay or get ahead of their rivals. Free and competitive markets operate
in an evolutionary fashion, with only the fittest surviving, and with the only ‘failure’ being the ‘creative
destruction’ of firms that are unable to produce at minimum cost, or create or imitate new technologies (for
example, Schumpeter, 1987; Nelson & Winter, 1982). Other economists would argue that market participants
can and often do overcome the problem of imperfect information themselves. We have already referred to the
creation of ‘search markets’ in which employers and job seekers spend time and money to acquire information
they need to secure their occupational well-being. Another example is the payment of efficiency wages by
employers in order to save on monitoring costs and, in the process, avoid an otherwise serious principal–agent
problem (Stiglitz, 1984). Yet another example is credit rationing on the part of banks when demand exceeds
supply at the prevailing interest rate. There is also a view that too little is known generally about the nature and
extent of market failures, and that governments have less of an incentive to fill informational gaps of this
nature than the private sector (Cowen & Crampton, 2011).

2.5 Enter the public sector: General approaches

Section 2.4 showed that various kinds of market failure could render the benchmark model of perfect
competition unworkable in the real world. ese failures provide a prima facie case for government
intervention. In this section, we focus brie y on the broad approaches that governments can follow to deal
with failures of this nature.
Economists conventionally distinguish between three broad functions of government: the allocative,
distributive and stabilisation functions.

2.5.1 Allocative function

e allocative function of government refers to the role of government in allocating scarce resources to
competing uses in the economy and stems from the distorting effects of market failure on the allocation of
resources in an economy. Market failures caused by incomplete and non-competitive markets are
particularly important sources of allocative distortions. Chapter 3 discusses incomplete markets in more
detail, but we will brie y point out the main manifestations of incomplete markets here.
ere are two manifestations of incomplete markets. e rst one involves characteristics of some goods
and services that prevent efficient supply by competitive markets. Market failures of this nature can take
various forms. In the case of pure public goods (such as national defence and street lighting), potential
consumers have a strong incentive not to reveal their demand. is makes it impossible to determine a
price, or to force users to pay for the bene ts they derive from using the good or service. As a result,
competitive markets cannot supply pure public goods at all, even though they are in great demand. A second
class of goods and services, known as mixed goods, has some public-good characteristics. Consumers would
either not reveal their demand for such goods and services or producers would nd it impossible to enforce
payment of the price. Mixed goods can be supplied by competitive markets, but neither the quantity
supplied nor the price resulting from market provision would be optimal. Examples of these goods are
subscription television services and certain healthcare services.
A second manifestation of incomplete markets is the existence of externalities. e activities of individual
consumers and producers often affect other parties. Failure to account for these external effects tends to
create divergence between actual market prices and quantities and their socially optimal equivalents.
Externalities can be either negative, as in the case of air or river pollution, or positive. An example of a
positive externality is the ‘additional’ bene ts that society derives from education, such as having a more
literate voter population and a lower crime rate. ese bene ts are additional to the private bene ts
accruing to individuals in the form of higher earnings.
Chapter 4 explains that non-competitive markets may take two forms. ‘Arti cial’ monopolies operate in
markets where perfect competition is technically feasible, but prevented by legal restrictions imposed by
government or professional bodies. By contrast, ‘natural’ monopolies emerge in industries characterised by
large capital outlays that give rise to economies of scale over the entire range of their output. Only one rm
can operate effectively in a market of this nature. Examples include the electricity transmission industry and
the local loop in the telecommunications sector. In both forms of non-competitive markets, rms maximise
their pro ts by supplying less than the optimal quantity of the good or service at too high a price.
Governments have various instruments at their disposal for correcting the allocative distortions arising
from incomplete and non-competitive markets. Section 2.6 refers to some of these instruments, while
Chapters 3 and 4 discuss them in more detail.

2.5.2 Distributive function

Section 2.4 pointed out that the neoclassical general equilibrium model is decidedly agnostic when it comes
to the distribution of wealth or income in a society. It yields a distributional outcome that is largely
determined by the initial distribution of labour and capital between the market participants – in fact, a
Pareto-optimal allocation exists for each possible distribution. Hence, the model is useful for determining
the Pareto-optimal allocation of resources for a given income distribution, but is silent on the fairness of the
outcome.
e distributions of income and wealth generated by markets are often highly unequal. e ubiquity of
the distributive function of government (that is, the use of combinations of taxes, transfer payments and
subsidies to alter distributional outcomes) suggests that no society regards the market-determined
distribution of income as being fair or just. Some commentators have even suggested that a redistribution
could improve the general well-being of society even if it carried a cost in terms of lower levels of
productivity or slower economic growth. While much disagreement remains about appropriate criteria for
evaluating policies to redistribute income or wealth, the vast majority of economists now agree that the
distributive function is a key aspect of the role of government in a modern capitalist economy. Chapters 5, 8,
9 and 10 investigate aspects of this function in more detail.

2.5.3 Stabilisation function

e stabilisation function of government has to do with its macroeconomic objectives, which include an
acceptable rate of economic growth, full employment, price stability, and a sound and manageable balance
of payments (see Chapter 18). An inability on the part of competitive markets to achieve these outcomes
would represent market failure on a grand scale, and often induces governments to implement corrective
monetary, exchange-rate and scal policies. While a detailed analysis of macroeconomic issues falls outside
the scope of this book, Chapters 17 and 18 discuss key aspects of scal policy. e purpose of this subsection
is to provide a brief overview of the stabilisation role of government.
e notion that governments have an important stabilisation function is associated mainly with the
Keynesian school of macroeconomic thought. e Keynesian approach to stabilisation rests on three
premises:
1 e market economy is inherently unstable.
2 Macroeconomic instability is a form of market failure that is extremely costly to an economy.
3 Governments are able to stabilise the economy by means of appropriate macroeconomic policies.

Hence, Keynesians propose active counter-cyclical policies to stabilise economic activity. eir proposed
measures mainly work on the demand side of the economy. In times of recession, governments should
reduce taxes, increase their expenditure and boost credit expansion in order to raise aggregate demand and
stimulate economic activity. Conversely, in ationary overheating of the economy should be addressed by
higher taxes and lower levels of state spending and credit expansion, thus moderating aggregate demand.
is Keynesian view of the stabilisation function of government is not without its critics. Economists
from the New Classical school of thought in macroeconomics argue that Keynesian economics is not
properly grounded in microeconomic principles and that its adherents fail to realise that rational agents may
anticipate the actions of government and act upon them even before they are executed or have their
intended effects. New Classical economists believe that the market economy is self-adjusting and that
government intervention worsens rather than improves matters. e implication of the rst belief is that
stabilisation policies are unnecessary in market economies. e second belief implies that attempts by
governments to stabilise economic activity will invariably fail, irrespective of whether such interventions are
warranted.
In response to these criticisms, the so-called Neo-Keynesian school of thought has tried to revive
Keynesian theory by providing it with credible microeconomic foundations. Neo-Keynesian economists
argue that many rigidities characterising the modern economy are perfectly consistent with rational
economic behaviour and that some of them, such as wage and price contracts, can prevent the economy
from responding or adjusting rapidly to exogenous shocks. ese theories provide some justi cation for
demand-management policies, although Neo-Keynesians do acknowledge the practical difficulties
confronting policy-makers, such as policy lags and shifting expectations.

2.6 Direct versus indirect government intervention

Another way of approaching the role of government in the economy is to look at the nature of government
interventions. Viewed very broadly, it is useful to distinguish between direct and indirect forms of
intervention.
Direct government intervention refers to the actual participation of government in the economy. It
includes the government’s constitutional right to tax individuals and companies, to borrow on the nancial
markets and to execute its budgeted spending programmes. As far as the latter programmes are concerned,
governments intervene directly when they respond to a market failure by producing or supplying a good or
service (such as national defence, waste disposal or electricity) or by nancing production undertaken by
the private sector on a contract basis (such as school textbooks and much of the state’s infrastructure).
Indirect government intervention refers to the regulatory function of government. Regulation entails
enacting a law or proclaiming a legally binding rule that gives rise to market outcomes that are different from
those that would have been obtained in the absence of the regulation. Examples abound. In South Africa,
they include the labour laws that are aimed at improving the working conditions of labour, the anti-tobacco
law through which it is hoped to curb tobacco smoking, the competition policy, which is aimed at
preventing abusive behaviour on the part of monopolies, and several environmental control measures.
Indirect taxes and subsidies, which change market outcomes, also constitute indirect scal measures.
e distinction between direct and indirect interventions can make an important difference to our
estimates of the size of the public sector and its effects on the economy. Conventional indicators of the size
of the public sector, which are based on the total tax burden, government expenditures and the budget
de cit or surplus, provide a reasonably accurate picture of the size and extent of direct government
intervention in the economy (or of the ‘resource use’ by government, as explained in Chapter 1). e
problem is that regulatory interventions, whose total effect on the private sector may well be as important as
that of direct measures, are not explicitly captured in the national or government accounts. Conventional
measures of the size of the public sector, such as those used in Chapter 1 and in Chapter 7, may well
underestimate the overall effect of public-sector activity on an economy. However, as yet there is no suitable
quantitative indicator that fully accounts for the impact of government regulations on the economy.

2.7 Note on government failure

It is important to realise that governments, like markets, can also fail. ose involved in the business of
government – politicians, bureaucrats and public employees – are ordinary human beings like the rest of us.
ey often pursue their own self-interest rather than the public interest, and the protected nature of their
business dampens their incentive to be X-efficient. ey make mistakes, wittingly or unwittingly, and some
are even corrupt (as are the citizens and corporate executives who offer them bribes and other inducements
to abuse public resources).
In Chapter 6, we argue that ‘government failure’, like market failure, is nothing sinister or extraordinary. It
is a perfectly natural outcome of the factors in uencing the behaviour of politicians and government
officials. Like their counterparts in the private sector, they are utility maximisers: politicians want to
maximise votes, virtually at all costs, while bureaucrats often strive to maximise the size of their
departmental budgets, or ‘empires’. e net effect is usually an excess supply of public goods and services or
a government that is bigger than its ‘optimal’ size.
It is therefore important – indeed imperative – for the tax-paying public as well as students of public
economics to know how efficiently their government is performing its various functions. However, in order
to make this judgment, it is not only necessary to know why governments intervene in the economy in the
rst place (which is the focus of the rest of Part 1), but also what it is that governments are supposed to do.
is is the focus of the remaining parts of this book. Only then will we be able to make an informed
judgment about the role of government in our economy.

Key concepts
• allocative efficiency (page 22)
• allocative function of government (page 31)
• asymmetric information (page 28)
• direct government intervention (page 34)
• distributive function of government (page 32)
• dynamic efficiency (page 27)
• indirect government intervention (page 34)
• market failure (page 28)
• Pareto optimality in consumption (page 22)
• Pareto optimality in production (page 23)
• Pareto-optimal top-level equilibrium (page 26)
• regulation (page 34)
• stabilisation function of government (page 33)
• technical efficiency or X-efficiency (page 27)
SUMMARY
• Over the years, the neoclassical theory of general equilibrium has become something of a benchmark
model or a frame of reference that can be used for analysing real-world problems.
• As such, the model is based on a set of ideal but unrealistic assumptions (for example, fully informed
agents, no externalities and perfectly competitive markets) that are necessary to generate a hypothetical
economy in which all resources are fully utilised to the satisfaction of all of the agents involved.
• e equilibrium properties are essentially threefold. First, each producer achieves a constrained
optimum, determined by his or her own resources and competitive factor prices. Second, each consumer
maximises utility subject to his or her own budget constraints and the market-determined competitive
commodity prices. e third condition follows from the rst two sets of equilibria occurring
simultaneously, thus establishing a Pareto-optimal ‘top-level’ equilibrium.
• e latter equilibrium represents a fully employed economy that is X-efficient in the sense of producing
the maximum output with its given resources. It is also allocatively efficient in the sense of producing the
optimal or most desired mix of commodities.
• With this model as the backdrop, the many market failures that characterise the real world are
highlighted. ese include a lack of information on the part of job seekers and employers in the labour
market as well as producers and consumers who are often unaware of the inherent qualities or
‘standards’ of the goods that they produce, export and consume.
• Markets are often incomplete in the sense that they cannot meet the demand for certain public goods, for
example, street lighting and national defence, on their own. Neither do they fully account for the external
costs and bene ts associated with individual actions, such as air and water pollution or the free
dissemination of useful information.
• Commodity markets are often dominated by monopolies and oligopolies capable of abusing their power,
while labour markets are in turn constrained by minimum wages imposed by trade unions, by
governments and by large corporations themselves. ese ‘failures’ all undermine allocative efficiency.
• An important shortcoming of the above model is the fact that it is entirely neutral about the distribution
of wealth and the related problem of poverty. It operates like a ‘black box’: what you get out of it is what
you put into it. e distributional outcome – as re ected by the precise top-level equilibrium on the PPC
– is determined largely by the initial distribution of capital and labour between the individuals.
• From a macroeconomic perspective, markets may take too long to adjust to changing external conditions
and it is often necessary for domestic policy-makers to take appropriate actions. e important role now
played by monetary and exchange rate policies can be viewed as an attempt on the part of governments
to deal with the problem of market failure at the macroeconomic level.
• Government intervention thus usually plays an allocative, distributive or stabilising role in the economy.
e nature of this intervention may be ‘direct’ in the sense of supplying public goods or ‘indirect’ in the
form of regulating the private sector.

MULTIPLE-CHOICE QUESTIONS
2.1 Which of the following statements is/are correct? e theoretical notion of a ‘general equilibrium’
implies equality between …
a. the marginal rates of technical substitution (MRTS) among producers.
b. the marginal rates of substitution (MRS) among consumers.
c. the MRTS and the MRS.
d. the MRS and the marginal rate of product transformation (MRPT).
2.2 Which of the following statements is/are correct? Technical or X-efficiency refers to an economy
operating at …
a. any point along its contract curve for production.
b. any point on its production possibility curve (PPC).
c. a point on or below its PPC.
 d. its top-level equilibrium on the PPC.
2.3 Which of the following statements is/are correct? In the general equilibrium model, allocative efficiency
represents …
a. any point along the contract curve for consumption.
b. any point on the PPC.
c. a point on the PPC where MRS = MRPT.
d. a situation where each consumer achieves maximum utility simultaneously.

SHORT-ANSWER QUESTIONS
2.1 Brie y distinguish between technical or X-efficiency and allocative efficiency.
2.2 In what sense can a lack of information be viewed as a market failure?
2.3 Distinguish between the allocative and the distributive functions of government.
2.4 Should governments have a stabilisation function?
2.5 Brie y explain the meaning and implications of government failure.

ESSAY QUESTIONS
2.1 Outline the conditions for a top-level general equilibrium and explain why they represent a Pareto-
optimal allocation of resources.
2.2 Distinguish between allocative efficiency, X-efficiency and economic growth (‘dynamic’ efficiency), and
brie y consider their relevance to Southern Africa.
2.3 Explain the meaning of market failure and provide a few pertinent examples.
2.4 Should governments necessarily intervene to correct so-called market failures?
Public goods and externalities

Philip Black and Krige Siebrits

Chapter 2 introduced a benchmark model that explains the allocation of scarce resources in a perfectly
competitive economy. e exposition emphasised that this model is not a realistic description of the real
world; instead, it is a normative standard against which the performance of real-world markets can be
judged. Against this backdrop, Section 2.4 of Chapter 2 introduced the notion of market failure, that is, the
inability of real-world markets to achieve the efficient outcomes of the benchmark model.
is chapter provides more detailed discussions of two important sources of market failure, namely
public goods and externalities. ese two types of market failures re ect the incompleteness of many real-
world markets. On their own, free markets cannot meet the demand for pure public goods or fully account
for the external costs and bene ts associated with individual actions. Hence, these market failures provide a
rationale for complementary government actions to improve the allocation of resources. In addition to
theoretical perspectives on public goods and externalities, this chapter also outlines policy implications and
options. ese include the regional and global dimensions of some public goods and the international
spillover effects of some externalities.

Once you have studied this chapter, you should be able to:
distinguish between private, public, mixed and merit goods
use supply and demand analysis to derive the conditions for the optimal allocation of private and public
goods
explain why competitive markets fail to provide public and mixed goods efficiently
explain the distinction between the financing and the production of public goods and services
explain the concept of an externality
identify the main types of externalities
use supply and demand analysis to explain the effects of positive and negative externalities
discuss policy options to correct for externalities
discuss cap-and-trade programmes
provide examples and discuss the policy implications of global or regional public goods.

3.1 Private goods and the benchmark model

Chapter 2 gave no information about the goods produced and consumed in the benchmark model. It simply
labelled these goods ‘X ’ and ‘Y ’ and proceeded to derive the conditions under which they would be
produced and consumed in a Pareto-efficient manner. We now have to consider the question whether the
nature of X and Y has any bearing on the outcome of that analysis. Can we substitute any actual commodity
or service for X and Y, and still achieve allocative efficiency? As suggested in Section 2.4 of Chapter 2, the
answer to this question is ‘no.’ Competitive markets cannot supply all commodities and services efficiently.
e requirement for efficient production under competitive conditions is that consumers reveal their
preferences (or demand) for goods and services. By doing so, they signal to producers what types and
quantities of goods they prefer. Producers use these signals to decide what and how much to produce.
Furthermore, competition among producers ensures that they produce at minimum cost. Provided that
consumer preferences are revealed fully, the market performs like a huge auction that achieves the third or
top-level condition for allocative efficiency: simultaneous achievement of equilibrium by producers and
consumers (see Chapter 2, Section 2.2, for a discussion of allocative efficiency).
Conversely, competitive markets will fail if there are no satisfactory mechanisms by means of which
consumers can reveal their preferences. Whether mechanisms of this nature exist depends on the nature or
characteristics of goods and services. ey certainly exist in the case of private goods, which have the
following two characteristics:
• Rivalry in consumption: Private goods are wholly divisible among individuals.is means that one
person’s consumption of the good reduces its availability to other potential consumers. For example, if
andi wears a particular dress, it is not possible for Christine to wear it simultaneously. Similarly, the
consumption of an apple by Christine reduces the quantity of apples andi can consume by one.
• Excludability: e consumption of a private good can be restricted to particular persons, typically those
who pay the indicated or negotiated price. Once private goods have been paid for, ownership (or the
assignment of property rights) is certain and uniquely determined. For example, if abo pays for a drink
in a restaurant, he gains the sole right to consume that drink and has legally excluded Charles from
enjoying it.

Hence, the bene ts of consuming private goods are restricted to individuals who reveal their preferences for
these goods. e rivalry and excludability of private goods force potential consumers to reveal their
preferences, which sets in motion the competitive processes that result in allocative efficiency.
Figure 3.1, which depicts the market for takeaway coffees, illustrates these points. DB and DJ are the
individual demand curves for the two consumers, Bongani and Joan (recall from Box 2.1 that the basic
benchmark model has only two individuals and two goods). Each demand curve depicts the quantities of
takeaway coffees that the respective consumer would demand at different prices. e market demand curve
– given by DB + J – is simply the horizontal sum of the individual quantities demanded at each price. Market
equilibrium occurs at point E, where market demand equals market supply. is equilibrium yields a single
equilibrium price at point P. Joan and Bongani cannot affect the equilibrium price they pay for takeaway
coffees and are therefore price-takers. e equilibrium output of takeaway coffees is 0Q, with the quantities
demanded by Joan and Bongani given by 0J and 0B respectively. Although 0J and 0B add up to 0Q, the
quantities demanded by the two persons are not necessarily equal. e respective quantities demanded at
the equilibrium price may differ depending on the tastes, incomes and other characteristics of the two
consumers. Hence, they are quantity-adjusters, in the sense that each one determines the quantity that he or
she demands, given the equilibrium price.
Figure 3.1 Equilibrium of a private good

e takeaway coffee example highlights two important characteristics of private goods:

• Marginal utility equals marginal cost for each consumer. Recall that the area underneath the demand
curve gives the total utility derived from the consumption of takeaway coffees (that is, the sum of the
marginal utilities derived from the consumption of each takeaway coffee), while the area under the
supply curve gives the sum of the marginal costs of producing each takeaway coffee. erefore, at the
equilibrium price 0P, the marginal utilities of Bongani and Joan (BF and JG respectively) both equal the
marginal cost QE. is is the condition for the efficient supply of a private good.
• e price of a private good equals its marginal cost. As is evident from Figure 3.1, this is the efficient
pricing rule for private goods.

3.2 Pure public goods: Definition

e fact that private goods have two de ning characteristics implies that there are three classes of ‘non-
private’ goods. Two of these (namely non-rival, excludable goods and rival, non-excludable goods) are
known as mixed goods. Section 3.5 focuses on such goods. e remainder of this section discusses goods
that exhibit neither of the two characteristics of private goods (in other words, goods that are neither rival
nor excludable). Such goods are called pure public goods or pure social goods.
Pure public goods such as street lighting and national defence are indivisible – that is, they cannot be
divided into saleable units – and are therefore non-rival in consumption. For a given level of production of a
public good, one person’s consumption does not reduce the quantity available for consumption by another
person.1
 If andi uses a street light to guide her during her walk to a postbox, Roger can use that same street light
to establish whether he has found the street in which a distant relative of his lives. Similarly, the protection
provided by the national defence force to the inhabitants of one city does not reduce the ‘quantity’ of
protection available to the inhabitants of other cities or towns.
Non-rivalry in consumption has two important implications. Firstly, the fact that one person’s
consumption does not reduce the quantity available to other consumers implies that the marginal cost (in
other words, the cost of admitting an additional user) is zero. e second implication follows from the rst: it
would be Pareto-inefficient to exclude anyone from consuming a non-rival good, even if it was feasible to do
so. e reason is straightforward: allowing Ibrahim to use the street light referred to earlier will clearly make
him better off than before, but will not detract from the utility that andi and Roger derive from that same
street light. Section 3.3 returns to these implications.
Apart from being non-rival in consumption, pure public goods are also non-excludable, that is, it is
impossible to exclude persons from consuming these goods.2 Put differently, it is not possible to assign
speci c property rights to public goods or to enforce them. Again consider national defence and street
lighting as examples. e inhabitants of one province in a country cannot be excluded from protective
services provided by the SANDF (South African National Defence Force). Neither can anyone strolling along
a seafront promenade be excluded from sharing in the bene ts of street lights.
e two criteria for pure public good status are quite stringent. In practice, it would seem that there are
very few goods that qualify as pure public goods. For example, the protection offered by an army becomes
less effective as more people or larger areas have to be protected. Hence, it is debatable whether national
defence is fully non-rival at all levels of provision.
Similarly, very few goods are non-excludable in the true or ‘technical’ sense of the term. New
technologies are constantly expanding the scope for applying the exclusion principle to more goods.
Consider lighthouses, which used to be one of economists’ favourite examples of non-excludable public
goods. However, the service offered by a lighthouse is navigational assistance, rather than a mere beam of
light. Nowadays, it is technologically feasible and cost-effective to provide this service in the form of an
electronic signal made available only to those willing to pay for it.
Non-excludability on cost grounds is perhaps more common. It would clearly be very costly to place
police officers at all street lights and expect them to chase away all those who are unwilling to pay for the
bene ts received. It is possible, however, that technologies may yet be developed that make such exclusion
viable in nancial terms.
Much like the model of perfect competition, the pure public good case is an important analytical
benchmark despite its limited applicability to real-life situations. In the context of this chapter, it serves as a
useful introduction to the sections that follow.

3.3 The market for public goods

is section focuses on the market for pure public goods from a partial equilibrium perspective. More
speci cally, it analyses the public good equivalent of the private good case discussed in Section 3.1. e
implications of the divergent characteristics of public and private goods will become apparent when the
results depicted in Figure 3.2 are compared with those shown in Figure 3.1.
Figure 3.2 Equilibrium of a pure public good

Assume that Bongani and Joan live in the same apartment block. Figure 3.2 depicts the market for street
lights in Joan and Bongani’s street. Also assume that they are the only ‘consumers’ of the light. Curves DJ and
DB in the gure give their respective demands for street lighting. Note that these are so-called ‘pseudo
demand curves’ because they can be drawn only if consumers accurately reveal the quantities that they
demand at different prices. As was indicated in Section 3.1, however, revelation of preferences occurs only
with private goods. is point and its implications are discussed further below, but assume for the moment
that Bongani and Joan do reveal their preferences for street lighting accurately. Given this assumption, the
individual demand curves and the total supply curve S are drawn similar to those in Figure 3.1.
e fundamental difference between the public and private good cases is the manner of deriving the
market demand curve. In Figure 3.1, the market demand for takeaway coffees was derived by adding the
demand curves of Joan and Bongani horizontally. Yet, in the case of a public good, which is indivisible,
horizontal summation of the quantities demanded by each consumer at each price is clearly inappropriate.
e non-excludability of street lighting implies that the full quantity supplied is available to both Bongani
and Joan. is implies that the consumers of pure public goods are quantity-takers. Hence, the market
demand for a pure public good (DB + J ) is derived by adding the demand schedules vertically. In effect, this
means adding the marginal utilities they derive from different quantities of street lighting (or, put differently,
the prices they are willing to pay for it), not the quantities they demand at different prices.
e equilibrium position is de ned in the usual manner at the point of intersection between the market
demand curve and the total supply curve. is occurs at point E. e equilibrium output 0Q is available to
both consumers. Price 0PB + J represents the total amount that the two consumers together would be willing
to pay for the equilibrium quantity of street lighting, 0Q. In the example of Figure 3.2, Bongani is willing to
pay a price or equivalent tax of 0PB (which equals his marginal utility), while Joan is willing to pay a price or
tax of 0PJ (which equals her marginal utility). Hence, Bongani and Joan are price-adjusters who can adjust
their willingness to pay for street lighting.
Lindahl (1958) developed a similar model in which each user paid a different price per unit of the public
good based on his or her tax share (or Lindahl price). rough an auctioneer-driven competitive process,
equilibrium is achieved when all users agree to their respective tax shares and the associated quantity of the
public good, for example, 0PB, 0PJ and 0Q in Figure 3.2.
e rules for the efficient allocation and pricing of public goods also differ from those for private goods.
Keep in mind that the areas under the demand and supply curves in Figure 3.2 show the sum of marginal
utilities and the sum of the marginal costs respectively. e equilibrium position implies that the condition
for the efficient provision of a public good is equality between the marginal cost and the sum of the marginal
utilities of the individual consumers. is condition makes it possible to derive the efficient pricing rule for
public goods: the sum of the individual prices should equal the marginal cost. If good X in the two-sector
model of Chapter 2 is the public good, then the equilibrium for sector X can be stated as follows:

where the two terms on the right represent the marginal utilities that Bongani and Joan derive from
consuming good X.
e other conditions for a general equilibrium remain the same as shown in Chapter 2. It is important,
however, to reiterate that the equilibrium shown in Figure 3.2 is basically a ‘pseudo’ one, because real-world
consumers of pure public goods will not reveal their true preferences.
Table 3.1 summarises the discussion so far by contrasting key characteristics of public and private goods.

Table 3.1 A comparison of key characteristics of public and private goods

Public goods Private goods

Property rights Non-excludable Excludable

Consumption Non-rival Rival

Vertical addition of individual

Aggregate demand curve Horizontal addition of individual demand curves
demand curves

The sum of marginal utilities

Partial equilibrium condition for Marginal utility of each consumer equals marginal cost (MUi =
equals marginal cost
optimum provision MC, with i the individual consumer)
(∑MUi = MC)

The sum of individual prices

Efficient pricing rule equals marginal cost Price equals marginal cost (P = MC)
(∑P = MC)

Source: Adapted from Freeman, A.M. 1983. Intermediate microeconomic analysis. New York: Harper, p. 482.

3.4 Who should supply public goods?

Why do private markets fail to supply goods and services characterised by non-rivalry and non-
excludability? Section 3.2 already touched on the effects of non-rivalry when it stated that the marginal cost
of admitting additional users of non-rival goods is zero. Hence, the condition for efficient pricing by
competitive markets (P = MC) requires the price to be zero as well. It is clear that pro t-maximising
producers cannot apply the efficient pricing rule to non-rival goods, because they would be unable to cover
the costs of providing the good or service if they charge a price of zero.
In principle, the alternative of setting a cost-covering price equal to the sum of the individual prices
would enable a competitive market to supply the good. However, such provision would be inefficient. Any
price other than zero exceeds the zero marginal cost of admitting additional users of a non-rival good and
consequently reduces its consumption. Such exclusion is Pareto-inefficient, because doing away with it
would increase the welfare of previously excluded consumers without reducing the welfare of those who
already enjoy access to the good. In summary, it is impossible to determine an equilibrium market price for
a non-rival good.
e non-excludability characteristic of public goods and services creates incentives for ‘free-riding,’ that
is, misrepresentation or full-scale hiding of preferences by consumers who believe that it may be possible to
enjoy the bene ts of the good without having to pay for it. Recall the example of street lighting. Being
rational individuals, Bongani and Joan know that they cannot be excluded from enjoying the bene ts once
street lighting is provided. Hence, both of them have an incentive to understate the intensity of their
preferences for street lighting in the hope that the other will reveal his or her demand and pay for the service.
If they respond to this incentive, they become free riders. If Bongani reveals his preference for street lights
while Joan attempts to ‘free ride,’ a competitive market will undersupply street lighting at the level where
Bongani’s marginal utility equals the marginal cost of provision; this is represented by point B in Figure 3.2.
In the extreme case where both Joan and Bongani attempt to ‘free ride,’ no street lighting would be provided
at all: their true preferences would then not be revealed. Box 3.1 discusses experimental evidence about the
extent of free-riding.

Box 3.1 Do people ‘free ride’? Experimental evidence

The notion of free-riding has recently been brought into question. Economists have conducted several
experiments among students to test one of the basic tenets of neoclassical economics, namely the
‘selfishness axiom’ or, in the present context, the temptation to free ride (see Kagel & Roth, 1995). In the so-
called public goods game (PGG), for example, the experimenter gives each participant or player an initial
endowment – not known to others – in the form of real money or a token that may or may not differ in value.
Each player is then asked to make an unspecified and entirely voluntary contribution to a public fund to be used
to the benefit of all. The players also know that once all contributions have been made, the experimenter will
augment the public fund (for example, by 50%), and then share it out equally among all of the players.

The dominant strategy in this game would be free-riding (that is, zero contributions to the public fund). Yet the
vast majority of PGG players actually contribute, on average, between 40 and 60% of their endowments.
Although responses tend to vary markedly and some players contribute nothing, free-riding is definitely not a
dominant strategy. In an experiment among secondary school learners in Cape Town with a highly unequal
distribution of endowments, Hofmeyr, Burns and Visser (2007) found that low and high endowment players
contribute roughly the same fraction of their endowments to the public fund. This result emerged after ten
rounds of the game, during which time players could adjust their contributions to what they considered to be a
‘fair share’. This and most other studies show that reciprocity and a sense of fairness do feature in the utility
functions of individuals.

Government provision of public goods and services can improve on the inefficient outcomes of the market.
However, it cannot ensure an optimal provision of public goods. Compared to the market, the government
has the advantage of coercive powers that can be used to enforce payment for public goods. Yet the
government is no more able than the market to get consumers to reveal their demand for these goods.
Hence, it cannot determine efficient prices either. Figure 3.2 can be used to illustrate these points. Assume
the government wishes to apply the efficient pricing rule for public goods, P = MC. To do so, however, it
would have to know the demand curves of the two consumers so that it can charge each consumer a price
that is equivalent to his or her marginal valuation of the bene ts of street lighting. In this case, the
government would have to charge Bongani 0PB and Joan 0PJ to recover the full marginal cost (0PB + J) of
providing street lighting (0Q). Optimal provision of a public good thus requires the application of price
discrimination, that is, the practice of charging different consumers different prices.
In practice, however, the government does not have the required knowledge about the preferences of
individuals to enable it to apply perfect price discrimination. is is the reason why governments typically
cover the costs of supplying public goods by collecting a ‘tax price’ from consumers. e mandatory nature
of tax payments eliminates the ‘free rider’ option and gives taxpayers a direct stake in revealing their
preferences for public goods. Once Bongani and Joan have been forced to surrender a part of their hard-
earned salaries to the government, they clearly have an incentive to participate in decisions on the use of
their tax contributions, for example, by insisting on proper funding of the defence force or the provision of
street lighting in their neighbourhood. As will be pointed out more fully in Chapter 6, such participation
could take the form of voting in a referendum on tax and expenditure measures or voting for political parties
in democratic elections.
e actual production of public goods need not necessarily be undertaken by government – it could be
on a contract basis by the private sector. It follows that the critical difference between public and private
goods lies in the nancing of these goods. When mention is made of public goods, it essentially refers to the
need for public or collective nancing (as opposed to private nancing via the private nancial sector). It
might even be argued that all public goods could be produced by the private sector as long as the public
sector undertakes the nancing. Many economists would argue that such a system would be more efficient
than one in which the government, which is not subject to the pro t and loss discipline of the market,
produces public goods.
ere are many examples of goods and services that are produced by the private sector but nanced by
the public sector: school textbooks, medicines, roads, dams and the like. In addition, governments often use
private goods and services to meet public demands, with the labour of public sector employees being the
only signi cant value that is added in the ‘production processes’. Examples include the equipment used by
the military to produce national defence systems, the equipment and buildings used by diplomats in
overseas embassies, and the electronic equipment used by administrators in the public sector. It is clear that
a substantial part of government activity is in a sense already ‘privatised.’ e much-debated issue of further
privatisation largely involves the service component and the very important nancing issue.
To be sure, government nancing presents problems of its own. Other chapters of this book will discuss
some of the efficiency implications of using taxes and debt to nance government spending.

3.5 Mixed and merit goods

As the name suggests, mixed goods possess characteristics of both private and public goods. ese goods
and services are common in the real world and raise several vital questions about the economic role of
government. Two classes of mixed goods and services can be distinguished.
• Non-rival, excludable mixed goods and services. Consider a highway running through a remote rural
area (for example, the part of the N10 highway between Britstown and Prieska in the Northern Cape
province in South Africa). e exclusion principle can be applied by installing a toll gate, as has already
been done on the N1 highway between Johannesburg and Bloemfontein and the N3 highway between
Johannesburg and Durban. On an average day, however, access to the N10 is non-rival as users do not
compete for road space. e public good characteristic of non-rivalry in access prevents competitive
markets from providing roads of this nature efficiently. As discussed in the case of pure public goods, the
problem is the impossibility of determining a competitive price. e competitive solution of setting the
price equal to marginal cost is inappropriate since the marginal cost of access to the road is zero, while
charging a cost-covering price will cause Pareto-inefficient exclusion. is class of mixed goods is also
known as ‘club goods’: as is the case with sports and other clubs, the bene ts associated with their use
tend to be non-rival up to a point. As the number of users grows, however, congestion eventually sets in
 and consumption becomes rival. Other examples are private parks, Wi-Fi networks and subscription
television channels.
• Rival, non-excludable mixed goods and services. On weekdays, main thoroughfares in the downtown
areas of large cities in Southern Africa are good examples of the class of mixed goods characterised by
rivalry in consumption and non-excludability. Rivalry in the form of competition for the scarce road
space is erce and the marginal cost of road usage increases as congestion increases. In theory, it
becomes possible to determine an efficient price at the level of the marginal cost. It is very difficult to
apply the exclusion principle, though – attempts to levy toll charges at the entrances and exits to the
central business districts of Johannesburg and other large cities would cause very costly congestion
effects. In this case, market failure arises from the non-excludability characteristic of the mixed good.
is characteristic also explains why rival, non-excludable mixed goods are sometimes called ‘common
pool resources’. Fish stocks in international waters and open grazing elds are also examples of this type
of mixed goods.

e excludability of mixed goods is in uenced strongly by technology. As was pointed out in Section 3.2,
technological innovation has changed the status of navigational aids from a pure public good (the
lighthouse) to a non-rival but excludable mixed good (electronic information signals). In the same way,
sensors that measure traffic volumes, detect licence plate numbers and bill road users effectively change
congested urban roads from rival but non-excludable mixed goods into private goods. is technology
underpins the e-toll system to fund the Gauteng Freeway Improvement Project (GFIP) in South Africa. e
implementation of the e-toll system has been thwarted by legal challenges and widespread non-payment,
though.
It remains an open question whether mixed goods should be supplied by the public or the private sector.
In many countries in Southern Africa and further a eld, mixed goods such as education and healthcare
services are provided by both sectors. Such countries have public and private hospitals, for example, and
wholly owned government schools as well as privately owned schools and training colleges. Likewise,
universities in most countries obtain revenue from government in the form of subsidies, but also charge
students registration and tuition fees. is split can be viewed as an attempt to share the costs of university
education in accordance with its public and private good components.
Various market-failure-related considerations affect the extent to which governments become involved
in the provision of mixed goods. For one thing, persons who buy or receive non-excludable mixed goods
such as education and healthcare services often confer external bene ts on others. Markets underprovide
and underprice goods and services that confer positive externalities, and one of the arguments for public
provision of mixed goods and services is that it can overcome such market failures. Section 3.6 discusses this
issue in more detail. A second consideration affects infrastructure services that are excludable and exhibit
non-rivalry in consumption until congestion sets in (such as rail transport and electricity supply). Only one
rm can operate efficiently in markets for such services. Hence, public provision can be justi ed as a means
to prevent the allocative and X-inefficiency associated with supply by monopoly rms. Chapter 4 returns to
this issue. Section 3.6 and Chapter 4 will show that public provision is not the only option for addressing
these two types of market failure, though.
In the case of some mixed and even private goods, it is possible to apply the exclusion principle, but the
goods in question are regarded as so meritorious that they are often provided via the national budget.
Examples of these merit goods are education and healthcare services. Similar considerations underpin
certain regulations, such as laws that make schooling and the use of seatbelts in cars mandatory and ones
that prohibit harmful practices (for example, comprehensive bans on smoking or on the use of certain
drugs). Merit good arguments for regulations often re ect the belief that individuals are unable to act in their
own best interest. ere is an undeniable element of paternalism to such arguments that makes them quite
controversial, particularly among those who fear that special interest groups would attempt to use the
government to further their own views of how people should behave.

3.6 Externalities
Externalities, or external effects, can be positive or negative. Positive externalities occur when the actions
of a producer or consumer confer a bene t on another party free of charge. In the same way, negative
externalities occur when such actions impose a cost on the other party for which he or she is not
compensated. External effects can be technological or pecuniary in nature. Technological externalities
occur when external effects directly affect the level of production or consumption of the other party. By
contrast, external effects that change the demand and supply conditions, and hence the market prices facing
the other party, are known as pecuniary externalities. In both cases the bene ciaries get windfalls by not
having to pay for bene ts, while the prejudiced parties get no compensation at all.
It can be argued that pecuniary externalities do not have net effects on society. e gist of this argument
is that such external effects merely transfer resources from one owner to another, and that markets adjust
efficiently to the changes in demand and supply conditions. Consider an area in which crime is becoming
rampant and house prices are falling rapidly. Current owners and sellers will be disadvantaged, but buyers
will have the bene t of lower house prices. ere is therefore little or no net loss to society, only a
redistribution from one group to another. Nonetheless, it remains true that house buyers enjoy a windfall
while house sellers – the losers – get no compensation.
External effects drive wedges between the private (or monetary) costs and bene ts and the social costs
and bene ts of everyday market transactions. e social costs of a transaction are simply the sum of the
private costs and the external costs; likewise, the social bene ts are the sum of the private bene ts and the
external bene ts. e section that follows focuses on the respective marginal equivalents.
Externalities can originate on the supply side or the demand side of the market. It is possible to
distinguish between four broad categories.
e productive activities of a producer can have one of the following effects on the supply side of a
market:
• A negative external effect on other producers or consumers, in which case the marginal external cost
(MEC) > 0 and the marginal social cost (MSC) > the marginal private cost (MPC)
• A positive external effect, in which case MEC , 0 and MPC > MSC.

Likewise, the consumption activities of a consumer can have one of the following effects on the demand side
of a market:
• A positive external effect on other consumers or producers, in which case the marginal external bene t
(MEB) > 0 and the marginal social bene t (MSB) > the marginal private bene t (MPB)
• A negative external effect, in which case MEB , 0 and MPB > MSB.

Demand and supply curves can be used to analyse the effects of externalities and the scope for remedial
government intervention. While each of the four types of externalities can be analysed in this way, this
chapter focuses on two cases: a negative production externality and a positive consumption externality.

3.6.1 Negative production externality

Assume that a coal- red power station pollutes the air and the water used by nearby livestock and crop
farmers. is example of a negative production externality can be analysed with the aid of Figure 3.3.
e diagram shows the normal private demand curve (DP 5 MSB) and the private supply or marginal
cost curve (SP 5 MPC) for the electricity generated by the power station. Recall that these curves represent
the consumers’ bene ts from using electricity and the supplier’s cost of providing it, respectively. In a typical
market situation, equilibrium would occur at point E0, with 0Q0 electricity supplied at a unit price of 0P0.
From the perspective of the community as a whole, the costs incurred by the supplier do not re ect the
full cost of providing the electricity. e external costs of pollution to nearby farmers are ignored, yet they are
in fact part and parcel of the full or social cost of providing electricity. e ‘social’ supply curve labelled SS =
MSC in Figure 3.3 re ects these external costs. is curve shows that the negative externality raises the
social costs of providing electricity above the private costs of the supplier. By producing 0Q0 units of
electricity, the supplier incurs a marginal private cost equal to Q0E0 and a marginal external cost of E0F, which
together make up the marginal social cost of Q0F. At the private equilibrium point E0, total private costs equal
0Q0E0K and total external costs are KE0F.
If the externalities were taken into account, the ‘social’ equilibrium would be at point E1, where social
supply (or MSC) equals demand (assumed to equal MSB). At point E1, only 0Q1 units of electricity are
supplied at a unit price of 0P1.
Two points are worth emphasising here. Firstly, 0Q1 represents a lower quantity of output than 0Q0,
whereas P1 is a higher price than P0. us, from a social point of view, the presence of a negative production
externality in a competitive market causes inefficiency in the form of overprovision and underpricing of the
good in question. Secondly, the movement from point E0 to point E1 did not eliminate the externality. It
merely reduced it from KE0F to its optimal level, KJE1. e latter is an optimal level because the farming
community is prepared to accept this negative externality from electricity generation in exchange for the
value that it adds to their personal comfort and farming activities.e opposite case of a positive production
externality implies the presence of negative external costs, that is, a situation in which the social supply
curve (MSC) lies below and to the right of the private supply curve (MPC). A good example is the classic case
of a bee farmer’s bees pollinating the apple blossoms on an adjacent farm.

Figure 3.3 External cost and Pigouvian tax

3.6.2 Positive consumption externality

e market for education provides an example of a positive consumption externality. In Figure 3.4, the curve
S represents the social supply of educational services, that is, the marginal social cost of providing education.
Curve DP depicts the private demand for education, which re ects marginal private bene ts in the form of
skills accumulation, expected higher earnings and the sheer enjoyment to be had from being more
knowledgeable. e market equilibrium occurs at point E0, with 0Q0 education being supplied at a ‘unit
price’ of 0P0.
In this case, the externality originates on the demand or consumption side of the market. e bene ts of
education are not restricted to the individual recipient. Society as a whole also derives considerable bene ts
from the effects of education. e educated individual may, for example, disseminate valuable information
to other producers and consumers free of charge, thus enabling them to become more productive or happier
citizens. Higher education levels also go hand in hand with lower birth rates and lower crime rates, which
relieve the pressure on government, and hence on taxpayers, to provide additional healthcare facilities and
policing. Hence, the marginal social bene ts from additional education exceed the marginal private
bene ts.3 is is shown in Figure 3.4 by the social demand or MSB curve DS, which lies to the right of and
above curve DP.

Figure 3.4 External benefit and Pigouvian subsidy

e equilibrium position thus moves from point E0 to point E1 when the external bene ts from education
are taken into account. is movement raises the price from 0P0 to 0H and increases the quantity from 0Q0 to
0Q1. Competitive markets clearly underprovide and underprice goods and services with external bene ts.
e opposite case of a negative consumption externality implies the existence of negative external
bene ts, with the social demand curve (MSB) lying below and to the left of the private demand curve (MPB).
For example, someone who plays hard rock music at high volume into the early hours of the morning would
impose a negative externality on a next-door neighbour who prefers classical music.
3.6.3 Concluding note
External effects – whether negative or positive – on individual consumers and producers are common. Some
externalities even affect the natural environment via their effects on biodiversity. Air and water pollution and
the provision of education are textbook cases because of their undeniable importance. But there are many
other examples, including the external costs associated with the consumption of tobacco products and
alcoholic beverages. Box 3.2 gives a summary of the estimated size and composition of alcohol-related
externalities.

Box 3.2 Size and composition of alcohol externality

Alcohol consumption by private individuals can impose various external costs on society at large. These costs
are usually divided into three broad categories. The first type of cost refers to ‘direct’ expenditures on policing,
healthcare and repairs to deal with consequences of alcohol abuse such as fatal and non-fatal traffic accidents,
person-to-person assault, damage to property and ‘avoidable’ illnesses (for example, liver and cardiovascular
disease). The second category refers to ‘indirect’ labour and productivity costs, which result from output losses
caused by alcohol-related premature deaths, injuries, illness and workplace absenteeism. It is customary to
refer to the two cost categories of direct and indirect costs as financial tangible costs, because they can be
measured in monetary terms if the requisite data are available.
The third category is non-financial intangible welfare costs. These costs include emotional pain, depression
and feelings of regret that come about as a result of alcohol-related harm. These non-financial costs can be
estimated even though they do not have a monetary value. This can be done by finding out, through appropriate
surveys, how much the affected individuals and institutions would be prepared to pay to avoid these costs.
Many attempts have been made to estimate the first two cost categories mentioned above. In a recent
overview of international studies, the weighted average tangible external cost of alcohol consumption was
found to be 1,58% of GDP per year, with individual estimates ranging between 1% and 2% of GDP (Mohapatra,
Patra, Papova, Duhig & Rohm, 2010).
Similarly, Parry (2009) estimated the tangible external cost of alcohol consumption in South Africa at 1,5%
of GDP. There have been very few attempts to estimate the intangible costs of alcohol abuse, but a recent study
estimated that intangible costs in the EU came to more than double the tangible costs (Anderson & Baumberg,
2006). If this was true of other countries, including South Africa, the total size of the alcohol externality may be
as high as 3% of GDP.

3.7 Possible solutions to the externality problem

What can governments do about the allocative inefficiencies caused by externalities? Several policy
interventions exist, including Pigouvian taxes and subsidies, direct government regulation, the creation of
regulated markets, support of alternative markets, and the establishment of appropriate property rights.

3.7.1 Pigouvian taxes and subsidies

One possible solution involves the introduction of Pigouvian taxes and subsidies, named after the famous
British economist A.C. Pigou. e aim of such taxes and subsidies is to internalise externalities, that is, to
force parties to include the external effects of their actions in their own cost and bene t calculations.4
e effect of a Pigouvian tax can be explained by revisiting the negative production externality illustrated in
Figure 3.3. Recall that a negative production externality leads to an overprovision and underpricing of the
good in question. By levying a Pigouvian tax on the party whose actions cause the externality, the
government can increase the producer’s marginal private cost to the level of the marginal social cost. In the
case of air and water pollution caused by the generation of electricity, this would be achieved by levying an
ad valorem tax on the price charged by the coal- red power station, that is, a percentage tax equal to the
value of the externality. In unit terms, the tax would equal E0F at price Q0F (5 0G) and JE1 at price Q1E1 (5
0P1). Hence, it would shift the private supply curve SP to SS 5 MSC. However, since supply exceeds demand
at point F, the new equilibrium will be at point E1, where MSC equals MSB. To be efficient, this tax must
equal the marginal external cost measured at the new social equilibrium. At Q1, the marginal external cost is
the vertical difference between MPC and MSC, in other words, JE1. At point E1, the tax per unit of output is
JE1 and the (internalised) externality equals KJE1, which is its optimal size. e net effect of the policy is that
a large portion of the original externality – given by the area JE0FE1 – has been eliminated.
Instead of levying a Pigouvian tax on each unit of output, it could be levied on each unit of the externality
itself. is is commonly known as an emission fee or a congestion tax. If a unit tax of this nature exceeded
the unit cost of reducing the polluted emissions, the polluting rm could reduce output or it could use
pollution-free inputs and new technologies in order to reduce its emissions. It would do so until the
marginal cost of these reductions equalled the unit tax. Cities like Singapore and London have been
imposing congestion taxes on vehicles entering highly congested urban areas, especially during rush hours,
for a number of years.5
Likewise, Figure 3.4 can be revisited to illustrate the use of a Pigouvian subsidy to correct a positive
consumption externality. If the government subsidises education by E0F (or GE1) per unit, again on an ad
valorem basis, it will bring the marginal private bene t in line with the marginal social bene t by shifting
private demand DP to DS 5 MSB. At point F, the price inclusive of the subsidy is high enough to induce a
positive supply response. At the new equilibrium point E1, where the marginal social bene t equals marginal
social cost, the price paid by consumers will be Q1G (50P1), the subsidy per unit will be GE1 (5 P1H) and the
supplier will receive Q1E1 (5 0H). e subsidy must equal the marginal external bene t at the optimal output
level Q1 to produce the optimal result. e nal equilibrium represents a more efficient outcome in which
more educational services are provided at a lower unit price.
Pigouvian taxes and subsidies are widely used in the real world. Nonetheless, they are subject to the
same informational constraints as most of the other regulatory options discussed below. e Pigouvian
options unrealistically assume that the authorities are perfectly informed about the size of the external
effects and, hence, about the slopes and shapes of the respective demand and supply curves. Moreover, it is
assumed that it is costless to obtain this information.
In the case of tobacco and alcohol products (the so-called ‘sin’ goods), the Pigouvian tax becomes a
crude measure in the sense that it may not in fact achieve its objective of reducing consumption and, with it,
the attendant externalities. It may even produce perverse results. is point is discussed further in Box 3.3,
and in chapters 12 and 13.

Box 3.3 Perverse effects of Pigouvian taxes

It is generally accepted that consumption of tobacco and alcohol products – the perennial ‘sin’ goods – inflicts
huge externalities on the broader community. Tobacco smoking is a major cause of heart and lung-related
illnesses that require expensive medical treatment, mostly paid for by the tax-paying public. The South African
government, in line with many other countries, has therefore introduced a range of control measures aimed at
reducing tobacco smoking and the attendant externalities. The most important of these measures are excise
tax hikes. These are supplemented by stipulations of the Tobacco Products Control Amendment Act of 1999,
including the prohibition of smoking at work and other public places; comprehensive bans on tobacco
advertising, promotion and sponsorships; and restrictions on the tar and nicotine content of tobacco products.
One of the unintended effects of tax hikes on tobacco products is an increase in smuggling or the
substitution of illicit and cheaper products, often imported, for the legally taxed local equivalents. This has
happened in Canada, the United States, England, Belgium and many other countries. Tobacco smuggling has a
long history in South Africa, owing partly to the size and extent of informal trading networks within the country,
and also because cheaper substitutes are readily available in neighbouring countries. It is estimated that the
illicit (and unrecorded) market comprises between 40 and 50% of the total tobacco market in South Africa
(Lemboe, 2010).
The potentially perverse nature of the problem is straightforward. To the extent that tax hikes do boost
smuggling activity, the government may in fact fail to achieve its objective of reducing tobacco consumption.
 International evidence also shows that smuggled cigarettes are typically of a lower quality than their legal
counterparts, thus potentially resulting in a larger negative externality per cigarette than before the tax hike.
Hence, the net effect of the tax hike may be an increase in the overall size of the externality rather than a
reduction.
A similar substitution effect also applies to tax hikes on alcohol consumption. Official figures usually fail to
capture a significant portion of the alcohol market. This unrecorded market comprises trading in privately
produced and imported alcoholic beverages, often at prices lower than those in the recorded market. The World
Health Organization estimates that unrecorded consumption makes up 27% of the global market in alcoholic
beverages, while it is estimated that unrecorded consumption in South Africa constitutes 26,3% of the total
market (Econex, 2010). As is the case with tobacco consumption, evidence shows that home-brewed sorghum
beer, which is highly toxic and unhealthy, is being substituted on a large scale for legal alcohol in response to
excise tax increases. To the extent that this happens, the excise tax is unlikely to fulfil its mission of reducing
alcohol-related harms to an acceptable level and may even have the perverse effect of adding to it.
Tax hikes on tobacco and alcohol consumption may also have unintended adverse effects on the distribution
of income within households, especially poor households. This may happen when tax hikes raise the amount
spent on tobacco and alcohol products relative to other goods making up the household budget. Such changes
in the composition of household budgets have negative effects on non-drinking and non-smoking members of
households.
The male heads of many patriarchal households are often egotistical and heavy users of tobacco and
alcohol products. These men, who control their households’ budgets, tend to maintain their consumption of
such products in the face of excise tax increases, often at the expense of other important household goods
and services such as food, healthcare and education (Black & Mohamed, 2006). In such cases, the tax hikes
cause other members of households to be worse off than before. As mentioned before, household heads may
also respond to tax hikes by diverting tobacco and alcohol spending to poor-quality substitutes that may
compromise their health status. In either case, the tax hikes may have the perverse effect of actually enlarging
the negative externalities experienced within households and in communities more generally.

3.7.2 Direct regulation

e option of direct regulation is particularly relevant to cases of negative production and consumption
externalities caused by air and water pollution, sub-standard food and other consumer goods, alcohol
abuse, tobacco smoking, hard drugs, gambling and prostitution. Direct regulation includes restricting the
quantities of ‘harmful’ goods and services that are produced or consumed, and is often used as a
supplement to Pigouvian taxes.
In addition to coal- red electricity plants, there are many industries polluting the air and water,
including oil re neries, smelters, and producers of textiles and chemicals. e direct option – also known as
command-and-control regulation – entails informing each polluting industry that it must reduce pollution
to a certain level or face legal sanction. In the example used in Figure 3.3, the power station would be
ordered to reduce its output from 0Q0 to the socially optimal level 0Q1. Proponents of this option assume that
the government is sufficiently well-informed to determine the output level 0Q1. As is the case with Pigouvian
cases, it is very difficult to get such information in real-world conditions. Furthermore, regulation of this
nature is not efficient in industries consisting of differently sized rms, since it requires each rm to reduce
its output by equal absolute amounts or proportions. Such reductions, however, may force rms to violate
the efficiency criterion of producing at the point where marginal social cost equals marginal social bene t.
e reason is that these costs and bene ts may vary signi cantly between rms within the same industry, so
that the application of a blanket rule would result in some of them producing too much and others
producing too little.
e regulatory role of the state stretches far and wide. Most countries have bureaux of standards whose
task it is to ensure that all goods and services adhere to certain minimum standards. One of the aims of such
standards is to address an externality problem: consumption of sub-standard food may cause illnesses that
require treatment paid for by the general public, while a defective automobile may cause accidents that
could harm the owner-driver as well as other innocent persons. Similar arguments also apply to a host of
measures regulating industries such as telecommunications, tobacco products, alcoholic beverages,
gambling and prostitution.
In many of these cases, more direct or regulatory measures are superior for controlling externalities to
Pigouvian taxes (or subsidies). In the case of tobacco and alcohol consumption, for example, reference has
been made already to the need for limiting trading hours, and imposing age and space restrictions. ere is
also the problem of smuggling, or the substitution of lower quality and cheaper surrogates for taxed legal
goods. It may therefore be worth allocating public resources to combating smuggling before Pigouvian taxes
are used in an attempt to reduce legal consumption. Similarly, the external costs of drunk driving seem to
constitute a signi cant portion of the total external costs caused by excessive alcohol consumption. Here
too, it would seem that higher penalties and more road blocks would be better means of addressing the
problem than excise tax hikes would be, provided that sufficient law enforcement capacity exists.

3.7.3 Creation of regulated markets

A third possible solution to the externality problem is an incentive-based option generally referred to as the
cap-and-trade programme. It originated in the context of the pollution problem and consists of the
creation of a market in which the government would issue or sell legal permits giving owners the right to
pollute. e rst step would be for government to establish the overall quantity of pollutants that it considers
to be an efficient level – for example, the area KJE1 in Figure 3.3 – and then to issue a limited number of
individual permits to rms in polluting industries. Permit holders are free to sell their permits or buy
additional ones and will do so by comparing the price of the permit with the marginal cost of reducing their
emissions, for example, by shifting to cleaner technologies. If the permit price exceeds the marginal cost of
pollution reduction, permits will be sold and the proceeds can be used to help fund the costs involved in
reducing pollution. ose who buy permits can increase their emission of pollutants and possibly contribute
to a more concentrated distribution of a xed total quantity of emissions. If new technologies or other
factors lower the cost of reducing pollution, however, more permits will be offered for sale at lower prices
until few or no takers remain. Given the xed (and theoretically optimal) total supply of permits, a declining
price is a sure sign that the quantity of emissions is also decreasing.
e market-creation approach also assumes that the government possesses perfect knowledge about the
sources of pollution and the level of pollution commensurate with the efficient level of output. In fact, some
pollutants (such as the emission of sulphur dioxide from power plants) can be measured and monitored. On
balance, the evidence suggests that cap-and-trade programmes are cost-effective and environmentally
effective alternatives to command-and-control regulation. To be effective, however, such programmes
should be well designed to prevent excessive volatility in permit prices and other potential problems.6

3.7.4 Support of alternative markets

Given the pollution problem discussed above, a case can be made for governments to support the
development of alternative energy markets that are pollution-free. e case would be similar to that made in
respect of a Pigouvian tax aimed at reducing the negative externality caused by a coal- red power station to
its optimal level (Figure 3.3). e aim would be to encourage polluters to reduce the quantity of coal-
generated electricity by switching to these alternative sources of energy.
e alternatives to coal-generated energy include nuclear power, natural (including shoal) gas, hydro-
generated energy, solar heating and wind turbines. ese all have the advantage of emitting little or no
carbon dioxide, methane and other greenhouse gases. Furthermore, all of them except for nuclear and
natural gas can be installed and developed at relatively low cost and in good time. Wind turbines, solar
panels and, to a lesser extent, shoal gas fracking, are all divisible and excludable, and can thus be viewed as
private rather than public goods.
e number of wind and solar-powered energy plants has grown very rapidly worldwide, with China’s
investment in wind energy currently exceeding its investments in fossil-fuel generation. e South African
Wind Energy Association (2019) reported that independent producers of wind power had 900 wind turbines
in operation in March 2019. ese turbines had a combined installed capacity of 2 078 MW connected to the
national grid. But the problem – or rather challenge – is that there are many more prospective suppliers of
renewable energies who all need the permission of government, in the form of highly priced licences, to start
up or even expand their businesses. While licensing fees of this nature are important, depending as they do
on acceptable environmental impact studies, they also need to take into account the urgent need to reduce
greenhouse gas emissions.

3.7.5 Legal solution: Property rights

Yet another approach to solving the externality problem derives from the view that regulation as well as
Pigouvian taxes and subsidies are costly to administer and, in any case, are not likely to achieve their
purpose. e Nobel Laureate Ronald Coase argued that the divergence between marginal private cost and
marginal social cost arises as a result of insufficiently de ned property rights over the use of resources.
Broadly de ned, property rights represent the legal speci cation of who owns what goods, including the
rights and obligations attendant upon this ownership. Hence, this approach rests on the view that the
problem of externalities boils down to disputes over the ownership of resources.
Consider an example. Does the owner of a at have the right to paint her front door a garish pink when it
not only offends her neighbour every time he passes it to reach the door of his own at, but may also reduce
the value of his apartment? Or, returning to an earlier example, do the producers of electricity possess the
right to discharge pollutants into the atmosphere? Do farmers possess rights to unpolluted air and water? If
property rights are well-de ned, they can be exchanged on a voluntary basis by means of straightforward
market transactions. If Prudence had the right to paint the outside of her front door any colour she wished,
then Bongani could offer to pay her a certain amount not to paint it pink. Likewise, if farmers on the
Mpumalanga Highveld had a right to clean air and water, nearby power stations could buy that right from
them by offering to compensate them for the damage caused. e externality problem can thus be resolved
without the need for potentially unsuccessful government intervention.
e Coase theorem holds that market incentives will generate a mutually bene cial exchange of
property rights through which externalities can be fully internalised, provided that property rights are well-
de ned and enforceable, and transaction costs are negligible. Transaction costs are the costs of the
exchange process. ey comprise the resources used when economic agents attempt to identify and contact
one another (identi cation costs), negotiate contracts (negotiation costs), and verify and enforce the terms
of contracts (enforcement costs). ese transaction costs, together with the costs of enforcing property
rights, may well be lower than the costs of regulation and taxing or subsidising goods and services. In the
Coasian approach, the government’s role in respect of externalities mainly consists of the maintenance of a
judicial system to de ne and enforce property rights and a market system to lower transaction costs.
e Coase theorem is crucially predicated on the twin assumptions of well-de ned and enforceable
property rights and zero or negligible transaction costs. As suggested above, the quality of the judicial system
largely determines the de nition and enforcement of property rights. If transaction costs are non-trivial,
however, they may prevent parties from exchanging their property rights and internalising the externality in
the process. In general, the higher the transaction costs, the lower the degree of internalisation of
externalities will be and the greater the resultant divergence between marginal private cost and marginal
social cost will be. Transaction costs are bound to be high when large numbers of people are involved, such
as the many persons experiencing respiratory illnesses from inhaling the polluted air of industrial areas. It
may be too costly for them to take the necessary steps to have their right to clean air legally enforced.

3.8 Global public and merit goods

Neighbouring countries often agree to take joint responsibility for and share the burden of providing certain
cross-border public, mixed and merit goods. ese are called global or regional public and merit goods.
Agreements of this nature can apply to global or regional defence systems and cross-border road and rail
networks, or to measures aimed at addressing externality problems and other forms of market failure.
 One country’s national defence system may confer bene ts on neighbouring countries free of charge,
unless they enter into an agreement that forces each country to contribute pro rata to the cost of the service.
In fact there may be an important economies-of-scale argument here: providing an effective defence system
or a subsidised electricity network at minimum unit cost may require high levels of production capable of
serving several countries at the same time. It is also pointless to construct a road or railway line that stops at
the border separating two countries that trade with each other. Countries making up the European Union
(EU) all derive huge bene ts from being connected by road and railway networks; likewise, the road linking
Gauteng Province in South Africa with Maputo in Mozambique clearly bene ts holiday-makers and
transport companies as well as importers and exporters in both countries.
In all of these cases, it is important to know who the bene ciaries are, how they are spread across
different countries and how payment – in the form of taxes and user charges – should be divided among the
bene ciaries. is point can be illustrated by referring back to Figure 3.2 and replacing the two individuals
(Bongani and Joan) with two countries (for example, South Africa and Mozambique). e optimal price-
discriminating solution again occurs when each country pays a price equal to its marginal value of the
service, that is, PJ and PB, where the subscripts now refer to the two countries. We already know that a pricing
strategy of this nature is not a feasible solution, partly owing to a lack of knowledge about preferences and
the attendant problem of free-riding. e only alternative is a ‘tax price’ paid by each country, presumably
based on some pre-determined formula. is clearly calls for a formal bilateral agreement or a multilateral
agreement if more than two countries are involved. An agreement of this kind would normally form part of a
more broadly based regional trade agreement, such as those associated with the Southern African Customs
Union (SACU), the Southern African Development Community (SADC), the Common Market for Eastern
and Southern Africa (COMESA) and the Economic Community of West African States (ECOWAS).
Similar arguments apply to externalities: air and water pollution may impose negative external effects on
producers and consumers in neighbouring countries. Ideally, these producers and consumers should have a
say in resolving the problem, including the task of de ning property rights and determining compensation
in terms of the Coase theorem. A truly global example is the emission of carbon dioxide into the atmosphere,
which happens virtually everywhere in the world where automobiles are used and fossil fuels are burnt. Box
3.4 shows that these emissions can give rise to dramatic climate changes in different parts of the world, with
adverse effects for producers and consumers that may put the well-being of future generations at risk.
Solving this problem evidently calls for a binding international agreement that aims to reduce CO2 emissions
on a global scale. e Kyoto Protocol and subsequent protocols were steps in that direction.

Box 3.4 Global warming

Much of our natural energy comes from the sun in the form of shortwave radiation that warms the earth and its
inhabitants. When some of this energy is again released into the atmosphere in the form of long-wave radiation,
the result is a delicate temperature range that is life-giving and makes the earth inhabitable. The incoming
short waves are capable of penetrating man-made greenhouse gases such as carbon dioxide, ozone, methane
and other industrial gases, but the outgoing long waves are not, and are absorbed by the same greenhouse
gases. If the emission of these gases reaches and surpasses a critical level, as many commentators are
claiming, the result is an enhanced greenhouse effect that raises average global temperatures, changes
precipitation levels and may give rise to extreme weather patterns (Stowell, 2005). The earth’s atmosphere can
be considered a free resource and a global public good in the sense that it is largely non-excludable, thus
benefiting either all or most inhabitants on earth or, in the case of a greenhouse effect, harming many innocent
individuals, countries and regions. Excessive carbon dioxide emissions in one region may have negative
spillover effects in many other regions (without any compensation changing hands). Conversely, steps taken to
reduce emissions in one country may benefit many others. These are examples of global externalities that call
for appropriate forms of intervention by a supranational or global institution, as discussed in the accompanying
text.

Yet another example is the so-called brain drain. When, for instance, a Namibian citizen, schooled and
trained in that country, accepts a job offer in Germany, the latter country may bene t greatly in that it
obtains human capital free of charge, but paid for by Namibian taxpayers. Such positive spillover effects
happen on a large scale in many parts of the world and can be viewed as a manifestation of an individual’s
right to choose his or her place of work. Over time, the net effect may be small for various reasons: Namibia
may also bene t from human capital created in Germany, the brain drain may be reversed when the
Namibian citizen returns home as a more experienced and productive worker or the brain drain can
contribute to the development of global networks that encourage international trade and capital ows
between the countries concerned.
A nal comment refers to the earlier discussion of market failures other than ‘missing markets’. Within a
sovereign country, an appropriate competition policy authority can prevent or minimise abuse of market
power by monopoly rms. Similarly, the problem of ignorance about the harmful properties of certain goods
and services can be addressed by means of a bureau of standards through which minimum standards can be
legally enforced. Likewise, a national government can apply tax and other measures to internalise positive or
negative externalities within its geographical borders. In the case of regional and global public goods,
however, there is no single government with jurisdiction over the entire set of activities. International
cooperation in the form of agreements, treaties and different forms of integration then becomes a
requirement for addressing the ‘problem’ of public goods. One possibility is a mutual agreement whereby
the public institutions of one country are also allowed to exercise their authority in neighbouring countries.
Arguably, this will ensure that they operate more efficiently (in other words, at lower unit cost). Due
cognisance should be taken of relevant differences between the countries, however, because one size does
not necessarily t all.
Alternatively, new regional or global institutions may be formed. For example, exogenous shocks at the
macroeconomic level usually have a similar negative contagion effect on economically integrated countries,
such as those of the SADC, and may call for a regionally coordinated approach to the conduct of
macroeconomic policy. In the case of the European Union, the extent of economic integration has actually
reached the advanced state of using a single monetary system.
Going beyond the regional dimension, many countries stand to bene t from a world that is becoming
more integrated by the day in terms of economics, culture and geopolitics. At the same time, however, these
countries are also becoming more vulnerable to a range of negative spillover effects emanating from
exogenous nancial shocks, technology-driven climatic changes, ethnically driven regional con icts,
international terrorism and a host of contagious diseases. In the absence of a world government, it is
therefore imperative that global institutions such as the International Monetary Fund (IMF), the World Bank
(WB), the World Trade Organization (WTO), the World Health Organization (WHO), the Food and
Agricultural Organization (FAO) and the United Nations (UN) take the necessary steps to eliminate or at
least minimise the incidence and extent of negative spillovers of this nature.
One of the major obstacles has been the reluctance of industrialised and high-polluting countries to
accept the need for a Coasian-type redistribution of resources (in other words, limiting their own levels of
carbon emissions while compensating poor and low-polluting countries at the same time). However, at a UN
conference involving 193 countries in Cancun in 2010, industrialised countries accepted the proposal to
reduce greenhouse gas emissions by between 25 and 40% of 1990 levels within the next ten years, while also
agreeing to create a Green Climate Fund aimed at helping developing countries to cope with the effects of
global warming.
e relevant greenhouses gases (GHGs) are carbon dioxide, methane, nitrous oxide, sulphur
hexa uoride and two groups of gases (hydro uorocarbons and per uorocarbons) produced by them, all of
which are measurable by known techniques. Furthermore, both the UN conference and the (latest) Kyoto
Protocol actively support the use of clean and green forms of renewable energy that can easily be replaced
and repeated. is speci cally refers to the use of solar energy, wind energy, biomass energy and geothermal
energy. Finally, it is also worth noting that although nuclear power has a waste disposal problem and is
vulnerable to earthquakes, it does not pollute the air or water, while natural gas causes less pollution than
coal- red electricity plants.

Key concepts
• cap-and-trade programme (page 56)
• club goods (page 47)
• Coase theorem (page 57)
• command-and-control regulation (page 55)
• common pool resources (page 47)
• direct regulation (page 55)
• emission fee (or congestion tax) (page 53)
• excludability (page 39)
• externalities (page 48)
• free-riding (free rider) (page 44)
• global or regional public and merit goods (page 58)
• merit goods (page 48)
• mixed goods (page 46)
• negative externality (page 49)
• non-excludable (page 41)
• non-rivalry in consumption (page 41)
• pecuniary externality (page 48)
• Pigouvian subsidy (page 53)
• Pigouvian tax (page 53)
• positive externality (page 48)
• private goods (page 39)
• property rights (page 57)
• pure public goods (page 41)
• regulatory measures (page 55)
• rivalry in consumption (page 39)
• technological externality (page 48)
• transaction costs (page 57)

SUMMARY
• is chapter discusses two important sources of market failure, namely public goods and externalities.
Both re ect the incompleteness of markets. On their own, free markets cannot meet the demand for
public goods or fully account for the external costs and bene ts associated with individual actions.
Hence, these market failures provide a rationale for complementary government actions aimed at
improving the allocation of resources.
• In contrast to private goods, pure public goods such as street lighting and national defence are indivisible
– that is, they cannot be divided into saleable units – and are therefore non-rival in consumption. For a
given level of production of a public good, one person’s consumption does not reduce the quantity
available for consumption by another. is implies that the marginal cost of the good is zero.
• Public goods are also non-excludable, that is, it is impossible to exclude individuals from consuming
these goods. In contrast to private goods, this implies that the market demand for public goods is derived
by adding the individual demand schedules vertically (as opposed to horizontally). is effectively
means adding the individuals’ marginal utilities or the prices they are willing to pay for different
quantities of the good, not the quantities they demand at different prices.
• Given differences in individuals’ demand for a public good, government would ideally like to charge
each consumer a price that is equivalent to his or her marginal valuation of the bene ts of the good. is
would enable the government to recover the full cost of providing the service. Hence, the optimal
provision of a public good requires the application of price discrimination, that is, the practice of
charging different consumers different prices.
• In practice, however, the government does not have the required knowledge about people’s preferences
to enable it to apply perfect price discrimination. is is the reason why governments typically recover
the costs of supplying public goods by collecting a so-called tax price from consumers.
• Mixed goods have characteristics of private goods and public goods. Examples include toll roads,
subscription television channels and healthcare services. Many mixed goods are provided by the private
sector and the public sector. Universities, for example, get their income partly from government and
partly from students in the form of registration fees. is split can be viewed as an attempt to share the
costs of university education in accordance with the public and private good aspects of the service.
• e external costs and bene ts associated with individual actions are also examples of incomplete
markets. Such externalities can be positive or negative. ey are positive when the actions of an
individual producer or consumer confer a bene t on another party free of charge, and negative when
those actions impose a cost on the other party for which he or she is not compensated.
• ese external impacts can be of a technological or a pecuniary nature. ey are technological when they
have a direct effect on the production or consumption levels of the other party, and pecuniary when they
change the demand and supply conditions, and hence the market prices, facing the other party. In either
case, however, the bene ciary gets a windfall by not having to pay for the bene t, while the prejudiced
party gets no compensation at all.
• External effects drive a wedge between the private (or monetary) costs and bene ts and the social costs
and bene ts associated with everyday market transactions. Marginal social cost (MSC) is simply the sum
of the corresponding marginal private cost (MPC) and the marginal external cost (MEC). In the same
way, marginal social bene t (MSB) is the sum of the corresponding marginal private bene t (MPB) and
the marginal external bene t (MEB).
• In the case of a coal-driven electricity plant polluting the air, the MSC will exceed the MPC of supplying
that electricity. is implies an oversupply and underpricing of electricity. Similarly, the MSB of
education will exceed the MPB when students share their knowledge of public economics free of charge
with ‘third’ parties. e result will be an undersupply and underpricing of education.
• One possible solution to externality problems is the use of so-called Pigouvian taxes and subsidies, by
means of which externalities can be internalised by forcing parties to include the external effects of their
actions in their cost and bene t calculations. Yet to do this effectively, governments would need to know
the size of the externality to begin with. Fiscal intervention of this nature may also have unintended
effects. Alcohol taxes discriminate against moderate drinkers, for example, and tobacco taxes give rise to
higher levels of smuggling.
• Apart from using direct measures such as licences, penalties, and age and spatial restrictions to address
pollution problems, governments could also regulate markets by adopting cap-and-trade policies
according to which it would issue legal permits that give private owners the right to pollute. ese
permits would be sold if the price exceeded the marginal cost of reducing pollution. If new technologies
make the latter cheaper, more permits will be sold at lower prices until there are few or no takers left. As
before, this option requires that government knows the optimal size of pollution to begin with.
• Another way to reduce greenhouse emissions is to support the development of alternative energies that
are pollution-free, including nuclear power, hydro-powered energy, solar heating and wind turbines.
• e externality problem can be viewed as the result of insufficiently de ned property rights over the use
of resources. It can be argued that affected parties can simply exchange property rights in order to
internalise an externality, provided that property rights are well-de ned and enforceable, and
transaction costs are negligible.
• Both public goods (for example, national defence) and externalities (for example, pollution) have
spillover effects on neighbouring countries and beyond. Global warming caused by carbon emissions is a
prime example. In the absence of a world government, global effects of this nature clearly require
international cooperation on a bilateral and a multilateral basis.

MULTIPLE-CHOICE QUESTIONS
3.1 Which of the following are characteristics of pure public goods?
a. Consumption is non-rival.
b. Consumption is non-excludable.
 c. e aggregate demand curve is obtained by horizontal addition of the individual demand curves.
d. All of the above options are correct.
3.2 Which of the following factors explain why private markets fail to supply pure public goods?
a. It is impossible to determine an equilibrium price for providing pure public goods via private
markets.
b. Private rms lack the technological know-how to provide pure public goods.
c. It is impossible to exclude those who are unwilling to pay for using pure public goods.
d. All of the above options are correct.
3.3 When you share your knowledge of public economics with a friend, you would be conferring a positive
externality if …
a. the marginal social bene t (MSB) exceeds your own marginal private bene t (MPB).
b. the marginal social bene t (MSB) equals the marginal social cost (MSC).
c. the marginal external bene t is positive.
d. All of the above options are correct.
3.4 When a pecuniary externality causes a decrease in the price of an asset being traded, …
a. the seller is made worse off.
b. the buyer secures a windfall gain.
c. there is little or no net effect on the well-being of society.
d. All of the above options are correct.

SHORT-ANSWER QUESTIONS
3.1 Explain the meaning of the terms ‘mixed goods’ and ‘merit goods’. Who should supply these goods?
3.2 Explain the concept of externality and distinguish between the different types of externality.
3.3 Critically discuss the cap-and-trade approach to addressing the externality problem.
3.4 Discuss the phenomenon of global or regional public goods.

ESSAY QUESTIONS
3.1 Explain the difference between a private and a pure public good. How do their respective equilibria
differ?
3.2 Explain why markets cannot supply pure public goods, and outline the implications of this reality for
the production and nancing of such goods.
3.3 Should Pigouvian taxes be used to internalise the negative external effects of tobacco and alcohol
consumption?
3.4 Discuss Coase’s theorem and consider its usefulness as a means of solving the externality problem.

1 It may be somewhat misleading to use the word ‘consumption’ in the context of pure public goods. Goods of this nature are not
really consumed; instead, they generate a stream of bene ts that can be enjoyed by all.
2 In his path-breaking analysis, Samuelson (1954 and 1955) de ned public goods in terms of non-rivalry in consumption only. Non-
rivalry is indeed a sufficient condition for public good status, because it makes exclusion inefficient even when it is feasible. Some
textbooks follow Samuelson by de ning public goods in terms of non-rivalry only, but this book follows the more conventional
approach of using non-rivalry as well as non-excludability as criteria for public good status.
3 ese bene ts may be subject to a form of decreasing returns. Most empirical estimates of the returns to education indicate that
the excess of the social bene t over the private bene t decreases as education becomes more advanced (Todaro & Smith, 2015:
388–391).
4 Internalisation of externalities could also be viewed as a means of conferring property rights on the individuals involved in order to
account for the divergence between private and social costs and bene ts. See Section 3.7.5.
5 For a discussion of congestion charges in these cities, see Santos (2005).
6 For a recent discussion of the effectiveness of cap-and-trade programmes, see Schmalensee and Stavins (2017).
Allocative efficiency, imperfect competition and regulation

Philip Black and Estian Calitz

Chapter 2 identi es a number of market failures that establish a rationale for government intervention to
promote allocative efficiency, equity and macroeconomic stability. ese include imperfect competition,
which is called ‘non-competitive markets’ in Section 2.4.4. e most common manifestation of imperfect
competition is market domination by monopolies and oligopolies. is chapter outlines the economic
effects of monopolies and discusses policy interventions to rectify one manifestation of this phenomenon,
namely electricity supply.
It is customary to distinguish between statutory monopolies and natural monopolies. A statutory
monopoly (also known as an ‘arti cial monopoly’) operates in a market where competition is technically
feasible, but prevented by legal restrictions imposed by the government or a professional body or by entry-
limiting behaviour on the part of the incumbent rm itself (for example, exerting control over critical
suppliers or temporarily setting price below its pro t-maximising level). By contrast, a natural monopoly
operates in a market where technical factors prevent competition. While this chapter focuses on natural
monopolies, the discussion of general effects of imperfect competition in Section 4.1 also refers to statutory
monopolies.
e chapter consists of ve sections. Section 4.1 highlights the social costs of imperfect competition by
contrasting outcomes under perfectly competitive and monopoly conditions. e two parts of Section 4.2
discuss natural monopolies, which are also known as ‘decreasing cost industries’. Section 4.2.1 outlines the
nature and economic effects of this type of imperfect competition, while Section 4.2.2 outlines traditional
policy approaches towards decreasing-cost industries and the more recent approach known as the ‘new
model’. e last three sections contain more information about the three elements of the new model, namely
competitive restructuring (Section 4.3), privatisation (Section 4.4) and regulation (Section 4.5).

Once you have studied this chapter, you should be able to:
discuss the social costs of statutory monopolies
outline the characteristic features of natural monopolies
outline and contrast traditional policy approaches towards natural monopolies and the more recent new
model
explain competitive restructuring or unbundling of vertically integrated state monopolies with appropriate
examples
discuss the economic rationale for and effects of the privatisation of state monopolies in developing
countries
discuss the characteristics and effects of the best-known forms of economic regulation
discuss the most important obstacles to effective economic regulation in developing countries.
4.1 The social costs of statutory monopolies 1

Figure 4.1 illustrates the familiar distinction between perfect competition and monopoly. Here we assume
that the demand function, D, and marginal cost, MC, are the same for the two market forms. ere is one
difference, though: under perfect competition, MC represents the sum of the marginal cost curves of the
individual rms making up the market, whereas under monopoly, it represents the marginal cost of the
monopolist only.
e perfectly competitive equilibrium occurs at point E in Figure 4.1 where supply equals demand and
0Qc of the good is produced at a price of 0Pc . Under monopoly, equilibrium occurs at point F where MC =
MR. e market here produces a smaller quantity, 0Qm , at a higher price, 0Pm , than it does under perfect
competition.
e loss in consumer surplus under monopoly is the area given by PmGEPc, part of which – PmGHPc – is a
straight transfer from consumers to the producer, with the remaining triangle, GEH, being the net welfare (or
‘deadweight’) loss. Although the usual assumption is that the value represented by the rectangle labelled
HEQcQm is transferred to other sectors in the economy, this is evidently easier said than done. e resources
contributing to this value, for example, labour and capital equipment, may remain unemployed for long
periods in a real-world economy with frictions and time lags, at least until they nd alternative employment.
e value thus lost represents yet another social cost of monopoly.

Figure 4.1 Monopoly versus perfect competition

e two-sector model discussed in Chapter 2 can also illustrate the difference highlighted in Figure 4.1.
e implications of monopoly for this model are straightforward. Recall from Equation 2.5 in Chapter 2 that
the marginal rate of product transformation (MRPT) equals the marginal cost and price ratios for the two
commodities, that is:

Assume Y is a monopolist and X a perfectly competitive industry. is implies that Py > MCy, while Px = MCx .
It follows that:
e above equations indicate that the second, and by inference, the third or ‘top-level’ condition for a Pareto
optimum has been violated. is is illustrated at point M0 in Figure 4.2 by the difference between the slope of
the production possibility curve, R0T0 (indicated by tt), and the slope of the commodity price line, PMPM. e
effect of introducing a monopoly here is to lower the output of good Y and raise its relative price, as is
evident from a comparison of the monopoly equilibrium at point M0 and the competitive equilibrium at
point C. e difference between the equilibria at points M0 and C in Figure 4.2 re ects the degree of
allocative inefficiency arising from the presence of a monopoly in one of the two sectors. In other words,
the economy as a whole produces too little of good Y relative to good X at point M0. Hence, a reallocation of
resources to increase the production of good Y relative to that of good X and to move the economy to point C
would be welfare-enhancing.

Figure 4.2 Efficiency implications of monopoly

Moreover, a monopoly may have an additional cost resulting from X-inefficiency.2 It may be the case that
monopolists do not utilise their existing resources as efficiently as rms operating under the constant
pressure of competitive markets. With no threat of entry, the monopolistic rm may lack the incentive to
maintain a high level of labour productivity or to spend sufficient time and effort searching for and acquiring
the necessary information. Such a situation would imply an equilibrium lying inside the production
possibility curve, for example at M1 in Figure 4.2. Consequently, the social costs of a monopoly may result
from both allocative inefficiency and X-inefficiency.
 On the other hand, some economists argue that monopolistic rms are in a better position to achieve
technological advancement than their perfectly competitive counterparts. e typical monopolist may
have an incentive and the means to initiate or imitate cost-saving technical inventions and innovations. If
the monopolistic rm is to satisfy its shareholders, it will have to make a healthy pro t that can then be used
for research and development purposes to improve productivity within the rm. In Figure 4.2, the effect of
technological progress will be to shift the production possibility curve from R0T0 to R1T1 and the monopoly
equilibrium from M0 to M2. In this case, technological progress has enabled the economy to move beyond its
original production possibility frontier.
is brings us to the important issue of deregulation, that is, attempts on the part of the authorities to
promote competition by removing certain barriers to entry. Such barriers can take various forms, including
prohibitive licensing fees, property taxes, restrictive labour laws and inappropriate health standards. e
analysis in this section suggests that the removal of barriers to entry will improve allocative efficiency and X-
efficiency in the relevant industry. is will move the equilibrium closer to point C in Figure 4.2 (for
example, to M3). However, it may also entail a cost in terms of the decreased pro ts made by competitive
rms and their resultant inability to initiate and carry out technical inventions and innovations. e
economic case for deregulation will therefore depend on whether the gains in allocative and X-efficiency are
sufficient to offset the slower pace of technological advancement among competitive rms.

4.2 Natural monopolies

4.2.1 Characteristics and effects

An industry is a natural monopoly if it is characterised by large capital outlays that give rise to economies of
scale over the entire range of its output. e minimum average cost of production may thus occur at a level
of output sufficient to supply the whole market, which means that only one rm can operate effectively in
such a market. e best-known examples of this type of industry are public utilities involved in the provision
of electricity, water, rail and road transport, and postal services.
Increasing returns to scale means that the long-term average cost (AC) of the rm decreases as output
increases. Its marginal cost (MC) curve will therefore lie below the AC curve over the entire output range.
is is illustrated by the curves labelled AC and MC in Figure 4.3. In a perfectly competitive market, each
rm will set marginal cost equal to the market price (for example, at point E in Figure 4.3 where Pe = MC).
With increasing returns to scale, however, the industry will make a unit loss equal to ES so that individual
rms will eventually close down until a natural monopoly emerges.
If the natural monopoly is not controlled by the government, it will maximise pro t at point M where its
marginal cost equals marginal revenue. At point M, the equilibrium price, 0Pm, exceeds the socially efficient
price, 0Pe, while the corresponding level of output, 0Qm, is smaller than the Pareto-optimal level, 0Qe. e
pro t-maximising behaviour of the monopolist may therefore results in too little output being produced at
too high a price, which would give rise to a concomitant loss of welfare. is welfare loss is the difference in
consumer surplus between the two equilibria. Under monopoly, consumer surplus equals the area AFPm,
which is evidently much smaller than the consumer surplus under the hypothetical competitive solution,
given by the area AEPe in Figure 4.3. Note that the existence of a natural monopoly has nothing to do with
whether the activity is undertaken by the government or the private sector. As will become apparent in
Section 4.4.2, the ownership issue has to do with the way in which the allocative and equity implications of a
natural monopoly are managed, but does not take the monopoly element away.
Figure 4.3 Decreasing cost industry

4.2.2 An overview of policy options

e question that arises is whether something should be done about the above-mentioned loss of welfare.
ere is no simple answer to this question, but the government does have several options at its disposal.
ese all depend on whether the natural monopoly provides goods or services that are important inputs for
many other sectors and industries in the economy (electricity and water supply are obvious examples).
Externalities are involved if this is the case. e terminology explained in Section 3.6 of Chapter 3 implies
that D and MC in Figure 4.3 represent marginal social bene ts (MSB) and marginal social costs (MSC),
respectively. Production at point M then clearly implies that MSB > MSC. e only way to create and confer
pecuniary externalities on other industries is then to expand production and lower the price, for example, to
0Qe and 0Pe . e lower price therefore signi es the internalisation of the social bene t.
One option is for government to establish or take ownership of the natural monopoly itself, as has
happened in many countries during the previous century. It could then apply marginal cost pricing at point
E in Figure 4.3 and cover the resultant loss by means of a unit subsidy equal to ES. But the required subsidy
would have to be paid for by government, and hence by the tax-paying public. If the government introduced
(say) a new or higher indirect tax for this purpose, it will drive a wedge between marginal cost and marginal
revenue elsewhere in the economy. Such a tax wedge will cause a loss in welfare in the form of an excess
burden. Alternatively, the government can borrow money to pay for the subsidy. is, however, could put
pressure on interest rates and crowd out private spending in the rest of the economy. Recall that point M0 in
Figure 4.2 represented the equilibrium of a private natural monopoly. e equilibrium of a nationalised
natural monopoly would lie closer to point C (the perfectly competitive equilibrium). However, it will be
inside the PPC, for example, at a point such as Ms, because of the distorting effect of the required tax and a
possible crowding out effect. us, while nationalising a natural monopoly may have an allocative advantage
vis-à-vis the private option, it also has a distorting effect on the rest of the economy.
Government-owned monopolies are also less X-efficient than private monopolies and have less of an
incentive to initiate and implement cost-saving innovations than their pro t-driven private counterparts.
After all, a privatised monopoly has little choice but to remain pro table and satisfy its shareholders if it is to
avoid the threat of a possible takeover. e nationalised monopoly, by contrast, is under no such threat and
can always rely on the seemingly endless resources of the state.
Nationalised natural monopolies failed in the vast majority of developing countries.3 It proved difficult to
implement the solution outlined above (a combination of marginal-cost pricing and unit subsidies) in
environments with much scope for government failure and weak incentives for X-efficiency. To boost their
popularity, many governments created unproductive jobs in state-owned monopolies and forced these
organisations to charge prices that did not cover their marginal costs. Compensatory subsidies drained
national budgets, yet in most cases were inadequate to cover the resulting unit losses. Widespread non-
payment by customers further compounded the nancial problems of most government-owned
monopolies. e resulting revenue shortages prevented monopolies from investing enough to maintain
facilities and equipment, let alone expand service provision to growing populations. X-inefficiency took the
two forms of low labour productivity and poor service quality. is re ected the reality that many managers
were political appointees who lacked skills to resolve such problems; others exploited information
asymmetries to maximise their budgets as predicted by models of bureaucratic failure (see Section 6.7.2 of
Chapter 6). Box 4.1 shows that many of these problems manifested recently in the South African electricity
supply rm Eskom.

Box 4.1. The Eskom experience4

Eskom is the dominant supplier of electricity in South Africa: in 2008, it generated 96% of South Africa’s
electricity, with the remainder coming from municipalities (1%) and independent power producers and other
entities (3%) (Department of Minerals and Energy, 2008: 8). During the 1970s and 1980s, it established itself
as a prominent name in the South African capital market when its bonds were priced better than those of the
government. This reflected Eskom’s sound finances and effective bond management strategy at the time. Yet in
the past two decades Eskom has experienced a number of problems. These included (in no particular order):
• Load shedding
The first two decades of this century brought periodic load shedding,5 which has had major implications for
economic activity. Various factors caused these interruptions in the supply of electricity, such as insufficient
production capacity, poor maintenance of existing production plants (including the nuclear power station at
Koeberg in the Western Cape), and periodic shortages of coal.
• Reversal of tariff trends
Between 2007 and 2016 average electricity tariffs increased by 374% in nominal terms, that is, by 16,8%
per annum. The corresponding real increases were 186% and 10,4%, respectively. This reversed the trend
from 1997 to 2007, when average electricity tariffs decreased by 21,6% in nominal terms (that is, by 62,2%
in real terms). The average annual decreases in this period were 2,4% in nominal terms and 9,4% in real
terms. The timing and size of Eskom’s investments in electricity supply capacity as well as changes to the
tariff-setting methodology caused this large change in tariff adjustments.
• Municipal arrears
Municipalities are wholesale suppliers of Eskom electricity. The municipal levy on household consumption of
energy, which is added to the Eskom levy, is a significant source of revenue for local governments. Eskom’s
debt problems have been worsened by the huge outstanding bills of cash-strapped municipalities. Eskom’s
rising municipal arrears reflect a form of moral hazard by municipalities that operate as if their debt is
explicitly or implicitly guaranteed by the national government or provincial governments.
• Corporate governance issues: irregular expenditure related to circumventing tender procedures
Business Day (2017) reported that Eskom headed the list of government departments and agencies that
sought permission to deviate from normal government procurement procedures in 2016–2017.6 In fact,
Eskom’s requests accounted for R31,3 billion of the R37 billion worth of deviations requested in 2016–
2017.7
• Nuclear energy
Some years ago, the South African Department of Energy entered into negotiations with Russia to
commission additional nuclear energy plants. Technical and financial experts soon questioned the cost
efficiency of such plants, especially when compared to alternative sources such as wind and solar energy.
The process eventually came to a halt when the Western Cape High Court declared a secret 2014 co-
 operation agreement between the South African government and the Russian state nuclear energy
corporation Rosatom unlawful. In February 2018 (then) Deputy President Ramaphosa said that South Africa
did not have the money to build nuclear power stations (Quartz Africa, 2018).
• Downgrade of Eskom’s credit rating
During the 1990s decisions to expand Eskom’s generation capacity were delayed. This was in part a
response to the over-expansion of capacity during the 1970s and 1980s, when unrealistically high economic
growth projections informed future planning. But it also reflected that the government’s attempts to reduce
the size of the public sector focused on reducing public investment, rather than government consumption.
The outcome was considerable strain on Eskom’s existing capacity as it struggled to ‘keep the lights on’ even
if it meant liquidity problems that caused it to struggle to service debts as well as postponement of much-
needed maintenance of plants. These concerns contributed to the downgrading of Eskom’s credit rating by
Standard and Poor and Moody’s in 2016. Moody’s (quoted in Eskom, 2018) justified its downgrade of Eskom
in January 2018 as follows: ‘Despite a number of improvements at the company in relation to its corporate
governance and liquidity, there is limited visibility at this juncture as to Eskom’s plans for placing its longer
term business and financial position on a sustainable footing.’
• Resistance by Eskom to modern methods of managing peaks and troughs and allowing access to the network
from other suppliers
Eskom’s control of access to the transmission network has enabled it to operate as a monopoly in the
production of electricity as well. Such control of market entry and exit has shielded Eskom from competitive
forces that would have yielded allocative efficiency gains in activities where natural monopoly conditions do
not apply. Efficiency is better promoted if electricity consumers are allowed to sell unused electricity back into
the network during low consumption periods for use during peak hours, for example, and if producers are
allowed to sell wind and solar energy into the network as a structural increase in the supply of electricity.
• Adopting a new process of price regulation
The period of low and declining real electricity tariffs ended around 2006/07. The Electricity Regulation Act of
2006 made the National Energy Regulator (NERSA) responsible for tariff determination, and the South African
Government adopted the Electricity Pricing Policy (EPP) in 2008. The EPP stated that electricity tariffs should
ensure the financial sustainability of the electricity supply industry and added that ‘… the industry has the
potential to generate strong cash flows to sustain a financially viable industry and the need for direct state
support and subsidies should, apart from funding social objectives, be minimal’ (Department of Minerals and
Energy, 2008: 15). This means that electricity tariffs should be set at levels that allow Eskom to recover the
costs of supplying electricity and to achieve a fair rate of return on its asset base. This emphasis on cost-
reflective pricing has advantages, but also carry risks (see Box 4.2).
• Corporate governance
In a significant and unprecedented step on 19 January 2018, more than 200 senior Eskom managers
submitted a letter to then Deputy President Cyril Ramaphosa that expressed concern over the lack of
‘decisive and bold actions against allegations of fraud, corruption and maladministration’ (Fin24, 2018). The
signatories requested Ramaphosa to ensure that a credible board is appointed with an ‘exemplary track
record in large complex organisations’ to rebuild trust in the parastatal. The memorandum stated that
lenders, investors and ratings agencies required this. The signatories also wanted the Eskom Executive
Management team to be reconstituted with a new group CEO and CFO who will be ‘well received by investors
and citizens of this country’. Soon after this the Government replaced Eskom’s Board (Sunday Times, 2018).
Comprehensive financing programme
A major outstanding matter is that Eskom has not released a comprehensive long-term plan that integrates
operational and investment expenditure with a comprehensive financing plan. The latter should include all
financial sources: electricity tariffs (indicating cross-subsidisation or financing from government), loans (from
government and bond markets) and equity (from government, if the only shareholder, and/or from private
investors, if applicable). The quoted justification for Moody’s most recent downgrade of Eskom confirms the
importance of dealing with this matter.
The serious need for reforms were acknowledged when the Finance Minister confirmed the State President’s
earlier announcement in his 2019 State of the Nation Address that Eskom will be subdivided into three
independent subdivisions, namely for the generation, transmission and distribution of electricity. This is bound
to set the electricity market ‘on a new trajectory, and allow for more competition, transparency and a focused
funding model’ (Minister of Finance, 2019: 9). The Government was not going to take over Eskom debt, but
 budgetary funds to the tune of R23 billion per annum in support of the reconfiguration of Eskom were
announced. This was followed up by further conditional financial support.

It is partly for these reasons that the traditional model of using state-owned monopolies to provide services
in decreasing-cost industries has fallen out of favour in various industrial and developing countries. Many
countries have switched, either partially or fully, to the so-called new model for the governance of such
industries. is model has three elements: (1) restructuring (or unbundling) of state-owned monopolies, (2)
privatisation of the components of these entities in which competition is possible, and (3) regulation of the
privatised components. e three purposes of such reforms in decreasing-cost industries have been to
improve allocative efficiency and X-efficiency, improve service delivery, and reduce the nancial burdens
these entities impose on governments and taxpayers. e next three sections of this chapter provide more
detailed discussions of the three elements of the new model and their effects. While con rming that this
model has had positive effects in many countries, these sections will also identify some risks as well as
requirements for success.

4.3 Competitive restructuring 8

Competitive restructuring, which is also known as ‘unbundling’, is at the heart of the new model for the
governance of decreasing-cost industries (natural monopolies). e state-owned monopoly model
assumes that all activities in decreasing-cost industries are affected by decreasing average costs, which
preclude pro table operation by more than one rm. e belief that there is no scope for competition led to
the establishment of state-owned monopolies to handle all activities in decreasing-cost industries.
More recently, economists and policymakers realised that most of the activities in decreasing-cost
industries are not natural monopolies. e electricity supply industry, for example, has four elements:
generation (production of electricity in power plants), transmission (long-distance transportation of
electricity at high voltage), distribution (transportation of electricity to end-users over shorter distances and
at lower voltages), and supply (selling of electricity to end-users). Only transmission and distribution are
true natural monopolies, because construction of the networks (‘grids’) that transport electricity requires
huge capital investments that give rise to decreasing average costs over large output ranges. Competition is
feasible in the other two elements of the industry: power plants can compete in generation, for example, and
retailers in supply. Similar distinctions exist in other decreasing-cost industries. e so-called local loop (the
networks that carry analog and digital signals from service providers to customers) is the natural monopoly
element in the telecommunications industry, while competition is possible in long-distance telephony,
mobile telephony, and various value-added services (live streaming, mobile money services, mobile
advertising, et cetera). Similarly, the tracks, signals and other xed facilities are the natural monopoly
elements in the railways industry, while the operation and maintenance of trains are potentially competitive
ones.
As suggested by the term ‘competitive restructuring’, unbundling involves separation of the potentially
competitive and natural monopoly components of government-owned monopolies in decreasing-cost
industries and the introduction of competition in the former. Privatisation of the potentially competitive
components is a key aspect of such reform processes. Governments usually retain the natural monopoly
components of the entities, but can establish regulatory authorities to oversee pricing and the quality of
service provision.
Competitive restructuring is a step towards reaping the allocative and X-efficiency bene ts of
competitive markets (as was pointed out in Section 4.1, industries in which rms compete should produce
more and sell outputs at lower prices than monopolistic ones do, ceteris paribus). ese bene ts should not
be assumed, however, especially in smaller developing countries in Southern Africa and elsewhere where
markets may be too small for effective competition. Hence, the long-run effects of unbundling depend
signi cantly on the degree of effective competition in the restructured industries and the government’s
ability to regulate the conduct of market participants. e next two sections discuss aspects of these issues.
4.4 Privatisation
Section 1.5.3 of Chapter 1 described privatisation as transferral of the production of goods and services
from the public sector to the private sector. It can take various forms. e most comprehensive form of
privatisation is when the government transfers all aspects of a particular activity – for example, the planning,
design, construction, nancing and maintenance of a road – to a private rm. No government involvement
whatsoever remains in such cases, apart perhaps from enforcement of certain regulatory standards. Partial
forms of privatisation, which involve sharing of roles, responsibilities and risks between public and private
sector entities, can take various forms. ese go under the general rubric of public-private partnerships (see
also Section 1.5.3 of Chapter 1).
In education, for example, teaching and the construction of buildings and other infrastructure may be
privatised, while curricula and standards remain government responsibilities. Medical services may be fully
privatised, while the government remains partially responsible for nancing to ensure that the most
vulnerable in society have access to proper healthcare. Some countries have used another quasi-
privatisation possibility known as the BOOT (build, own, operate and transfer) model for toll roads, among
other things. is model entails privatisation of construction, nancing and maintenance for a speci ed
period, after which ownership of the asset reverts back to government with maintenance sometimes allotted
by tender to a private company. Public utilities present another variation on the theme. As we have seen, in
the case of electricity, generation and supply can be privatised, while transmission and distribution (the real
natural monopoly elements) may remain government responsibilities. is section discusses the purposes
and effects of privatisation of former state-owned monopolies in decreasing-cost industries in more detail.
Financial considerations have often featured strongly in decisions to privatise state-owned monopolies:
many governments have been keen to rid themselves of loss-making entities and to earn revenue from the
transactions. However, the efficiency considerations discussed in Sections 4.1 and 4.2.2 have been at least as
important. Private rms, whether in competitive or non-competitive industries, have to remain pro table to
survive; hence, they are likely to be more X-efficient than government-owned monopolies subsidised from
the national budget. But a change from public to private ownership as such will not necessarily bring
allocative efficiency gains. Such bene ts will realise only when the privatised rm experiences competition
— the spur for innovation, effective use of resources, high-quality service delivery, et cetera. is suggests
that the economic effects of privatisation may well depend on contextual factors such as the size of the
market, the openness of the economy and the competition policy framework and its application.
Equity considerations invariably also come into play in privatisation decisions.9 Budget constraints often
force newly privatised rms to reduce bloated workforces and to raise previously subsidised prices to cover
their costs. ese steps raise difficult questions. e reality is that decreasing-cost industries expend
resources to provide services to consumers, and have to recover these costs to remain viable. e equity
question is whether these costs should be borne in full by the users of the services or subsidised (i.e. shared
with taxpayers). e issue becomes even more complex when consideration is given to factors such as
distinctions between richer and poorer users of services, the opportunity costs of subsidies and the
economic effects of the taxes raised to nance them. e costs of job losses in privatised entities should be
weighed against the cost-raising effects of retaining unproductive workers, which will be re ected in the
prices charged to other rms and to consumers (and possibly also the subsidies needed to keep entities
a oat). Be that as it may, its immediate costs have made privatisation an extremely controversial policy in
many countries, despite the likelihood of larger longer-run efficiency bene ts such as expanded and more
sustainable service provision and more efficient use of the funds used to subsidise loss-making entities.
South Africa has been no exception. A few privatisation transactions have occurred, including the listing
of Iscor on the Johannesburg Securities Exchange (JSE) in 1989 and the partial sale of Telkom to two
international companies in 1997. However, strong opposition rooted in equity considerations and the
preference for state-led economic development among socialists and adherents to developmental-state
thinking brought an end to such transactions. More recently, opportunists harnessed the opposition against
privatisation as a convenient veil to maintain opportunities in the public enterprises for rent-seeking, state
capture, nepotism and outright fraud.
Although hampered by various methodological problems, econometric and case studies of the effects of
privatisation have yielded several important ndings.10 ese studies have compared the pre- and post-
privatisation performance of speci c rms and industries or the performance of state- and privately owned
entities in speci c sectors of different countries. Various aspects of the performance of rms and industries
have been studied, including pro tability, operational efficiency, capital investment, employment, prices
and dividends.
Several of these sectoral and rm-level studies have suggested that privatisation often leads to increased
production and improvements in productive efficiency, prices and service delivery. As was suggested earlier,
however, the extent of the bene ts has depended markedly on the scope for competition after privatisation
of state-owned monopolies and on governments’ capacity to regulate the industries in question. Other
elements of the general business environment (such as the depth and liquidity of capital markets, the quality
of the legal system and competition policy, and the number of quali ed managers) have apparently also
affected the success of privatisation attempts. ese ndings suggest that investment in regulatory
capabilities and the establishment of sound business environments in which competition can ourish are
prerequisites for obtaining major efficiency bene ts from privatisation in Southern Africa and elsewhere.
e equity effects of privatisation have been mixed. Successful privatisation accompanied by effective
competition and regulation have often had positive equity effects, including improved access to services for
the poor, lower prices and, in the longer run, expanded employment. In other cases, combinations of
inequitable divestment methods (e.g. selling of state-owned entities to rich individuals at favourable prices),
sharp price increases and large-scale job losses have caused adverse distributional effects. e evidence has
suggested, however, that targeted subsidies can do much to lessen the effects of price increases on poor
consumers.

4.5 Regulation 11

Sections 4.3 and 4.4 referred to the important role of regulation in the new model for the governance of
decreasing-cost industries. e term ‘economic regulation’ refers to rules that are made and enforced by
government agencies to control entry and prices in particular sectors of the economy. Regulatory authorities
are independent agencies that operate in speci c sectors in accordance with enabling legislation. Two well-
known South African examples are ICASA (the Independent Communications Authority of South Africa)
and NERSA (the National Energy Regulator of South Africa), which regulate the telecommunications and
energy sectors, respectively.
In decreasing-cost industries, the aim of regulation is to complement competitive restructuring and
privatisation by preventing privatised rms in weakly competitive sectors from imposing costs of allocative
inefficiency on the rest of the economy. Put differently, the aim of regulation is to limit privatised natural
monopolies to reasonable rather than maximum (abnormal) pro ts, partly for efficiency reasons and partly
to protect the interests of consumers and producers using the relevant product or service. In terms of Figure
4.3, such a regulated equilibrium would lie somewhere between the two extremes of marginal cost pricing
and pro t-maximising monopoly pricing.
e remainder of this section outlines three models for regulating the prices of privatised natural
monopolies, namely rate of return regulation, price-cap regulation and sliding-scale regulation, and
comments on regulation in developing countries. Box 4.2, which summarises the process for determining
Eskom’s electricity tariffs, complements this subsection.

4.5.1 Rate-of-return regulation

e aim of rate-of-return regulation, which is also known as pro t regulation, is to provide a regulated
rm with a degree of certainty about pro t outcomes. Hence, this form of regulation derives a price for a
good or service from efficient levels of capital and operating costs and a satisfactory rate of return on the
rm’s asset base. We illustrate this approach with reference to Figure 4.4, which is an extension of Figure 4.3.
e additions are shown in bold print and thicker lines.
Figure 4.4 Pricing under rate-of-return regulation

Given a regulated rm’s supply and demand curves, the regulator determines the regulated rm’s capital
and operating costs as the area THQr0 and deems that a total pro t of PrGHT (i.e. a pro t per unit of GH)
represents a satisfactory rate of return on its asset base. Hence, the required revenue of the rm is the sum of
the areas PrGHT and THQr0, that is, area PrGQr0. is implies a price Pr at production level Qr. Note that the price
will vary depending on the rate of return allowed by the regulator; in principle, it can assume any value
between the monopoly price Pm and the perfectly competitive price Pe. e same applies to the
corresponding production level.
is type of regulation is intended to serve various ends, namely the nancial integrity of regulated rms,
their ability to attract new capital and adequate compensation for new investment. At the same time, the
rms are expected to provide goods and services on a non-discriminatory basis and charge just and
reasonable prices. In practice, however, regulators may well be wary of letting rms go bankrupt, and may
try to prevent this by setting the allowed rates of return too high. is may well incentivise production
inefficiency, because it encourages the adoption of capital intensive production technologies by effectively
subsidising capital. On the other hand, such technologies may contribute to dynamic efficiencies if they
embody innovations. Information asymmetries have also undermined many attempts to apply rate-of-
return regulation. Note that the sizes of rms’ asset bases directly in uence the satisfactory rates of return
allowed by regulators. is creates incentives for cost-padding, for example, unnecessary capital
expenditures that expand asset bases and enable rms to claim that they need higher pro ts to achieve the
allowed rates of return. Regulators seldom have enough information about rms’ operating and capital costs
to prevent such inefficient use of resources.

4.5.2 Price-cap regulation

As suggested by the name, price-cap regulation caps the prices charged by regulated rms. In the case of
the regulated rm depicted in Figure 4.3, for example, a regulator applying this type of regulation would
simply cap the allowed price at Pr. (Recall that a regulator who applies rate-of-return regulation would
derive the allowed price from the revenues the rm requires to cover its costs and achieve a satisfactory rate
of return on its asset base.) e major advantage of this type of regulation is that it creates incentives for
rms to cut costs and boost their pro ts. In the United Kingdom, where this method is used extensively,
regulators apply the following formula to implement price-cap regulation:
where P denotes the regulated price, RPI the retail price index and X the rate of productivity growth of the
production factors used by the rm. If the in ation rate is 5% per annum and the rate of productivity growth
3% per annum, for example, regulated rms would be allowed to increase their prices by no more than 2%
per annum. Note the built-in incentive for increased efficiency: a rm’s pro ts will increase automatically if
it achieves a higher rate of productivity growth than provided for in the price-cap formula. Price-cap
regulation therefore partly mimics conditions in perfectly competitive markets where rms are price-takers
and aim to maximise their pro ts.
ese attractive features of price-cap regulation should be weighed against two potential drawbacks. e
rst is that it can be as administratively burdensome as rate-of-return regulation, because regulators must
model the nances and economic environments of regulated rms in considerable detail. e information
needed to make forecasts of future trends in demand, productivity and input costs, among other things, is
hard to come by. A second potential problem is that it requires close monitoring of the operations of
regulated rms that lack strong competitors, because the cost-cutting encouraged by this form of regulation
may take the form of reductions in the quality of service provision. Many regulatory agencies do not have the
capacity to undertake such monitoring.

4.5.3 Sliding-scale regulation

Sliding-scale regulation combines elements of rate-of-return regulation and price-cap regulation. e basis
of this form of regulation is a price cap that creates the efficiency incentives outlined above. is price cap is
reduced as soon as the pro ts of the regulated rm rise above a pre-determined level. e effect of this
adjustment is to share the additional pro ts resulting from the efficiency gains of the rm between the
producer and consumers. e price-adjustment mechanism therefore dampens but does not eliminate the
efficiency incentives of the rm; at the same time, it prevents excessive pro t-making at the expense of
consumers. Unsurprisingly, however, the information requirements for successful sliding-slide regulation
are no less formidable than those of rate-of-return regulation and price-cap regulation.

4.5.4 Regulation in developing countries

e efficiency gains associated with effective regulation imply an outward shift in the PPC similar to that
shown in Figure 4.2, where the unregulated monopoly equilibrium occurs at point M2 or M3. e fact that
any form of regulation entails additional administrative costs means that the regulated equilibrium will
occur closer to point C but inside the (new) PPC, for example, at a point such as M3. Regulation can thus
strike a neat compromise between the need for economic growth (as re ected in the shift of the PPC) and
the need for a higher level of allocative efficiency than under the unregulated scenario.
A growing number of developing countries, including South Africa, have been introducing regulatory
mechanisms to reap such efficiency gains. is section has highlighted the potential as well as the
complexity of effective regulation of decreasing-cost industries. Box 4.2, which outlines the variant of rate-
of-return regulation used to determine Eskom’s electricity tariffs, provides more examples of the information
problems faced by regulators.12 Political economy factors often compound such problems. Apart from those
hinted at in Box 4.1 in Section 4.2.2, such factors include the phenomenon of regulatory capture: regulatory
agencies created to promote the public interest sometimes come to advance the commercial or political
interests of regulated rms instead. is can be the result of effective lobbying by rms (which usually stand
to gain much from in uencing regulators), the opportunities that rms may have to reward sympathetic
officials from regulatory authorities with lucrative job opportunities, or outright bribery.

Box 4.2 The determination of electricity tariffs in South Africa

 Decision process in terms of the electricity pricing policy

Source: Amra (2013).

The above outline of the process illustrates how various factors can derail attempts to pursue efficiency and
equity in a balanced manner. These include the following (the numbers in parenthesis refer to the numbered
blocks of the pricing decision process):
• Scope to misrepresent the value of assets (2) to artificially generate the desired amount of revenue (4) – this
becomes particularly likely if external auditors are not alert to irregular, unauthorised or even fraudulent
expenditure.
• Scope to manipulate expenditure items (3) to artificially generate the desired amount of revenue (4) – this
becomes particularly likely if external auditors are not alert to irregular, unauthorised or even fraudulent
expenditure.
• Scope to manipulate the cost of human resources (12) by maintaining unproductive employees, to make it
possible to apply for a higher amount of allowable revenue (4) – this becomes particularly likely if external
auditors are not alert to irregular, unauthorised or even fraudulent expenditure, and if excessive wage
demands by labour unions can be additional disruptive forces.
• Scope to inflate or state lower sales figures to present a desired average revenue ratio (13) – this becomes
particularly likely if external auditors are not alert to irregular, unauthorised or even fraudulent expenditure.
• Scope for NERSA to approve revenues (6) that are biased towards efficiency (at the expense of equity
considerations) or equity (at the expense of allocative or X-efficiency).

ese considerations underscore that effective regulation always requires clear and transparent policy
frameworks, adequate technical capacity, and political support. e expertise required for effective
regulation at the sectoral level is likely to be a signi cant constraint in many developing countries, and
policymakers should keep this in mind when taking decisions about the feasibility of the various models of
regulation.

Key concepts
• competitive restructuring (page 73)
• deregulation (page 68)
• economic regulation (page 76)
• increasing returns to scale (page 68)
• natural monopoly (page 65)
• new model for the governance of decreasing-cost industries (page 73)
• price-cap regulation (page 78)
• privatisation (page 74)
• profit regulation (page 77)
• rate-of-return regulation (page 77)
• regulatory capture (page 79)
• sliding-scale regulation (page 79)
• statutory monopoly (page 65)
• X-inefficiency (page 67)

SUMMARY
• Perfect competition and the monopoly case are extreme market types. Monopoly often results from
governments intervening in the market using policy instruments such as licenses and regulations. We
therefore refer to these monopolies as arti cial or statutory monopolies. Statutory monopolies cause
social costs (a loss in consumer surplus). ese social costs can be reduced by deregulating the industry,
that is, by removing the arti cial barriers to entry imposed by government.
• e natural monopoly is another market failure caused by imperfect competition. is is a market form
that results from industries exhibiting decreasing average costs over the entire production range for
which there is market demand. Public utilities such as energy supply, bulk water providers and rail
transport are examples of natural monopolies. If these industries were to produce at the perfectly
competitive market equilibrium, they would be making losses. Government therefore has an important
role to play in ensuring that production and pricing are at a point that is close to the social optimum
level.
• Governments traditionally took ownership of natural monopoly industries. In theory, a combination of
marginal-cost pricing and unit subsidies by a government-owned natural monopoly should yield better
outcomes than a private monopoly would achieve, albeit with distorting effects on the rest of the
economy and increased X-inefficiency. In practice, however, nationalised natural monopolies failed in
the vast majority of developing countries.
• A growing number of countries have switched to the so-called new model for the governance of
decreasing-cost industries. is model has three elements: competitive restructuring (or unbundling) of
state-owned monopolies, privatisation of the components of these entities in which competition is
possible, and regulation of the privatised components.
• e new model holds considerable promise, but also brings risks. Moreover, the effectiveness of
competitive restructuring and privatisation depends signi cantly on the degree of effective competition
in the restructured industries and governments’ ability to regulate the conduct of market participants.
ese are non-trivial requirements.
• Rate-of-return regulation, price-cap regulation and sliding-scale regulation are three well-known models
of economic regulation that have been applied to privatised components of decreasing-cost industries.
Each of these models has advantages and disadvantages. Clear and transparent policy frameworks,
adequate technical capacity and political support are necessary to apply them to good effect.

MULTIPLE-CHOICE QUESTIONS
4.1 An arti cial or statutory monopoly …
a. is a form of government failure.
b. is the result of inefficiencies in the market and government policies.
c. causes social costs that can be reduced by deregulating the industry.
d. only results in allocative inefficiency.
4.2 A natural monopoly …
a. is characterised by decreasing returns to scale.
b. is characterised by large capital outlays.
c. implies that average costs lie below marginal costs over the entire output range for which there is
demand.
d. should be deregulated to ensure a social optimum level of production.
4.3 e new model for the governance of decreasing-cost industries …
a. argues that competition is possible in some parts of these industries.
b. implies that the local-loop component of the telecommunications industry can be privatised.
c. includes regulation of privatised components of decreasing-cost industries.
d. offers a simple framework for improving the performance of decreasing-cost industries.

SHORT-ANSWER QUESTIONS
4.1 Distinguish between arti cial and natural monopolies.
4.2 What is meant by competitive restructuring of vertically integrated natural monopolies?
4.3 Explain the nature and causes of regulatory capture.

ESSAY QUESTIONS
4.1 Illustrate the effect on general equilibrium of introducing a monopoly into the two-sector model. What
are the efficiency implications?
4.2 Outline the two main sets of policy options for governing decreasing-cost industries.
4.3 Discuss the theoretical and empirical cases for and against privatisation of decreasing-cost industries.
4.4 Compare the aims, advantages and disadvantages of the three best-known models of economic
regulation.

1 is and the next section are partly based on Black and Dollery (1992: 10–16).
2 Section 2.3 of Chapter 2 explains the notion of X-efficiency.
3 is paragraph draws on Kessides (2005: 82–83).
4 Information in this Box is mostly sourced from the Parliamentary Budget Office (2017).
5 e term ‘load shedding’ refers to planned and controlled interruptions in the supply of power when power station capacity is
insufficient to supply the demand from all customers.
6 A forensic investigation report by Fundudzi for and released by National Treasury (2018c) indicated that Eskom did not advertise a
competitive bid to supply Majuba Power Station with coal and allowed the supplier to make an offer outside the competitive
bidding process. e process that was followed was allowed by Eskom’s 2008 Medium-Term Procurement Mandate, but Tegeta, to
whom the tender was allotted, did not comply with all the requirements of the mandate.
7 is is not a new phenomenon. In 1979, the demand for electricity increased in South Africa and the so-called reserve margin (the
available electricity supply capacity over and above that needed to meet normal peak demand) dropped. e urgency of the
situation led to the decision to by-pass the prescribed tender processes, and the construction of two new power stations (Lethabo
and Tutuka) were expedited by ordering turbines and boilers from existing suppliers. See Steyn (2006: 15).
8 is section draws heavily on Kessides (2005: 83–85, 85–86).
9 Chapter 5 discusses some well-known criteria for assessing the welfare effects of policy measures.
10 e following review draws on the ndings of Estrin and Pelletier (2018), Kessides (2005), and Parker and Kirkpatrick (2005a).
While more recent studies have overcome some of the methodological pitfalls identi ed by Parker and Kirkpatrick (2005a: 526), it
remains prudent to avoid strong general conclusions about the economic effects of privatisation and never to underestimate the
value of case-by-case analysis and interpretation.
11 is section draws on Den Hertog (2010: 38–40), Parker and Kirkpatrick (2005b), and Parker (2002a: 501–503).
12 is section has discussed models for regulating the prices charged by private monopolies. Eskom, however, is a regulated public
enterprise. Any reforms to the regulatory regime that may accompany the application of the new model in the South African
electricity supply industry may well have profound effects on Eskom’s managerial model and organisational incentives.
Equity and social welfare

Philip Black & Krige Siebrits

In Chapter 2, we noted that one of the most important shortcomings of the neoclassical model of general
equilibrium is its neutrality in respect of the distributional issue. e fair distribution of wealth or income
within a community can be viewed as a potential market failure. In Section 2.4.6, we refer to the ‘black box’
nature of our benchmark model of general equilibrium, or the fact that its predictions are basically
determined by the initial assumptions on which it is based. e same applies to the distributional issue: the
model predicts a particular distributional outcome that is a mirror image of the distribution assumed
initially. If the initial distribution is deemed unacceptable, so too will be the nal distribution.

Once you have studied this chapter, you should be able to:
distinguish between the Pareto and Bergson criteria for a welfare improvement
discuss Nozick’s entitlement theory and its relevance to the recent history of South Africa
explain how a redistribution of income can be justified in terms of the theory of externalities
distinguish between the cardinal and ordinal social welfare functions
discuss the efficiency implications of policies aimed at redistributing income from rich to poor people.

5.1 Introduction
In terms of the two-sector model illustrated in Figure 5.1, all points along the PPC are Pareto-efficient as
de ned in earlier chapters. Each of these points also corresponds to a particular distribution of income
between the individuals participating in the economy. is means that the distribution at point S is different
from that at point C. In particular, one individual is in a better position relative to the other at point S than he
or she is at point C.
Two important implications arising from our familiar two-sector model are relevant here:
• A competitive economy producing the output mix given by point C will not necessarily also yield the
most preferred distribution of income; the latter may, for example, occur at point S.
• A policy-induced movement along the PPC, for example, from point C to point S, will necessarily change
the distribution of income and thus place one individual in a worse position compared to the other.

Economists normally distinguish between two criteria when assessing the welfare effects of public policy:
the Pareto criterion and the so-called Bergson criterion. e Pareto criterion implies that a policy-induced
change is justi ed only if it improves the well-being of at least one person without harming any other. e
Bergson criterion is much broader and allows for a welfare improvement even if one or more individuals
are harmed in the process. In this chapter we shall consider both criteria: Section 5.2 focuses on Robert
Nozick’s (1974) well-known entitlement theory, which provides a Pareto-based justi cation for a
redistributive policy aimed at redressing past injustices; Section 5.3 examines several other Pareto criteria
and does so in terms of the familiar theory of externalities; Section 5.4 considers the Bergson criterion and
introduces what is conventionally referred to as ‘welfare economics’; and nally, Section 5.5 looks at some of
the efficiency implications of policies aimed at redistributing income from rich to poor people.

Figure 5.1 Potential top-level equilibria

5.2 Nozick’s entitlement theory

e Pareto criterion is commonly associated with the libertarian approach to public policy, according to
which individual freedom is viewed as the primary goal of the community. is is usually de ned in terms of
the maximisation of ‘negative freedom’, or protection of the right not to be coerced by others (Hayek, 1960;
Nozick, 1974). e libertarian school thus advocates a laissez faire system in which the role of government is
reduced to that of a caretaker charged with the responsibility of protecting individual freedom. Libertarians
are in principle opposed to distributional policies that infringe upon the freedom of individuals.
ere is, however, an important exception to the libertarian rule that derives from Robert Nozick’s (1974)
entitlement theory. Nozick distinguished between three ‘principles of justice’ in which he sets out the
conditions for a just distribution. e rst two principles can be de ned as follows:
• Principle 1: Justice in acquisition, which states that individuals are entitled to acquire things that do not
belong to others or do not place others in a worse position than before. Such ‘things’ refer to property
and capital goods only – not to labour income, which Nozick regards as an inalienable individual right.
• Principle 2: Justice in transfer, according to which material things can be transferred from one
individual to another on a voluntary basis, for example, in the form of gifts, grants, and bequests, or
through voluntary exchange.

In terms of these principles ‘… a distribution is just if it arises from a prior just distribution by just means’
(Nozick, 1974: 58). Violating either of the rst two principles gives rise to Nozick’s third principle of justice:
• Principle 3: Recti cation of injustice in holdings, in terms of which a redistribution of wealth is
potentially justi ed only if one or both of the rst two principles have been violated.
Nozick’s third principle provides his only justi cation for a policy aimed at redistributing resources between
individuals. But it is evidently easier said than done. Nozick (1974: 67) himself recognises the practical
difficulties involved when he asks: ‘… how far back should one go in wiping clean the historical slate of
injustice?’ is question is clearly relevant to many peace initiatives and con ict resolutions undertaken in
many parts of Africa and, in particular, to South Africa’s recent past. Nozick’s principles presumably formed
one of the cornerstones of the investigations undertaken by the Truth and Reconciliation Commission
(TRC) in South Africa. e TRC limited its focus to the apartheid era, in particular to the period between 1
March 1960 and 5 December 1993 during which the National Party was in power, though it did recognise the
many historical precedents set by earlier regimes. Some of the functions of the TRC are outlined in Box 5.1.

Box 5.1 Rectification in South Africa

The overriding objective of the Truth and Reconciliation Commission was to develop a human rights culture in
the country and to bring about national unity and reconciliation. The Commission was divided into three
committees whose primary functions can be summarised as follows:
• The Committee on Human Rights Violations was responsible for identifying victims of ‘gross human rights
violations’ committed during the period 1 March 1960 to 5 December 1993. The committee assessed the
nature and magnitude of those violations, and referred legally prepared reports on the victims to the
Committee on Reparations and Rehabilitation. During its deliberations the former committee also identified
alleged perpetrators, as well as persons already being prosecuted or found guilty in a court of law, and
submitted reports on them to the Committee on Amnesty.
• The Committee on Amnesty considered applications from persons who had committed human rights
violations and based its decision on whether these were committed for political reasons, rather than for
personal gain or any other reason. The Committee also looked at the nature and the degree of seriousness
of violations before reaching a decision. Those who were granted amnesty could not be held liable for
damages and could not be criminally charged in a court of law, while those who were unsuccessful were
liable for prosecution.
• The Committee on Reparations and Rehabilitation had to submit proposals to the Cabinet for reparations in
the form of monetary payments to individual victims, as well as ‘community rehabilitation programmes’
aimed at providing a range of basic services to communities that had suffered under apartheid. The TRC Act
required Parliament to establish a President’s fund from which payments were to be made.
After consulting across a broad spectrum of the community, and adopting a strict legalistic approach to its
deliberations and findings, the TRC concluded that an amount of R2,8 billion was needed to pay reparations to
apartheid-era victims. The Department of Land Affairs had been involved in a process of investigating and
legally processing a number of land claims arising from past violations of individual and communal property
rights (another legacy of the Group Areas Act and human resettlement programmes implemented during the
apartheid era). In addition, several politicians, including former president Nelson Mandela, appealed to the
private sector to make voluntary contributions to the government’s own job creation fund – aimed at ‘achieving
racial reconciliation in South Africa’.

According to Nozick, proper recti cation requires a thorough analysis of the historical events that gave rise
to the violation of his rst two principles. It also calls for an accurate assessment of the distributional pattern
that would have emerged in the absence of the violation. While both tasks are needed to determine the
extent of recti cation, neither can be said to be straightforward. Both require a wealth of historical data,
much of which will be largely hypothetical and based on anecdotal evidence, and neither is likely to produce
outcomes that are free of human prejudice.
On the other hand, Nozick’s third principle is restricted to a redistribution of improper holdings of xed
property and capital only – not of labour income. e latter re ects a person’s innate endowments and
cannot therefore be taken from him or her, either for equity or efficiency reasons. While this restriction is
perhaps highly contentious, it does at least simplify the practical application of Nozick’s recti cation
principle.
 e Pareto avour of Nozick’s recti cation principle is straightforward: if Tom enriched himself at
andi’s expense and did so against her will, the principle demands that Tom should give back to andi
what rightfully belonged to her so that both parties would be in the same position as they would have been
in the absence of the injustice.

5.3 Other Pareto criteria

Policies aimed at redistributing income from rich to poor people can be justi ed on Pareto grounds in terms
of the theory of externalities, that is, the externality argument for redistribution, as discussed earlier in
Chapter 3. In communities characterised by a high degree of inequality it is possible that the poor may
impose certain negative externalities on the rich. High levels of crime and violence often go hand in hand
with widespread poverty, and these may undermine the quality of life of the rich. Likewise, a lack of
sanitation and other health-promoting services among the poor may give rise to a variety of contagious
diseases that may ultimately threaten the health status of the rich.
Under these conditions, rich people may be prepared to transfer part of their income to the poor in an
attempt to reduce poverty and minimise its negative external effects. However, no single rich person can do
so alone and it is partly for this reason that the distribution of income is often viewed as a public good: all or
most rich people stand to bene t from a reduction in poverty, and hence in the level of crime and violence
or in the incidence of disease, but individuals acting on their own cannot bring about such changes. Poverty
relief thus calls for appropriate government action aimed at bringing down the negative external effects of
poverty to an optimal level. Government policy could take the form of direct transfer payments to the poor,
or it could be used to provide basic services or strengthen the security system, in which case both poor and
rich people stand to bene t from a healthier and more secure environment.
A related justi cation for redistribution derives from the so-called insurance motive. Individuals may
view their tax payments as a relatively inexpensive means of insuring themselves against a possible future
loss of income or ill-health. On becoming unemployed, for example, they may qualify for support from a
state-run unemployment insurance fund. If they should become ill, they could likewise avail themselves of
health services provided by the state. ese individuals may view tax payments as a superior or cheaper
alternative to taking out private insurance.
In all these cases there is no charity involved, but rather a quid pro quo principle: rich people give up part
of their income for distribution among the poor because they expect to derive commensurate material
bene ts from such actions.
By contrast, a redistribution of income can be justi ed on Pareto grounds if one or more individuals are
assumed to be altruistic, that is, both concerned and generous (Hochman & Rodgers, 1969). Such
individuals could experience a net increase in utility from a policy that taxes their own income and
redistributes it in favour of another (non-altruistic) individual. In terms of our two-sector model, a
movement along the PPC would then improve the welfare of both individuals.
It is important to note that the notion of altruism implies the existence of external effects in
consumption. Reverting back to our earlier examples in Chapters 2 and 3, if individual a is an altruist but b is
not, we can write a’s utility function (Ua) as follows:

where the rst derivatives are all positive. Equation [5.1] simply states that a derives utility not only from her
own income, Ma, but also from individual b’s level of utility [Ub(Mb)] – presumably up to some maximum
level. Individual b derives utility only from his own income, Mb.
Simplifying Equation [5.1] by setting

we can state the condition for a ‘Pareto-efficient’ redistribution of income from our altruist, a, to the non-
altruist, b. Mathematically it implies that:
erefore, the increase in a’s utility resulting from b’s higher income (g’(Mb) > 0) must exceed the decrease
in a’s utility resulting from the drop in her own income (0 > g’(Ma)). In other words, the fact that b is in a
better position gives rise to a net increase in a’s utility.
ere are presumably many real-world examples of Pareto-efficient redistributions. When people
contribute towards charitable organisations, or give money to beggars, they presumably do so because it
makes them feel better – this is why we view altruism as a Pareto-based justi cation for redistribution. Such
transactions are, of course, voluntary while redistribution via the scal process is not. Nonetheless, it can be
suggested that some taxpayers do derive utility from that part of their taxes earmarked for the relief of
poverty and the care of old and disabled people (see Box 5.2).

Box 5.2 Experimental games

In the field of behavioural economics, we have witnessed the results of a large number of experimental games
aimed at testing the ‘self-interest hypothesis’ that lies at the root of many of our standard economic models.
These games include the so-called ultimatum game, the gift exchange game, the trust game and, as we saw in
Chapter 3, the public goods game.
In the ultimatum game, for example, the experimenter gives an amount of money to a so-called proposer
with the instruction of sharing it with a responder. If the proposer gives some portion to the responder who then
rejects it, both players get nothing; but if the responder accepts it, the proposal stands. Now, in terms of the
self-interest hypothesis, the smallest possible amount would be offered and accepted. And yet in many such
games played across the world most proposers offer in excess of 20 per cent of the original amount, while the
probability of rejection falls dramatically with the size of the offer. The conclusion here is that both players have
a sense of ‘fairness’. Other experimental games show that people are generally altruistic, or have social (as
opposed to only individual) preferences, and also have an ‘aversion’ to inequality (e.g. Bowles, 2004). From a
public economics perspective, one implication of these experimental findings is that people would be willing to
give part of their income to the fiscus if it will, directly or indirectly, benefit the poor.

5.4 Bergson criterion

In terms of the Bergson criterion, a redistribution of income can be justi ed on welfare grounds even if it
places one or more individuals in a worse position. e Bergson criterion is best explained in terms of the
familiar social welfare function, according to which a community’s welfare is de ned in terms of the
utilities of all the individuals making up that community.
We can distinguish between two such welfare functions. e rst is the so-called ‘cardinal’ or additive
welfare function:

where W represents the level of community welfare, and Ua and Ub are individual utilities. According to
Equation [5.4], community welfare equals the sum of individual utilities – these are assumed to be
measurable on the same scale.
e additive welfare function does illustrate the Bergson criterion very neatly: it allows for the Pareto
criterion insofar as W will increase if either Ua or Ub increases, or if both Ua and Ub increase at the same time.
It also allows for a welfare improvement consequent upon a decrease in either Ua or Ub – W will increase so
long as the increase in Ua (or Ub) exceeds the decrease in Ub (or Ua). In addition, if a poor person, say
individual a, derives greater additional utility from an extra R10 than does his or her rich counterpart,
individual b, then the stage is set for a welfare-improving redistributive policy: taking R10 away from b and
giving it to a will raise Ua by more than it will reduce Ub . is conclusion assumes that an individual’s
marginal utility from income diminishes as his or her income increases.
Equation [5.4] represents a very restrictive welfare function. Apart from the measurability issue, it
assumes that individual utility functions are identical and depend only on their incomes. It is also highly
debatable whether increases in income engender smaller increases in utility at higher levels of income.
It is partly for these reasons that economists prefer the more generalised welfare function:

is function is ordinal in nature and does away with the assumption of measurability. By letting W take on
different (constant) values, it is possible to derive a set of social or community indifference curves, such as
those labelled W1, W2, and W3 in Figure 5.2. ese functions have the same properties as individual
indifference curves: they are convex with respect to the origin, cannot intersect, and exhibit diminishing
marginal rates of substitution.
Figure 5.2 also shows a utility possibility frontier (also known – rather grandly – as the ‘grand utility
possibility frontier’) that is directly derived from the familiar PPC. e utility possibility frontier, QR, gives
the utility combinations associated with all the top-level equilibrium points along a conventional PPC. In
the example of Figure 5.2 the community prefers the combination at point H – the welfare maximum – to any
other point along the frontier.

Figure 5.2 Utility possibility curve and welfare maximum

is analysis crucially depends on two closely related assumptions. Firstly, the community is assumed to
be able to choose between different points along the utility possibility frontier, for example, point H as
opposed to point G; the question of how a community makes such choices has given rise to a vast literature –
referred to as public choice theory – and we shall return to this issue in the next chapter. Secondly, when
choosing a particular point on the frontier, the community is making an explicit value judgement about the
relative worthiness of the two individuals a and b. is is illustrated in Figure 5.3 where two alternative sets
of welfare functions are shown, one labelled W1, W2, …, and the other W1’, W2’, … . e similarly numbered
subscripts indicate the same level of social welfare. It is clear that the former function embodies a relatively
strong preference for individual a, generating a top-level equilibrium at point I; whereas the function W1’,
W2’, …, embodies a strong relative preference for individual b. Individual a is in a better position at point I
than at point J, while the opposite is true of individual b, and yet the two social indifference curves, W2 and
W2’, represent the same level of welfare.
It is but a small step to show that the difference between the two welfare functions in Figure 5.3 also
implies a difference in the community’s assessment of the worthiness of the two sectors, X and Y. All we
need to assume is that the individuals have different tastes. us, if individual a has a strong relative
preference for good X, and individual b has a strong relative preference for good Y, then it follows that the
function labelled W1, W2, … indicates, by implication, that the community has a strong relative preference for
sector X. Similarly, the function W1’, W2’, … would indicate a strong relative preference for sector Y.

Figure 5.3 Alternative welfare functions

Consider the following two individual utility functions:

Given that X = Xa + Xb and Y = Ya + Yb , application of the ‘function-of-a-function’ rule to Equation [5.5]

provides the following:

where V indicates a different functional relationship from our earlier function. Equation [5.8a] simply states
that the welfare of society depends on the production levels of the two commodities X and Y. Equation [5.8a]
is a very general version of the welfare function (de ned in terms of individual commodities) and it can be
speci ed in many ways. However, most of these speci cations will embody the above value judgement, that
is, that the community has to make a judgement about the relative worthiness of the two sectors and, by
implication, of the two individuals as well.
e commodity-based welfare function is shown in Figure 5.4 where we assume that the top-level
competitive equilibrium occurs at point C. But this does not coincide with the welfare maximum at point S:
the community prefers point S to point C, thus establishing a prima facie case for appropriate state
intervention to move the economy to point S.
is analysis has shown that a top-level competitive equilibrium is only a necessary condition for a social
welfare maximum, not a sufficient condition – this is indicated by the difference between points S and C in
Figure 5.4. Two questions therefore arise:

Figure 5.4 A competitive equilibrium versus welfare maximum

• What kinds of policy could be used to bring about an inter-sectoral or inter-personal redistribution of
income, that is, a movement from point C to point S?
• What are the implications for economic efficiency of such policies?

As far as the rst question is concerned, government has several options at its disposal: it can tax sector Y
and subsidise sector X, or it can tax one individual and subsidise the other; it can also redirect its own
spending towards one sector or individual. ese options will be discussed in greater detail in subsequent
sections of this book. Note, however, that a movement from C to S entails an improvement in social welfare
even though Ub may be reduced on account of the reduced supply of good Y. is is the difference between
the Bergson welfare function and the Pareto criterion discussed earlier.
In the meantime we turn our focus to the second question raised above.

5.5 Efficiency considerations

It would be surprising if a redistributive policy comprising a suitable tax-subsidy mix had no effect on
efficiency levels in the economy. As we shall see in subsequent chapters, most taxes and subsidies do have
distortional effects on markets, and the real question is whether the perceived bene ts from such policies
justify these distortions.
We shall brie y introduce two such distortions here. e rst is the possible effect that individual taxes
and subsidies might have on the willingness to work. Speci cally, does a new or higher tax or subsidy cause
people to work longer or fewer hours per day or per month? And similarly, does a higher tax or subsidy
cause people to work more or less productively, that is, produce more or fewer units of output per hour? In
other words, can one expect an increase or decrease in the supply of labour or in its productivity – or, in
terms of our two-sector model, an optimal or sub-optimal top-level equilibrium?
e evidence on this issue – referred to as ‘income and substitution effects’ or the ‘(dis)incentive effect’ –
is anything but conclusive. What it does show is that, above a certain level, increased taxes tend to have a
small negative effect on the willingness to work. is is illustrated in Figure 5.5 where the initial competitive
equilibrium again occurs at point C0 on the PPC labelled M0N0, representing an initial welfare level of W1.
Now, a policy aimed at redistributing resources from sector Y to sector X (for example by taxing sector Y
and/or subsidising sector X) will move the economy in the direction of point S – the policy target. However, if
the policy does have a small disincentive effect, labour productivity may fall below its potential and the
economy may end up at a sub-optimal allocation lying inside the PPC – for example point F in Figure 5.5.
Although point F is inferior to point S, it nevertheless represents a higher level of welfare than the
original allocation at point C0 – W2 as opposed to W1. A stronger disincentive effect would have a bigger
negative impact on labour productivity and hence on the level of social welfare. us, the policy could move
the economy to a point such as E on the social indifference curve W0, which is clearly inferior to the original
allocation at point C0.
e second possibility referred to above concerns the dynamic consequences of a distribution policy
aimed at taxing the rich and subsidising the poor. e imposition of a new or higher tax (on sector Y or on
individual b) may limit savings and investment and hence economic growth and, in the limit, keep the
economy on its existing PPC, for example, at point S in Figure 5.5, representing the higher welfare level W3
(compared to the original allocation at point C0). In the absence of such a policy, however, the economy may
experience positive economic growth over time – indicated by the outward shift of the PPC from M0N0 to
M1N1 and the concomitant change in the competitive equilibrium from point C0 to point C1. e latter
change – though not achieving a welfare maximum – nevertheless represents an improvement in the level of
social welfare from W1 to W4 and is evidently superior to the static policy-induced movement to point S.
Figure 5.5 Disincentive and savings effects of redistribution

On the whole, the above analysis indicates that there are many good reasons why societies might want to
achieve a more equitable distribution of income. But whatever the justi cation might be, it is clearly
important that the expected bene ts of a policy of redistribution should be carefully weighed against the
possible negative effects it might have on labour supply and on savings and investment, and hence on
economic growth in the long term. We will return to this theme in subsequent chapters.

Key concepts
• additive welfare function (page 89)
• altruism (page 87)
• Bergson criterion (page 84)
• entitlement theory (page 85)
• externality argument for redistribution (page 87)
• insurance motive (page 87)
• justice in acquisition (page 85)
• Pareto criterion (page 84)
• rectification of injustice (page 85)
• redistribution (page 87)
• social indifference curves (page 90)
• social welfare function (page 88)
• Truth and Reconciliation Commission (page 85)
• willingness to work (page 92)
SUMMARY
• Economists normally distinguish between two criteria when assessing the welfare effects of public
policy: the Pareto criterion and the so-called Bergson criterion. e Pareto criterion implies that a policy-
induced change is justi ed only if it improves the well-being of at least one person without harming any
other. e Bergson criterion is much broader and allows for a welfare improvement even if one or more
individuals are harmed in the process.
• Robert Nozick’s entitlement theory distinguishes between the principles of justice in acquisition and
justice in transfer, arguing that ‘a distribution is just if it arises from a prior just distribution by just
means’ (Nozick, 1974: 58). Violating either of the rst two principles gives rise to Nozick’s third principle
(that is, justice in recti cation), in terms of which a redistribution of wealth is potentially justi ed only if
one or both of the rst two principles have been violated.
• Nozick (1974: 67) himself recognises the practical difficulties involved when he asks, ‘… how far back
should one go in wiping clean the historical slate of injustice?’ is question is clearly relevant to various
peace initiatives and con ict resolutions undertaken in many parts of Africa, and, in particular, to South
Africa’s recent past. Nozick’s principles of justice in effect formed one of the cornerstones of the
investigations undertaken by the Truth and Reconciliation Commission (TRC) in South Africa. e TRC
limited its focus to the apartheid era, in particular to the period between 1 March 1960 and 5 December
1993, during which the National Party was in power, although it did also recognise the many historical
precedents set by earlier regimes.
• Policies aimed at redistributing income from rich people to poor people can be justi ed on Pareto
grounds in terms of the theory of externalities. High levels of crime and violence often go hand in hand
with widespread poverty, which may undermine the quality of life of the rich. Likewise, a lack of
sanitation and other health-promoting services among the poor may give rise to a variety of contagious
diseases that may ultimately threaten the health status of the rich.
• Under these conditions, rich people may be prepared to transfer part of their income to the poor in an
attempt to reduce poverty and minimise its negative external effects. However, no single rich person can
do so alone and it is partly for this reason that the distribution of income is often viewed as a public good:
all or most rich people stand to bene t from a reduction in poverty, and hence in the level of crime and
violence or in the incidence of disease, but individuals acting on their own cannot bring about changes
of this nature.
• e Bergson criterion can be explained in terms of the familiar social welfare function, which represents
the relative preferences of a group or community of individuals and which, inter alia, re ects the optimal
or most preferred distribution of income. is welfare maximum also coincides with a particular point
lying on the production possibility curve (PPC), but may well be different from the toplevel equilibrium
derived in Chapter 2. When choosing this point on the PPC, the community is making an explicit value
judgement about the relative worthiness of the two (groups of ) individuals, a and b.
• In terms of the two-sector model of general equilibrium, the government can tax the rich (for example,
individual b) and subsidise the poor (individual a) or it can tax luxury goods consumed mostly by the
rich (for example, sector Y) and subsidise basic consumer goods (sector X). It could also redirect its own
spending towards one sector or individual.
• A policy comprising a tax-subsidy mix of this nature may have distortional effects on markets. e real
question is whether the perceived bene ts from policies such as this justify these distortions. In terms of
the effect on the willingness to work, for example, does a new or higher tax or subsidy cause people to
work more or fewer hours per day or per month? Similarly, does a higher tax or subsidy cause people to
work more or less productively, that is, to produce more or fewer units of output per hour?
• In addition, the imposition of a new or higher tax (on sector Y or on individual b) may limit savings and
investment, and hence also economic growth.
• It is clearly important that the expected bene ts of a policy of redistribution be weighed carefully against
the possible negative effects that it might have on labour supply as well as on savings and investment,
and hence on economic growth in the long term.
MULTIPLE-CHOICE QUESTIONS
5.1 A policy-induced movement along the production possibility frontier can be justi ed in terms of …
a. the Pareto criterion.
b. the Bergson criterion.
c. the Pareto and Bergson criteria.
d. All of the above options are correct.
5.2 e Bergson criterion implies that a policy-induced change can be justi ed if …
a. the well-being of at least one individual is increased without anyone else being harmed.
b. the well-being of at least one individual is improved and that of another is diminished.
c. individual bene ts exceed individual losses.
d. All of the above options are correct.
5.3 Robert Nozick’s entitlement theory is based on the following principle or principles of justice:
a. Justice in acquisition
b. Justice in transfer
c. Recti cation of an injustice
d. All of the above options are correct.
5.4 Taxing the rich and subsidising the poor could have the effect of …
a. an increase or a decrease in labour supply, depending on the magnitude of the tax or subsidy.
b. an increase or a decrease in labour supply, irrespective of the magnitude of the tax or subsidy.
c. an increase or a decrease in labour productivity.
d. an increase or a decrease in savings and investment.

SELF-ASSESSMENT EXERCISES
5.1 Distinguish between the so-called Pareto and Bergson criteria for a redistribution of income between
rich and poor people.
5.2 Explain why altruistic behaviour could provide a Pareto-based justi cation for a policy of income
redistribution.
5.3 Distinguish between the ‘additive’ and ‘ordinal’ (or Bergson) social welfare functions.

ESSAY QUESTION
5.1 Discuss the proposition that Pareto optimality is a necessary but not a sufficient condition for a welfare
maximum.
5.2 Discuss Nozick’s entitlement theory and brie y consider its relevance for South Africa.
5.3 Discuss the efficiency implications of an inter-sectoral (or interpersonal) policy of income
redistribution.
Public choice theory

Philip Black and Estian Calitz

In the previous chapters, we looked at a range of so-called market failures and highlighted the point that
private markets cannot, on their own, supply public goods or make allowance for external costs and bene ts.
Similarly, although it is conceptually possible to regulate imperfectly competitive markets or to evaluate the
desirability of various income distributions by means of the Pareto and Bergson normative criteria, we have
not as yet addressed the question of how a community can practically express its collective preferences on
these matters. In this chapter, we focus on the way in which communities make public choices.

Once you have studied this chapter, you should be able to:
discuss the Rawlsian theory of justice and comment on its relevance to recent political developments in
South Africa
explain the median voter theory, and indicate its potential strengths and weaknesses
discuss the meaning and importance of Kenneth Arrow’s impossibility theorem
consider whether logrolling (or vote trading) is an efficient means of improving the outcomes of a majority
voting system
explain the theory of optimal voting rules and consider the question of whether it does indeed provide an
optimal rule for majority voting
discuss the maximising behaviour of politicians and bureaucrats, and consider the implications of this
behaviour for majority voting
explain the origins and consequences of rent-seeking.

6.1 Introduction
Many public issues are discussed and resolved in the political marketplace, where the quantities of public
goods and services and the levels of taxes and subsidies are decided. e decisions are usually made by
elected politicians on behalf of the voting population. In this chapter, we look at the way in which these
public decisions are made from a strict economic perspective. We investigate the mechanisms – or social
choice rules – by which individual preferences are transformed into social or public preferences and ask
whether these will ensure that governments act in accordance with the wishes of the people they represent.
e most straightforward method of effecting a transformation of this nature lies in simply imposing the
ethical views of some dictator or central politburo on society, and then adjusting economic policy
accordingly. Viewed historically, however, dictatorial rule has often led to abusive behaviour and clearly
does not represent a desirable social choice rule. Most countries today prefer some method of collective
decision-making based on, and legitimised by, mass participation on the part of their citizenry. Indeed, the
characteristic social choice rule employed in most democracies is majority rule, whereby citizens elect
representatives who must act within certain constitutional parameters.
us, in the absence of dictatorial rule, the range of social choice rules varies between the unanimity
rule, in terms of which a proposal requires 100% support before being accepted, and the ordinary majority-
voting rule, for which 50%-plus-one-vote is required. In this chapter, we rst consider the unanimity rule
(Section 6.2), focusing speci cally on John Rawls’s well-known theory of justice (1971), partly because of its
relevance to the dramatic shift towards a full democracy in South Africa. is is followed by a discussion of
the ordinary majority-voting rule as expressed in the median voter hypothesis in Section 6.3. Sections 6.4
and 6.5 look at some of the problems associated with majority voting, focusing speci cally on its inability to
allow for differences in the intensities of individual preferences. Section 6.6 deals with an important
variation on the ordinary majority-voting rule, that is, Buchanan and Tullock’s (1962) notion of optimal
voting rules. In Section 6.7, we shift our attention to positive public choice theory, and consider the
behaviour of politicians, bureaucrats and private interest groups within a typical representative democracy.
Section 6.8 concludes.

6.2 The unanimity rule and the Rawlsian experiment

e unanimity-voting rule means that each member or representative group within a community must
support a proposal before it becomes the collective decision. e unanimity rule is the only voting rule that
will lead to a Pareto-optimal outcome and, since it requires that collective decisions be in the interest of all
parties, it can be viewed as a positive sum game.
A good, if somewhat unusual, example of the unanimity principle is provided by John Rawls, or the
Rawlsian theory of justice (Rawls, 1971). e theory is essentially a normative one setting out the
conditions under which ‘free and rational’ persons will choose certain principles of justice that govern the
‘basic structure of society’. ese principles determine the fundamental rights and duties of individuals, and
regulate the institutional framework within which allocative decisions are made on a collective and
unanimous basis. e ‘social contract’ that ultimately emerges is described by Rawls as a case of ‘justice in
fairness’.
What sets the Rawlsian theory apart from others is its focus on the process by which individuals reach
unanimity over the principles of justice. He thus asks his readers to imagine a situation in which the
contracting parties, representing the whole community, all step through a ‘veil of ignorance’ and enter a
hypothetical ‘original position’, which is a position that is reached through intensive personal discourse
during which all barriers are broken down, and each party is able to rid itself of all prejudices and prior
knowledge. In the original position, each party is wholly unaware of its ‘place in society’ and has no
knowledge of the probability distribution of expected outcomes, let alone its current well-being. ey all
stand as equals on the same playing eld. It is important to emphasise that Rawls only asks his reader to
imagine an original position of this nature, not actually to create it!
What is important for our purpose is that each party in the original position is assumed to be equally risk-
averse. All parties would thus support a risk-minimising social welfare function that effectively insures them
against the worst possible outcome. A welfare function of this nature could take the following form:

indicating that social welfare depends on the lower of the two individual utilities. us, if Ua < Ub, then
welfare reduces to W = Ua. is implies that in order for W to increase, there must be an increase in Ua. e
latter condition does not preclude an increase in Ub; it is perfectly in order for Ub to increase as long as Ua
increases as well. e Rawlsian welfare function is therefore perfectly consistent with a Pareto-based policy
bene ting both parties.
From a Rawlsian perspective, Equation [6.1] implies that all parties in the original position will adopt a
so-called maximin strategy, which gives priority to the party potentially nding itself in the worst-off
position or with the minimum utility. is will provide them with a measure of protection in case they end
up being that party. All parties will therefore choose and unanimously approve a political constitution
embodying the Rawlsian welfare function, and, by implication, also an institutional system and a set of
policies aimed at allocating and distributing resources in accordance with that function. Public policy
should thus be aimed at bene tting the poorest of the poor at all times, irrespective of whether it harms or
bene ts other parties. e conclusion is a good example of the Bergson criterion for a welfare improvement,
according to which a redistribution of income is justi ed on welfare grounds – that is, if social or additive
welfare is enhanced – even if it places one or more individuals in a worse position (see the discussion in
Chapter 5).
It hardly seems prudent to suggest that negotiations for a new constitution and democratic dispensation
in South Africa, Namibia, Mozambique and several other African countries even approximated a Rawlsian
original position. None of the negotiating parties can be said to have been completely impartial, while the
underlying principle would appear to have been one of compromise rather than unanimity. Nonetheless,
the growing concern for the poor and historically disadvantaged in South Africa, as re ected in several
empowerment charters and the redistributive nature of recent budgets (see for example Chapter 7), does
seem to come close to acceptance of a Rawlsian welfare function, that is, one that affords priority status to
the disadvantaged members of the community.
e unanimity rule is not without its shortcomings. Reaching a unanimous decision may take a very long
time, partly because of the divergent nature of individual preferences and the number of issues involved.
Consider, for instance, the drawn-out nature of the (‘Kempton Park’) negotiations that preceded the drafting
of a new constitution in South Africa. e point here is that the costs of reaching consensus may be
inordinately high. It is important that cognisance be taken of the opportunity costs associated with the
unanimity rule. e time spent lobbying and in uencing individuals and groups of individuals could
arguably be spent more productively elsewhere in the economy. We return to this issue in Section 6.6.
Viewed more practically, the unanimity rule could give rise to strategic actions on the part of certain
individuals or groups who might want to enter into bargaining contracts with other parties in order to secure
special bene ts. Under these conditions, parties may be persuaded to engage in logrolling (or vote trading)
by forfeiting something that they want in exchange for something about which they feel particularly strongly.
We shall return to the issue of logrolling in Section 6.5, but suffice it to say here that the outcome of a process
of this nature is hardly likely to be Pareto-efficient.
A nal objection to unanimity rule is that it gives minorities the right of veto; in the extreme case, the last
unpersuaded voter has the decisive vote. Exercising a veto right of this nature will clearly render the
unanimity rule obsolete as it will lead to a situation in which effectively a small minority rules.

6.3 Majority voting and the median voter

e most common social choice rule is the ordinary majority rule (or majority voting). Every individual is
given one vote and the issue, policy or party receiving the most votes wins the day. Under a direct
democracy (direct democratic dispensation) where each voter reveals his or her preferences directly via a
referendum, the majority-voting rule requires that a proposal receives 50%-plus-one-vote support before it
can be imposed on the community. If South Africa had a direct democracy and the public were asked in a
referendum to vote for or against an increase in the rate of value-added tax (VAT), the rate would not be
increased if (say) 12 500 001 voters out of a total of 25 million voted against the increase.
In a representative democracy, individual voters elect representatives who make decisions on their
behalf. A representative democracy is generally less costly to administer than a direct democracy and it is
largely for this reason that the former is most widely used in the world today. However, some countries (for
example, Switzerland) combine the two systems, utilising the direct method when important national
decisions have to be made.
Voters’ interests in a representative democracy are represented by several in uential actors, including
elected politicians, bureaucrats, and private and public interest groups. e role of politicians is paramount
and one can reasonably ask: What are they in the ‘market’ for? What do they want to maximise? According to
Anthony Downs (1957), politicians act like any other utility-maximising consumer or pro t-maximising
entrepreneur; the only difference is that they are in the business of maximising the number of votes they
collect in an election. In an ideal world, the vote-maximising behaviour of politicians is an important means
of transforming individual preferences into a logically consistent set of social preferences.
What is required to maximise votes in a representative democracy? In order to answer this question, we
must examine the median voter theorem. We begin our explanation of this theorem by de ning the median
voter as the voter whose set of preferences divides the voting community exactly into two. Let’s assume we
have a community of ve people: Ndlovo, Mary, andi, Johan and Ibrahim. We assume furthermore that we
know their precise preferences concerning the size of the national health budget. Table 6.1 sets out the
budgets for which each of them will vote.

Table 6.1 Voter preference for size of health budget

Voter Amount (R million)

Ndlovo 50

Mary 200

Thandi 400

Johan 600

Ibrahim 800

Let us adopt a step-by-step approach towards discovering the majority decision on the size of the budget,
beginning with a zero budget. Assuming that there are no extreme preferences (see below), all ve voters will
prefer a R50 million budget instead of a zero budget. A movement from R50 million to R200 million will win
the support of everyone except Ndlovo; that is, everyone except Ndlovo prefers a budget larger than R50
million. A movement from R200 million to R400 million will be approved by andi, Johan and Ibrahim,
while only two voters will support an increase from R400 million to R600 million, and so on.
It is clear from Table 6.1 that three of the ve options will enjoy majority support: all ve voters (or 100%)
will support a budget of R50 million, four voters (80%) will support a budget of R200 million and three voters
(60%) will support a budget of R400 million. But which is the optimal one? Which one will make our voting
population happiest or cause the least harm?
e answer is provided by the median voter theorem: the best option is that of andi’s – our median
voter – whose preference divides the voters exactly into two. e reason is that both Johan and Ibrahim
would prefer andi’s option to that of Ndlovo and Mary; Ndlovo and Mary will likewise have a relative
preference for andi’s option vis-à-vis those of Johan and Ibrahim. It follows that our larger-budget
supporters, Johan and Ibrahim, will rather give their support to a politician campaigning for the median
voter’s choice than to a politician promoting any other potential majority choice.
We can now formulate the median voter theorem: under a majority voting system in which preferences
are not extreme, it is the median voter’s preferred option that will win the day, since that is the option that
will produce a minimum welfare loss for the whole group.
e median voter model provides a simpli ed explanation of the rational behaviour of politicians under
ideal conditions. e model assumes that politicians interact with voters to determine their relative
preferences. By doing so, they are able to identify the median voter, act upon his or her preferences and ful l
the wishes of the majority at minimum cost in the process.
Of course, the real world of politics is a bit different from what the median voter theory would have us
believe. Not all politicians are vote maximisers responding passively to individuals’ demands. Some might
pursue the public interest rather than vote-maximising strategies, while others may appeal to voters because
of their vision and personality, rather than any tangible bene ts that they might promise. e model also
presumes that the median voter can be identi ed. is is not easy, especially since different political issues
may have different median voters. e model furthermore assumes that voters are rational and that
everyone will vote. Politicians and voters are often far from being perfectly informed, which renders rational
choice unlikely, if not impossible. We return to this issue in Section 6.7.
 On the whole, the majority-voting rule – although not Pareto-efficient – does have two important
advantages vis-à-vis the unanimity rule:
• Reaching majority approval takes much less time and is therefore less costly than achieving unanimous
support. A parliamentary majority vote would normally be regarded as sufficient to approve the national
budget, but a two-thirds majority is needed to change the constitution. e latter entails a more
expensive and time-consuming decision-making process and requires much more effort to achieve,
especially in a multiparty democracy with strong opposition.
• Under majority rule, it is much less likely that a minority will be able to prevent the majority from getting
their proposals accepted; on the other hand, majority rule can be criticised for its winner-takes-all
consequences, and for its potential to ignore minority interests and enable the majority to tyrannise
minorities.

6.4 The impossibility theorem

A potentially serious shortcoming of the majority-voting rule is the fact that it can lead to results that are
logically inconsistent. Before we explain this, we note that an optimal result under majority voting implicitly
assumes that the choice of a public good is subject to declining marginal bene ts, similar to private goods.
is implies that individuals are consistent in their choice of alternative public choices, which we refer to as
exhibiting single-peaked preferences. We demonstrate this with the aid of Figure 6.1, which is based on
Hyman (1999: 178–181). e important point is that the marginal bene t for a voter decreases as the
quantity of the public good increases. is is shown in Figure 6.1(a). Each higher output level consistently
generates less additional bene t, as shown by the negatively sloped line MB. is rules out a multi-peak
preference line such as MB*. e horizontal line t represents the taxpayer’s constant amount of tax per unit
of public good. e difference between MB and t shows the net bene t for each amount of public good. At
quantity Q* the difference is zero, which means that no extra net bene t accrues when the extra amount of
public good reaches Q*. Beyond Q* net bene t is negative. In Figure 6.1(b) we show the total net bene t. e
bell-shaped curve reaches its maximum level (i.e. total net bene t) at Q*, after which it starts to decline
because the tax payment exceeds the bene t. Only when all voters have similar single-peaked preferences,
will majority rule generate an efficient supply of public goods. is is one of the important prerequisites of
the Arrow theorem, discussed next.
Nobel Laureate, Kenneth Arrow (1951) proved that, unless majority rule meets with certain conditions, it
will not lead to consistent outcomes. is became known as Arrow’s impossibility theorem. Arrow proved
that it is not always possible to derive a logically consistent set of social preferences from a corresponding set
of individual preferences on the basis of an ethically acceptable (or democratic) social choice rule. He came
to the conclusion that there is not a single voting rule – not even majority voting – that would meet all of the
minimum ethical conditions that he set for an acceptable social choice rule.
Arrow’s theory crucially depends on these ethical conditions, which can be summarised as follows:
• e rst is the rationality assumption, according to which individuals and the community must either
prefer one option (X) to another (Y) or be indifferent between the two; this assumption is formulated as
follows:
Figure 6.1 Declining marginal benefit of a pure public good, reflecting single-peaked preferences.

X > or Y > X or X = Y

where the symbols > and = stand for, respectively, ‘prefer to’ and ‘indifferent between’. e community must
also adhere to the familiar transitivity condition, that is:

If X > Y and Y > Z, then X > Z

which means that if X is preferred to Y and Y is preferred to Z, then X must be preferred to Z. is result is
exactly what single-peaked preferences achieve, as explained in Figure 6.1.
• Next is the independence of irrelevant alternatives, which implies that if the choice is between two
options, X and Y, then the effect of Z is irrelevant.
• e Pareto principle has to apply, that is, if voter a prefers X to Y and voter b is indifferent between the
two options, then the (two-person) community must prefer X to Y; or in algebraic terms:

If (X >Y)a and (X =Y)b then X >Y

• ere has to be an unrestricted domain, that is, it should be possible for all eligible voters to vote.
• Finally, non-dictatorship should apply.

Although these conditions seem very reasonable indeed, it is important to note that Arrow’s theorem is
especially relevant to voters who have widely divergent preferences or who choose extreme alternatives. We
can illustrate a case of this nature by considering three voters, Brenda, Christelle and Abdullah. Each of them
has to choose between three alternative budgets, that is, a large budget (denoted by L), a moderate budget
(M) and a small budget (S). Let the individual preferences be as follows:

Brenda: L>M>S
Christelle: M>S>L
Abdullah: S>L>M

It is clear from this example that if alternative budgets are voted for in pairs, a majority of the voters (in other
words, Brenda and Abdullah) would prefer L to M. Another majority (in other words, Brenda and Christelle)
would prefer M to S. (It is appropriate at this point to pause in order to mention, with reference to Figure
6.1(a), that decreasing marginal utility should be interpreted here in the sense that utility falls consistently in
the case of each voter as he/she moves away from the rst choice, rather than visualising a linear array of
public good outputs on the horizontal axis from small to large.) To return to our example: according to the
transitivity condition, therefore, a majority should prefer L to S. Yet this is not the case: a majority (in other
words, Christelle and Abdullah) actually prefer S to L. Hence we have a logically inconsistent or intransitive
outcome.
e reason for this outcome is straightforward. Abdullah has extreme preferences: he prefers a small
budget to a large one, but also prefers the large one to a moderate one. Line MB* in Figure 6.1(a) actually
portrays this. When given the choice between any two budgets, a voter of this nature does not consistently
prefer a larger budget to a smaller budget or vice versa. Abdullah’s preferences may for example stem from
the following.1 Let’s say the above budget is for public schooling, which is nanced by taxes. Abdullah has
the possibility of enrolling his daughter in a private school at a substantial cost, while still having to pay taxes
to nance the budget for public schooling. If the budget for public schools (denoted here as S) is
substantially lower than the cost of a private school, he would prefer S to L. In a choice between M (a
medium-sized budget) and L, however, he may prefer L if that could result in a quality of schooling that may
cause him to enrol his daughter in the public school, possibly still at a somewhat lower rate than the cost of
private schooling.
It is not that decisions cannot be taken by means of majority voting, but that decision-making has less
credibility under circumstances such as outlined above and optimal resource allocation is unlikely. In a
system of majority voting where voting often occurs in a pair-wise fashion,2 the presence of extreme
preferences3 can have a number of consequences. e rst is that there may be a new winner every time the
sequence of voting is changed. Consequently, it is impossible to get a consistent winner. While the
preferences of voters may individually be consistent or rational, their combined preferences or the
community’s preferences (as re ected in their voting) will not be consistent if the group of voters includes
people with extreme preferences. e phenomenon is referred to as the voting paradox. is kind of voting
can continue inde nitely, in a phenomenon called cycling.
Furthermore, if the organisers of the election have prior knowledge of the existence of these extreme
preferences, it is possible to organise the sequence of voting in such a way that a desired result is obtained.
is is known as agenda manipulation. Any result that can be changed if the order of voting is changed is
not consistent and is not a true re ection of voters’ preferences. Such a result therefore cannot claim to
represent a choice for an optimal allocation of resources.
What is the practical relevance of all this? In a strict sense, Arrow’s theory implies that the majority-
voting rule does not necessarily produce outcomes that enjoy the support of the majority, which is
something of a contradiction in terms. All is not lost, however. e theory only proves that logical
inconsistency is a possible outcome. In the above example of three voters choosing between three
alternatives, for instance, the probability of a logical inconsistency arising is only 6% (Frey, 1978: Chapter 2).
As one increases the number of voters and the number of alternatives, this probability rises only very
gradually, reaching about 31% when the number of voters becomes very large and the number of
alternatives reaches six. It is thus tempting to conclude that perhaps too much is made of the impossibility
theorem.
Nonetheless, the problem of inconsistency is compounded by the fact that majority voting does not allow
for differences in the intensity of individual preferences and by the attendant problem of logrolling, as we
shall see in the next section.
6.5 Majority voting and preference intensities
Another major shortcoming of the majority-voting rule is the fact that it cannot account for differences in the
intensity of voters’ preferences or at least it cannot do so in a cost-effective way. Under these conditions, it is
quite possible that a small majority may have a relatively weak preference for a particular candidate, whom
they nevertheless vote into power. If a large minority opposes the same candidate very strongly, it is possible
that, in net welfare terms, the community as a whole will be in a worse position: given an additive welfare
function (implying measurability of individual utilities), the cumulative decreases in individual utilities will
exceed the cumulative increases.
Under majority voting, there are two ways in which preference intensities can be accommodated. e
rst is to ask people to vote in the form of intensity units. Instead of a straight yes-no vote, each voter can, for
example, be given a total of 100 points that he or she can allocate between competing candidates. In this
way, the ordering of preferences is weighted and the weights can be taken into account in the voting
procedure. It is thus possible that a candidate who would have come last under a straight yes-no system
fares better under the weighted system. But weighting is a normative procedure: andi’s 80% may mean
something completely different from Johan’s 80%. Weighting is also difficult to implement and
administratively very costly, and it must be asked whether the additional bene ts from introducing a
procedure of this nature are worth the additional costs.
Another – some would say better – way of accounting for preferences under a majority-voting rule is
through logrolling or vote trading. Logrolling may occur between and among minority parties and the
majority party. For example, an intense minority may trade its support for an issue enjoying strong support
among the majority in exchange for majority support of the minority issue. Alternatively, the same exchange
can be based on amendments being made to the issues involved. In practical terms, the latter exchange
would only be feasible if the minority had a particularly strong preference for the minority issue and the
majority did not feel too strongly about it. It would also not help if the minority took a minority view on each
and every issue: if the majority could never count on a minority’s vote on any issue, there would be no point
in trading votes. Nonetheless, to the extent that logrolling enables a better expression of consumer
preferences in respect of public goods, it may increase the social welfare of society.
Logrolling can also take the form of an exchange of votes between different minority groups. ese
groups could gang up against the majority by supporting each other’s causes. Suppose two minority parties
in parliament each have a bill that it will never get approved on its own. By voting for each other’s bill, they
could get both bills passed if the majority party had less than 50% of the votes in parliament. Vote trading
may be bene cial if it leads to the approval of economically viable projects that, together, may otherwise not
have seen the light of day. It could, however, equally lead to the adoption of non-viable projects or to
projects that do not meet with the approval of the majority. While vote trading on the part of minorities does
reveal the intensity of individual preferences, it can either increase or decrease the ability of a majority
voting system to re ect the wishes of the majority accurately.

6.6 Optimal voting rules

Is there anything in between an ordinary majority-voting rule (50%-plus-one-vote) and a 100% unanimity
rule? Is there anything more efficient than ordinary majority voting? e answer can be found in the theory
of optimal voting rules, as propounded by James Buchanan and Gordon Tullock (1962).
eir main hypothesis is that the optimal voting majority varies in accordance with the particular public
issue in question and that these optimal majorities depend on the costs involved in the act of voting. Voters
are faced with two kinds of costs: external costs and decision-making costs.
External costs arise when a community takes a decision that goes against the interest of an individual
voter or group. In other words, the greater the number of people not supporting a public decision, the higher
the external costs will be, or the higher the degree of unhappiness among the voting public will be. e
expected external cost will be highest when public decisions are made by one person – a dictator – since
these decisions will potentially ignore and undermine the interests of all other voters. By contrast, under a
unanimity voting system, where public decisions require 100% support, the external cost can be expected to
be zero. In other words, the closer one comes to unanimity, the smaller the risk of harming minorities.
External costs are inherent in all decision-making rules except the unanimity rule. In Figure 6.2, which
depicts majority size as a function of cost, the external cost curve falls from top left to bottom right.
Decision-making costs refer to the costs involved in persuading voters to support a particular public
issue. e smaller the community of voters, the easier it will be to reach a majority or unanimous decision
and the lower the decision-making costs will be. As unanimity is approached, however, it becomes
increasingly difficult and costly to harness the support necessary to pass a new law or resolution. e reason
is that it takes time and (often) extra resources to debate with and convince more decision-makers to
support a proposal, especially if there are many members of opposition parties whose votes will have to be
obtained. One also nds that the opportunities to act as a free rider increase as the size increases of the
group whose consent is sought. us, as the size of the required support base increases, it becomes
increasingly expensive to induce individuals to reveal their preferences accurately. In Figure 6.2, the
decision-making cost curve rises from left to right as the number of individuals required for a collective
decision increases.

Figure 6.2 Optimal majority and the cost of democratic decision-making

We assume that voters take both types of costs into consideration when casting their votes. However,
since these costs will vary with the particular issue in question, the optimal – that is, cost-minimising –
voting rule will vary likewise. If the two kinds of costs are summed vertically, we obtain a total cost curve, as
shown in Figure 6.2. e lowest point on the total cost curve, coinciding with the percentage given by M*,
determines the optimal majority for the particular public issue in question. ere is likely to be a different set
of cost curves for each issue on which a vote has to be cast.
e characteristics of the optimal majority point are as follows:
• the higher the external cost (curve), ceteris paribus, the greater M* becomes
• the higher the decision-making cost, ceteris paribus, the lower M* will be.

e shape of the curves will differ according to the type of activity that is to be decided on and the optimal
majority need not be an ordinary majority (in other words, 50%-plus-one-vote).
 A factor that in uences both decision-making and external costs is the degree of homogeneity among
the voting community. In a homogeneous community, it is probably easier to achieve a majority outcome or
unanimity since individual tastes are relatively similar and, consequently, decision and external costs are
relatively low. One implication is that a community characterised by sharp differences among individual
citizens and groups may not be able to afford the high decision-making costs involved in near-unanimity
rules for collective choice. However, the potentially adverse consequences – or high external costs – for
minority individuals and groups may be the very factor that will prompt these individuals to press for costly
near-unanimity rules.
e above analysis can be used to determine voting rules for different kinds of collective actions. All of
these actions need not be organised under the same voting rule. Voters, for example, will choose a voting
rule with a high majority requirement for collective actions that incur high external costs (or a major loss of
utility) for them. Conversely, there are collective actions that incur low external costs, which means that a
lower majority-voting rule will probably suffice. However, one should bear in mind that individuals and
groups will normally not deem it in their interests to support a voting rule that may promote the interests of
other groups. e customary rule when important issues such as amendments to a constitution are
concerned is that more than just an ordinary majority is required before these can be executed. Changes to
the Bill of Rights in the South African Constitution, for example, require a two-thirds majority in the National
Assembly.

6.7 Government failure

A necessary condition for the success of both direct and indirect (or regulatory) intervention by government
is the presence of an adequate and efficient institutional framework. Important institutions include the
legislative authority (for example, parliament), the judiciary, law enforcement, tax collection or revenue
services and regulatory bodies. ese institutions should ideally ensure that agents comply with the rule of
law, and honour their contractual and regulatory commitments. A sufficient condition for success refers to
the value system of the community, including behavioural norms and customs, which should ideally
entrench high levels of trust between and among consumers, producers and government institutions
(North, 1991).
Both of these conditions should ultimately ensure that government performs its functions in a
transparent, accountable and consistent manner (Parker, 2002a). But experience shows that such utopian
outcomes are rare, especially in developing countries (Kahn, 2006), owing to what economists refer to as
government failure. As we show below, governments fail because of the ‘rational’ behaviour of politicians
and bureaucrats, and also because of the in uence that special interest groups have over government.

6.7.1 Politicians
As mentioned in Section 6.3, politicians can be viewed as entrepreneurs who engage in vote-maximising
strategies in order to secure and retain political office. It is important to consider the implications for
resource allocation resulting from behaviour of this nature. e likely consequences can be more readily
determined given two further characteristics of the majority-voting rule:
• Voters are rationally ignorant of much of what politicians stand for, since they usually do not have a
sufficient incentive to acquire all of the information necessary to determine the desirability of all of the
relevant public issues.
• Politicians are elected on the basis of a package of policies and therefore do not have to please a majority
of voters on each separate policy issue.

ese characteristics can give rise to implicit logrolling, favouring special interest legislation. We can
illustrate this proposition by means of a hypothetical example, as shown in Table 6.2.
Imagine a politician standing for election for a political party with diverse interests. e politician
explicitly supports three special-interest programmes: the relocation of the South African parliament to
Pretoria, a rugby development programme and a subsidised loan scheme for students. Each of these special
interest programmes is likely to attract strong support from a particular group within the voting population.
Civil servants will strongly support the relocation of parliament, rugby lovers will likewise support the
proposed development programme and students will lend strong support to the proposed subsidy scheme.
But none of the programmes is likely to bene t a majority of the voters. e subsidy scheme, for example,
will not bene t civil servants or rugby lovers directly, or at least not attract their strong support; when faced
with a choice between the three programmes, they may have a relatively weak opposition to the proposed
subsidy. Civil servants and rugby lovers may therefore rationally decide to remain ignorant about the full
cost of the subsidy scheme.
Meanwhile, the politician in question who supports all three programmes has an incentive to make the
bene ts of these policies clear to the three unrelated recipient interest groups in order to form a coalition
through implicit logrolling. He or she knows that the strong preference that civil servants have for relocating
parliament to Pretoria probably outweighs their mild opposition to the student loan scheme and the rugby
development programme: they would rather have all three programmes than none. And the politician can
make sure of this by disguising or understating the cost of each of the programmes to the electorate as a
whole, that is, by creating scal illusion. Table 6.2 illustrates the nature of this implicit logrolling.

Table 6.2 Coalition-forming and implicit logrolling

Policy Strongly favoured by Weakly opposed by

Civil servants Rest of electorate

Relocation of parliament
(33,3%) (66,7%)

Rugby lovers Rest of electorate

Rugby development
(33,3%) (66,7%)

Students Rest of electorate

Student loan scheme
(33,3%) (66,7%)

It is evident that the politician supporting all three minority programmes will defeat an opponent who
opposes them, or who supports only one or two of them, by mobilising the strong preferences of civil
servants, rugby lovers and students by means of implicit logrolling.
Two important consequences for resource allocation ow from this example:
• We can anticipate a preponderance of special interest legislation producing a variety of relatively
unpopular public goods. is is particularly likely if special interest groups with rather obscure or
idiosyncratic programmes – which will on their own and despite very intensive preferences of a small
minority probably never gain a majority vote – need to be courted.
• We can expect an aggregate oversupply of public goods in society.

It is clear that vote-maximising behaviour on the part of politicians can lead to outcomes that are not in
accordance with the wishes of the majority of voters. Some writers, most notably Buchanan and Tullock
(1962), argue that this phenomenon is a consequence of constitutional failure and can only be dealt with by
constitutional reform, for example, by limiting the proportion of scarce resources expended on public goods
to some xed percentage of national income and specifying the distribution of these resources between
alternative kinds of public goods. is represents a form of scal rule that will be discussed in more detail in
Chapter 18.

6.7.2 Bureaucratic failure

A second source of government failure stems from the maximising behaviour of government employees or
so-called bureaucrats. In essence, bureaucratic failure results from rational responses on the part of utility-
maximising civil servants to the incentives presented to them by politicians and the institutional structures
within which they operate.
omas Borcherding (1977: xi) notes that:

Individuals in the bureaucracy, like the rest of us, do react to different incentive schemes; they do have
various preferences, and have the capacity, will and desire to ful l these preferences. ey prefer more
rather than less income, power, prestige, pleasant surroundings and congenial employees.

e rational behaviour of bureaucrats can be analysed in terms of the demand for and supply of public
goods. e demand for public goods in a representative democracy is generated by the decisions of vote-
maximising politicians, while the supply of public goods is usually the responsibility of the state
bureaucracy. Unlike private rms, however, bureaucracies do not maximise pro t. Instead, they receive
annual lump sum payments from the legislature based on estimates – prepared by bureaucrats – of the costs
of providing speci ed (and usually monopolised) public goods. Consequently, bureaucracies do not face
any market test. William Niskanen (1971) argued persuasively that since higher salaries, more power, greater
prestige and other favourable attributes are positively related to bureau size and hence to bureau budgets,
bureaucrats have an incentive to maximise their budgets.

Figure 6.3 Bureaucratic failure

e resultant bias towards the excess provision of public goods is illustrated in Figure 6.3. Part (a) of the
gure shows the total social cost (TSC) and total social bene t (TSB) curves for a public good. e usual
marginal principles apply here: total cost rises at an increasing rate as output expands (owing to the
principle of diminishing marginal productivity), while total bene ts increase at a decreasing rate as output
expands. e rates of change of these curves, or their slopes, determine the shapes of the corresponding
marginal curves shown in Figure 6.3(b).
e socially optimal level of output of the public good is given by 0Q0, where marginal social bene t
(MSB) equals marginal social cost (MSC) (in Figure 6.2(b)) or where the difference between TSB and TSC is
maximised (in Figure 6.2(a)). But a budget-maximising bureaucrat would attempt to justify output 0Q1 > 0Q0,
where TSB equals TSC and where MSC exceeds MSB by the distance FG. e result is that the total value to
consumers increases from 0AEQ0 to 0AFQ1 (in the (b) part) or by the area Q0EFQ1, while total cost increases
from 0BEQ0 to 0BGQ1 or by the larger area Q0EGQ1. e increase in total cost thus exceeds the increase in
total bene ts by the area EGF, which is the net welfare loss to society.
Moreover, given the absence of a pro t motive, bureaucrats may supply public goods inefficiently, for
example, by using more than the required inputs to produce a given level of output. e resultant
inefficiency (or welfare loss) thus stems from an excessive use of inputs rather than from excessive
production by the bureau.
In essence, the problem of bureaucratic failure is simply an example of a principal–agent problem. In
the public sector, bureaucrats act as agents for their principals, the tax-paying public, who are in turn
represented by elected politicians. Since each individual bureaucrat bene ts directly from a large budget, he
or she has a strong preference for a high level of expenditure. Individual taxpayers, on the other hand,
generally bene t only marginally from that expenditure and therefore remain rationally ignorant about it.
Owing to the size and divergent nature of the tax-paying public, there is little individual taxpayers can do to
lobby against higher levels of state spending or in favour of lower taxes. us, there is a poor correlation
between the objectives of agents and the objectives of principals. Put differently, bureaucrats have a greater
incentive to increase spending than taxpayers have to reduce taxes. It follows that one should expect
spending levels on public goods to exceed their corresponding optimal levels.
Although Niskanen’s model of bureaucratic failure provides a plausible explanation for excessive state
intervention in the economy, it can be questioned on several grounds. Is it realistic to assume that
politicians and the tax-paying public are entirely at the mercy of bureaucrats? Budgetary procedures have
become more transparent everywhere, including South Africa and other African countries, while the salaries
and perks of senior public servants are not usually linked to the size of their budgets. A recent trend has been
to determine remuneration in relation to the extent to which pre-set performance goals are met. In a
democratic environment, buttressed by strong institutions, it seems unlikely that bureaucrats will be allowed
to get away with practices that run against the public interest, at least not for long. It is therefore difficult to
con rm or deny the alleged empire-building motives and actions of bureaucrats.

6.7.3 Rent-seeking and corruption

As discussed in Chapter 2, one of the important tasks of a government bureaucracy is to sub-contract, by
tender, the provision of (public) goods and services, legally approved by the legislative authority, or to limit –
through regulation – the production of certain goods and services deemed not to be in the public interest.
Both forms of intervention can give rise to rent-seeking behaviour on the part of special interest groups in
the private sector. is is a type of behaviour whereby interest groups attempt to capture or compete for the
arti cial rent created by government through interventions in the market and is perfectly legal, although it
often adds to social waste attributable to the rent created by government (Buchanan, Tollison & Tullock,
1980).
e concept of economic rent is usually de ned as that part of the reward accruing to resource owners
over and above the payment that the resource would receive in any alternative employment. For example, if
government yields to pressures from domestic producers to place a tariff on imports when competitive
market prices apply, producers get a bene t (extra ‘rent’) at the expense of importers and domestic
consumers who have to pay higher prices for imported products. Rent is similar to monopoly pro t: it
cannot be competed away. In a perfectly competitive world, however, market forces would ensure the
dissipation of rent in a manner that produces socially desirable outcomes. e existence of positive rent in a
competitive market will attract resources in the same way as the existence of (potential) pro ts, and
consequently result in the erosion of this rent through an efficient reallocation of resources. However, once
we adjust the mechanisms through which this process occurs, the consequence of rent-seeking behaviour
can be harmful to society at large.
e theory of rent-seeking deals with the origins of, and competition for, arti cially created rent. e
latter usually results from government-protected monopoly power. Numerous examples abound in Southern
Africa, ranging from quantitative restrictions on licences for hawkers, liquor outlets and taxi drivers, to
qualitative restrictions on purveyors of food and on people wishing to enter occupations such as real estate
sales, IT repairs, legal services and alternative medicine.
e consequences of rent-seeking under competitive conditions may be illustrated by means of a simple
diagrammatic example (see Figure 6.4)
Figure 6.4 shows a typical competitive market characterised by constant returns to scale, shown by the
supply curve S0 and a normal demand curve that yields an equilibrium price 0P0 and quantity 0Q0. e
demand curve is also the average revenue curve. In the absence of government intervention, it would not be
easy for any number of producers to enforce a higher price than where the demand curve intersects the
supply curve, that is, where average revenue (AR) = price (P0). is is at point E0. e reason is that it would
always be possible for any supplier to thwart attempts by other suppliers to extract extra economic rent by
asking a higher price than P0. How can state intervention favour some suppliers? One way is for the state to
limit the supply of the product, which will result in a regulated higher price. Assume that the state intervenes
to limit output to 0Q1. As a result, the price of the good rises to 0P1, causing a loss of consumer surplus equal
to area P0E0E1P1. According to conventional economic theory, the area P0AE1P1 denotes a socially costless
wealth transfer from consumers to producers, while AE0E1 indicates the deadweight welfare loss to society.
However, this may understate the net loss to society, as we explain below.
Under our present assumptions, the area P0AE1P1 is available to potential suppliers to capture in an
attempt to boost their pro ts. ey would thus be prepared to incur additional costs – by lobbying
government, for example – in an effort to capture a share of this total rent. Two possibilities exist here.
If participating suppliers – taxi drivers, for instance – undertook the lobbying function themselves and
the additional costs were internal to them, then the suppliers could, like a cartel, share the higher pro ts
among themselves. But given the lobbying-induced exhaustion of P0AE1P1, the social cost of state
intervention will equal both areas P0AE1P1 (or the social waste from rent-seeking) and AE0E1 (the deadweight
loss). e point is that the taxi drivers will have transferred part of their own resources away from productive
activities in favour of non-productive rent-seeking activities. Similar arguments may also apply to real estate
agents, powerful stakeholders in the shing industry and many other pro t-seeking special interest groups.
Figure 6.4 Rent-seeking

Rent-seeking in the form of lobbying, as explained above, while wasteful, would not necessarily be
against the law and would thus not count as a criminal activity. But the borderline between rent-seeking and
corruption is quite thin, if not blurred. No wonder corruption has been depicted as a special branch of rent-
seeking and described in a rather vague manner as occurring ‘when competition for preferential treatment
is restricted to a few insiders and when rent-seeking expenses are valuable to the recipient’ (Lambsdorff,
2002: 120).
Irrespective, therefore, of who is doing the lobbying – suppliers or concerned citizens – the creation of
rents by government, albeit legal, does establish a potential for corrupt behaviour on the part of bureaucrats.
It is not beyond rent-seekers to offer bribes and kickbacks in an attempt to achieve their objectives more
rapidly. If these offers were accepted by bureaucrats, they would be guilty of committing corruption, which
is, of course, illegal. ey would do so only if there were a net bene t in it for them, in other words, if the
expected gain (of the bribe) exceeded the expected cost of being detected and punished. e fact that
corruption is widely practised in many developing countries (Kahn, 2006; Merriman, 2003) would seem to
point to an inadequate legal system that arguably constitutes the critical core of the institutional framework
referred to above.

6.8 Concluding note

In broad terms, then, the application of the microeconomic paradigm of rational maximisation provides
further insight into the social costs attendant upon policy intervention by the state. e inefficiencies
introduced into the market economy by vote-maximising politicians and budget-maximising bureaucrats
generally take the form of an oversupply of public goods. Bureaucrats also have the responsibility of creating
rents aimed at increasing or reducing the provision of certain goods and services, especially if these actions
are deemed to be in the public interest. But these very actions often induce wasteful rent-seeking on the part
of special interest groups in the private sector, and may also be the source of high levels of corruption if
bureaucrats accept bribes and kickbacks offered by rent-seekers. A necessary condition for avoiding or
minimising government failure seems to be an institutional system that includes a judiciary capable of
applying the rule of law to both public and private citizens. However, this may not be sufficient: individuals
and groups also need to have a natural predilection to behave in an honest and trustworthy manner.

Key concepts
• agenda manipulation (page 105)
• bureaucratic failure (page 110)
• corruption (page 113)
• cycling (page 105)
• decision-making costs (page 106)
• direct democracy (direct democratic dispensation) (page 100)
• economic rent (page 112)
• external costs (page 114)
• government failure (page 108)
• implicit logrolling (page 108)
• impossibility theorem (page 102)
• institutional framework (page 108)
• logrolling or vote trading (page 108)
• majority rule or majority voting (page 100)
• majority-voting rule (page 98)
• median voter theorem (page 100)
• optimal voting rules (page 106)
• preference intensities (page 105)
• principal–agent problem (page 111)
• Rawlsian theory of justice (page 98)
• Rawlsian welfare function (page 99)
• rent-seeking (page 112)
• representative democracy (page 100)
• unanimity rule (page 98)
• voting paradox (page 105)

SUMMARY
• Decisions regarding the levels of taxes and other economic policies are made in the political market by
elected politicians. e social choice rules by which governments make decisions range from the
dictatorial rule, where one person’s preferences determine outcomes, to the unanimity rule, where 100%
support from the electorate is required before a decision is carried.
• e unanimity rule nds expression in the Rawlsian theory of justice, which requires communities to
make decisions behind a ‘veil of ignorance’ and to enter a hypothetical ‘original position’. From the
Rawlsian perspective, all parties would unanimously approve a political constitution where priority is
given to the person potentially nding him- or herself in the worst position. e so-called maximin
strategy protects each person in case he or she ends up being in that position. One objection to the
unanimity rule is that it effectively gives minorities veto power.
• e most common voting rule is the simple majority rule, which requires the support of 50%-plus-one-
vote in order for a decision to be accepted. In a representative democracy, where voters elect
representatives to make decisions on their behalf, the preferences of the median voter (or group of
voters) may determine voting outcomes. Although the simple majority-voting rule may have advantages
in comparison to the unanimity rule, it also has shortcomings. According to Arrow’s impossibility
 theorem, it can lead to logically inconsistent results and it does not account for differences in the
intensities of voters’ preferences.
• Buchanan and Tullock (1962) proposed an optimal voting rule theory that can be applied to different
kinds of collective decisions. e optimal majority is determined by two kinds of costs: external costs
(that decreases as percentage of votes required increases) and decision-making costs (that increases as
the percentage of votes required increases). Where the sum of these costs is a minimum, the optimal
voting percentage is obtained.
• e perfectly competitive market system fails in a number of respects and consequently requires some
form of government intervention to correct for these failures. Likewise, in the political market,
government fails to provide adequate and efficient institutions as well as public goods and services.
Government failure is ascribed to the rational behaviour of politicians, bureaucrats and special interest
groups. Politicians seek to maximise votes and do not necessarily promote the interest of voters.
Bureaucrats try to maximise their budgets, often leading to inefficient levels of public goods provision.
Special interest groups seek to maximise their own interests and engage in rent-seeking, which again
leads to social waste. Rent-seeking often results in corrupt behaviour on the part of bureaucrats and rent-
seekers.

MULTIPLE-CHOICE QUESTIONS
6.1 e simple majority voting rule …
a. requires 50% of the votes for a decision to carry.
b. implies that voters take decisions behind a veil of ignorance.
c. is a less costly voting rule than the unanimity rule for reaching majority approval.
d. is logically inconsistent when voter preferences are extreme.
6.2 According to Niskanen’s bureaucratic failure module …
a. bureaucrats maximise pro ts when supplying public goods.
b. bureaucrats’ salaries, bonuses, prestige and power are linked to the size of their departmental
budgets.
c. bureaucrats determine the quantity of public goods by setting total social costs equal to total social
bene ts.
d. bureaucrats maximise their budgets by setting marginal social bene ts equal to marginal social
costs.
6.3 Government may fail to be efficient because …
a. taxpayers generally avoid paying taxes to nance high government expenditures.
b. bureaucrats have more of an incentive to increase government spending than taxpayers have to
lobby for reduced taxes.
c. politicians tend to exploit the ignorance of voters.
d. bureaucrats try to use only a few inputs to obtain the maximum level of services.
6.4 Refer to Figure 6.4. Which of these statements is correct?
a. As a result of rent-seeking, the consumer surplus increases to BP0E0.
b. Rent-seeking behaviour will result in the price decreasing to P0.
c. Rent-seeking results in spending on non-productive activities (for example, bribes), causing the
consumer surplus to increase from BP1E1 to BP0E0.
d. e deadweight welfare loss caused by rent-seeking is equal to the area E1E0A.

SHORT-ANSWER QUESTIONS
6.1 Describe the unanimity voting rule, and compare its advantages and shortcomings.
6.2 Distinguish between a direct democracy and a representative democracy.
6.3 De ne the following concepts:
 a. Vote trading
b. Median voter
c. e voting paradox
d. Rent-seeking
6.4 In what way could the behaviour of bureaucrats cause government failure?

ESSAY QUESTIONS
6.1 Discuss the Rawlsian theory of justice and brie y comment on its relevance to the political economy of
South Africa.
6.2 Outline the median voter theorem and explain its importance to the successful application of a majority
voting system.
6.3 Do you agree with the contention that majority rule does not necessarily produce outcomes
representative of the majority view? Discuss with reference to the impossibility theorem and the
phenomenon of vote trading.
6.4 Discuss the theory of optimal voting rules and brie y comment on its relevance to the social choice rule
of majority voting.
6.5 Explain how the maximising behaviour of politicians could contribute to an oversupply of public goods.
6.6 Explain the meaning of the term ‘rent-seeking’ and show how it could undermine efficiency in the
economy.

1 e idea of this example comes from Hyman (1999: 179–180).

2 Suppose there are three candidates for a position. Pair-wise voting entails that the rst vote is between the rst two candidates. e
one who wins is then paired with the other candidate in the second election. If the third candidate wins the second election, a third
election will be required between the winner of the second election and the loser of the rst election to determine the ultimate
winner.
3 In the example of extreme preferences given above, the result is quite confusing if the voter is asked to vote in a pair-wise manner.
If the rst vote is between large and small, small wins. In small versus moderate, moderate wins. In the nal vote between
moderate and large, large now beats the candidate that conquered its own superior.
Public expenditure and growth

Philip Black, Krige Siebrits and eo van der Merwe

Part 1 of this book explained in theoretical terms why governments should mobilise and spend scarce
resources in a typical market-oriented economy. In this chapter, we turn to real-world aspects of
government expenditure in South Africa. We want to nd answers to the following pertinent questions: What
does the Constitution have to say about government expenditure? How much money is spent annually by
government entities? What forms does this spending take and what services are provided? How has the
composition of government expenditure changed over time and how do these trends compare with trends in
other countries? Why has government expenditure grown so much over time? And, most importantly, how
does government expenditure affect the economy over the medium to long term?
As indicated in Chapter 1, these questions are particularly topical in view of the economic challenges
facing South Africa and the global context within which they have to be addressed. e levels of government
expenditure, revenue and the de cit before borrowing also in uence the decisions of rating agencies and
investors, especially when they consider the economic outlook of a country. In this chapter, we rst look at
the growth of government expenditure and at the changes in its composition, both in South Africa and
elsewhere in the world, before we discuss some of the theories explaining these changes. In the last section,
we turn our attention to the long-term effects of government activity on the economy.

Once you have studied this chapter, you should be able to:
comment on the implications that the Constitution has for government expenditure in South Africa
discuss salient trends in the size, growth and composition of government expenditure in South Africa
identify the main similarities and differences between government expenditure patterns in South Africa and
other countries
compare and contrast two or more of the theories that explain the growth of the government sector and
indicate whether they have any relevance for South Africa
consider the long-term effects of government spending in terms of new growth theory
discuss the role of infrastructure investment in the economy.

7.1 The constitutional framework

e Constitution is the supreme law of the Republic of South Africa and provides important guidelines on
how South Africans should conduct themselves in the political and economic spheres. Hence, its provisions
for taxation and government expenditure are the basic paradigm within which the government formulates
its budgetary policies. is highlights why state capture and corruption are such serious issues, since it
jeopardises and contradicts the intentions of the constitution when individuals and certain groupings
misuse state funds to their own personal advantages, at the expense of those individuals who are targeted in
the Constitution to receive government services.

7.1.1 The Constitution and public goods

e South African Constitution contains many provisions that affect the extent and composition of
government expenditure directly or indirectly. At a very general level, these provisions depend on how the
government sector and its primary functions are de ned in the Constitution. Government functions are
derived from, and structured according to, the constitutional distinction between the legislative, executive
and judicial branches of government; the national, provincial and local levels of government; the security
services and certain constitutional entitlements (see Section 7.1.2) and statutory bodies such as the Public
Protector, the Human Rights Commission, the Auditor-General and the Independent Electoral Commission.
By granting powers and assigning functions to institutions of this nature, the Constitution implicitly charges
government with the task of maintaining them and providing for the necessary public funding. Failing to do
so would indeed be unconstitutional. It is therefore very important that the constitution be respected and
protected by South Africans and that it should not be changed at any whim.
In addition, the government of the day is constitutionally obliged to provide or extend the provision of
speci ed basic goods and services. e clearest examples of these provisions are found in the Bill of Rights,
which entrenches the right of each citizen to adequate housing, healthcare, food, water, social security and
education. However, Section 26(2) of the Constitution of the Republic of South Africa, No. 108 of 1996, does
recognise the need for adequate resources in this regard: ‘… the state should take reasonable legislative and
other measures, within its available resources, to achieve the progressive realisation of this right’ (italics
added).
It is worth noting that these rights generally pertain to mixed and merit goods, rather than only to pure
public goods (see Chapter 3), which partly con rms our earlier point that pure public goods are extremely
rare in practice. It is also indicative of modern thinking about the relative importance of the public sector in
promoting sustained economic growth. We return to this point in Section 7.6 of this chapter.

7.1.2 Constitutional entitlements

e rights to certain goods and services conferred by the Constitution could be regarded as constitutional
entitlements. What are the practical implications of these entitlements for the way in which government
manages its own budget? As we have pointed out, the wording of these entitlements in the Constitution does
acknowledge that government is subject to budgetary constraints, but many economists feel uneasy about
the vagueness of phrases such as ‘reasonable measures’ and ‘within its available resources’. ey are of the
opinion that this wording may provide government with too much discretionary power and may lead to
excessive expenditure, which could threaten the macroeconomic sustainability of scal policy. Experts in
human rights law also differ in their opinions on how enforceable these rights are in practice. To what extent
is it realistic to keep the government of a developing country like South Africa responsible for the provision
of housing, social security and other basic services to all of its citizens? Which of these rights should be
accorded priority when trade-offs become necessary?
From time to time, the Constitutional Court has to respond to some of these very difficult questions. In
an early case of this nature, the Court ruled in November 1997 that a South African kidney patient was not
entitled to receive expensive treatment at the state’s expense. Two constitutional principles were involved in
this case: the right to live and the right to have access to healthcare services. One of the grounds for the
ruling was that the latter right is subject to the ‘availability of resources’. In a number of subsequent cases, the
Constitutional Court also refrained from granting individual claims and rather took the view that the
government should adopt a comprehensive plan that provides in a reasonable manner for the progressive
realisation of the rights in the Constitution.
From a scal point of view, a strong case can be made for a ruling that the government’s obligations to its
citizens should be extended to also include future generations. is means that any current attempt at
ful lling these obligations should take full account of the impact it is likely to have on the future growth of
GDP, and hence on the growth of government revenue, since the latter is clearly a necessary condition for
maintaining the supply of public services over time. At the same time, however, it can be argued that the
future growth of GDP depends on the current provision of services such as education and healthcare. e
matter is therefore far from simple. A steady and consistent supply of these services whereby backlogs and
future demands are met within reasonable timeframes may help to resolve the con ict between
constitutional entitlements and macroeconomic affordability. In this regard, medium-term expenditure
frameworks may ful l an important role, as we shall see in Chapter 18.

7.2 The size, growth and composition of public expenditure

In Chapter 1, we provided some indication of the size of the public sector in South Africa. Table 1.1 showed
that the public sector has expanded considerably during the past few decades. In this Section, we take a
closer look at trends in general government expenditure since 1992. As we explained in Section 1.5.1 of
Chapter 1, general government expenditure includes the outlays of the national government, provincial
governments, local authorities and extra-budgetary institutions, but excludes the spending of public
corporations such as Eskom, Telkom and Transnet.

7.2.1 Size and growth of public expenditure

In this section, we provide data for two measures of general government expenditure: the total amount of
resources used and the total amount of resources mobilised (the concepts of resource use and resource
mobilisation are explained in Section 1.5.2 of Chapter 1).1
We rst look at the size and growth of general government resource use at constant 2016 prices.2 is
measure suggests that general government expenditure grew signi cantly in South Africa after 1990: real
resource use increased from R376,4 billion in 1992 to R1273 billion in 2017. On average, real resource use
therefore grew at a relatively high rate of 5% per annum. Table 1.1 in Chapter 1 indicates that resource
mobilisation by the public sector (which includes public corporations) have increased sharply from an
average of 24,1% of GDP for the period 1960–1969 to an average of 39,8% for the period 2008–2017.
How did this pattern of growth affect the size of the general government sector relative to that of the
South African economy as a whole? Figure 7.1 answers this question by showing general government
resource use and resource mobilisation as percentages of the gross domestic product (GDP) at current
market prices from 1992 to 2017. Over the period as a whole, general government expenditure clearly grew
signi cantly in relative terms: resource use increased from 21,1% of GDP to 27,4% and resource mobilisation
grew from 28,2% to 36,8%. It is also evident from the gure that the share of general government expenditure
has decreased for a rather brief period in the mid-1990s and again in the early 2000s, due to scal restraint
and the relatively good economic growth experienced during these periods. e graph also indicates that
government spending is increasing steadily as a share of GDP, which of course implies that the share of the
private sector of the economy is shrinking.
To summarise: general government expenditure grew signi cantly in South Africa from 1992 to 2017,
both in real terms and as a percentage of GDP. us, in 2017, the general government sector was responsible
for a markedly higher portion of aggregate expenditure in South Africa than was the case in 1992. In Sections
7.4 and 7.5 of this chapter, we outline various theories that try to explain why government spending grows in
absolute terms and/or relative to the rest of the economy. Chapter 18 explains why the growth rate of general
government spending slowed so markedly in South Africa until 2001, before accelerating to growth rates last
seen before the start of the democratic dispensation.
Figure 7.1 The size and growth of general government expenditure in South Africa

Source: Calculated from South African Reserve Bank, Quarterly Bulletin (various issues). Pretoria: South African Reserve
Bank [Online]. Available: https://www.resbank.co.za/Publications/QuarterlyBulletins/Pages/.

Signi cant changes in the composition of general government expenditure accompanied these trends in
its size and growth. We now provide a synopsis of these changes.

7.2.2 The changing economic composition of public expenditure

Table 7.1 summarises trends in the economic classi cation of government expenditure, which
distinguishes between the current and capital components of total outlays. e various expenditure
categories are expressed as percentages of total general government expenditure and of GDP in the scal
years 1994, 2006 and 2017.3
In 2017, cash payments for operating activities (formerly known as current expenditure) accounted for
92,8% of general government spending in South Africa and purchases of non- nancial assets (outlays for the
acquisition of xed capital assets, land, stock and intangible assets, which was formerly known as
investment) for the remaining 7,2%. In descending order, the largest categories of cash payments for
operating activities were compensation of employees (the government’s wage bill), purchases of goods and
services (for example, stationery, computers, vehicles and maintenance of capital assets), social bene ts
paid (mainly welfare payments such as old-age pensions or child-support grants), interest on public debt
and other payments.
Table 7.1 shows that cash payments for operating activities (in other words, current outlays) increased
slightly from 92,2% of total cash payments in 1994 to 92,8% in 2017. Over the same period, purchases of non-
nancial assets (in other words, capital spending) decreased from 7,8% of total cash payments to 7,2%. is
represented a continuation of a long-term trend in the expenditure of the public sector highlighted in
Chapter 1 of this book: From Table 1.1 it can be deduced that current expenditure by the public sector (the
sum of consumption expenditure and transfer payments) increased from an average of 14,5% of GDP in the
period 1960–1969 to an average of 32,6% for the period 2008–2017, while investment fell rather sharply from
an average of 9,6% of GDP in the period 1960–1969 to 7,2% in the period 2008–2017. e economic effects of
government investment depend not only on its extent but also on its nature: in the long run, the
construction of power stations and roads, for example, is clearly more important for the promotion of
economic growth than the construction of new buildings for government departments. Nonetheless, most
economists believe that the decline in general government spending on economic and social infrastructure
has hindered economic growth in South Africa. Despite the fact that the purchases of non- nancial assets by
the general government have decreased from 7,8% of total cash payments in 1994 to 7,2% in 2017, its share of
GDP has increased marginally from 2,4% of GDP to 2,6%. Investment by government is therefore still very
important even though it may not be addressing the most important needs of the economy sufficiently.

Table 7.1 The economic composition of general government expenditure in South Africa, 1994–2017a

Notes: a Government nance statistics data, scal years ending 31 March

b Figures may not add up owing to rounding.

Source: Calculated from South African Reserve Bank, Quarterly Bulletin (various issues). Pretoria: South African Reserve
Bank. [Online]. Available: https://www.resbank.co.za/Publications/QuarterlyBulletins/Pages/QuarterlyBulletins-
Home.aspx [Accessed 16 July, 2018].

From 1994 to 2017, total cash payments of the general government increased its share of GDP from 31,4% to
36,8%. is was as a result of an increase in the share of GDP of all economic categories of government
spending, with the exception of interest on public debt and other payments. e decrease in interest on
public debt can be attributed to scal prudence up to the Global Financial Crisis of 2008 and to relatively low
interest rates in the latter period.4 Hence, it could be argued that government has been losing its grip on
scal discipline since 2007. From a scal policy perspective economic stimulation was required during the
crisis, but unfortunately no return to previous levels of government expenditure occurred afterwards (see
Section 7.4.2 for an explanation of this phenomenon). e increases in the de cit before borrowing during
and after the crisis are responsible for the rise of the share of interest on public debt from 2,9% of GDP in
2006 to 3,4% in 2017. is increase follows on the decrease from 4,6% in 1994 to 2,9% in 2006.
 e GDP share of compensation of government employees has also increased from 10,6% in 1994 to
13,1% in 2017, indicating a sharp rise in the pay packages of civil servants accompanied by an increase in
employment by government, while many of the other sectors of the economy were shedding employment
opportunities. Efforts to restrain the growth of the wage bill at times led to con ict between the government
and public sector labour unions. e reality that total general government cash payments increased by 5,4
percentage points of GDP for the period under consideration meant that the increases in the total spending
shares of the other expenditure components (purchases of goods and services, subsidies and social bene ts)
also represented growth relative to the size of the economy as a whole.

Table 7.2 The functional composition of general government expenditure in South Africa, 1994–2016a

Notes: a Government nance statistics data, scal years ending 31 March

b Figures may not add up owing to rounding.

Source: Calculated from South African Reserve Bank, Quarterly Bulletin (various issues). Pretoria: South African Reserve
Bank. [Online]. Available: https://www.resbank.co.za/Publications/QuarterlyBulletins/Pages/QuarterlyBulletins-
Home.aspx [Accessed 18 July 2018].

7.2.3 Functional shifts in public expenditure

e functional classi cation of government expenditure refers to the amounts spent on the different
goods and services provided by government. Table 7.2 contains data on the functional composition of
general government spending in the scal years 1994, 2005 and 2016 expressed as percentages of the total
and GDP.5
 In 2016, social services accounted for 50,6% of total general government expenditure in South Africa,
making it by far the largest functional category. Social services expenditure includes spending on education,
health, social protection, housing and community services, and recreation, culture and religion. e second
and third largest functional categories were general public services (interest payments and other public debt
transactions as well as the outlays of departments such as the Presidency, Foreign Affairs, Public Works, and
Cooperative Governance and Traditional Affairs, including transfers from the central government to local
authorities) and protection services (which consists of government spending on defence, policing,
correctional services and the justice system). e other expenditure category, economic services,
encompasses the outlays of government departments that oversee and assist various sectors of the economy
(such as agriculture, mining, manufacturing and transport) as well as those that are charged with cross-
sectoral issues such as labour and technology. Table 7.2 also shows that environmental protection is
becoming more important, although spending on this function remains at a low level.
After the 1994 elections South Africa experienced relatively good economic growth and government also
followed a policy of scal restraint, but this good fortune changed after the Global Financial Crisis of 2008
that required scal stimulus. e situation was exacerbated by the corruption, state capture and patronage
of the Zuma administration that led to great economic uncertainty and the ring of the minister of nance,
Mr Nhlanhla Nene. e mining charter, land expropriation without compensation, and the possible change
of the Constitution, are also not contributing towards certainty and an environment conducive for economic
growth. e economic situation is aggravated by international uncertainty (possible trade wars) and Brexit.
It is therefore not surprising that South Africa experienced a deterioration in economic growth and relatively
sharp rises in the share of some functions of government expenditure expressed as percentages of GDP.
Expenditure on social services has for example increased from 14,8% of GDP in 1994 to 17,6% in 2016.
e increase in the share of education spending-to-GDP ratio raises concerns of whether South Africans
are receiving value for their money in the light of the large number of learners leaving school early and the
limited mathematical and language competencies of some school leavers. Education is the largest
expenditure item of social services with a share of 6,6% of GDP in 2016. Additional pressure will also be
added to this expenditure item due to government’s commitment to provide free tertiary education to needy
individuals from 2018. Social protection, which consists mainly of social assistance (grants) is the second
largest expenditure item within social services and in 2016 its share of GDP was 4,7%. According to the
National Treasury (2018a: 61) the number of bene ciaries will reach 18,1 million in the 2020/21 scal year.
is implies that more or less one out of every three South Africans will be receiving grants.
From 2005 to 2016, the other functional categories of government experienced a decrease in their GDP
shares. e net results of these developments have been that general government spending is now markedly
more oriented towards the delivery of social services and public order and safety. Simultaneously, the
priority given to defence against external threats has clearly declined. Low economic growth levels put the
ability of government to generate revenue under pressure and add additional pressure on government to
decrease expenditure. e authorities also need to turn their attention to serious initiatives aimed at
improving the efficiency and effectiveness of government programmes.
e links between trends in the economic and functional compositions of total outlays are now apparent.
ere is a close relationship between rates of growth in labour-intensive and supplies-intensive functions,
such as public order and safety and education, and rates of growth in the government wage bill. Similarly,
the extent of growth in transfers to households is closely associated with the importance assigned to social
protection services. Changes in the government spending shares of subsidies and purchases of non-
nancial assets are major determinants of rates of growth in outlays on economic services.

7.3 Comparisons with other countries

For much of the twentieth century, increases in per capita incomes were accompanied by absolute and
relative growth in government spending. One time-series study (Tanzi & Schuknecht, 1995) found that
general government expenditure in the industrialised countries increased from levels of 10% of GDP or less
in 1870 to between 45% and 50% of GDP in 1994. Government spending also increased markedly in most
developing countries in the 1960s, 1970s and 1980s (Lindauer & Velenchik, 1992). We discuss some of the
theoretical explanations for this growth in government expenditure in the next two sections.
By contrast, the last quarter of the twentieth century was characterised by a fundamental reconsideration
of the economic role of government. Various factors gave rise to this development, including the collapse of
the controlled economies of the former Soviet Union, Central and Eastern Europe, and parts of the
developing world (notably sub-Saharan Africa). Mounting evidence has suggested that the expansion of the
public sector beyond a certain point is more likely to contribute to macroeconomic problems, as well as the
erosion of economic incentives and personal freedoms, than to meaningful improvement in the values of
indicators of social and economic progress. Halting or reversing the expansion of the public sector became a
policy objective in many countries, and empirical studies (for example, Fan and Rao, 2003; Tanzi, 2005)
show that government spending did in fact decrease as a percentage of GDP in several of them. From the
Global Financial Crisis of 2008 onwards, many governments launched scal stimulus packages to counteract
the effects of severe economic downturns on production, demand and employment. Most of these packages
included public spending programmes that raised government expenditure-to-GDP ratios. It remains to be
seen if these increases will be reversed as planned (the intention usually was that stimulus packages should
be of a temporary nature) or provide more longer-lasting boosts to public spending levels via the
displacement effect discussed in Section 7.4.2. Such an outcome could even turn out to be a recent
manifestation of the ratchet effect that caused Keynesian scal activism to lose its appeal during the second
half of the previous century (see Chapter 18, Section 18.3.2).
Van der Berg (1991) summarised the international cross-section evidence on the composition of
government expenditure at different levels of development as follows:
• In low-income countries, the bulk of government spending is typically directed at capital investment in
the infrastructure, stimulation of industrial development through export subsidies and other incentives,
and the establishment of primary education and healthcare systems.
• Middle-income countries give priority to education, healthcare, and research and development, and
usually also start to develop a social security system.
• High-income countries are characterised by huge increases in the share of transfer payments (especially
social security-related ones), which are typically compensated for by reduced public investment.

Economic development is thus associated with a shift in the economic composition of government
expenditure from capital expenditure to consumption spending, including transfer payments. e
functional counterpart of this trend is a shift from protection and economic services to social services (see
the discussion in Section 7.4.1). In an authoritative study on the issue, Saunders and Klau (1985: 16) reached
the following conclusion:

e structure of government expenditure has thus shifted away from the provision of more traditional
collective goods (defence, public administration and economic services) towards those associated with the
growth of the welfare state (education, health and income maintenance), which provide bene ts on an
individual rather than collective basis and where redistributive objectives are more dominant.

e longer-run pattern of change in government outlays in South Africa broadly corresponds with these
international trends. As indicated, South Africa has also experienced a shift from capital expenditure and
economic services towards consumption spending and social services. ese compositional trends have
obvious distributional effects, but recent advances in the theory of economic growth suggest that
government spending on education and health also brings efficiency gains in the form of higher levels of
human capital (see Section 7.6).
However, there are some interesting peculiarities to the evolution of government spending in South
Africa. ese include the following:6
• e shift in the economic composition of expenditure from capital to current outlays has left South Africa
with a relatively high level of government consumption spending. In 2016, the ratio of general
government .consumption to GDP was 20,8% in South Africa compared to the world average of 17%, and
 the net investment in non nancial assets by the general government in South Africa accounted for 0,9%
of GDP compared to a world average of 1,3%.
• As far as the functional composition of expenditure is concerned, public spending on education is also
relatively high in South Africa. In 2016, public spending on education in South Africa amounted to 6,6%
of GDP. e ratio exceeded the world average in 2014 of 4,9% of GDP. South Africa’s level of public
spending on healthcare in 2015 (8,2% of GDP) was below the world average of 9,9%. It must be kept in
mind, however, that in high-income countries public spending on healthcare is signi cantly higher than
in middle-income countries such as South Africa. It is also notable that South Africa is spending a
relatively small portion of its national income on defence. In 2016, this spending amounted to 1% of GDP,
which was signi cantly lower than the average of 2,2% for all countries.

Hence, international comparisons tentatively suggest that the scope for further reallocation of resources
from capital to current spending and from protection and economic services to social services in South
Africa is limited. Before further increases in social spending are considered, the interest on the public debt
will have to be reduced and the crime situation be brought under control. However, it is questionable
whether South Africa needs to invest even larger portions of its national income in social services.
Government spending on these items is already relatively high by international standards, and the real
challenge for the future is for government to utilise its existing expenditure more efficiently than before and
to eliminate corruption and state capture. Future increases in expenditure could perhaps more
appropriately be achieved through additional revenues obtained from sustained real growth in per capita
income, as illustrated in Box 7.1.

Box 7.1 Why economic growth is crucial for expanding service delivery

The relative effects on per capita government expenditure of sustained economic growth can be illustrated with
the aid of an example. Let us compare public spending on education in South Africa and Japan (the data are
from World Bank, 2018b). In 2014, the ratios of public expenditure on education to GDP were 6,0% for South
Africa and 3,6% for Japan. South Africa’s GDP per capita was US$6 161 and that of Japan was US$38 109. In
Japan, the public sector therefore expended US$1 372 per person on education, compared to South Africa’s
US$370. To achieve the same level of public spending on education in per capita terms, South Africa would
have been required to spend almost 22% of its national income on education. Since the situation is much the
same for other government functions such as healthcare, social security and many other services, it is clear
that South Africa (and other developing countries) cannot reach the same per capita levels of government
spending by simply increasing the portion of national income devoted to spending of this nature.
That would require extremely high tax rates or unsustainable levels of borrowing. To achieve higher levels of
government spending in per capita terms, it is clearly more important to raise per capita income than to
increase the share of national income devoted to government social spending. Note, however, that per capita
expenditure only measures educational outputs. Unfortunately, education outcomes in South Africa lag behind
those of some other African countries with lower levels of per capita expenditure on education (see Section
10.4.2 in Chapter 10).

7.4 Reasons for the growth of government: Macro models

In Chapter 1, we pointed out that there are different ways of measuring the size of the government; one is to
express government expenditure as a percentage of gross domestic product. From the information provided
for South Africa at the beginning of this chapter, it appears that the share of government in terms of both use
and mobilisation of national resources had steadily increased over the period under consideration, with
relatively short periods of decrease in the mid 1990s and early 2000s.
South Africa is not unique in this regard. Both industrial and developing countries have experienced an
increase in the share of government in the economy, and various theories have been developed in an
attempt to explain this global phenomenon. Before we discuss these theories, two important quali cations
should be made. Firstly, government expenditure growth is not necessarily the same as a growing share of
government in the economy. Some of the theories (for example, that of Brown and Jackson, which is
discussed in Section 7.5.2) provide an explanation for the phenomenon of expenditure growth, irrespective
of whether it is accompanied by an increasing share. Most of the theories under discussion, however, deal
with the issue of an expanding government sector relative to the rest of the economy.
Secondly, it is important to distinguish between the empirical issue of expenditure growth and the
reasons for it, on the one hand, and the normative question of what the appropriate size of government
should be. e question of whether government is too big or too small is clearly a complex if not ambiguous
one: the answer is bound to vary among different countries and cultures, and, for any given country, is also
likely to vary over time, as the sections below explain. Following Brown and Jackson (1990: 120), we
distinguish between macro models of government expenditure growth, which tend to explain the broad
patterns of government expenditure with regard to aggregate variables such as GDP (Section 7.4), and micro
models of government expenditure growth, which focus on the decision-making behaviour of public
individuals and institutions (Section 7.5).

7.4.1 Wagner and the stages-of-development approach

Both Adolph Wagner (1883) and the subsequent stages-of-development model, developed by Musgrave
(1969) and Rostow (1971), explain how government expenditure tends to increase when a country develops
from a subsistence or traditional economy to an industrialised economy. Wagner emphasised the need for
government to create and maintain internal and external law and order during the rst or early stages of
development, and also to set in place the critical legal and administrative institutions necessary to cut the
costs of doing business. According to the stages-of-development approach, it is also important for
government to get investment going in the rst place. During the rst stage, the formal sector of the
economy is still relatively small and, as a result, government may have to participate actively by providing
the basic infrastructure (for example, roads, railways and harbours) necessary to create an environment
conducive to economic development. e implication of the rst stage for government expenditure is that
capital expenditures will feature prominently.
During the middle stages of development, government will continue to supply investment goods while
private investment will also start to take off, partly as a result of the positive pecuniary external effects of
government investment undertaken during the rst stage (see Section 7.6). However, the development of the
private sector may cause certain market failures, including externalities, monopoly pricing and high-density
living conditions associated with growing urbanisation, that government would have to address, thereby
giving rise to further increases in government expenditure. Chapters 2, 3 and 4 discussed such market
failures.
In the last stage of development, capital expenditure by government, expressed as a percentage of GDP,
usually decreases because most of the necessary infrastructure is already in place. At this stage, however,
expenditure on education, health, welfare programmes and social security will tend to increase owing to the
high income elasticity of demand for expenditures of this nature. e result is a continuous increase in the
share of government in the economy.
During the early stages of South Africa’s industrialisation, the government (through public enterprises
and corporations) was heavily involved in major capital projects such as railways, harbours, roads and
public works, as well as in (state-owned) enterprise activities such as the creation of Sasol and Iscor, both of
which were subsequently privatised. e growth in public investment expenditure between the 1930s and
1960s, for example, corresponds with the rst and middle stages discussed above. e subsequent decline in
public investment is not, however, an indication that South Africa had entered the last stage of development:
it is perhaps better understood in terms of Peacock and Wiseman’s displacement effect or in terms of
Meltzer and Richard’s hypothesis (see the next two sections). Note that many rural areas in the country can
hardly be said to have entered the rst stage of development in the sense that public investment has scarcely
begun in these areas.
7.4.2 Peacock and Wiseman’s displacement effect
Peacock and Wiseman (1967) used a political theory to explain the in uence of political events on public
expenditure. ey acknowledged a point made by Wagner, that is, that ‘… government expenditure depends
broadly on revenues raised by taxation’ (Peacock & Wiseman, 1967: 26). Governments would therefore be in
a position to continue increasing their own expenditures and expanding their role in the economy, provided
their economies continued to grow through industrialisation. On the other hand, individuals may not be
prepared to continue paying higher taxes in order to nance this increased expenditure. In a democracy,
government has to respect the wishes of the majority (50% of voters plus one). Under normal circumstances,
government expenditure would therefore only increase when it is strictly necessary and it can be expected
that governments would take the possible resistance of taxpayers against higher tax rates into account.
However, social upheavals or disturbances may change the established conceptions of the public.
Examples of upheavals of this nature are the two world wars, the Great Depression of the 1930s and the
recent Global Financial Crisis. National and international crises of this magnitude may cause a rapid
increase in government expenditure, since they may convince taxpayers that higher taxes are necessary to
prevent a disaster. Peacock and Wiseman called this phenomenon the displacement effect, as certain
government expenditures (for example, war-related expenses) displace private expenditures as well as other
government expenditures (such as non-war-related expenses).
After a crisis had subsided, government expenditure could be expected to return to its pre-crisis level.
According to Peacock and Wiseman (1967: 27), however, this was unlikely and government expenditure
could even remain at the new post-crisis level, the reason being that taxpayers would become accustomed to
the higher levels of taxation and expenditure and accept them as a part of life.
e displacement effect described by Peacock and Wiseman may help to explain the growth in
government expenditure in South Africa. e previous political dispensation in South Africa resulted in a
massive military build-up and incursions into neighbouring countries amidst widespread social unrest
inside the country, especially post-1976. ese crises triggered rapid increases in expenditures on protection
services (defence and police) and also on social services. Expenditure on protection services at one stage
(1983) reached almost 20% of the national budget. Also, the share in total expenditure of education started to
increase as a result of the unrest in schools in 1976. It is, therefore, possible to argue that both social
upheaval and the con ict situation on the borders contributed to the growth in government expenditure in
South Africa during the period up to 1994. However, one may also argue that displacement pressures in
South Africa were dampened by a redirecting of government expenditure (for example, from economic
services to social services). Lusinyan and ornton (2007) concluded that there is some evidence that the
displacement effect did in fact occur in South Africa in the period following 1960.

7.4.3 The Meltzer-Richard hypothesis

Redistributive policies may have an important impact on the growth of government expenditure. Meltzer
and Richard (1981) developed a general equilibrium model in which majority voting determines the
magnitude of income distribution and, as a result, also the share of government expenditure in the
economy. According to this model, the most important reason for the increase in government expenditure is
an extension of the franchise, which brings about a change in the median (or the decisive) voter (see the
discussion of the median voter theorem in Chapter 6).
Meltzer and Richard (1981: 43) argued that the median voter plays an important role in determining the
size of the government sector in a democracy. For example, if all voters were ordered from left to right
according to their income, with the individual with the lowest income on the far left-hand side and the
individual with the highest income on the far right-hand side, the median voter would be the one right in the
middle. erefore, if there were, for example, only ve voters in a country, with incomes of R3 000, R4 000, R6
000, R15 000 and R25 000 per annum, the median voter would be the individual with an income of R6 000.
e median voter is important because in a two-party democracy, he or she will determine which party will
win the election. Both parties therefore try to gain the support of the median voter and will need the support
of three individuals (including the median voter) for this purpose.
 According to the model of Meltzer and Richard, there will be pressure for redistributing income if the
income of the median voter lies below the average income. Redistribution would bene t the median voter
and he or she will therefore vote for the party that proposes a programme of redistribution. In our example,
the income of the median voter (R6 000) lies below the average income (R10 600). To gain the support of the
median voter (and the majority), parties will emphasise redistributive policies that could result in higher
taxes and higher expenditure on social services. e median voter effectively determines the tax level.
However, Meltzer and Richard’s model (1981: 916) does not allow for unlimited redistribution because
they assume that voters are aware of the disincentive effects associated with high taxes and redistribution.
e rational median voter will thus choose the tax rate that maximises his or her utility, while taking into
account the possible impact on the economic behaviour of other individuals (in other words, their decisions
to work, consume and invest). If the median voter chooses a relatively high tax rate, other individuals may
decide to work less (consume more leisure), with the result that the economic pie may become smaller.
Following this, the scal scope for redistribution may also become smaller.
e hypothesis would suggest that the extension of suffrage (which accompanied democratisation in
South Africa) should have resulted in a major increase in the share of government expenditure in the South
African economy. But this did not happen, despite the fact that the election of 1994 resulted in the
appearance of a new median voter with an income well below the average income. It may be argued, of
course, that the median voter model does not apply to South Africa or that a substantial degree of
redistribution had already occurred earlier in an attempt to counter social upheaval (as already explained).
Expenditure on social services increased between 1994 and 2016 from 45,3% to 50,6% of total general
government expenditure and from 14.8% to 17,6% of GDP (Table 7.2), while defence expenditure declined
dramatically both prior to and after the 1994 elections. Also, the overall scal restraint imposed by the need
for macroeconomic stability forced the government to meet many of its distributional goals by means of a
reallocation of social spending. e picture, however, changed during the Zuma administration due to low
economic growth rates and rising employment by government and rampant corruption and state capture.

7.5 Micro models of expenditure growth

Having discussed macro models of government expenditure growth, we now turn our attention to micro
models of government expenditure growth. e focus here is on how the decision-making behaviour of
public individuals and institutions may impact on increasing the share of government in the economy.

7.5.1 Baumol’s unbalanced productivity growth

Government expenditure may also increase disproportionately owing to an increase in the prices of inputs
used by the public sector relative to those employed in the private sector. is phenomenon has drawn the
interest of William Baumol (1967), who developed a microeconomic model of unbalanced productivity
growth to explain the growth in government expenditure. He divides the economy into two broad sectors: a
progressive sector and a non-progressive sector. e progressive sector is characterised by technologically
progressive activities, such as innovation, capital formation and economies of scale, which all contribute to a
rise in the level of output. An important feature of this sector is a cumulative increase in the productivity of
employees that justi es increases in wages and salaries. e inherent characteristics of the non-progressive
sector, on the other hand, only permit sporadic changes in productivity.
e technological structure of a sector will therefore determine the increase in the productivity of labour
inputs used. In the progressive sector, labour is only one of the inputs in the production process, while in the
non-progressive sector labour is often the end product. Baumol (1967: 416) illustrates this difference with
the aid of several examples. Consumers are usually not interested in the labour used to produce an air
conditioner, as they only care about the end product, that is, cold air. However, the labour input is of great
concern when one has purchased a ticket to attend a one-hour concert of a Beethoven string quartet. Any
effort to increase the overall productivity of the concert to below four human hours (for example, by
doubling the tempo of the music) may upset listeners and detract from the end product (that is, the concert).
In this case, there is clearly a limit to productivity increases, which is greatly determined by the labour-
intensive nature of the service.
e non-progressive sector usually consists of services that also constitute a large component of the
public sector. In this sector, labour plays an important role, for example, in respect of functions such as
education and public order and safety, where a certain teacher–learner ratio or a certain number of law-
enforcement officers per thousand of the population is usually the aim. According to Baumol (1967),
technological changes do not have such an important effect on productivity in the non-progressive sector as
they do in the progressive sector. It could be counterproductive, for example, to try to increase productivity
in health services by halving the time individuals spend in an operating theatre or hospital. As a result, there
are only sporadic improvements in productivity in the non-progressive sector compared to relatively rapid
increases in the progressive sector.
Baumol (1967) argues further that there cannot be too big a difference in the wages and salaries between
the two sectors, otherwise employees would be leaving the non-progressive sector to join the progressive
sector. is raises the relative costs of the non-progressive sector because salary increases are not
accompanied by the same increases in productivity as in the progressive sector. Baumol (1967: 426) thus
came to the conclusion that ‘the costs of even a constant level of activity on the part of government can be
expected to grow constantly higher’. Furthermore, if production in the non-progressive sector has to be
maintained relative to that in the progressive sector, it will imply that a larger share of the labour force will
have to be employed in the former sector, which could have negative effects on economic growth. One point
of critique, however, is that Baumol may have underestimated the opportunities for technological
advancement in the public sector, for example, through the computerisation of certain services.
As we saw in Table 7.1, a large share of government expenditure in South Africa goes towards the
remuneration of employees. Education, health services and policing, to mention but a few, all require labour
inputs as an end in itself. e structure of government expenditure thus corresponds well with Baumol’s
(1967) notion of unbalanced productivity growth. However, the Baumol (1967) hypothesis has not been
tested empirically in South Africa or in other African countries; no rm conclusions can therefore be drawn
about its relevance to government expenditure growth on the African continent. Unfortunately, for South
Africa, government has become the main creator of employment, with the result that productivity in the
government sector deteriorated and added additional pressure on government expenditure.

7.5.2 Brown and Jackson’s microeconomic model

e purpose of microeconomic models of growth in government expenditure is to study the factors
in uencing the demand and supply of public goods and services. Brown and Jackson (1990: 127) have
developed a microeconomic model to derive the levels of publicly provided goods and services by, inter alia,
taking determinants of demand such as the preferences, the income and the tax rate of the median voter into
account, as well as the costs of the goods and services in question. Since the scope of this book does not
permit a detailed description of this model, we will only discuss brie y some of the factors that may have an
in uence on the demand and supply of these goods, and hence on the level of government expenditure.
It often happens that government expenditure increases without a corresponding change in the level or
perceived quality of service. Although this state of affairs can be viewed as a sign of inefficiency, Brown and
Jackson (1990: 137) argue that there may be other forces at work. It may in fact be the result of changes in the
service environment. For example, an increase in the level of crime, as we have witnessed in South Africa,
calls for increased policing and additional funding merely to counter this increase and ensure that the same
level of security as before is maintained. If a sufficient increase in expenditure did not occur, the service
could be regarded as deteriorating and the service environment became worse.
Changes in the size and density of the population and its age structure may also have an in uence on the
service environment. Population growth may lead to an increase in the demand for publicly provided goods
and services. As far as pure public goods such as defence are concerned, we have seen that the marginal
social cost of one additional consumer is by de nition zero. However, governments do not only provide pure
public goods, but also mixed and merit goods such as education and health services. In such instances,
increases in the population will lead either to higher levels of expenditure or to a drop in standards (see
Chapter 10). On the other hand, population growth may also imply a decrease in the unit cost of these
services because the payment for these services should be shared by more individuals (taxpayers); in other
words, economies of scale may occur. If the relative price of services decreases, it may be an incentive for
government to supply more of those services. However, population growth and human migration may also
lead to changes in the density of the population, which may cause congestion and add to the costs of
government. e ght against HIV/Aids, malaria and tuberculosis should also put pressure on government
expenditure.
Another factor that may in uence the level of government expenditure is the quality of goods demanded
by the median voter. To de ne quality is not always easy. According to Brown and Jackson (1990: 139),
however, a good or service is of superior quality if it requires more inputs in its production process than a
good requiring fewer of those inputs, ceteris paribus. For example, a hospital that has 500 patients and 100
nurses should provide a service of superior quality to a hospital with 500 patients and 50 nurses. Increases in
quality may therefore put further pressure on government expenditure if the additional costs cannot be
recovered from the users.
In a nutshell, it would seem that the micro model of Brown and Jackson (1990) has identi ed and
combined important and relevant factors that may in uence the supply and demand of publicly provided
goods and services. eir model provides a useful point of departure for anyone interested in modelling
government expenditure in African countries and identifying the explanatory variables that will determine
future trends in government expenditure.

7.5.3 Role of politicians, bureaucrats and interest groups

We have already discussed the role played by individual agents and interest groups (see Section 6.7 of
Chapter 6) in in uencing government decisions and government expenditure. Politicians, bureaucrats and
other interest groups are often powerful enough to pressure policy in a direction that is detrimental to the
social welfare of the broader community. eir behaviour may result in a higher than optimal level of
government expenditure, thereby contributing to the growth of government’s share in the economy at the
expense of the private sector.
Apart from engaging in various forms of vote trading, as discussed in Chapter 6, there are many other
ways in which the vote-maximising behaviour of politicians can bring about an increase in government
expenditure. ey may grant wage and salary increases to state employees in order to win the support of
what is often a numerically signi cant and powerful constituency. From a macroeconomic point of view, the
ruling party may be tempted to introduce populist policies by relaxing scal and monetary policies in an
attempt to stimulate the economy before an election. A relaxation of this nature may well please voters who
stand to secure short-term gains in the form of lower interest rates, lower taxes and better job prospects, and
they may well respond by lending their support to the ruling party. However, populist macroeconomic
policies are not sustainable in the long term, giving rise to in ationary pressures and balance of payments
problems, and ultimately necessitating even tighter monetary and scal policies than the measures that
were in place prior to the election.
We also saw in Chapter 6 that bureaucrats tend to maximise the size of their departmental budgets, partly
to help build their own personal empires, and that they are able to convince politicians of their actions as
they are better informed about the activities of their own departments. e net result is a bureaucracy that is
larger than the optimal size. In South Africa, we have seen how the so-called homelands policy of the past
and the increase in the number of provinces after the 1994 election have contributed to a duplication of
government activities that has put additional pressure on government expenditure.
Individuals with shared interests are often able to organise themselves into powerful interest groups, and
put pressure on government to implement programmes and pass legislation that will meet their own
parochial interests. South African farmers have formed an important lobbying group in the past that has
bene ted greatly from agricultural subsidies and other forms of support. e same is true of labour unions
and organised business, and the government may well be tempted to grant them special favours, such as tax
allowances, that will erode the revenue base of government. e persistent recurrence of actions of this
nature can also contribute to the growth of government expenditure in the economy.
7.6 Government and the economy: Long-term effects
In the preceding sections, we looked at the growth of the public sector in the economy and discussed some
of the reasons for this growth. We saw that there are various theories explaining the secular growth of
government in virtually all economies across the world. We now shift our focus to the economic effects –
rather than the causes – of government activity and deal mostly with the long-term effects that changes in
government expenditure can have on the economy.
e short-term impact of government activity on the economy depends on the value of the spending
multiplier (that is, the effect of the change in expenditure on the GDP) and on how government spending is
nanced on an annual basis (for example, taxes or borrowing). ese issues are discussed further in
Chapters 17 and 18.
e long-term role of government can be viewed within the context of the growing body of literature
known as endogenous or new growth theory (NGT) (Romer, 1986). Adherents to this school have all called
attention to the contribution that government can make towards stimulating and reinforcing sustained
economic growth. Several kinds of public investment and expenditure programmes are reputed to confer
signi cant positive externalities (of both a pecuniary and a technological kind) on private producers and
consumers, and it is to these programmes that policy-makers should turn in their quest to promote
economic growth and development.
e origin of NGT can be traced to a general dissatisfaction with the classical and neoclassical theories of
growth, according to which savings and private xed investment or additions to the physical capital stock
play a pivotal role in fostering economic growth. According to NGT, however, this is an oversimpli cation of
what is really a very complex process. is school of thought adopts a much broader de nition of the
concept of capital and focuses attention on the role played by each of the components of capital in the
growth process.
In addition to privately owned physical capital (for example, factories, equipment, office buildings and
luxury homes), capital also includes the following three components:
• e existing economic or physical infrastructure (for example, roads, rail, street lighting, sewage systems,
water and electricity)
• Accumulated human capital acquired through education, training and healthcare, in other words, social
infrastructure
• e stock of technical expertise acquired through learning-by-doing and research and development
(R&D).

e main thrust of NGT is that additions to any of the components of capital may yield increasing returns
because they create externalities that bene t a range of sectors and industries in the economy. Investment in
the physical infrastructure creates (pecuniary) externalities by lowering production costs and boosting
returns in the private sector. Likewise, recipients of education may transfer their skills free of charge to third
parties, healthier citizens will be more productive and limit the spread of disease, new users of electricity will
boost the demand for electrical appliances and so on. Each of the components of capital can be in uenced
by government through appropriate policy intervention. e case for intervention of this nature is again
based on market failure. Since the marginal private bene ts from investments in the economic
infrastructure, human capital and R&D are lower than the corresponding marginal social bene ts, free
markets will underprovide these services. NGT thus provides a strong justi cation for government
intervention in these areas, as it will create favourable conditions for private investment and economic
growth. Investment in a country’s economic infrastructure usually boosts the productivities of a range of
inputs used by existing private enterprises (see Aschauer, 1989). It may do so by lowering transport and
communication costs, thus cutting down on travel time, reducing the time and costs involved in
negotiations, and facilitating the discovery of new input and output markets. It may also ‘crowd in’ new
private investment, including foreign direct investment (FDI), in the sense of new rms starting up or
existing rms expanding their operations. e point here is that it is the new or improved infrastructure that
triggers both the (mostly pecuniary) externalities and the new private investment, neither of which would
have happened in the absence of infrastructure investment.
 From a macroeconomic perspective, the role played by government capital investment can be brie y
illustrated with the aid of a Solow-type aggregate production function. For example, let aggregate supply,

where Kp and Kg represent the levels of capital owned by the private and public sectors respectively, e is a
measure of labour productivity and N is the quantity of labour (with eN being Solow’s ‘effective’ labour).
While an increase in Kg will directly boost Y in Equation [7.1], it may also either crowd in new private capital
investment or, if the addition to Kg is nanced by means of borrowing, putting upward pressure on interest
rates, crowd out some private investment (see Chapter 17). In addition, an increase in Kg (for example, in the
form of new school buildings or healthcare facilities) may boost labour productivity, e. We may therefore
extend Equation [7.1] as follows:

where the relationship, Kp (Kg), is either positive or negative, depending on whether there is a net crowding-
in or crowding-out effect, and where the rst derivative of e (Kg) is positive.
Infrastructure investment is usually nanced by means of borrowing on the open market, but it is
generally accepted that investments of this nature are more likely to have a net crowding-in effect than a
crowding-out effect. e reason is that government borrowing usually constitutes a small portion of total
borrowing, so that it is unlikely to have much of an effect on interest rates, either directly or indirectly. But
even if it did cause a rise in the interest rate, private investors would not necessarily be deterred by it. To
them, other things, including a country’s infrastructure, are often more important when it comes to making
important decisions about their own investments.
Empirically, too, several recent studies have produced evidence con rming this proposition. Although
government spending as a whole tends to have a negative impact on economic growth – partly because it
consists mostly of recurrent consumption expenditure – most studies show that public infrastructure
investments do have a positive impact on factor productivities in the private sector. In its World
Development Report 1994, for example, the World Bank (1994: 14) summarised the results of these studies
and concluded that ‘… the role of infrastructure in growth is substantial, signi cant and frequently greater
than that of investment in other forms of capital’.
Much the same can be said of public investment in education, training and healthcare. When these
spending categories are suitably disaggregated, one nds signi cant differences between the constituent
components of each category. Vocational training often produces higher returns than poor-quality formal
schooling, and so too do good quality primary and secondary education in relation to certain categories of
tertiary education. Other categories of tertiary education, in turn, provide the specialised skills that are
necessary for sustained economic growth (Bhorat, 2004).
ese ndings are clearly important in helping governments to prioritise their spending programmes,
especially in view of the important redistributive role that programmes of this nature play today. We will
return to this theme in the next chapter. e prioritisation of state expenditures, based on efficiency criteria,
can help governments to do two things: achieve a more equitable distribution of income and create the
conditions for sustainable economic growth over the long term.

Key concepts
• displacement effect (page 131)
• economic classification of government expenditure (page 123)
• endogenous or new growth theory (page 136)
• functional classification of government expenditure (page 125)
• functional composition (page 125)
• general government expenditure (page 121)
• macro models of government expenditure growth (page 130)
• micro models of government expenditure growth (page 130)
• service environment (page 135)
• stages-of-development model (page 130)
• unbalanced productivity growth (page 133)

SUMMARY
• e South African Constitution explicitly charges government with the task of maintaining critical
government institutions, including national, provincial and local governments, and providing them with
the necessary public funding. Failing to do so would in fact be unconstitutional.
• In addition, the government of the day is constitutionally obliged to provide or extend the provision of
speci ed basic goods and services. e clearest examples of these provisions are found in the Bill of
Rights, which entrenches the right of each citizen to adequate housing, healthcare, food, water, social
security and education. However, Section 26(2) of the Constitution does recognise the need for adequate
resources by stating explicitly, ‘… the state should take reasonable legislative and other measures, within
its available resources, to achieve the progressive realisation of this right’ (italics added).
• General government expenditure grew in South Africa from 1960 to 2017, both in real terms and as a
percentage of GDP. us, in 2017, the general government sector comprised some 27,4% of GDP in terms
of resource use and 36,8% in terms of resource mobilisation.
• In terms of the economic classi cation, cash payments for operating activities (formerly known as
current expenditure) accounted for 92,8% of general government spending in 2017 in South Africa, while
purchases of non- nancial assets (outlays for the acquisition of xed capital assets, land, stock and
intangible assets, formerly known as investment) accounted for the remaining 7,2%. e largest
categories of total cash payments were the government’s wage bill, purchases of other goods and services
(for example, stationery, computers, vehicles and maintenance of capital assets), and social bene ts
paid.
• In terms of its functional composition, government spending on public order and safety, education,
health, social protection, housing and community amenities, and transport all increased during this
period as percentages of both total general government expenditure and of GDP.
• e long-run pattern of change in government outlays in South Africa broadly corresponds with
international trends, with South Africa also having experienced a shift from capital expenditure and
economic services towards consumption spending and social services.
• It is questionable whether South Africa needs to invest even larger portions of its national income in
publicly nanced education and healthcare. Government spending on these items is already relatively
high by international standards, and the real challenge for the future is for government to utilise its
existing education and health budgets more efficiently than before. Future increases in these budgets
could perhaps be achieved more appropriately through additional revenues obtained from sustained
real growth in per capita income.
• Growth in the government sector does not necessarily imply an increase in the share of government in
the economy. If government expenditure increases at the same rate as the (real) GDP, no change in its
relative share will occur. However, if government expenditure increases at a higher rate than GDP, its
share in the economy will increase.
• Macro models explain the growth of government expenditure in the economy by taking aggregate
variables such as the GDP into account. According to these models, the development process goes
through several stages and will cause an increase in the share of government in the economy. Social
upheaval (such as war and social unrest) may cause instability and may require an appropriate response
by government, which may lead to a potentially permanent increase in the share of government in the
economy. Furthermore, democratic processes may also play a role, especially in a society with high levels
of inequality. Voters may elect a party to parliament with the speci c aim of addressing these inequities.
• Micro models explaining the growth of government expenditure focus mainly on productivity
differences, factors in uencing the demand and supply of public goods and services, and the decision-
making behaviour of public individuals and institutions. Differences in productivity growth between the
 private and public sectors may cause an increase in government expenditure owing to the need to pay
higher wages and salaries in an attempt to retain employees who have scarce skills in the government
sector. Increased demand for policing, HIV/Aids treatment and education, for example, may place
additional pressure on government expenditure. Politicians, bureaucrats and special interest groups
often pursue their own interests at the expense of the national interest, resulting in higher levels of
government expenditure.
• In terms of endogenous growth theory, public investment in the economic and social infrastructure may
yield increasing returns because it creates positive externalities that bene t a range of private sectors and
industries in the economy. It may also crowd in new private investment, including foreign direct
investment (FDI), in the sense of new rms starting up or existing rms expanding their operations.

MULTIPLE-CHOICE QUESTIONS
7.1 Which of the following statements is or are correct?
a. Resource-use measures of the size of government expenditure are always larger than resource-
mobilisation measures.
b. In the economic classi cation of government spending, purchases of goods and services is a
component of purchases of non- nancial assets.
c. e functional classi cation distinguishes between the current and the capital elements of
government expenditure.
d. None of the above.
7.2 Which of the following statements is or are correct?
a. General government spending has grown without interruption as a percentage of GDP in South
Africa from 1992 to 2017.
b. Long-term trends suggest that general government spending on infrastructure remains relatively
low in South Africa.
c. International comparisons suggest that South Africa still has considerable scope for increasing
social spending as a percentage of GDP.
d. All of the above.
7.3 Which of the following statements is or are correct?
a. Growth in government expenditure does not necessarily imply an increase in the share of
government in the economy.
b. According to the stages of development model, urbanisation will not put additional pressure on
government expenditure.
c. e increase in the share of government in the economy of South Africa can be attributed only to
the displacement effect, which is caused by social upheaval.
d. All of the above.
7.4 Which of the following statements is or are correct?
a. e median voter will always favour the redistribution of income.
b. Government expenditure increases because government tends to use too much new technology.
c. Politicians, bureaucrats and interest groups are always inclined to put the national interest before
their own self-interest.
d. None of the above.
7.5 To promote GDP growth, governments should invest in:
a. the economic infrastructure
b. human capital
c. research and development (R & D)
d. All of the above options are correct.

SHORT-ANSWER QUESTIONS
7.1 What guidelines does the South African Constitution give in respect of the role of government in the
economy?
7.2 Explain the displacement effect in Peacock and Wiseman’s model of government expenditure growth.
7.3 Explain how unbalanced productivity growth may affect government expenditure and brie y comment
on its relevance to South Africa.
7.4 Distinguish between the notions of ‘crowding in’ and ‘crowding out’ (of private investment).

ESSAY QUESTIONS
7.1 Discuss trends in the size, growth and composition of government spending in South Africa since 1994.
7.2 Distinguish between the various macro and micro models of public sector growth.
7.3 Explain the median voter hypothesis of Meltzer and Richard, and indicate whether it can explain the
growth of government expenditure in South Africa.
7.4 Consider the implications of new growth theory for the role of government in the economy.

1 e data were obtained from the electronic database of the South African Reserve Bank (available online at
http://www.resbank.co.za/Publications/QuarterlyBulletins/Pages/QBOnlinestatsquery.aspx).
2 Data at constant 2016 prices are not available for the resource mobilisation measure.
3 In South Africa, scal years run from 1 April to 31 March. e scal year 2009 therefore represents the period from 1 April 2008 to
31 March 2009.
4 Legislation decrees that the government should service the public debt before undertaking other expenditures.
5 .Data for the scal years before 2005 were compiled by Statistics South Africa on the basis of the 1986 version of the Government
Finance Statistics (GFS) Manual, whereas data for later years re ect the conventions of the 2001 version. Hence, the two parts of
the series are not strictly comparable and caution should be exercised when interpreting the information in Table 7.2. Note also
that the total expenditure–GDP ratios in Tables 7.1 and 7.2 differ. According to the South African Reserve Bank, the main cause of
the discrepancy is the exclusion from the functional classi cation of general government expenditure of data on the trading
accounts of local governments.
6 is section reports World Bank data for South Africa, which may differ from the South African statistics used elsewhere in this
chapter.
Government intervention to reduce inequality and poverty

Krige Siebrits

Sections in earlier chapters of this book introduced equity-related arguments for government intervention to
reduce inequality and poverty. Chapter 2 (Section 2.4.6) argued that highly unequal distributions of income
and wealth and the related problem of high levels of poverty are examples of market failure that establish
prima facie cases for policy intervention. Section 2.5.2 con rmed the real-world relevance of this argument
by stating that the income distributions generated by markets tend to exhibit considerable inequality. ese
theoretical and practical considerations underpin governments’ distributive function. As was also pointed
out in Section 2.5.2, this function is important to governments, all of whom attempt to change the
distributions of income and wealth and to reduce poverty. Apart from discussing criteria for assessing the
welfare effects of policies to make economic outcomes more equitable, Chapter 5 commented on the
efficiency implications of such interventions. Section 5.5 showed that redistributive measures might harm
the long-term growth potential of economies by distorting markets. However, such efficiency-equity trade-
offs are sometimes avoidable. Well-designed redistributive measures can enhance human capital
development and boost investment by reducing the instability associated with distributional con ict.
Redistributive and anti-poverty policies that achieve these ends make the allocation of resources more
efficient and contribute to higher rates of economic growth.
is chapter, as well as the next two chapters, develop these arguments further by discussing practical
aspects of the role of government in addressing inequality and poverty. e present chapter discusses broad
conceptual and programme-design aspects of the redistributive and poverty-reducing roles of government.1
e next two chapters will examine speci c elements of these roles, namely income security (Chapter 9) and
the provision of social services (Chapter 10). All three chapters focus mainly on government spending
programmes. We also pay some attention to tax issues, however, because the revenue and expenditure sides
of the budget are deeply intertwined aspects of governments’ endeavours to reduce inequality and poverty.
Part 3 of this book discusses some of the tax-related issues raised in these three chapters in more detail.
e structure of this chapter is as follows. Section 8.1 introduces important measures of inequality and
poverty and uses these to provide information about the extent of these phenomena in South Africa and
other Southern African Development Community (SADC) countries. Against this backdrop, Section 8.2
explains how governments can use scal policy instruments to change the primary and secondary
distributions of income. Section 8.3 discusses an important aspect of the design of policies to reduce
inequality and poverty, namely targeting mechanisms that channel the bene ts of public programmes to
those in greatest need. e chapter concludes with Section 8.4, which explains the concept of ‘ scal
incidence’ and illustrates the usefulness of incidence studies for gauging the efficacy of policies to reduce
inequality and poverty.

Once you have studied this chapter, you should be able to:
 distinguish between primary and secondary income and discuss policy approaches to change the primary
and secondary distributions of income
discuss broad considerations in the design of government interventions to reduce inequality and poverty
define targeting and distinguish the most common mechanisms to target the benefits of government
expenditure programmes
define the concept of ‘fiscal incidence’ and explain how it is used to assess the effectiveness of policy
interventions to reduce inequality and poverty.

8.1 Inequality and poverty in SADC countries

Poverty and inequality are very important policy issues in Southern Africa. Table 8.1 contains measures of
the severity of these problems in the SADC member states.2
e most widely used indicators of the extent of poverty are poverty rates (also known as poverty
headcount ratios). A poverty rate simply shows the percentage of the population of an area whose incomes
are below the poverty line. e international poverty line of $3.20 per person per day used by the World Bank
in the 2018 edition of the World Development Indicators database is the basis of the poverty rates in Table
8.1. e second column in the table shows that poverty is widespread in all SADC countries except Mauritius
and Seychelles (with only 3,2% and 2,5% respectively of the population living in poverty in these countries).
South Africa’s poverty rate of 37,6% is relatively high for a middle-income country. Countries that have levels
of per capita income similar to that of South Africa tend to have markedly lower poverty rates. For example,
the most recent poverty rates for Brazil, Colombia and Indonesia at the time of writing were 8,0%, 11,8% and
30,9%, respectively.
e Gini coefficient is a well-known summary measure of the distribution of income. It is derived from
the Lorenz curve, which shows the cumulative income shares of cumulative fractions of a population. Figure
8.1 shows a Lorenz curve based on the income distribution data for South Africa in the July 2018 version of
the World Bank’s World Development Indicators. e income shares of the ve quintiles are as follows: rst
(poorest) quintile – 2,4%, second quintile – 4,8%, third quintile – 8,2%, fourth quintile – 16,5%, and fth
(richest) quintile – 68,2%. is implies that 7,2% of total income accrued to the poorest 40% of the
population, 15,4% to the poorest 60%, and 31,9% to the poorest 80%. e dark line in Figure 8.1 is a Lorenz
curve based on these cumulative income shares. e Gini coefficient is a measure of the extent to which an
actual income distribution, such as the one summarised by the Lorenz curve in Figure 8.1, deviates from a
perfectly equal distribution. e dashed line in Figure 8.1 depicts such an equal distribution: Exactly 20% of
total income accrues to each of the quintiles. To calculate the Gini coefficient, the area A (that is, the area
between the dashed reference line of perfect equality and the Lorenz curve) is divided by the area A + B (that
is, the entire triangular area beneath the reference line of perfect equality). e resulting coefficient can
range from zero (perfect equality, which implies that the Lorenz curve coincides with the reference line) to
one (perfect inequality, which implies that all income accrues to one member of the population). In
practice, Gini coefficients tend to range from 0,20 to 0,70.
Table 8.1 Measures of poverty and inequality in SADC countries

Source: World Bank (2018b).

Hence, it is clear from the data in Table 8.1 that several Southern African countries have exceptionally
high levels of income inequality. South Africa’s Gini coefficient of 0,630 is the highest among the 163
countries for which the World Bank published gures in June 2018. e reality that only 2,4% of income
accrued to the poorest 20% of South Africans in 2014 while fully 68,2% accrued to the richest 20% con rms
the extreme skewness of the distribution of income. Namibia, Botswana, Zambia, Lesotho, Mozambique and
Eswatini also have very high Gini coefficients. In these countries, too, the gaps between the income shares of
the poorest 20% and the richest 20% of the populations are very large.
A discussion of the reasons for the high levels of income inequality and poverty in South Africa and most
other SADC countries falls outside the scope of this book. e remainder of this chapter focuses on another
important question raised by the data in Table 8.1: What can governments do to reduce poverty and extreme
inequality in the distribution of income? To this end, it discusses policy approaches and tools for assessing
their effectiveness.
Figure 8.1 The Lorenz curve and the calculation of the Gini coefficient

Source: Based on data for South Africa in World Bank (2018b).

8.2 The budget, redistribution and poverty alleviation

Taxes, government expenditure programmes and regulation are important policy instruments for reducing
poverty and changing the distribution of income. In the short run, the national budget is the most important
direct mechanism available to any government to in uence the distribution of private earnings. In the
longer run, the scal system also has important indirect effects: government spending and taxes in uence
economic growth rates and investment in human and physical capital, which are major determinants of the
distribution of earnings from assets and labour market activity.
e distinction between primary income and secondary income can be used to outline the scope for
using scal policy instruments for these purposes. e primary income or personal income of an
individual, household or group is the actual value of income received in cash or in kind. It includes the value
of subsistence production activities such as subsistence agriculture. e secondary income of the
individual, household or group is obtained by subtracting direct taxes paid from primary income (which
leaves disposable income) and adding the value of government services consumed. Hence,

where Ys represents secondary income, Yp represents primary income, Td represents direct taxes and G
represents government expenditure.3
 Sections 8.2.1 and 8.2.2 provide overviews of two policy approaches aimed at changing the distributions
of primary income and secondary income, respectively. e two approaches can be used separately or at the
same time. Poverty reduction is invariably also an important objective of both approaches. Section 8.2.3
introduces some key issues policymakers should take into account when they design and assess the effects
of measures to reduce inequality and poverty, including the risk of unintended consequences caused by
behavioural responses to such policies. e discussions of redistributive public spending programmes in
Chapters 9 and 10 as well as those of taxes in Chapters 11 to 15 provide more detail on these issues.

8.2.1 Changing the distribution of primary income

e rst approach relies on interventions to make the primary distribution of income less unequal. Put
differently, the aim of this approach is to establish rules and incentives that generate more equitable market
outcomes. Governments often use regulations for this purpose. Possibly the best-known examples are
minimum wage regulations, two aims of which are to establish oors for the remuneration of lower-paid
workers and to reduce pay gaps between these workers and higher-paid ones. Several Southern African
countries have minimum wage regulations for particular or for all sectors of their economies, including
Botswana, Mozambique, South Africa, Zambia and Zimbabwe. Governments can also use public spending
programmes and tax measures to in uence the distribution of primary incomes. Given that job creation is a
powerful remedy against poverty and a rst step towards reducing inequality, governments can pay wage
subsidies to rms to encourage them to hire unemployed members of the labour force. A similar reduction
in the cost of employing labour can be obtained by granting tax incentives to rms that hire jobless persons.
South Africa’s Employment Tax Incentive, which reduces the employees’ tax payable by rms that hire young
and less experienced persons, is an example of such a measure.

8.2.2 Changing the distribution of secondary income

e aim of the second approach is to in uence the secondary distribution of income. Hence, it involves the
use of scal instruments ex post to reduce disparities in the primary distribution of income generated by
market forces.
Government expenditure programmes are the backbone of this approach. Such programmes can take
the form of cash transfers, in-kind transfers and the provision of social services. Cash and in-kind transfers
directly increase recipients’ command over resources, while effective provision of social services raises the
living standards of recipients and improves their ability to access income-generating opportunities.4 A few
examples of cash transfers programmes in SADC countries are the Social Cash Transfer Programme in
Malawi, the Child Support Grants and Grants for Older Persons in South Africa, the Basic Social Subsidy
programme in Mozambique, and the Harmonised Social Cash Transfer Programme in Zimbabwe. In-kind
transfer programmes include the Food for Education Programme in Tanzania, the Food Security Pack in
Zambia, and the school feeding schemes in Angola, Botswana, Lesotho and Namibia. Like their counterparts
elsewhere, the governments of Southern African countries also provide social services such as education,
health care and social welfare services.
Governments can also reduce inequality by imposing taxes on the incomes and assets of the rich. Of
course, taking from the rich in itself does not reduce poverty– taxes contribute to poverty reduction only to
the extent that the proceeds are used to nance transfer payments and social service provision to the poor.5
is con rms that the redistributive and poverty-reducing effects of the revenue and spending sides of
budgets should be assessed jointly. Section 8.4 also emphasises this point.

8.2.3 Broad considerations for the design and assessment of policies

is subsection outlines two interrelated sets of considerations policymakers should be mindful of when
designing policies to reduce inequality and poverty: rst, whether the policies will achieve their aims and,
second, how the policies might in uence the behaviour of intended bene ciaries. ese considerations can
also be used to formulate criteria for assessing the effects of such policies.
One key factor that determines the effectiveness of government spending programmes to reduce poverty
and inequality is whether such programmes reach the intended bene ciaries. In practical terms, this means
that bene ts of transfer and social service programmes should be provided to the largest possible fraction of
the poor and that bene t leakage to the affluent should be avoided. is requires appropriate targeting
mechanisms (the topic of Section 8.3) and sufficient administrative capacity within the government agency
responsible for policy implementation. When it comes to taxes aimed at reducing disparities in income and
wealth, the equivalent effectiveness issue is the extent to which tax burdens are borne by the affluent. is
issue matters because those responsible for paying certain taxes may succeed via legal or illegal means to
shift the actual burdens to others ( rms, for example, can sometimes pass on tax burdens to other rms,
their workers or customers). Taxes imposed on wealthy individuals or highly pro table rms fail to achieve
redistributive purposes if these taxpayers succeed in shifting burdens to less affluent individuals. Chapter 11
provides a detailed discussion of this issue, which is known as the incidence of taxes.
e usefulness of the bene ts to recipients also in uences the effectiveness of poverty-focused public
spending programmes. e forms such bene ts take – cash, goods or social services – and how recipients
use them are relevant considerations.6 For example, Section 9.3.1 comments on the validity of the argument
that in-kind transfers and social services programmes are preferable to cash transfers because poor people
might use cash to purchase luxuries and ‘sin goods’ (such as tobacco products and alcoholic beverages).
Another important issue is that the utility recipients derive from publicly provided social services depends
heavily on the quality of such services. e link between the quality and the poverty-reducing potential of
government spending on education and healthcare is a major theme in Chapter 10.
A more general effectiveness-related issue is whether the impact of government spending programmes
extends beyond short-term poverty relief to the promotion of sustainable livelihoods. In the past, many cash
transfer programmes enabled bene ciaries to meet essential consumption needs, but failed to prevent the
transmission of poverty from generation to generation. Hence, a growing number of countries have adopted
conditional cash transfer programmes, which link eligibility for cash transfers to the performance of actions
such as regular school attendance by children and visits to clinics for health check-ups (the Community-
Based Conditional Cash Transfer Programme in Tanzania is a Southern African example). e purpose of
the conditions is to break inter-generational poverty cycles by building the human capital of children in poor
households. Section 9.3.2 comments on the effectiveness of conditional cash transfer programmes.
Policies to reduce inequality and poverty can in uence the behaviour of bene ciaries and taxpayers in
ways that distort the allocation of resources and counteract intended effects. For example, rms might react
to the wage subsidy mentioned in Section 8.2.1 by replacing existing workers with subsidised ones. e
outcome would be that the subsidy fails to create new jobs and only bene ts rms by reducing their labour
costs. is subsection has already referred to two other examples: Recipients might squander cash transfers,
and the rich and highly pro table rms may shift the burdens of some taxes to less affluent individuals. e
problem of moral hazard is at the heart of many actual or perceived behavioural distortions associated with
government spending programmes to ght poverty. Moral hazard occurs when individuals and families do
not provide for their own needs or do not protect themselves against negative income shocks because they
have access to government assistance in times of need. Section 9.2.2 contains a more detailed discussion of
the nature and effects of moral hazard problems. In addition, attempts by individuals and rms to avoid or
reduce the burdens of taxes imposed to nance transfer payments and social service provision can give rise
to various forms of inefficiency. Distortion of time allocations for work and leisure is a well-known example
of this. Chapter 12 discusses the efficiency effects of behavioral responses to various taxes.

8.3 Targeting of government spending programmes

Targeting is the practice of using mechanisms to identify those in greatest need and to channel the bene ts
of government spending programmes to them. ree broad categories of targeting mechanisms exist:
1 Means testing. is involves using levels of income or wealth to distinguish poor individuals who are
eligible for government assistance from more affluent ones who are not. Means tests determine
 eligibility for the bene ts of the three largest cash transfer programmes in South Africa (the Grant for
Older Persons, the Child Support Grant and the Disability Grant) and the Child Maintenance Grant in
Namibia, among others.
2 Indicator targeting. Means testing is sometimes complicated by measurement problems and can
create perverse incentives (for instance, individuals might forgo or hide labour-market earnings to pass
means tests and qualify for cash transfers). In such circumstances, government might use indicator
targeting instead. is involves basing eligibility for bene ts on one or more indicators that are highly
correlated with income yet easier to measure. Such indicators might include ownership of land or other
assets, levels of education, age, and areas of residence. Southern African examples of the use of
indicator targeting mechanisms are programmes to assist vulnerable groups such as children (including
the school feeding schemes listed in Section 8.2.2, the National Orphan Care Programme in Botswana,
and the Basic Education Assistance Module in Zimbabwe), individuals with disabilities (including the
Basic Invalidity Pension in Mauritius and the Disability Grant in Namibia) and the elderly (including
the Old-Age Pension in Botswana and Eswatini’s Old-Age Grant).7
3 Self-targeting. Self-targeted programmes have requirements or offer bene ts that are unattractive to
the affluent. us, the wages offered by public works programmes might be set too low to attract
workers who hold other jobs, and in-kind transfer programmes might provide foodstuffs or other goods
that are consumed only or mainly by the poor. is targeting mechanism underpins the Ipelegeng
Public Works Programme in Botswana, among others. Effective self-targeting has a clear advantage over
other targeting mechanisms: It is unnecessary to spend resources to identify intended recipients,
because aspects of the design of the programme prevent bene t leakage to affluent individuals.

8.3.1 The benefits and limits of targeting

e major bene ts of effective targeting are budgetary savings and strengthening of the poverty-reducing
effects of government expenditure programmes. Figure 8.2, which compares the effects and budgetary costs
of targeted and universal (non-targeted) cash transfer programmes, illustrates these bene ts.
e pre-transfer incomes of the inhabitants of a country are arranged from the smallest (Ymin) to the
largest (Ymax) along the horizontal axes of panels A and B in Figure 8.2. e vertical axes of the two panels
show the post-transfer incomes of these individuals, arranged in the same way. e two horizontal lines
labelled z represent the country’s poverty line: All individuals who earn less than z are poor, while those who
earn more are not. Hence, the triangles Yminaz depict the country’s poverty gap, that is, the amount of money
needed to raise the incomes of all poor individuals to the poverty line and eliminate poverty.
We rst focus on Panel A, which compares the effects of universal and perfectly targeted cash transfer
programmes. e line Ymin ad represents the initial income levels of the country’s inhabitants. Assume the
government introduces a cash transfer programme that disburses an equal amount T (equal to c – Ymin) to
everyone. is non-targeted transfer causes a parallel upward shift of the incomes line from Yminad to cbe.
e programme clearly reduces poverty: It lifts all individuals whose pre-transfer earnings ranged between
Y1 and Y2 out of poverty. However, the reality that the transfers are too small to lift individuals whose pre-
transfer earnings ranged between Ymin and Y1 out of poverty means that it fails to eradicate the problem
completely. Hence, the poverty gap does not disappear; instead, it merely shrinks from Ymin az to cbz.
Figure 8.2 The benefits and limits of targeting

Source: Adapted from Hoddinott (2007: 91).

Universal cash transfer programmes also produce substantial bene t leakage. Panel A in Figure 8.2
shows the two types of leakages associated with such schemes. First, everyone who was lifted from poverty
except those whose pre-transfer incomes were exactly Y1 received more than what was required to raise their
incomes to z. Second, the bene ciaries of the programme included all non-poor individuals (that is,
everyone whose pre-transfer incomes exceeded Y2). Bene t leakage tends to be the main reason why the
bene ts provided by some universal transfer programmes are inadequate to eliminate poverty gaps.
A perfectly targeted cash transfer programme would yield markedly better outcomes. Such a programme
would provide everyone whose initial income was less than Y2 with a transfer that exactly equals his or her
individual poverty gap (that is, the difference between z and the person’s pre-transfer income). e result
would be the complete eradication of absolute poverty and, by implication, elimination of the poverty gap.
Moreover, no bene t leakage would occur. Put differently, perfect targeting avoids two important errors of
targeting mechanisms. ese are type 1 errors (or errors of exclusion, which occur when the targeting
mechanism excludes poor people whom the programme was supposed to bene t) and type 2 errors (or
errors of inclusion, which occur when the targeting mechanism fails to exclude rich people who were not
supposed to be among the bene ciaries).
It is likely to be very difficult or expensive to determine individual poverty gaps and to manage systems of
differentiated transfers. Hence, countries tend to use less sophisticated targeting mechanisms. Panel B in
Figure 8.2 uses the example of a simple means tested transfer programme to show that such mechanisms
still yield better outcomes than non-targeted programmes do. is programme provides a cash transfer T
(equal to c – Ymin) to all poor individuals (that is, those whose pre-transfer incomes are less than Y2). Hence,
it avoids type 2 targeting errors by excluding everyone whose pre-transfer incomes exceed the poverty line.
For a given programme budget, a means tested programme that excludes the affluent makes a larger transfer
to each poor person possible than a universal programme would. is is shown by the difference between
the extent of c – Ymin in Panel A and Panel B. e size of the programme budget would determine whether
poverty is eradicated or merely reduced. e means-tested transfer in Panel B does not eradicate absolute
poverty, and the result of this type 1 error is a small poverty gap cbz. Panel B also illustrates a major
drawback of this targeting mechanism: Individuals who formerly earned between Y1 and Y2 receive more
than what is required to raise their post-transfer incomes to z. Hence, the programme raises their incomes
above those of some others who are ineligible for government assistance because their pre-transfer incomes
exceed Y2. is change in the distribution of income seems unfair. In addition, it creates perverse incentives,
as these non-qualifying individuals are now able to increase their incomes by working less and becoming
eligible for transfers.

8.3.2 The costs of targeting

e bene ts of targeting should be weighed against its costs. Such costs can be classi ed into three types:
1 Administrative costs. e costs of identifying, reaching and monitoring the target population can be
substantial: Grosh (1994: 36–37) found that such costs make targeted transfers in Latin American
countries 3–8% more expensive than universal ones are. Hence, trade-offs exist between curtailing
administrative costs and avoiding the costs of poor targeting. Strictly speaking, the administrative costs
of targeted transfer programmes should include compliance costs for participants, including those
associated with interviews and documents to prove eligibility.
2 Incentive costs. Targeting can have positive incentive effects. To name but one example: A food subsidy
programme targeted at parents whose children attend school may raise school enrolment. Negative
incentive effects (that is, incentive costs) are more common, however. Distortions of the consumption-
saving and work-leisure choices are cases in point. Means testing might reduce incentives to save for
retirement, for instance, while generous unemployment bene ts might discourage job search.8 Non-
productive use of time and other resources that arises from moral hazard problems is another
important category of negative incentive costs associated with some targeting mechanisms. Examples
of this include the problem of so-called ‘rotten parents’ who may not provide adequately for their
children if they know that the state will do so by means of school feeding programmes or transfer
programmes.
3 Stigma costs. e stigma costs of targeting refer to the loss of self-esteem sometimes fostered among
participants by the negative attitudes of non-participants and officials. e desire to avoid
stigmatisation of poor children is the reason why some school-feeding programmes are universal
instead of targeted. By contrast, the reality that eligibility for the grant for older persons is almost
universal in many parts of South Africa means that there is hardly any stigma associated with being a
recipient.

8.3.3 The effectiveness of targeting

How effective is targeting in practice? A study of 86 targeted transfer programmes introduced in middle- and
low-income countries between 1985 and 2003 (Coady, Grosh & Hoddinott, 2004) nd that such programmes
generally, but not always, channel more bene ts to the poor than equivalent universal ones do. e criterion
used by the authors is the bene t ratio, which is the share of programme bene ts transferred to a certain
segment of the population divided by that segment’s share of the total population. Consider the poorest 40%
of the population (the segment used by Coady et al., 2004) as an example. A universal transfer programme
would yield a bene t ratio of one, because it would provide the same bene ts to everyone (for example, the
poorest 40% of the population would receive exactly 40% of the bene ts). By contrast, a targeted programme
with a bene t ratio of more than one provides more than 40% of the programme resources to the poorest
40% of the population. In such cases, the targeting mechanism enhances the redistributive effect of the
programme.
Fully 61 (or 70%) of the 86 programmes studied by Coady et al. (2004) had bene t ratios higher than one.
In fact, the ratios of 45 of the programmes were at least 1,25. Hence, the targeting mechanisms effected
signi cant redistribution in more than half of the sample of programmes. Another four programmes had
ratios of exactly one – this means that these transfer schemes were not more effective than non-targeted
ones would have been at channelling bene ts to the poor. However, the reality that 21 (almost one quarter)
of the programme had bene t ratios smaller than one, con rms that targeting does not always work well. It
should also be kept in mind when interpreting these ndings that bene t ratios ignore the targeting costs
discussed in Section 8.3.2.
One of the reasons for the disparate outcomes is the variation in the effectiveness of the categories of
targeting mechanisms. e authors found that the bene t ratio of the median programme with means
testing or some other form of individual targeting was 1,50, while those for the median programmes with
indicator targeting and self-targeting were 1,32 and only 1,10, respectively. Effective targeting depends in
part on choosing appropriate mechanisms and good enforcement of programme rules.

8.4 Fiscal incidence

Evaluation is an essential element of successful policymaking and policy implementation. Fiscal incidence
analysis, which draws on information from household surveys and administrative datasets to identify the
bene ciaries and the bearers of the tax burdens and other sources of nancing of government expenditure
programmes, is a useful tool for evaluating the effectiveness of policies to reduce inequality and poverty.
Well-executed incidence studies assist policymakers to establish whether policy measures reach intended
bene ciaries and achieve their redistributive goals.
It is possible to study the incidence of government expenditure programmes and that of taxes separately.
However, a clear picture of the redistributive effects of policies requires comprehensive scal incidence
studies, which estimate the net result of the incidence of spending bene ts and tax burdens. Such analyses
can be done for speci c programmes (for example, government spending on education) or for the scal
system as a whole. Identi cation of bene ciaries and bearers of scal burdens is normally done for income
groups such as deciles or quintiles. In some countries, however, historical or present-day factors make other
forms of scal incidence analysis important. In South Africa, for instance, the distorting effects of apartheid
on the distribution of bene ts of government expenditure programmes have motivated many race-group-
based scal incidence analyses.9

Figure 8.3 Income concepts in fiscal incidence analysis

 e remainder of this section presents a conceptual framework developed by the Commitment to Equity
Institute at Tulane University and selected ndings of studies for South Africa and Namibia to illustrate key
principles and the value of scal incidence analyses.10 It should be stated at the outset that various methods
exist for undertaking such studies. ese methods all have advantages and disadvantages. We will return to
some of the weaknesses of the method illustrated here after presenting the conceptual framework and
ndings.
Figure 8.3 summarises the conceptual framework of the Commitment to Equity Institute’s scal
incidence studies. e starting points for such analyses are market incomes of individuals, which consist of
salaries and wages, income from capital (such as dividend income), private transfers (such as private
pensions and remittances) and contributory pensions (pension scheme bene ts funded from compulsory
contributions by employers and workers). Market incomes are the real-world equivalent of the concept of
primary income in Section 8.2. Net market incomes are market incomes less personal income tax payments
and employee contributions to old-age pension schemes. Adding direct transfers (such as cash transfers)
and near-cash transfers (such as free school textbooks and school feeding scheme meals) to net market
incomes yields disposable incomes. Post- scal incomes are calculated by subtracting value-added tax,
excise tax and other indirect tax payments from disposable incomes and then adding indirect subsidies on
foodstuffs, energy and other goods and services. e nal major income concept is nal incomes, which
consist of post- scal incomes plus in-kind transfers (government-provided education and health
services).11 Final incomes are similar to secondary income, which was introduced in Section 8.2.

Table 8.2 Findings of fiscal incidence analyses for South Africa (2010/11) and Namibia (2009/10)

Note: a In 2010/11, the lower bound poverty line for South Africa was R443 per person per month. In 2009/10, the
lower bound poverty line for Namibia was N$277,54 per person per month.
Sources: Inchauste et al. (2015: 31); World Bank and Namibia Statistics Agency (2017: 52).

Researchers often present the results of scal incidence studies as Gini coefficients and poverty rates for
the income concepts in Figure 8.3. Table 8.2 shows results of scal incidence studies for South Africa and
Namibia (Inchauste, Lustig, Maboshe, Pur eld & Woolard, 2015; World Bank and Namibia Statistics Agency,
2017) in this format. Note that the poverty rates are based on national poverty lines. Hence, they are not
comparable.
e Gini coefficients for the nal incomes of both countries (0,596 for South Africa and 0,429 for
Namibia) are markedly smaller than the ones for market incomes (0,771 for South Africa and 0,635 for
Namibia), which implies that scal measures make their income distributions signi cantly less unequal.
Hence, the data con rm the redistributive potential of scal systems. Yet the South African data also
emphasise the limits of scal redistribution: e Gini coefficient for nal incomes of 0,596 remains
exceptionally high despite the strong redistributive effects of the scal system.12 Fiscal redistribution might
not suffice to obtain a fair distribution of nal incomes if the distribution of market incomes is extremely
skewed.
Redistributive scal systems usually also reduce the incidence of poverty. is is true for both countries:
e poverty rates in South Africa are 46,2% for market incomes and 39,6% for post- scal incomes, while
those for Namibia are 22,2% for market incomes and 16,7% for post- scal incomes. Note that poverty rates
are not provided for nal incomes. Fiscal incidence studies exclude the monetary value of education and
health services when calculating the poverty-reducing effects of scal systems because households ‘are
unlikely to be willing to pay as much as the government spends on these services and as a result do not view
these services as part of their income’ (Inchauste et al., 2015: 29).
Fiscal incidence studies usually also provide information about the contributions to redistribution and
poverty reduction of the various components of scal systems. e ndings of Inchauste et al. (2015: 15–28)
and the World Bank and Namibia Statistics Agency (2017: 30–49) yield the following conclusions for South
Africa and Namibia. Both countries have progressive income tax systems that reduce income inequality by
levying higher tax rates on the richer strata of their populations.13 Indirect taxes are somewhat regressive in
South Africa and distribution-neutral in Namibia. e main reason why these taxes are not as regressive as
might be expected is that both countries zero-rate various items consumed mainly by the poor.14 e net
result is that the tax systems of both countries are slightly progressive. Especially the indirect taxes increase
poverty, though. Hence, the data for South Africa and Namibia con rm the statement in Section 8.2.2 that
tax systems often reduce income inequality, but do not contribute directly to poverty alleviation.
Both countries maintain large cash transfer programmes (mainly pensions for the elderly, child grants
and disability grants). South Africa’s programmes are better targeted, however, and have stronger
moderating effects on income inequality and poverty (see World Bank and Namibia Statistics Agency, 2017:
38–40). Indirect subsidies – free water, electricity, sanitation and refuse removal services in South Africa and
water and housing subsidies in Namibia – and publicly provided education and health services augment
these effects. ese outcomes underscore another important statement in Section 8.2.2: Well-targeted cash
and in-kind transfers can markedly reduce inequality and, in the case of the former, poverty as well.
Table 8.3 illustrates another way of presenting results of scal incidence analyses. For each of the income
concepts, it shows average per capita income levels for the deciles of the income distribution of South Africa.
e market income column in Table 8.3 is a stark reminder of the extreme nature of income inequality in
South Africa. Recall from Figure 8.3 that personal income tax payments and employee contributions to
pension schemes constitute the difference between market incomes and net market incomes. Hence, the
per capita income gures in the net market income column con rm the progressivity of direct taxes in South
Africa by showing that the burden of personal income taxes rests heavily on the three richest deciles. By
contrast, it transpires from the disposable income column that individuals in the poorer deciles bene t
markedly more (in relative and absolute terms) from cash transfers than individuals in the richest deciles.
e post- scal incomes column shows the effects of indirect taxes. Such taxes reduce absolute incomes in all
deciles.
Table 8.3 Average per capita incomes by market income deciles in South Africa (Rands, 2010/11)

Note: e disposable income gures exclude the value of free basic services (water, electricity, refuse removal and
sanitation).
Source: Inchauste et al. (2015: 30).

Various issues should be kept in mind when interpreting the results of scal incidence studies of this nature.
ese include the following:
• Data and other problems usually restrict the scope of such studies to parts of scal systems. For example,
the South African study discussed in this section analysed the incidence of 64,5% of general government
revenue and 43% of general government expenditure. e results of many such studies might have been
quite different if it had been possible to include additional parts of scal systems.
• e reliability of the ndings depends markedly on the quality of the household and other surveys that
generate the core data for scal incidence analyses.
• Most scal incidence analyses do not incorporate the effects of taxes and government spending bene ts
on the behaviour of bene ciaries and non-bene ciaries. In trying to identify the effects of scal systems
on inequality and poverty, such analyses compare the secondary and primary incomes of individuals (or
households) and assume that the primary incomes would have been the same if the government did not
levy taxes and did not provide any spending bene ts. In reality, howver, scal systems are likely to affect
the labour supply, consumption and saving behaviour of individuals and households. Incidence analyses
that do not incorporate behavioural effects of taxes and bene ts can signi cantly under- or overestimate
the in uence of scal systems on inequality and poverty.
• Fiscal incidence analyses typically measure the amounts of money governments spend on individuals or
households in various income groups. ese amounts are not necessarily reliable indicators of the actual
bene ts of the recipients; the services provided by governments may be of poor quality, for instance. is
is a major issue in scal incidence analysis in South Africa, where public education and healthcare
spending is high by international standards and well targeted (see Section 7.3), yet highly inefficient in
terms of outcomes. Chapter 10 returns to this important issue.

Key concepts
• benefit ratio (page 152)
• disposable incomes (page 153)
• errors of exclusion (type 1 targeting errors) (page 150)
• errors of inclusion (type 2 targeting errors) (page 150)
• final incomes (page 154)
• fiscal incidence analysis (page 152)
• indicator targeting (page 148)
• market incomes (page 153)
• means testing (page 148)
• net market incomes (page 153)
• post-fiscal incomes (page 154)
• primary income (page 145)
• secondary income (page 145)
• self-targeting (page 149)
• targeting (page 148)

SUMMARY
• Indicators such as poverty rates, Gini coefficients and income shares show that poverty and inequality
are severe problems in most SADC countries.
• Taxes, government expenditure programmes and regulations are important policy instruments for
reducing poverty and changing the distribution of income. ese instruments can be used to change the
distribution of primary income, the distribution of secondary income, or both. Governments should be
mindful of two sets of issues when designing policies for these purposes: whether policies will achieve
their aims (which has to do with whether policies reach the intended bene ciaries, how useful the
bene ts are to them, and what the developmental effects of policy measures are), and the ways in which
policies might change the behaviour of the intended bene ciaries.
• Targeting is the practice of using various mechanisms (such as means testing, indicator targeting and
self-targeting) to identify those in greatest need and to channel the bene ts of government spending
programmes to them. e poverty-reducing effects of public spending programmes can be enhanced
markedly if appropriate targeting mechanisms are chosen and if the programme rules are enforced
properly. Targeting can bring administrative, incentive and stigma costs, though. Hence, decisions
whether to use targeting should always be based on comparisons of bene ts and costs.
• Fiscal incidence analysis is a useful tool for evaluating the effectiveness of policies to reduce inequality
and poverty. Such analyses use information from household surveys and administrative datasets to
identify the bene ciaries and the bearers of the tax burdens and other sources of nancing of public
spending. Clear pictures of the redistributive effects of policies require comprehensive scal incidence
studies, which estimate the net result of the incidence of spending bene ts and tax burdens. Fiscal
incidence studies tend to have various shortcomings, however, and these should be kept in mind when
results are interpreted and used for policymaking purposes.

MULTIPLE-CHOICE QUESTIONS
8.1 Which of the following statements is/are correct?
a. Income from subsistence agriculture is included in primary incomes.
b. Secondary incomes include bene ts from government expenditure programmes.
c. Regulations are the only policy measures that can be used to make the primary distribution of
income less unequal.
d. Cash transfer programmes are examples of policy measures to make the secondary distribution of
income less unequal.
8.2 Which of the following statements is/are correct?
 a. Not all targeting mechanisms require substantial expenditures to identify intended bene ciaries.
b. Universal cash transfer programmes produce substantial bene t leakage.
c. Targeting mechanisms cannot have positive incentive effects.
d. Targeting hardly ever fails.
8.3 Which of the following statements is/are correct?
a. Fiscal incidence analysis ignores the distributional effects of taxes.
b. Disposable incomes re ect the redistributive and poverty-reducing effects of cash transfers, among
other things.
c. Fiscal incidence studies usually do not provide poverty rates for nal incomes.
d. Final incomes do not include the redistributive and poverty-reducing effects of indirect subsidies.
8.4 Which of the following statements is not true?
a. Fiscal incidence studies show that the scal systems of South Africa and Namibia reduce income
inequality.
b. Incidence studies con rm that South Africa and Namibia have progressive income tax systems.
c. Most scal incidence analyses ignore the behavioural effects of taxes and government expenditure
programmes.
d. One of the strengths of scal incidence analysis is its ability to measure the actual bene ts of
government expenditure programmes.

SHORT-ANSWER QUESTIONS
9.1 Explain the difference between primary incomes and secondary incomes.
9.2 Distinguish between three broad categories of targeting mechanisms.
9.3 Explain the difference between disposable incomes and post- scal incomes.

ESSAY QUESTIONS
9.1 Use the distinction between primary incomes and secondary incomes to identify two broad approaches
to changing the distribution of income, and brie y discuss two sets of issues that government should
keep in mind when implementing either of these approaches.
9.2 Use a graphical analysis to demonstrate the advantages of targeted cash transfer programmes over
universal ones.
9.3 Explain why potential bene ts and costs should be taken into account when policymakers decide
whether to use targeting mechanisms.
9.4 Discuss the purpose and most common shortcomings of scal incidence studies.
9.5 Fiscal incidence studies for Country A and Country B yield the following information:
a. What do these statistics reveal about the overall effects of the two scal systems on inequality and
poverty?
b. Brie y state the most likely effects of direct taxes, indirect taxes, cash transfers, indirect subsidies
and in-kind transfers on income inequality and poverty.
c. Use these propositions to suggest possible explanations for the gures in the table.
 Findings of fiscal incidence analyses for Country A and Country B

1 Some sections in this chapter refer to speci c social protection programmes in Southern African countries. We drew on Cirillo and
Tebaldi (2016) for information about such programmes.
2 At the time of writing this chapter, the gures in Table 8.1 were the most recent ones in the World Bank’s World Development
Indicators database.
3 Strictly speaking, social transfers form part of primary incomes (Yp). If included in Yp, such transfers should be omitted from
government spending (G) to prevent double counting.
4 Government expenditure on social services (for example, education and healthcare) forms part of general government
consumption in Table 1.1 in Chapter 1. Hence, it is included in the resource use measure of the size of the public sector. By contrast,
cash and in-kind transfers are included in the resource mobilisation measure under the heading ‘Subsidies, grants and other
transfer payments’.
5 Section 11.4.1 discusses the pitfalls of dedicated or earmarked taxation, among other things. To avoid these pitfalls, decisions about
the allocation of tax revenues for the nancing of programmes to reduce inequality and poverty should be integrated into the
government’s expenditure prioritisation process.
6 e second of these issues is linked closely to the effects of such programmes on the recipients’ behaviour, to which we return
below.
7 Indicator targeting is sometimes combined with means testing. In South Africa, for example, eligibility for the bene ts of most cash
transfer programmes depends on personal characteristics (such as disability and age) as well as income or wealth.
8 Chapter 9 discusses these incentive distortions in more detail.
9 Van der Berg (2014) provides a brief review of such analyses that shows the highly unequal distribution of government expenditure
bene ts in the apartheid era and the large resource shifts towards blacks that started in the mid-1970s and were completed after
the political transition in 1994.
10 For more detail on the conceptual framework and studies of other countries, see the website of the Commitment to Equity Institute
at http://commitmentoequity.org/.
11 Strictly speaking, nal incomes should exclude user fees and co-payments for government services (for example, healthcare).
However, the scal incidence studies for South Africa and Namibia discussed below do not provide information about such fees
and payments.
12 Inchauste et al. (2015: 31) compare scal redistribution in twelve middle-income developing countries (Armenia, Bolivia, Brazil,
Costa Rica, El Salvador, Ethiopia, Guatemala, Indonesia, Mexico, Peru, South Africa and Uruguay). South Africa’s scal system
reduces its Gini coefficient more than those of the other countries do. Yet because of South Africa’s exceptionally unequal
distribution of market incomes, its Gini coefficient for nal incomes remains the highest.
13 e tax rate structures of personal income tax systems are progressive if the average tax rate increases as income increases – see
Section 12.2.2.
14 Indirect taxes are levied on commodities and market transactions (see Section 12.2.4). Poorer individuals typically consume larger
percentages of their incomes and save smaller percentages than richer individuals do. e reality that consumption falls as a
percentage of incomes as incomes increase means that taxes on the consumption of commodities tend to be regressive – average
tax rates tend to fall as incomes increase.
Social insurance and social assistance

Krige Siebrits

e International Social Security Association (2018) de nes social security as ‘… any programme of social
protection established by legislation, or any other mandatory arrangement, that provides individuals with a
degree of income security when faced with the contingencies of old age, survivorship, incapacity, disability,
unemployment or rearing children. It may also offer access to curative or preventive medical care.’ is
chapter discusses the income security-related components of social security systems, which are core
elements of governments’ efforts to reduce poverty and income inequality. Such arrangements serve two
purposes. e rst is to protect individuals who normally earn enough to support themselves against
income losses or uctuations that would put their welfare at risk, for example, those caused by bouts of
illness or unemployment. e other purpose of such arrangements is to support individuals who cannot
earn enough to meet their basic consumption needs, for example, children in poor households, disabled
persons who cannot work, and elderly persons without enough savings. is chapter focuses on income-
augmentation programmes nanced or administered by governments. However, the exposition also gives
some attention to private arrangements with similar purposes, which are closely interlinked with public
ones.
is chapter consists of ve sections. Section 9.1 distinguishes between the two types of arrangements
that make up the income security-related components of social security systems, namely social insurance
programmes and social assistance programmes. Section 9.2 discusses social insurance programmes in more
detail. It provides an overview of the economic theory of insurance and explains the need for government
intervention to overcome the inability of private markets to provide insurance against the effects of certain
income losses. e topic of Section 9.3 is social assistance programmes. is section discusses the choice
between providing such assistance as cash or goods and services, as well as the pros and cons of adding
conditions to cash transfer programmes. Section 9.4 reviews the incentive problems associated with income
security programmes. In closing, Section 9.5 outlines and comments of various aspects of the South African
income security system.

Once you have studied this chapter, you should be able to:
distinguish between social insurance programmes and social assistance programmes
use aspects of the economic theory of insurance to explain the need for social insurance systems
discuss the choice between cash and in-kind transfer programmes
discuss the nature and effectiveness of conditional cash transfer programmes
discuss the incentive problems associated with social insurance and social assistance programmes
outline and discuss the South African income security system.
9.1 The two components of income security systems
As was pointed out earlier, the income security-related parts of public social security systems have two
components, namely social insurance programmes and social assistance programmes. Social insurance
programmes are funded from mandatory contributions by workers and employers. ese contributions,
which are known as social security taxes, are examples of payroll taxes (see Section 9.2). Only those who
have made such contributions may access the bene ts of social insurance programmes when affected by
income losses or uctuations. Social assistance programmes, on the other hand, are cash transfer
programmes funded from general tax revenues. Consequently, eligibility for social assistance is not
restricted to those who have contributed to dedicated funds from which bene ts are paid.1 e cash transfer
programmes referred to in Section 8.2.2 are examples of social assistance schemes.
Social insurance schemes have long been the core elements of income security systems in European
welfare states. Such schemes initially were established for industrial workers and were extended to other
groups over time. Social assistance schemes have played a residual role in these systems by protecting
members of vulnerable groups who do not qualify for social insurance bene ts. Insurance-dominated
systems have sufficed in these countries, because the vast majority of working-age individuals have formal
sector jobs that enable them to access bene ts linked to payment of social security taxes. e income
security systems of most developing economies are still evolving. On balance, the unemployment rates and
informal sector participation rates in these countries have been higher than those of the advanced
economies. is has limited governments’ scope to use social insurance arrangements linked to formal
sector jobs as income protection mechanisms. Although many developing economies have been expanding
the social insurance components of their income security systems, most have remained more reliant on
social assistance programmes than advanced economies have been. e extent and growth of formal sector
employment have not been the only determinants of the weights of these two parts of countries’ income
security systems. Factors ranging from the availability of scal resources to ideological preferences,
distributions of political power and even accidents of history have all shaped the structures of such systems.
While the remainder of this chapter focuses on social insurance and social assistance programmes, some
sections also refer to two private income security mechanisms that interact with the equivalent public ones.
e rst of these is occupational insurance. In some countries that lack payroll tax-funded social insurance
programmes, legislation or collective bargaining agreements make it mandatory for private sector workers to
belong to their employers’ pension or provident funds. Such workers receive retirement bene ts from the
returns on the contributions they and their employers make to these funds. e second type of private
income protection arrangements referred to in this chapter is inter- and intra-household transfers (for
example, remittances from migrant workers). ese arrangements are known as informal, traditional or
indigenous income security systems. Section 9.5.2 refers to the interplay between public and private
income security arrangements in South Africa.

9.2 Social insurance

e rst part of this section brie y outlines the income protection role of insurance. is is followed by an
overview of the economic arguments for social insurance schemes. e roots of these arguments are equity-
and efficiency-related market failures.

9.2.1 The income protection role of insurance

Assume an individual has to choose between two consumption bundles. e rst option is to consume R300
000 in each of the next two years, while the second is to consume R400 000 in the rst and R200 000 in the
second year. Which option would the individual choose? Recall the principle of diminishing marginal
utility, which maintains that the marginal utility of consumption decreases as the level of consumption
increases. One of the implications of this principle is that the above-average level of consumption in the rst
year of the second option would not increase the utility of the individual as much as the below-average level
of consumption in the second year would reduce it. It follows that the individual is likely to derive more
utility from the rst option than from the second even though the values of the two consumption bundles
are the same. Put differently, the principle of diminishing marginal utility implies that individuals would
derive more utility from a smooth path of consumption than from a variable one of the same total
magnitude, ceteris paribus.
e conclusion that individuals prefer to avoid uctuations in incomes and consumption levels also
holds in real-world conditions of uncertain outcomes. To be sure, some individuals may be able to sustain
their levels of consumption when they experience unexpected income losses by drawing down their own
savings, borrowing from banks or moneylenders, and obtaining assistance from charitable organisations,
friends or family members. ese options may be not be available to all individuals, however, or may be
inadequate if the income losses are large or permanent. Insurance against income losses caused by
retirement, illness, work-related injuries and periods of unemployment can be an alternative form of income
protection. is would involve making payments known as ‘insurance premiums’ to insurers who, in turn,
pay amounts of money when the insured suffer income losses caused by events covered by the insurance
contracts. Hence, insurance amounts to consumption smoothing: individuals reduce their consumption in
times of plenty to nance premiums that would enable them to sustain their levels of consumption by
means of payouts in times of economic hardship.

9.2.2 The rationale for social insurance

In well-developed market economies, private rms provide insurance against various risks, including losses
of possessions caused by theft and natural disasters. Individuals can also obtain retirement insurance from
private rms (Section 9.1 pointed out that this is known as occupational insurance). e question therefore
arises whether there is economic justi cation for government intervention in the form of social insurance
systems. is subsection outlines the efficiency- and equity-related market failures in private insurance
markets that justify such intervention.
Efficiency arguments for social insurance are based on market imperfections that prevent insurers from
determining actuarially fair premiums for cover against some forms of income losses. An actuarially fair
premium equals the expected loss of the insured. If, for example, there is a one per cent probability that a
worker who earns R300 000 per annum will become unemployed in the next year, the expected income loss
is R3 000 (1% of R300 000). Hence, the actuarially fair unemployment insurance premium will be R3 000 per
annum. In general terms, this can be written as follows:

where ∏ represents the actuarially fair insurance premium, p represents the probability of the loss and L
represents the potential size of the loss (see Barr, 1989: 62).2
To determine actuarially fair premiums, insurance providers must either know or be able to make
reliable estimates of p and L for individual clients. Probabilities of and income losses associated with
retirement are readily available or estimable, which explains why private rms offer retirement insurance in
South Africa and elsewhere in Southern Africa. e information needed to determine actuarially fair
premiums is not available for all causes of income losses, though. Workers’ probabilities of suffering
unemployment, for example, are determined by many complex factors, including technological change,
changes in the sectoral composition of economies, and characteristics such as skills, work experience and
personal motivation. e probabilities of career-interrupting or -terminating injuries at work also vary from
person to person because of differences in the nature of jobs, workers’ attitudes to safety issues and the
extent to which rms adhere to workplace safety laws, among other things. While insurance providers can
sometimes use historical data about injuries and unemployment rates in speci c occupations or sectors to
estimate probabilities of income losses, unexpected technological change and other factors can severely
reduce the reliability of such estimates. e effects of information problems that prevent determination of
actuarially fair premiums are that private rms either do not provide cover against the income losses in
question at all or restrict cover to low-risk market segments (for example, workers in occupations that are
least likely to be affected by unemployment or work-related injuries). ese outcomes are examples of the
effects of violations of the full information assumption of the benchmark model in Section 2.1.
Section 2.4.1 distinguished between situations in which both parties to a transaction lack information
and instances of asymmetric information (recall that information asymmetries occur when one party to a
transaction has more or better information than the other does).3 Both situations cause market failure. Two
types of information asymmetries have particularly pernicious effects in insurance markets, though:
• Adverse selection. Equation 9.1 implies that the determination of actuarially fair premiums is hampered
when workers can hide information about factors in uencing their probabilities of suffering losses from
insurance providers (p). e term ‘adverse selection’ describes situations in which persons who face
high probabilities of suffering losses – and are therefore keener than others to insure themselves against
such events – can hide their high-risk status from insurers. Adverse selection is common in markets for
unemployment and work-related injury insurance: insurers often cannot access all the information
possessed by workers about generic risk factors in speci c jobs or sectors as well as relevant personal
characteristics (for example, the effort levels of individual workers and the extent to which they adhere to
safety regulations). e only option available to insurers who cannot distinguish between low- and high-
risk clients is to charge everyone the same premium based on the average risk. However, this would
mean that workers with high probabilities of suffering income losses would face inefficiently low prices
for insurance and in all likelihood would buy more than the optimal quantities. By contrast, low-risk
workers would have to pay inefficiently high prices and would be likely to buy too little insurance. ese
distorted prices and quantities would represent allocative inefficiency. Low-risk workers may even stop
buying any insurance if they deem the premiums excessive. Such ‘opting-out’ would make insurance
provision by private rms impossible, because insurers would lose money if they only serve high-risk
workers whose premiums are lower than their expected losses.
• Moral hazard. Another implication of Equation 9.1 is that insurers cannot determine actuarially fair
premiums if insured parties have scope to behave in ways that increase the size of an insured loss (L) and
the probability of its occurrence (p). Moral hazard exists when insured parties can affect the liabilities of
insurers via such behaviour without the knowledge of the latter. Moral hazard is a distinct possibility in
markets for insurance against income losses. Some insured workers may undertake actions that increase
p. Persons with comprehensive unemployment insurance may be more inclined to work in sectors with
volatile employment levels, for example, and may be less motivated to perform to the best of their
abilities to avoid becoming redundant than those who lack such cover. In addition, full insurance may
affect some workers’ resolve to avoid work-related injuries.4 Actions that increase L are also possible:
compared to those without cover, insured persons may put less effort into job search activities after
becoming unemployed or into rehabilitation programmes after injuries at work. Moral hazard would
have the same effects that adverse selection has: insurance provision would be inefficient (usually when
insured parties increase L) or not provided at all (usually when potential clients can increase p to such an
extent that pro table provision would be impossible).

Note that adverse selection and moral hazard are market failures caused by information asymmetries at
different stages of insurance transactions. Adverse selection (which is also known as the problem of hidden
characteristics) arises before insurance contracts are signed when potential clients hide information about
their probabilities of suffering losses from insurers. By contrast, moral hazard occurs after the signing of an
insurance contract when clients engage in hidden actions that affect the size of the loss or the likelihood of
its occurrence from insurance companies. Hence, it is sometimes called the problem of hidden actions.
Government intervention in the form of social insurance systems can overcome some efficiency-related
insurance market failures. Most notably, such systems provide solutions to the adverse selection problems
that prevent private rms from providing any insurance against some causes of income losses or restrict
provision to low-risk clients. Governments can force all high-risk and low-risk workers as well as their
employers to join social insurance schemes. is makes it possible to provide insurance against income
losses caused by unemployment and work-related injuries to all workers in the formal sectors of economies
(Section 9.5 refers to examples of such schemes in South Africa). Social insurance schemes that cover entire
economies may also have other efficiency-related bene ts. For one thing, there may be economies of scale
in the administration of insurance schemes (this argument assumes that such economies are not
outweighed by X-inefficiency or other forms of government failure). In addition, it may be the case that
individuals would not insure themselves adequately against some risks of income losses if governments do
not force them to do so. is argument, which clearly has a paternalistic bent, implies that social insurance is
an example of the class of goods described in Section 3.5 as ‘merit goods’.
Social insurance schemes remain prone to other efficiency-related market failures, though. e payroll
taxes that fund social insurance schemes can be linked to the earnings of individual workers (L in Equation
9.1). It is impossible to link such taxes to individual-speci c loss probabilities (p in Equation 9.1), though,
because governments cannot obtain more information than private insurance providers can about relevant
risk factors. e reality that these adverse selection problems cannot be overcome implies that social
insurance systems exhibit inefficiencies very similar to those that undermine private provision of insurance
against the same causes of income losses: individual risk pro les do not determine the ‘tax price’ paid nor
the degree of cover provided to each worker. Moreover, moral hazard problems caused by information
asymmetries are as common in social insurance programmes as in private insurance markets (see also
Section 9.4).
e efficiency-related arguments for social insurance systems are signi cantly strengthened by equity
considerations. e basis of the equity argument for social insurance is that individuals without adequate
access to savings, borrowing opportunities or informal income protection systems cannot maintain their
consumption levels when they suffer income losses. Negative income shocks often push such individuals
and their dependents into poverty, and dispersed income losses – such as those caused by deep recessions –
can exacerbate income inequality. Some of these individuals (for example, workers who earn small or
irregular incomes in the informal sector) cannot afford any insurance against income losses. Governments
can use non-contributory social assistance programmes to enable these individuals to meet their basic
consumption needs when they suffer negative income shocks (see Section 9.3). Social insurance schemes
nanced by payroll taxes protect individuals who have formal-sector jobs but lack the means to afford full
insurance, that is, insurance that fully compensates them for income losses. Mandatory social insurance
programmes that collect payroll taxes from all workers (including those with high earnings) can mobilise
enough funds to bring lower-income individuals closer to the full insurance or consumption-smoothing
outcomes than private insurance products can.
Income losses caused by unemployment and work-related injuries have featured prominently in the
summary of the efficiency justi cation for social insurance schemes in this section. Equity considerations
closely related to those raised in the previous paragraph underpin the case for social insurance against a
third cause of income losses, namely retirement. Private rms offer occupational insurance, but their
annuity products do not necessarily provide full insurance to persons who earned low incomes during their
working lives, especially if they live long after retiring and if unexpectedly high rates of in ation erode the
values of their retirement savings. In fact, the retirement incomes of some individuals with occupational
insurance may be insufficient to enable them to meet their basic consumption needs.5 Compared to the
products offered by private rms, mandatory social insurance programmes funded from payroll taxes on all
workers can provide former workers who earned low incomes with more generous retirement bene ts.
As was pointed out in Section 5.3, the income redistribution inherent to social insurance systems can be
justi ed on Pareto grounds. It is also fully compatible with the externality argument for Pareto-efficient
redistribution outlined in that section.

9.3 Social assistance

Section 9.1 stated that the social assistance parts of income protection systems disburse cash transfers to
needy individuals on a non-contributory basis. Such schemes are very important anti-poverty instruments,
both as complements to contributory social insurance systems and in their own right. Moreover, social
assistance programmes can be strongly redistributive if nanced from progressive taxes (a tax is progressive
if the average tax rate increases as consumption, income or wealth increases – see Section 11.2.1). Sections
9.3.2 and 9.5 will show that empirical evidence con rms the poverty-reducing and redistributive effects of
social assistance.
 Social assistance programmes were traditionally justi ed with reference to livelihood protection effects,
that is, effects on poorer individuals’ ability to maintain minimum standards of living, even when they lose
other income sources. Debates on social assistance now also focus on livelihood promotion effects, which
have to do with sustainable poverty reduction via improvements in living standards over time. Livelihood
promotion effects become possible when social assistance programmes promote investment in education,
healthcare, nutrition, social networks, and income-generating assets. Apart from strengthening the well-
established equity arguments, the possibility of such effects adds an important efficiency argument to the
case for social assistance: investments in education, healthcare and nutrition can boost economic growth
(the dynamic efficiency of economies – see Section 2.3) by making the labour force more productive.
To be sure, the possibility of livelihood protection and livelihood promotion effects depends strongly on
how recipients use social assistance money. Two sets of risks come to mind. e rst and most obvious is
that the recipients may squander the money on luxuries and so-called ‘sin consumption’ (expenditure on
gambling, harmful drugs, alcoholic beverages, tobacco products, et cetera). Alternatively, they may use all
the money for consumption smoothing purposes (livelihood protection) instead of investing some of it to
obtain livelihood promotion bene ts. Hence, the question arises whether it is possible to design social
assistance programmes in ways that enhance their bene ts and reduce such risks. e remainder of this
section discusses two design issues that may affect the effectiveness of social assistance programmes. ese
are the choice between providing the bene ts in the form of cash or goods and services (Section 9.3.1) and
the effects of adding behavioural conditions to cash transfer programmes (Section 9.3.2).

9.3.1 The choice between cash and in-kind transfers

Social assistance programmes can provide cash or in-kind transfers to bene ciaries. In-kind transfers can
take two forms: direct provision of goods and services to targeted groups, or subsidisation of the prices of
market goods and services. Direct provision of some goods and services (such as foodstuffs and housing)
can serve the same purpose as cash transfers, namely to enable poor persons and households to meet basic
consumption needs.
e core theoretical argument for cash transfers is that recipients who are allowed to choose how to use
social assistance bene ts can attain higher levels of utility than those who are given goods and services
selected by government officials. Figure 9.1 illustrates this choice-related argument. It uses standard tools for
analysing consumer choices, namely budget lines and indifference curves. A budget line shows
combinations of two goods or services that a consumer can afford given prices and his or her income. Its
slope is the negative of the price ratio of the goods or services. An indifference curve depicts combinations of
two goods or services that yield the same level of utility to a consumer. e marginal rate of substitution
between the two goods or services gives the slope of an indifference curve. e horizontal axis in Figure 9.1
shows quantities of food consumed per month, while the vertical axis shows quantities of other goods and
services consumed per month.
Figure 9.1 The choice-related argument for the superiority of cash transfers over in-kind transfers

Source: Adapted from Rosen and Gayer (2014: 265).

e initial equilibrium in Figure 9.1 is at E1, where the consumer’s budget line, AB, is tangent to
indifference curve I1. At E1, the consumer’s monthly consumption of food amounts to 0QF1 and that of other
goods and services to 0QO1. Assume that the government introduces a new social assistance programme that
provides a food parcel valued at 0QF2 to low-income individuals. e consumer satis es the income means
test that determines eligibility for this bene t. e in-kind bene t enables the individual to consume more;
hence, it is equivalent to an income increase and can be depicted by an outward shift of the budget line. e
individual’s cash income, however, remains the same. At the prevailing prices, this level of cash income does
not allow spending in excess of 0QO2 on other goods and services (the Y-intercept of the budget line depicts
how much the consumer can afford if he or she uses all cash income to purchase other goods and services).
Hence, line ACD represents the budget constraint after the introduction of the food parcel scheme. e
individual is now at E2, where the values of the monthly consumption bundles of food and of other goods
and services are 0QF2 and 0QO2, respectively.
Note that the in-kind transfer has increased the individual’s consumption bundles of food and of other
goods and services, and that the indifference curve associated with E2, namely I2, represents a higher utility
level than that associated with the initial equilibrium E1, namely I1. Yet, unlike an equivalent cash transfer,
the in-kind transfer does not enable the bene ciary to select combinations of food and other goods and
services along the hypothetical line segment EC. e preferred combination of this individual (monthly
consumption bundles of food and of other goods and services of 0QF3 and 0QO3, respectively, which would
have yielded a utility level shown by indifference curve I3) lies on this line segment. Indifference curve I3 lies
above indifference curve I2, which implies that the individual would have attained a higher level of utility if
he or she had been given cash instead of the food parcel. e reality that many in-kind transfer programmes
are likely to yield such sub-optimal outcomes underpins the choice-related argument for the superiority of
cash transfers.
is argument is not always decisive, though. Various counterarguments emphasise potential bene ts of
in-kind transfers. For one thing, such transfers can reduce the costs of targeting by serving as self-targeting
mechanisms (see Section 8.3). As an alternative to costly means testing, governments can provide staple
foods consumed mainly by the poor (such as maize meal) at subsidised prices to anyone who wishes to
purchase it. e reality that very few affluent consumers buy such products makes this a low-cost way to
achieve some of the bene ts of targeting discussed in Section 8.3.1. It should be kept in mind, however, that
interventions of this nature invariably distort prices and market outcomes. A second counterargument is
that policymakers may nd it easier to mobilise funding for in-kind transfer programmes, because public
support for programmes providing essential goods and services is often stronger than for cash transfer
schemes. Finally, Section 10.2 will outline the powerful case for in-kind provision of goods and services with
positive external bene ts (recall from Section 3.6 that the marginal social bene ts of such goods and services
exceed their marginal private bene ts because bene ts accrue to non-users as well).
Some proponents of in-kind transfers also use an argument that follows from the signi cant in uence of
recipients’ use of bene ts on the effectiveness of social assistance schemes. It states that in-kind transfer
programmes are more likely to achieve social assistance goals than cash transfer programmes are, because
at least some recipients of cash transfers will squander the money on luxuries or ‘sin goods’. A survey of 30
studies on African, Asian and Latin America countries (Evans & Popova, 2017) reported that the evidence for
this argument was weak.6 ese studies found that cash transfers were generally not spent on alcoholic
beverages and tobacco products; if anything, most participating households reduced their consumption of
‘sin goods’ after the introduction of the transfers. Two factors possibly caused this response: most of these
programmes were targeted at women (who generally are more likely than men are to spend cash bene ts on
the needs of children) and accompanied by strong social messaging that discouraged misuse.
It is in any case not true that governments can prevent all misuse of social assistance bene ts by only
providing in-kind transfers. Social assistance recipients with a strong preference for luxuries and ‘sin goods’
can sometimes sell in-kind bene ts – such as foodstuffs – to get cash, and attempts to prevent this can
signi cantly increase the administrative costs of in-kind transfers. Moreover, Figure 9.1 suggested that the
problem would not necessarily be solved by preventing sales of in-kind bene ts. Recall that the introduction
of food parcels increased the consumption bundles of food as well as other goods and services even though
the recipient did not resell any foodstuffs. e reason why the recipient could afford larger quantities of
other goods and services after the introduction of food parcels was that the income he or she previously
spent on food was freed up for other uses. ese other uses may or may not have included purchases of
luxuries and ‘sin goods’.
On balance, the case for using cash transfer programmes as income protection mechanisms is strong. It
rests on the theoretical argument about the utility bene ts to recipients, the lack of compelling evidence of
large-scale squandering of cash transfers, and the reality that in-kind programmes are not fail-safe
mechanisms to prevent misuse of social assistance bene ts. More generally, it is important that choices
between cash and in-kind transfer programmes should be informed by rigorous empirical analysis of data
on the use and effects of both types of transfers in speci c contexts.

9.3.2 Conditional cash transfer programmes

e introduction to this section emphasised that cash transfer programmes are very important instruments
for reducing poverty in developing countries. Governments traditionally used cash transfers for livelihood
protection purposes, and relied on other policy instruments (including education, healthcare and asset-
redistribution programmes) for livelihood promotion effects. As stated in Section 8.2.3, several developing
countries have recently introduced conditional cash transfer programmes (CCT programmes), which are
cash transfer programme with livelihood protection as well as livelihood promotion goals.
e twin aim of CCT programmes is to combat current poverty (by providing income support that
enables consumption smoothing) and future poverty (by encouraging human capital accumulation to break
inter-generational poverty cycles). To achieve this aim, CCT programmes provide cash transfers to
households that meet certain requirements. e most common purpose of the conditions is to increase the
human capital of children. Mexico’s ‘Prospera’ programme, for example, provides cash transfers and food
supplements to households if the children attend school regularly and if all family members attend health
workshops and make regular visits to health centres. e Community-Based Conditional Cash Transfer
Programme in Tanzania, which is discussed in more detail in Box 9.1, has similar conditions. Several other
developing countries – including Brazil, Colombia, Jamaica, Nicaragua and the Philippines – also have CCT
programmes.
e aims of CCT programmes make perfect sense. Basic economic theory, however, postulates that all
attempts by policymakers to force rational individuals to change their behaviour distort their decisions and
reduce their utility. (is is a variant of the choice-related argument for cash transfers in Section 9.3.1.)
Hence, an important question arises: are conditions appropriate measures to enhance the ability of cash
transfer programmes to achieve livelihood promotion effects? Economists have identi ed two sets of
considerations that justify behavioural conditions:7
• Efficiency-related market failures. e marginal social bene ts of parents’ investment in the education
and healthcare of their children may exceed the marginal private costs: such investment can increase
communities’ stocks of knowledge and skills and prevent the spread of disease (cf. Section 3.6.2).
Parents, however, may not consider these external bene ts when they take decisions about investing in
the health and education of their children. CCT programmes can overcome such externalities by
increasing investment in the human capital of children to socially optimal levels. Policymakers can also
use such programmes when mismatches between the preferences of parents and the interests of children
result in inefficiently low levels of investment in education. Some parents underinvest in the education of
their children simply because they prioritise own consumption needs, but economic motives often come
into play as well. In households that depend on subsistence agriculture, for example, the short-term
bene ts to parents of using their children as labourers may outweigh the longer-term bene ts of sending
them to school. In such cases, cash grants tied to school enrolment and attendance can enhance
efficiency and welfare.
• Equity-related targeting considerations. Conditions can be useful targeting mechanisms for cash transfer
programmes when means testing is not feasible. e requirement to visit public health facilities on a
regular basis, for example, would impose high opportunity costs on richer people who usually use
private health facilities. Hence, these conditions could make transfers more redistributive by tilting the
cost-bene t calculation of higher-income groups against participation. Policymakers should keep in
mind, however, that households without school-age children cannot bene t from programmes with
school attendance and child health conditions. In addition, poor households may conclude that the
costs of participation exceed the bene ts and opt out of cash transfer programmes if the conditions seem
too onerous to them. In such cases, potential equity and efficiency gains are foregone.

On balance, empirical studies of CCT programmes have found positive effects on household consumption,
reductions in child labour and, in some cases, increases in saving and investment (see, for example, Box 9.1).
Studies also show that conditions boost school enrolment and the use of health facilities. However, the
effects on education and health outcomes depend on the quality of the services provided by governments.
While most unconditional cash transfer programmes yield similar bene ts, the few direct comparisons of
the effects of the two types of cash transfer programmes have suggested that conditions usually strengthen
the bene ts.8

Box 9.1 Tanzania’s Community-Based Conditional Cash Transfer Programme9

 The Government of Tanzania established the Tanzania Social Action Fund (TASAF) in 2000 as a major
community-driven development initiative. The Community-Based Conditional Cash Transfer Programme, which is
one of the components of TASAF, was introduced as a pilot programme in three particularly poor districts in
2010. The Programme disbursed cash to poor households if the members satisfied three conditions: children
up to the age of five had to visit a health clinic six times per annum, children aged seven to fifteen had to be
enrolled in school and achieve attendance rates of at least 80%, and elderly people had to visit a health clinic
once a year. One of the notable features of this programme was that communities elected representatives who
helped to select participating households and monitored their compliance with the conditions. This feature
reflected the community-driven orientation of TASAF.
The pilot phase of the conditional cash transfer programme was designed with scientific assessment of its
effects in mind. It provided the conditional benefits to households in 40 randomly selected villages in the three
districts, and researchers then compared education, healthcare and other outcomes in those villages to those
in the 40 other villages where households did not receive cash transfers. An assessment published in 2014
(Evans et al., 2014) showed that the conditional cash transfers had clear positive effects. Compared to their
peers in households that did not receive the conditional transfers, the members of participating households
were markedly healthier and had higher primary school attendance and completion rates. The beneficial effects
on school completion rates were especially large for girls. Analysis of spending patterns indicated that the
participating households mainly used the cash transfers to buy health insurance, shoes for children and
livestock (especially chickens). The poorest households also sharply increased their non-bank savings.
Concern that the involvement of community members in the selection and monitoring of participating
households would have jeopardised the cohesion of communities proved unfounded. On the contrary,
households that received conditional transfers were more likely than others to have attended community
meetings, participated in community projects and expressed trust in other community members.
These positive outcomes led to the creation of a much larger safety net programme known as the
Productive Social Safety Net (PSSN). This intervention, which was introduced in 2013 to support the poorest
15% of Tanzanians, combines conditional cash transfers with labour-intensive public works programmes and
other livelihood enhancement initiatives to support sustainable income-generation activities among poor
households.

It is too soon to say if long-term livelihood promotion bene ts will complement the promising short- to
medium-term effects of CCT programmes. However, the effects of such programmes on human capital
accumulation are likely to be negligible in countries where school attendance among poor children is very
high already or where schools and hospitals cannot provide high-quality education and healthcare services.

9.4 Incentive effects of income protection systems

us far, this chapter has identi ed various bene ts of income protection systems. However, many income
protection programmes also create disincentives or perverse incentives. omas Schelling (1985: 6–7), the
joint winner of the Alfred Nobel Memorial Prize in Economic Sciences in 2005, explained this as follows:

Policy issues are preponderantly concerned with helping, in compensatory fashion, the unfortunate and
the disadvantaged. An unsympathetic way to restate this is that a preponderance of government policies
have the purpose of rewarding people who get into difficulty. ere is no getting away from it. Almost any
compensatory programme directed toward a condition over which people have any kind of control …
reduces the incentive to stay out of that condition and detracts from the urgency of getting out of it. It is a
rare ameliorative programme that has no visible way, by its in uence on behavior, to affect the likelihood
or the duration or the severity of the circumstances it is intended to ameliorate. And most commonly – not
always but most commonly – the effect on behavior is undesired and in the wrong direction. To keep the
issue in perspective we can observe that private insurance … can have the same adverse in uence on
behavior.
Section 9.2.2 discussed a well-known example of such incentive problems, namely moral hazard in social
insurance schemes. is section provides two other examples and brie y comments on the implications of
such problems for income protection systems.
e rst example is the possibility that income transfers – such as universal income grants nanced from
general tax revenue –may create a disincentive to work. Figure 9.2 explains the effect of such transfers on the
work effort of an economically active person who is also subject to income tax. e horizontal axis of the
gure shows the 24 hours available to the person per day, and the vertical axis his daily income. Line AX is
vertical, because the time endowment of 24 hours per day is xed at all income levels. He can use this time
endowment for work or other activities (here labelled ‘leisure’). e person’s income is determined by his
allocation of the time endowment between work and leisure. If we assume that the person does not earn
non-wage income,10 we can write the relationship between his income and allocation of the time
endowment between work and leisure as follows:

Figure 9.2 The effect of an income transfer on work effort

where I is the daily income of the person, w his hourly wage, and L the number of hours per day he devotes
to leisure. If the hourly wage is R100, for example, the individual will earn R500 per day if he works ve hours
and uses the remaining 19 hours for leisure. A longer working day of eight hours (which implies 16 hours of
leisure) will yield a bigger income of R800 per day. e budget constraint BA depicts such relationships
between the person’s income levels and allocations of the time endowment. Its slope is the hourly wage w,
which also represents the opportunity cost of leisure. e indifference curves in Figure 9.2 (I0, I1 and I2)
depict the person’s preferences for work and leisure. Initially, the individual is in equilibrium at E0, where
the budget constraint BA is tangent to indifference curve I0. At E0, he works AC hours and devotes 0C hours
to leisure. is allocation of the time endowment yields income level 0D.
Assume the government now introduces a universal income grant, that is, a cash transfer to all citizens
of the country irrespective of age or economic status. Its value is BG per person per day. Hence, the budget
constraint shifts parallel to the right to GF. Note that the grant provides the individual with an income of AF
(which is equal to BG) when he devotes no time to work. is income oor is augmented by wage income as
he works more hours from A towards the origin along the horizontal axis. e new equilibrium is at E1 on the
new budget constraint GF and the higher indifference curve I1. Despite working fewer hours and devoting
more time to leisure than before the introduction of the transfer – the number of hours worked decreased
from AC to AH – the individual now has a higher income (0J > 0D) and a higher level of welfare. ese
changes re ect the income effect of the cash transfer.
Next, the government imposes an income tax to nance the universal income grant programme. Unlike
the cash transfer, this policy change has an income effect as well as a substitution effect. e tax changes the
slope of the budget constraint, because it reduces the hourly wage w received by workers. Recall that such a
reduction also implies a drop in the opportunity cost of an additional hour of leisure. Hence, the budget
constraint swivels inwards from GF to KF. e reduction in the relative price of leisure induces an increase in
the number of hours of leisure, which is the substitution effect. e income effect of this tax-induced wage
change increases the hours worked.
e nal equilibrium is at E2 on indifference curve I2, where the individual works even fewer hours per
day than before the introduction of the tax (AL, as opposed to AH). Reductions in the levels of income (0M <
0J) and welfare now accompany the reduction in work effort.
eoretical analysis therefore suggests that income tax- nanced cash transfers can reduce the incentive
to work in two ways: the income effect of the cash transfer (shown by HC) as well as the net impact of the
income and substitution effects of the income tax (shown by LH) are negative.
A second example of disincentive effects associated with social assistance transfers are the ‘poverty traps’
caused by some means tests. us, persons’ incentives to save for retirement during their working lives are
weakened by means tests that base eligibility for and the value of old-age pension on their non-pension
incomes. Such means tests also discourage them from working after they reach the legal retirement age.
Figure 9.3 illustrates these effects of some means tests for a hypothetical old-age pension scheme. Line
AEFG shows the monthly incomes of elderly persons from all sources other than social assistance
pensions.11 e pension scheme works as follows. Elderly persons whose non-pension incomes range from
zero to R2 400 per month receive the full social assistance pension of R1 600 per month. is implies that the
total income of an elderly person without non-pension income is R1 600 per month, while that of an elderly
person with non-pension income of R2 400 per month is R4 000 per month. Segment BC of line BCDEFG
depicts the pension-inclusive total incomes associated with non-pension incomes in this interval. For non-
pension incomes ranging from R2 401 to R3 600, the value of the pension drops by R1 for every additional
rand of non-pension income the elderly person has.12 Hence, the pension-inclusive total incomes of all
elderly persons with non-pension incomes in this range are R4 000 per month, as shown by the horizontal
segment CD of line BCDEFG. e income threshold of the means test that determines eligibility for old-age
pensions is a monthly income of R3 600 per month; furthermore, the minimum pension paid to an eligible
elderly person is R400 per month. is implies that an elderly person with non-pension income of R3 600
per month would have a pension-inclusive total income of R4 000 per month. Someone with non-pension
income of R3 601, however, would not be eligible for a pension. e near-vertical segment DE of line
BCDEFG depicts the ‘drop-off’ in total incomes when the pension falls away. e two lines share the
segment EFG, because the incomes of elderly persons who earn R3 601 per month or more from other
sources are not augmented by social assistance pensions.
Figure 9.3 Disincentive effects of means tests for old-age pensions

Source: Adapted from Van der Berg (2001a: 129).

e following example illustrates the disincentive effects of the sliding-scale bene t values and the
income threshold of the means test. Assume that a worker saves enough to have a non-pension income of
R2 401 per month after retirement. When deciding whether to increase her retirement savings, she has to
consider two things. First, the implication of the ‘drop-off’ associated with the income threshold is that the
pension-inclusive total incomes of retired persons with monthly non-pension incomes from R3 601 to R3
999 (segment EF of line BCDEFG) would be less than those of their peers with monthly non-pension
incomes from R2 400 to R3 600 (segment CD of line BCDEFG).13 Second, the sliding-scale bene t values
mean that a retired person with non-pension income of R3 600 per month would be no better off than one
with a non-pension income of R2 401: both would have a pension-inclusive total income of R4 000 per
month. Hence, she will not be better off in retirement, and may even be worse off, unless she can increase
her retirement savings sufficiently to generate a non-pension income of at least R4 001 per month. She may
well regard such a large increase as infeasible and refrain from saving more. Persons who reach the legal
retirement age but have the option to remain in paid employment would face similar disincentive effects.
To be sure, only persons with prospective or actual non-pension incomes in the relevant income
intervals would be affected by these disincentives. Moreover, such effects can be mitigated by appropriate
reforms. For example, the sliding scale bene t system can be changed so that the value of the pension
decreases by only 50 cents for every additional rand of non-pension income the bene ciary has, while the
extent of the income drop at the income threshold can be moderated by reducing the minimum pension
paid. Such reforms may reduce the effectiveness of means tests as targeting mechanisms, however, or limit
the number of poor households assisted with a given programme budget. Such trade-offs are common in the
design of social assistance programmes.
In the statement quoted earlier, omas Schelling pointed out that many of the disincentive effects of
income protection programmes are inherent to such schemes. Some, however, are effects of design errors.
To avoid such errors, policymakers should anticipate possible incentive effects of targeting mechanisms,
conditions, and bene t levels and keep these in mind when they design income protection programmes.
Once such programmes are in place, policymakers should strive to obtain reliable empirical information
about incentive effects of speci c design features and, where feasible, change those that distort behaviour.

9.5 The South African income security system

is section has three parts. Section 9.5.1 discusses the coverage of the South African income security
system in the three major life stages of individuals and families (childhood, working age, and old age),
Section 9.5.2 comments on the scope and adequacy of the system, and Section 9.5.3 reviews empirical
research on the effectiveness of its social assistance components.
Recall from the introduction to this chapter that income security-related programmes are components of
social security systems. Figure 9.4 contextualises the following discussion of South Africa’s income security
programmes by summarising its full social security system. e gure shows the three formal parts of the
system and the elements of each, as well as their types of nancing mechanisms and sources of funding:14
• Social insurance programmes. South Africa has three types of social insurance programmes that provide
income support or compensation for de ned-risk events. ese are the Unemployment Insurance Fund;
the Compensation Funds, which assist workers who sustain work-related injuries or diseases; and the
Road Accident Fund, which provides compensates to victims of road accidents caused by wrongful or
negligent driving. e Unemployment Insurance Fund and the Compensation Funds are nanced by
payroll taxes levied on workers and employers. e Road Accident Fund receives a portion of the fuel
levy on sales of petrol and diesel, which is paid by road users.
• Social assistance programmes. e social assistance component of South Africa’s social security system
consists of seven cash transfer programmes: child-support grants, care-dependency grants, foster-care
grants, disability grants, old-age pensions, war veteran’s grants, and grants-in-aid. Apart from the foster-
care grants, these cash transfer programmes are all means tested. Some also have health status-based
eligibility criteria. General tax revenues fund the various cash transfer programmes. is means that all
taxpayers – workers and rms, but also recipients of cash grants whose purchases of goods and services
are subject to value-added tax and excise taxes (cf. Sections 16.1 and 16.3) – contribute to the nancing of
these programmes.
• Occupational insurance programmes. Section 9.1 stated that the occupational insurance programmes in
South Africa are examples of private income security arrangements. Binding industrial-council and other
agreements between employers and employees have made it mandatory for most workers in the formal
sectors of the South African economy to join medical aid schemes and their employer’s pension fund or
provident fund. Hence, the term ‘occupational insurance’ refers to contributory retirement and health
insurance tied to workers’ conditions of employment. Legislation merely enables employers and
employees to enter into such agreements, and government involvement is limited to regulation to
protect contributors to provident funds, pension funds and medical aid schemes.

e remainder of this section discusses the income security-related components of this system in more
detail. e discussion covers all the programmes listed in Figure 9.4 except medical aid schemes (as was
stated in the introduction to this chapter, medical insurance is not an income security programme), the
Road Accident Fund (which is not primarily focused on poverty alleviation or income redistribution), and
grant-in-aid (which is provided to recipients of old-age pensions, war veteran’s grants and disability grants
who require full-time care because they suffer from physical or mental disabilities).15
Figure 9.4 The South African social security system

Source: Adapted from Woolard and Leibbrandt (2010: 3).

9.5.1 Coverage
is subsection discusses the components of the South African income security system listed in Figure 9.4 in
more detail.16 e discussion revolves around coverage in the three major life stages of individuals and
families, namely childhood, working age, and old age.

9.5.1.1 Childhood
South Africa has three grant programmes to assist children in poor families. e child- support grant, which
replaced the child maintenance grant in April 1998, is the largest of the three. e South African Social
Security Agency (SASSA) pays these grants to the primary caregivers of children aged 18 years or younger. In
the 2017/18 nancial year, the child-support grant amounted to R380 per month and reached 12 269 084
children. e means test stipulated that single and married caregivers must have earned less than R45 600
and R91 200 per annum, respectively, to have been eligible for child-support grants.
Care-dependency grants are paid to the parents or caregivers of children between the ages of one and 18
years who suffer from severe physical and mental disability and are in permanent home care. (Disabled
persons between the ages of eighteen and the retirement age receive state disability grants, while those over
the retirement age are eligible for old-age pensions). e parents or caregivers of 147 467 care-dependent
children received such grants in the 2017/18 scal year. At the time, the value of the grant was R1 600 per
month. e income thresholds that determined eligibility for the grant were R192 200 per annum for single
and R384 000 per annum for married parents or caregivers.
Children deemed by the courts to be in need of care are placed in the custody of foster parents. e aim
of foster-care grants is to reimburse these parents for the cost of caring for children who are not their own.
Hence, the grants are not means-tested and fall away if the foster parents formally adopt the children. In the
2017/18 scal year, SASSA disbursed 416 016 foster-care grants of R920 per month.

9.5.1.2 Working age

e Unemployment Insurance Fund (UIF), which is administered by the Department of Labour, provides
unemployment bene ts to qualifying workers. Employees and employers each contribute 1% of employees’
earnings up to a maximum of R124,78 per month to the UIF and the proceeds are used to pay bene ts to
contributors or their dependents in instances of unemployment, illness, death, maternity and adoption of a
child. Access to UIF bene ts is limited to one day for every six completed working days, up to a maximum of
365 days per period of four years. Furthermore, the bene ts vary from 60% of the earnings of low-income
earners to 38% of the earnings of higher-income earners. e net asset value of the UIF has been increasing
steadily in recent years: in the 2017/18 nancial year, for example, its total assets and total liabilities were
R160,1 billion and R13,4 billion, respectively (National Treasury, 2019a: 96). In that year, the UIF collected
contributions of R18,7 billion and earned R9,5 billion in investment revenue, while its bene t payments
totalled R9,2 billion (Unemployment Insurance Fund, 2018: 15).
It is well known that unemployment is exceptionally high in South Africa. In the four quarters that made
up the 2017/18 scal year, the narrowly de ned numbers of unemployed persons of working age ranged
between 5,9 million and 6,2 million (Statistics South Africa, 2019b: Table 2).17 e reality that the UIF
received only 679 988 claims for unemployment bene ts in that year (Unemployment Insurance Fund, 2018:
23) con rmed that the system plays an important but limited role as far as providing relief to unemployed
persons and their dependents are concerned. Two aspects of the rules of the UIF limit its ability to assist the
unemployed. e rst is that many unemployed persons have never worked in the formal sectors of the
economy. Because such persons have never contributed to the UIF, they are not eligible for unemployment
bene ts. In addition, many persons of working age remain unemployed for long periods and exhaust their
time-limited bene ts before they nd new jobs.
e Compensation Funds provide income bene ts and medical care to workers who are injured on the
job, funding for the rehabilitation of disabled workers and survivor bene ts to the families of victims of
work-related fatalities. e main Compensation Fund, which is administered by the Department of Labour,
covers workers in sectors other than mining and construction, while the Department of Health administers
the Mines and Works Compensation Fund, which provides bene ts to victims of work-related lung diseases
in the mining sector. Private rms licensed by the Compensation Commissioner administer two other funds:
the Rand Mutual Association for workers in the mining industry and the Federated Employers’ Mutual
Assurance for workers in the building industry. In 2017/18, the main Compensation Fund received
contributions to the tune of R7,3 billion and interest and dividend income of R4,2 billion, while it disbursed
bene ts amounting to R3,6 billion (Compensation Fund, 2018: 10,77). e Fund then had assets of R67,3
billion and liabilities of R34,3 billion (National Treasury, 2019a: 96).
Disability grants are available to people who are disabled in circumstances other than road accidents
and work-related accidents. Disabled persons between the ages of eighteen and the retirement age who do
not receive other state grants and who are not cared for in state institutions are eligible for such grants
(disability grants are converted to old-age pensions when recipients reach the retirement age). Strict medical
criteria determine eligibility: the disability should be permanent and sufficiently severe to prevent the
affected person from working. Hence, the purpose of the grant is to compensate disabled persons for loss of
income. In the 2017/18 scal year, the disability grant amounted to R1 600 per month. SASSA disbursed such
grants to 1 061 866 persons with disabilities. e means test stated that eligibility was restricted to single
persons whose annual incomes and assets did not exceed R73 800 and R1 056 000, respectively, and married
persons whose incomes and assets did not exceed R147 600 and R2 112 000, respectively. Persons with
disabilities who passed the means test did not necessarily receive the full grant amount of R1 600 per month,
though. Disability grants were paid on a sliding scale: the more income from other sources disabled persons
had, the smaller the grants they received.
9.5.1.3 Old age
South Africa has a sophisticated retirement fund market that provides coverage to some 60% of employees in
the formal sectors of the economy.18 is coverage rate as well as the ratio of pension fund assets to GDP are
comparatively high by international standards. is shows the extent to which membership of an
occupational fund is accepted as an obligatory condition of employment despite the quasi-voluntary nature
of the system.19
Coverage rates vary markedly across income categories, though: while most middle- and high-income
earners are well covered – partly because generous income tax incentives exist for retirement saving – many
lower-income earners do not belong to retirement, pension or provident funds. Moreover, the retirement
incomes of large numbers of pension and provident fund members are inadequate: only some 10% of retired
South Africans can maintain the levels of consumption they enjoyed while they were working. Various
factors contribute to this state of affairs, including the low earnings of many workers, low preservation rates
(many employees withdraw all their retirement savings when they change jobs) and the eroding effects on
bene ts of the high fees structure of the South African retirement industry. Hence, lower-income South
Africans (including many formal sector workers) depend on social pensions in old age.
Persons aged 60 and above who pass the means test receive old age pensions. In the 2017/18 scal year,
the old-age pension (which is also known as the grant for older persons) amounted to R1 600 per month for
elderly persons between the ages of 60 and 75 and to R1 620 per month for those above the age of 75. In that
year, SASSA disbursed 3 423 337 old age pensions. e means test for single persons limited eligibility to
those whose incomes and assets did not exceed R73 800 and R1 056 000, respectively, and married persons
whose incomes and assets did not exceed R147 600 and R2 112 000, respectively. As was the case with
disability grants, old age pensions were paid on a sliding scale: elderly persons with more income from other
sources received smaller grants than those who did not.
South Africans who fought in the Second World War or the Korean War are eligible for war veteran’s
grants. In 2017/18, SASSA disbursed 134 such grants, which amounted to R1 620 per month. e war
veteran’s grant had the same means test and sliding scale payment system as the old-age pension.

9.5.2 Scope and adequacy

On balance, South Africa has a relatively advanced social security system for a middle-income country.
When assessing the scope of the income security components of this system, it should be kept in mind that
international comparisons of social security expenditures underestimate its size. As was pointed out earlier,
many countries levy payroll taxes on employers and employees to nance social insurance bene ts. Such
taxes and bene ts are re ected in the social security components of governments’ accounts. In South Africa,
the occupational insurance contributions of private-sector employers and employees do not ow through
the coffers of the state; hence, neither these contributions nor the bene ts they nance show up in the
accounts of the public sector.
In 2017, the South African retirement fund industry consisted of 5 158 funds with 16 945 651 members
and assets of R4,3 trillion (Financial Services Board, 2018: 17, 18).20 In that year, it mobilised R238,5 billion in
contributions and disbursed bene ts amounting to R314,6 billion to its members (Financial Services Board,
2018: 20–21). By comparison, the South African government expended only R150,3 billion on social grants in
the 2017/18 scal year. As was pointed out in Section 9.5.1, however, the size and sophistication of the
industry belies the reality that many South Africans lack adequate retirement provision. Hence, National
Treasury has embarked upon a wide-ranging retirement reform process. e goals of this initiative are to
raise the level of savings in South Africa (including savings for retirement), increase transparency in the
retirement savings industry and improve the overall efficiency of the retirement system. To ensure that the
intended reforms achieve these goals, National Treasury has been engaging with various stakeholders and
has undertaken research on aspects of the retirement system in South Africa.
Table 9.1 Numbers of social grants (1996/97–2017/18)

Sources: South African Social Security Agency (2009: 20); South African Social Security Agency (2018: 21).

Many of the intended reforms will affect only the regulation of the private retirement insurance industry.
Some of these reforms, however, are likely to have direct or indirect implications for social pensions and,
hence, the public nances more generally. One of the major proposals has been to introduce mandation or
auto-enrolment. e intention of this proposal has been to make retirement provision mandatory – that is, to
convert the occupational insurance system to a social insurance system – though there are still questions
about the implications of such a step for low-income and vulnerable workers. Measures to preserve the
retirement insurance bene ts of workers when they change jobs are also under consideration.
e social assistance component of the South African social security system is particularly large by
international standards. Table 9.1 shows that the number of social grants disbursed by SASSA increased from
2 408 742 in 1997 to 17 509 995 in 2018. Almost one-third of the South African population now receive a
social assistance grant, which is a remarkably high gure for a middle-income country. Although all of the
grant types except the war veteran’s grant experienced signi cant growth in numbers of recipients during
the past two decades, the major driver of growth in the system as a whole has been the introduction and
subsequent expansion of the coverage of the child-support grant.21 Fully 70% of all grants disbursed in
2017/18 were child-support grants, while 19,6% were old-age pensions and 6,0% disability grants. Because it
is the smallest in money terms, however, the child-support grant does not dominate government spending
on social assistance programmes. In 2017/18, for example, 42,7% (R64,1 billion) of social assistance
spending took the form of old-age pensions, 37,1% (R55,8 billion) of child-support grants, and 13,9% (R20,9
billion) of disability grants (National Treasury, 2019b: 351).
Government expenditure on social grants increased in nominal terms from R16,0 billion in 1997/98 to
the already mentioned gure of R150,3 billion in 2017/18. Figure 9.5 shows that such expenditure increased
markedly as percentages of total general government spending and of GDP from 2000 to 2005 – largely
because of the expansion of the coverage of the child-support grant – but stabilised thereafter. In 2017/18,
spending on grants amounted to 8,2% of general government expenditure and 3,2% of GDP. e social grant
system should remain nancially feasible if the number of grants grows in line with the population (which
implies that new grants programmes should not be introduced and that the scope of existing ones should
not be expanded) and the rand values of the various grants remain constant in real terms.
Historical factors explain the evolution as well as the scope of the South African social security system.
e initial spur for the creation of this system was the desire to create an embryonic welfare state for whites
during the apartheid period. Following a pattern common to modern economies, the authorities rst
introduced social insurance schemes and established regulatory frameworks for occupational insurance.
Means-tested social assistance for the poor and vulnerable (for example, the elderly and the disabled)
complemented these schemes. Social assistance bene ts were eventually extended to other population
groups, albeit initially at much lower values. e government adopted the general principle of moving
towards parity in social spending programmes in the late 1970s, and social grants levels reached equality
even before the political transition from apartheid.
is process gave South Africa a relatively advanced social security system for a semi-industrialised
country, but also tied its scope to the needs of apartheid-era whites. Preferential access to education and job
reservation measures meant that most whites had employment security. Hence, they were adequately
protected by a system that combined occupational insurance with targeted social assistance programmes
and insurance against short spells of unemployment. Even in the apartheid era, however, this system did not
meet the needs of the other South African population groups. For one thing, many members of these groups
lacked access to occupational insurance. In addition, these groups generally were more exposed to
unemployment than whites were. e inadequacy of the unemployment insurance component of the
income protection system was severely exacerbated by the sharp increase in structural unemployment from
the mid-1970s onwards.
Van der Berg (1997) used needs and coverage by existing programmes as criteria to judge the adequacy
of the South African income protection system for four income classes. e affluent (largely the richest
quintile of the population in per capita income terms) generally have high levels of education that provide
relative job security. Members of this income class usually have occupational and private retirement
insurance, as well as cover against cyclical unemployment through the UIF system. Members of the stable
urban working class (quintile 4 of the population in per capita income terms) also have relatively high levels
of education and skills. ey rely more on occupational insurance than on social assistance transfers,
although some receive old-age pensions after retiring. eir main income protection concern is the risk of
relatively long unemployment spells. Persons in the insecure formal sector (quintile 3 of the population in
terms of per capita incomes) have relatively low levels of education and limited skills. Some members of this
income class experience upward social mobility, but face signi cant risks of losing their jobs. Hence, they
are inadequately protected against unemployment. In addition, limited access to occupational insurance
leaves them reliant on social assistance bene ts. Members of the fourth group, the outsiders (quintiles 1 and
2 of the population in terms of per capita incomes), generally have low levels of education and skills, and
high dependency burdens (in other words, relatively large numbers of non-working dependents).
Unemployment is rife among members of this group. Furthermore, workers in these quintiles often have
insecure jobs that pay low wages. Elements of the social assistance system (such as old-age pensions and
child-support grants) are crucial sources of income for these workers. Viewed from the perspective of the
outsiders, however, the absence of a separate grant for the unemployed is a serious gap in the social
assistance system.
Figure 9.5 Government spending on social grants in South Africa (1997/98–2017/18)

In summary, the South African social security system contains large gaps despite its unusual breadth
compared to the social security systems of other developing countries. Severe structural unemployment
limits the adequacy of the country’s social assistance programmes and its social security system as a whole.
Unemployment not only limits individuals’ access to occupational insurance schemes, but also exacerbates
the unmet social security needs of poor households.

Table 9.2 Income sources of five types of South African households (2008–2017)

Source: Zizzamia, Schotte and Leibbrandt (2019: 25).

9.5.3 The effectiveness of South Africa’s social assistance programmes

is subsection focuses on three aspects of the effectiveness of South Africa’s social assistance programmes:
the effects of these programmes on the welfare of poor households, the labour market, and population
growth.

9.5.3.1 Social grants and the welfare of poor households

Social grants have become an increasingly important source of household income in South Africa. One of
the many surveys that have con rmed this has been Statistics South Africa’s annual General Household
Survey: the portion of surveyed households that reported income from one or more grants increased from
27,5% in 2002 to 44,6% in 2017 (Statistics South Africa, 2002b: 67; 2017: 170). In fact, 20,1% of households
reported that grants were their main sources of income in 2017, compared to 18,1% in 2002 (Statistics South
Africa 2002b: 65; 2017: 54). Another survey, the National Income Dynamics Study (NIDS), has shown that
grant income is particularly important for poorer households. is is evident from Table 9.2, which is based
on one of the ndings of a study that pooled all the data from the rst ve waves of NIDS.22 e table shows
the income sources of ve types of households in South Africa: the chronic poor, the transient poor, the
vulnerable, the middle class and the elite. e dominant income sources of the nancially secure middle
class and elite households were salaries and other forms of labour-related remuneration.
Elite households also obtain considerable income from their investments. By contrast, vulnerable and
poor households obtain signi cantly smaller fractions of their incomes from paid employment. ese
households rely more heavily on social grants and, to a much lesser extent, on remittances from other
households. Grants constituted 50,2% of the incomes of chronically poor households, and labour income
only 41,0%.
e scal incidence study for 2010/11 discussed in Section 8.4 provided evidence of the extent to which
social grants augment the incomes of poor households in South Africa (cf. Table 9.3).23 It also made it
possible to draw preliminary conclusions about the effects of the grants on poverty and income inequality.

Table 9.3 Average per capita incomes by market income deciles in South Africa (2010/11)
Note: Section 8.4 de ned the concepts ‘net market income’ and ‘disposable income’. e disposable income gures
exclude the value of free basic services (water, electricity, refuse removal and sanitation).
Sources: Inchauste et al. (2015: 30).

It is clear that South Africa’s social grants are well targeted at poor households: in 2010/11, the value of
cash transfers decreased from an average of R2 163 in the poorest decile and R2 262 in the second decile to
only R283 in the richest decile. In the poorest and second deciles, the respective average values of the cash
transfers received by each person was almost eleven times and slightly more than three times the average
net market incomes. On average, the value of such transfers also exceeded the net market income of persons
in the third decile, and amounted to 70% of the average net market income of persons in the fourth decile
and 35% of the average net market income of persons in the fth decile. ese gures show that social grants
are vital mechanisms for augmenting the incomes of poor households in South Africa and for protecting
them against negative shocks to their other income sources.
Several studies have con rmed that the grants markedly reduce poverty and income inequality. e
World Bank (2018a: 72–73), for example, estimated that social assistance transfers reduced the poverty rate
by 8%, the poverty gap by 32%, and the Gini coefficient by 10,5% in 2014/15.24 ese were large reductions
compared to those achieved by the social assistance programmes of other countries. Exercises of this nature
compare the actual incidence of poverty and income inequality to the incidence that would have been
obtained if social assistance programmes had not existed. e results are indicative only, because they are
sensitive to the choice of a poverty line and rest on the assumption that the existence or non-existence of
social assistance programmes does not in uence aspects of the behaviour of bene ciaries, such as their
labour supply and household formation decisions. It should also be noted that the last two columns of Table
9.3 con rm an important conclusion in Section 8.4: the gaps between the incomes of rich and poor South
Africans remain extremely large even after the moderating effects of social grants have been taken into
account.
Providing well-targeted cash transfers to the poor is at best a necessary condition for reducing poverty.
e actual impact of these transfers depends crucially on how poor people use the money. In this regard,
Section 9.3 identi ed two threats to the poverty-mitigating potential of cash transfers: bene ciaries may
squander the money on luxuries and so-called sin goods, or use all of it for livelihood protection purposes
that do not generate sustainable improvements in living standards. South Africa’s social grants programmes
are currently not structured as livelihood-promoting interventions, as the intended bene ciaries (children,
the disabled and the elderly) are not economically active persons. However, the grants could contribute to
the future productivity of children in poor households if the income is invested in their sustenance and
education. Several studies have explored the impact of grant income on the spending patterns of recipient
households in South Africa, with particular emphasis on the nutrition and school attendance of children.
Many households pool income from grants and other sources (e.g. remittances and labour market
activity) to nance consumption spending. Hence, it is often difficult to establish which goods and services
households buy with grant income. No evidence of large-scale squandering of grant money has emerged,
however; studies have found that households’ use of grant money does not differ greatly from their use of
other incomes. Hence, social grants mainly boost the food spending of bene ciaries (Van der Berg &
Siebrits, 2010: 22). Research has highlighted the nutritional bene ts to children of increases in food
spending associated with receipt of child-support grants and social pensions (cf. Coetzee, 2013; Woolard &
Leibbrandt, 2010: 19–23). Studies using height-for-age ratios as indicators of nutritional inputs in early
childhood found that child-support grants improved the nutritional inputs of children in KwaZulu-Natal.
Furthermore, each grant received by a household markedly reduces the probability that any child in that
household goes hungry. With regard to old-age pensions, it appears that the gender of the recipient
in uences the nutrition and health-status effects of social grants. For example, the weight-for-height ratios
of girls living in households with female pension recipients increased after the large increases in social
pensions during the late 1980s and early 1990s, while no increases were discernible for boys or girls living in
households with male pension recipients.
Evidence also suggests that income from grants encourages school attendance among recipients of
child-support grants and children who live with recipients of old-age pensions. However, these positive
effects were small owing to the already high school enrolment and attendance rates in South Africa. is also
suggests that the bene ts of adding school attendance conditions to the child-support grant are likely to be
small (cf. Leibbrandt & Woolard, 2010: 26).

9.5.3.2 Labour market effects of social grants

As was indicated earlier, the vehicle for the provision of unemployment bene ts to able-bodied South
Africans is the Unemployment Insurance Fund, which is a contribution-based social insurance institution.
e only members of the working-age population who qualify for social grants are persons with disabilities
who also pass the means test. is does not mean that the South African social assistance system does not
affect labour-market participation. However, such labour market effects of social grants do not arise mainly
because of the mechanism emphasised in Section 9.4, namely distortion of the relative prices of work and
leisure. In fact, survey responses indicate that poor South Africans generally prefer wage income to the
currently available grants (cf. Van der Berg & Siebrits, 2010: 23). Instead, the grant system seems to in uence
the supply of labour through direct and induced effects on retirement decisions, household formation and
job search activities.
Direct effects have to do with the incentives faced by the actual recipients of grants. e means tests that
determine eligibility for the old-age pension and the disability grant have income thresholds that have ‘drop-
off’ effects such as those discussed in Section 9.4; furthermore, the values of the grants disbursed to eligible
elderly and disabled persons are reduced on a sliding-scale basis in accordance with their incomes from
other sources. Empirical studies have con rmed that these characteristics reduce the work incentives of
disabled and elderly persons (Ranchhod, 2006; Mutasa, 2010). Such disincentive effects are worsened by the
very high unemployment rate in South Africa and by other labour-market disadvantages confronting elderly
and disabled South Africans, many of whom have limited skills and reside in rural areas where job
opportunities are scarce. An additional factor that affects people with disabilities is that the available job
opportunities tend to be temporary and badly paid. e resulting small differential between the disability
grant and available market wage means that there is little incentive for many disabled persons to take up
paid work.
Empirical studies have also explored the indirect or induced labour-market effects of elements of the
South African social assistance system on persons other than the actual recipients. e social pension has
become a major source of support for unemployed South Africans of working age, especially in rural areas.
Unemployed youths and younger adults often delay forming new households, or return to their parents or
relatives to share in the pension income of the elderly. is is an example of interaction between public and
private income security systems (cf. Section 9.1). is attachment to households with pension recipients
apparently affects labour-market participation in two ways (cf. Leibbrandt, Lilenstein, Shenker & Woolard,
2013: 8–10). Some individuals stop looking for work when they join these households, often because they are
located in rural areas where job opportunities are scarce. On the other hand, there is evidence that access to
pension income may stimulate labour-market participation by enabling some household members to search
for jobs away from home. Positive effects of this nature appear to be particularly strong for women.
e available evidence indicates that receipt of child-support grants has boosted caregivers’ labour-force
participation, but has not affected their search behaviour or actual employment (Williams, 2007). is is not
unexpected, as the small value of the child-support grant makes major labour-supply effects unlikely.

9.5.3.3 Fertility
Public discourse has shown concern about the possibility that child-support grants have been encouraging
needy women (especially teenagers) to have more children. e incidence of teenage pregnancy did
increase markedly during the mid-1990s, but then levelled off around the turn of the millennium (cf. Van der
Berg & Siebrits, 2010: 25). Hence, it does not appear as if the introduction of the child-support grant in 1998
had a strong impact on teenage pregnancy. Moreover, the increase in teenage pregnancy was not especially
pronounced among those in the lower income groups who are eligible for means-tested child-support
grants.
 Incentives matter at the margin; hence, the availability of the grant may have tilted the cost-bene t
calculations of some in favour of having more children. In all likelihood, however, a small grant of this nature
would not have been decisive in the reproductive decisions of many people. e introduction of the child-
support grant probably at most slightly slowed the ongoing decline in the fertility rate compared to what
would have happened otherwise.

Key concepts
• actuarially fair premium (page 162)
• adverse selection (page 163)
• asymmetric information (page 163)
• conditional cash transfer programmes (page 169)
• consumption smoothing (page 162)
• diminishing marginal utility (page 162)
• full insurance (page 165)
• informal (traditional or indigenous) income security systems (page 161)
• livelihood promotion effects (page 166)
• livelihood protection effects (page 166)
• moral hazard (page 164)
• occupational insurance (page 161)
• social assistance programmes (page 161)
• social insurance programmes (page 161)
• social security (page 160)
• universal income grant (page 173)

SUMMARY
• Social security systems consist of programmes that provide protection against various contingencies,
including income losses causes by unemployment, disability, illness, work-related injuries and old age.
Such systems consist of social insurance programmes funded from mandatory contributions by workers
and employers and social assistance programmes funded from general tax revenues. Social insurance
programmes are usually complemented by private income security mechanisms such as occupational
insurance and informal income security schemes.
• Insurance-based income protection systems enable consumption smoothing. Efficiency- and equity-
related market failures in private insurance markets provide the economic rationale for government
intervention in the form of social insurance systems. e efficiency-related market failures have to do
with problems of hidden characteristics (adverse selection) and hidden actions (moral hazard). Well-
functioning social insurance schemes can overcome some, albeit not all, such effects of information
asymmetries in private insurance markets. Equity considerations strengthen the efficiency-related case
for social insurance systems. e income redistribution effected by social insurance systems can be
justi ed on Pareto grounds.
• e traditional economic justi cation for social assistance programmes based on livelihood protection
effects (poorer individuals’ ability to maintain minimum standards of living) are now complemented by
livelihood promotion effects (sustainable poverty reduction via improvements in living standards over
time).
• e justi cation for using cash transfer programmes as income protection mechanisms rests on the
utility bene ts to recipients, the lack of compelling evidence of large-scale squandering of cash transfers,
and the limited ability of in-kind programmes to prevent misuse of social assistance bene ts. Although
powerful, these considerations are not necessarily decisive: valid targeting, political economy and
externality-based arguments exist for in-kind transfer programmes.
• Conditional cash transfer schemes provide cash transfers to households that meet certain requirements.
e aims of such programmes are to combat current poverty (by providing income support that enables
consumption smoothing) and future poverty (by encouraging human capital accumulation to break
inter-generational poverty cycles). Efficiency- and equity-related market failures (such as externalities
and targeting considerations) justify the use of conditions to change the behaviour of recipients.
Empirical studies of conditional cash transfer programmes have found positive effects on household
consumption, reductions in child labour and, in some cases, increases in saving and investment.
Conditions also boost school enrolment and the use of health facilities, but their effects on education
and health outcomes depend on the quality of the services provided by governments.
• Income protection systems have various bene ts, but many also create disincentives or perverse
incentives. Income transfers nanced from general tax revenues, for example, may create a disincentive
to work. In the same way, means tests that base eligibility for and the value of old-age pension on non-
pension incomes weaken persons’ incentives to save for retirement during their working lives. Such
means tests also discourage them from working after they reach the legal retirement age. While some of
the disincentive effects of income protection programmes are inherent to such schemes, others are
preventable effects of design errors.
• South Africa has a well-developed income security system that consists of a few social insurance
schemes and an unusually large suite of social assistance programmes. Almost one-third of the South
African population now receive a social assistance grant. Despite its unusual breadth compared to the
social security systems of other developing countries, the South African social security system contains
major gaps: severe structural unemployment limits the adequacy of the country’s social assistance
programmes and its social security system as a whole. Unemployment limits access to occupational
insurance schemes and exacerbates the unmet social security needs of poor households.
• Empirical studies show that social assistance grants improve the welfare of poor households in South
Africa. Apart from markedly raising the incomes of such households, grants also seem to improve the
nutrition, health outcomes and school attendance of children. Direct effects have to do with the
incentives faced by the actual recipients of grants. Researchers have found that the means tests for and
sliding-scale values of the old-age pension and the disability grant reduce the work incentives of disabled
and elderly persons. Grants also have signi cant household formation effects. In a clear example of
interaction between public and private income security systems, the social pension has become a major
source of support for unemployed South Africans of working age, especially in rural areas. Some
working-age individuals stop looking for work when they join households to share in the pension income
of the elderly. In other cases, however, access to pension income seems to boost labour-market
participation by enabling some household members to search for jobs away from home.

MULTIPLE-CHOICE QUESTIONS
9.1 Which of the following statements is/are correct?
a. Social insurance programmes are nanced from payroll taxes.
b. Social assistance programmes are nanced from mandatory contributions by workers and rms.
c. Occupational insurance are nanced from general tax revenues.
d. Informal social security systems include transfers between and within households.
9.2 Which of the following statements is/are correct?
a. An actuarially fair insurance premium equals the potential size of the loss.
b. Adverse selection is also known as the problem of hidden characteristics.
c. Moral hazard arises after the signing of insurance contracts.
d. e income redistribution inherent to social insurance programmes can be justi ed on Pareto
efficiency grounds.
9.3 Which of the following statements is/are correct?
a. Livelihood protection effects have to do with consumption smoothing.
b. Price subsidies are examples of in-kind transfers.
 c. e bene ts of in-kind transfer programmes can be used to nance purchases of ‘sin goods’.
d. Conditions can improve the targeting of social assistance programmes.
9.4 Which of the following statements is not true?
a. South Africa’s Unemployment Insurance Fund is an example of an occupational insurance
scheme.
b. Comparisons with industrial countries underestimate the size of South Africa’s social security
system.
c. e child-support grant is South Africa’s most expensive social assistance programme.
d. e means test for South Africa’s old age pension reduces elderly persons’ incentive to work.

SHORT-ANSWER QUESTIONS
9.1 Distinguish between social insurance programmes and social assistance programmes.
9.2 Discuss the economic rationale for conditional cash transfer programmes.
9.3 Explain why some types of means tests reduce the incentive to work.

ESSAY QUESTIONS
9.1 Outline the economic rationale for social insurance systems.
9.2 ‘eoretical considerations and empirical evidence suggest that social assistance bene ts should
always be provided in the form of goods and services.’ Do you agree with this statement? Explain your
answer.
9.3 Explain why income tax- nanced cash transfer programmes can reduce the incentive to work.
9.4 Discuss the welfare and labour market effects of South Africa’s social assistance programmes.

1 As was explained in Section 8.3, however, means testing and other targeting mechanisms may restrict access to the bene ts of
social assistance programmes.
2 Equation 9.1 ignores transaction costs (that is, the administrative costs and normal pro ts of the insurance company). Hence, a
more realistic representation of an actual insurance premium is ∏ = pL + T, in which all transaction costs are included in T.
3 Uncertainty about the effects that arti cial intelligence technologies may have on many existing jobs is a contemporary example of
an information problem affecting both parties to insurance transactions (that is, workers and potential providers of unemployment
insurance). Such uncertainty complicates the estimation of unemployment probabilities by potential insurers as well as the career
prospects of prospective and existing workers.
4 To be sure, rational insured persons will not engage in such behaviour unless the perceived bene ts exceed the perceived costs.
Some insured office workers may spend too much time at their desks and ignore the risk of back problems, and some professional
athletes may ignore the risk of injuries caused by overtraining. Insurance cover is unlikely to cause life-threatening behaviour,
though.
5 Section 9.5 shows that this is a signi cant problem in South Africa.
6 Section 9.5 summarises evidence from South African studies.
7 Das, Do and Özler (2005: 64–66, 69–71) discuss these considerations in more detail.
8 Bastagli, Hagen-Zanker, Harman, Barca, Sturge, Schmidt and Pellerano (2016) provide a comprehensive review of the effects of
conditional and unconditional cash transfer programmes.
9 e source of the information in this box is Evans, Hausladen, Kosec and Reese (2014: 1–8).
10 Non-wage income includes investment income (e.g. interest and dividend income) and private transfers (e.g. remittances and
gifts).
11 ese sources would include salaries and wages, interest and dividend income, and annuities bought with accumulated savings.
12 Sliding-scale bene t values is a feature of many social assistance programmes, including the old-age pension and disability grant
programmes in South Africa (cf. Section 9.5.1.2. and Section 9.5.1.3). e purpose of restricting the amounts paid in this way is to
assist more poor persons with a given programme budget.
13 is effect of an income threshold is a well-known example of the incentives costs of targeting mechanisms (cf. Section 8.3.2).
14 Recall from Section 9.1 that informal income security systems, which are also known as traditional or indigenous income security
systems, consist of transfers between and among households.
15 In 2017/18, the monthly value of grant in aid was R380. It was provided to 192 091 persons.
16 is subsection reports features of the social grants system in the 2017/18 scal year. e rand values of the various grants are from
National Treasury (2018a: 62) and the means tests from the South African Social Security Agency (2017: 7). e South African
Social Security Agency (2018: 21) is the source of the numbers of and expenditures on the various grants. e Minister of Finance
 announced in his Budget Speech on 20 February 2019 that the monthly values of the various grant will be as follows from 1 April
2019: state old age grant – R1 780 (under 75 years of age) or R1 800 (over 75 years of age), war veteran’s grans – R1 800, disability
grant – R1 780, foster-care grant – R1 000, care-dependency grant – R1 780, and child-support grant – R425 (National Treasury,
2019a: 57).
17 For statistical purposes, only persons of working age who are actively seeking employment are classi ed as unemployed. In reality,
however, many unemployed persons in South Africa have given up trying to nd jobs. A more complete estimate of the extent of
joblessness can be obtained by adding these so-called ‘discouraged work seekers’ to official unemployment numbers. In the four
quarters of the 2017/18 scal year, this broader measure of unemployment ranged from 8,4 million work seekers to 8,8 million
(Statistics South Africa, 2019b: Table 2).
18 e following overview of retirement savings in South Africa draws heavily on National Treasury (2014b: 6–9).
19 Employers are not required by law to make retirement provision for their workers. If employers choose to set up retirement
schemes, however, employees must join as a condition of employment (National Treasury, 2014b: 6).
20 e membership gure is in ated by double counting, because some individuals belong to more than one fund.
21 Eligibility was limited to children younger than seven when the child-support grant was introduced in 1998. It was extended in
steps to children under the ages of nine (2003), eleven (2004), fourteen (2005), fteen (2008) and eighteen (2012).
22 e aim of NIDS has been to track and improve understanding of trends in poverty in South Africa. To this end, it has been
surveying the same nationally representative sample of more than 28 000 individuals in 7 300 households every two years since
2008.
23 Whereas the information in Table 9.2 is for households, that in Table 9.3 is for individuals.
24 Section 8.1 de ned poverty rates and Gini coefficients, while Section 8.3.1 de ned poverty gaps. e effects estimated by the World
Bank were based on three poverty lines for 2015 determined by Statistics South Africa: a food poverty line of R441 per person per
month, a lower bound poverty line of R647 per person per month, and an upper bound poverty line of R992 per person per month
(World Bank, 2018a: 8–9).
Social services

Krige Siebrits

e Government Finance Statistics system of the International Monetary Fund, which is the best-known
system for classifying government expenditure, distinguishes ve functional categories of social services.1
ese are education; health; social protection; housing and community amenities; and recreation, culture
and religion.2 Section 8.2.2 pointed out that effective provision of social services raises the living standards of
bene ciaries and improves their ability to access income-generating opportunities. Hence, governments
rely heavily on social services to reduce poverty and disparities in the primary distributions of income
generated by market forces.
is chapter discusses public provision of education and healthcare services. ese social services are
particularly important for economic development (cf. Todaro & Smith, 2015: 382). On the one hand,
education and healthcare services are vital inputs in the aggregate production functions of developing
economies – recall the discussion of the connection between social infrastructure and long-run economic
growth in Section 7.6. On the other hand, contemporary development economists emphasise that outcomes
of effective provision of social services, such as knowledge and long and healthy lives, are also ends or
objectives of economic development. Yet the link between the extent of government spending on education
and healthcare and developmental outcomes in these areas is often tenuous. Hence, this chapter
emphasises the importance of effectiveness in the delivery of social services.
e remainder of the chapter consists of ve sections. Section 10.1 discusses the market failures that
justify government involvement in the provision of education and healthcare. e case for in-kind subsidies,
which are important policy instruments in the provision of social services, is outlined in Section 10.2.
Section 10.3 discusses effectiveness in the provision of social services and introduces the notion of the
‘service delivery chain’. e last two sections of the chapter discuss aspects of public provision of social
services in South Africa. Section 10.4 focuses on education and Section 10.5 on healthcare.

Once you have studied this chapter, you should be able to:
discuss the market failure-based arguments for government involvement in the provision of education
discuss the market failure-based arguments for government involvement in the provision of healthcare
explain the excess burden of a subsidy
show that an in-kind subsidy can be welfare-enhancing if the consumption of the good or service in question
yields positive externalities
explain the importance of the ‘service delivery chain’ in the provision of social services
discuss public provision of education in South Africa
discuss public provision of healthcare in South Africa.
10.1Arguments for government intervention in education and
healthcare
Section 3.5 categorised education and healthcare as mixed services. Mixed goods and services exhibit
characteristics of private and public goods and services, and can be supplied by the public sector or the
private sector. As was also pointed out in Section 3.5, this explains the co-existence of public and private
schools and hospitals in Southern African countries and further a eld. e reality that private rms can
pro tably supply education and healthcare services raises the question whether economic justi cation
exists for public provision or nancing. Put differently: are education and healthcare markets prone to
market failures? is section shows that these markets are characterised by various market failures that
establish prima facie cases for government intervention.

10.1.1 Education
Externalities, information problems and capital market failures cause allocative inefficiency in education
markets. Equity considerations further strengthen the case for government intervention in such markets.
As was pointed out in Section 3.5, individuals are unlikely to take external bene ts to other persons into
consideration when deciding how much they should spend on education services. ese external bene ts
include free dissemination of valuable information and the reduced pressure on government expenditure
caused by the decreases in birth rates and crime levels that usually accompany improvements in education
levels. Section 3.6.2 argued that failure to consider these bene ts would cause market failure in the form of
underprovision and underpricing of education services. It also demonstrated how policymakers could use
Pigouvian subsidies to internalise such positive externalities and increase allocative efficiency.
Imperfect information also gives rise to allocative inefficiency in education markets. According to human
capital theory, spending on education and training by individuals represents investments in human capital
(i.e. characteristics that determine workers’ productivity such as knowledge and skills). In deciding whether
to make such investments, individuals and households have to estimate and compare private bene ts and
costs. Although education brings social, psychological and other bene ts as well, human capital theory
focuses on its nancial bene ts, that is, additional earnings owing from access to higher-paying jobs.
Education also has two types of costs: direct costs such as fees, books, computers, and accommodation and
transport expenses; and indirect or opportunity costs in the form of earnings foregone while studying. Figure
10.1 is a stylised representation of two possible outcomes of such cost-bene t decisions.
Panel A (the gure on the left) shows the costs and earnings of persons who receive no schooling, start
working at the age of six and retire at 62. Such persons avoid the direct and opportunity costs of education
and have relatively long careers in paid employment. Yet their starting wages are likely to be relatively low
and their earnings grow relatively slowly because progression to higher-paying jobs often depends on
academic quali cations. e persons whose costs and bene ts are depicted in Panel B (the gure on the
right) go to school at the age of six and enter into paid employment at the age of 22 after obtaining tertiary-
level academic quali cations. ey incur the direct and indirect costs listed above while studying, and have
fewer years in paid employment than their counterparts who joined the labour force at younger ages.
However, they are rewarded with relatively higher starting wages and relatively steeper earnings growth.
Figure 10.1 The private costs and benefits of investing in education

Source: Adapted from Connolly and Munro (1999: 387).

e cost-bene t calculations that underpin choices among these and other education and career paths
are complex, and some young people, parents and guardians may lack the information needed for optimal
investments in education. is information problem is the basis of the merit good argument for government
intervention in education markets outlined in Section 3.5. In addition, economic and other factors may tilt
the incentives of some persons or households against human capital accumulation (recall the example in
Section 9.3.2 of parents who may gain more in the short run from using their children as workers than from
sending them to school). To counter the effects of such information problems and distorted incentives,
governments combine education subsidies with regulations that ban child labour and specify minimum
ages at which children may leave schools.
Because the costs of education and training must be defrayed before the monetary bene ts are reaped,
households with limited savings have to borrow to nance human capital accumulation. Capital markets,
however, often fail to provide adequate education nancing: unlike physical assets such as buildings and
land that can be repossessed and sold by banks, human capital cannot serve as collateral to secure loans.
Such barriers to human capital accumulation distort the allocation of resources and further bolster the case
for government intervention in education markets.
e equity argument for public provision or nancing of education follows from the strong in uence of
human capital on labour market earnings and the distribution of personal incomes. In all countries, labour
market earnings are much bigger portions of personal incomes than transfers from governments are, and,
hence, more important determinants of income inequality. Leibbrandt, Woolard, Finn and Argent (2010:
23), for example, found that inequality in wage incomes explained between 85% and 90% of the total
inequality in household incomes in South Africa in 1993, 2000 and 2008. Hence, attempts to improve highly
skewed distributions of income are doomed to fail unless they empower the poor to obtain jobs and earn
higher salaries and wages. Education systems can be powerful mechanisms for enabling poor persons to
obtain well-paying jobs. If high direct and indirect costs prevent poor people from accessing education and
training programmes, however, education systems perpetuate, instead of reduce, poverty and inequality.
e capital market failures referred to earlier tend to exacerbate this problem, because richer persons and
households are more likely than poorer ones to have assets that can be offered as collateral to obtain bank
loans. ese considerations imply that subsidies and other measures that enable poor persons to
accumulate human capital should be cornerstones of governments’ efforts to reduce income inequality.
10.1.2 Healthcare
Externalities and imperfect information cause allocative inefficiency in unregulated healthcare markets.
Moreover, richer persons are served more effectively by such markets than poorer ones are. ese market
failures are the elements of a powerful case for government intervention in healthcare markets.
Healthcare services that prevent the emergence or spreading of contagious diseases confer positive
external bene ts to parties other than those receiving treatment. is implies that the marginal social
bene ts of such services exceed their marginal private bene ts. Section 3.6.2 explained that unregulated
markets underprovide and underprice goods and services that exhibit positive consumption externalities
and illustrated the efficiency-enhancing potential of Pigouvian subsidies. Section 10.2 returns to the case for
and policy implications of such subsidies in healthcare markets.
Speci c features of healthcare services make unregulated healthcare markets prone to severe
information problems. Healthcare per se has little value for consumers, who spend money on such services
to obtain something else that matters to them, namely good health. Put differently, healthcare services are
inputs into good health. e relationship between good health and spending on healthcare services is
complex, though, because health outcomes also depend on other inputs such as nutrition, lifestyle choices
and the physical and human environments. Individuals typically have limited information about the speci c
factors that determine their health outcomes, the efficacy of treatments, and the knowledge and skills of
providers of healthcare services. Hence, governments regulate the training, accreditation and activities of
medical practitioners to protect consumers of healthcare services.
e nature of the demand for healthcare services also causes information problems. Unpredictable
illnesses and injuries are major sources of the demand for such services, and the costs of treating many
health problems can be very high. Recall from Section 9.2.1 that insurance products enable individuals to
maintain their levels of consumption when they are affected by unexpected income losses. Large medical
bills caused by unexpected illnesses and injuries affect the nances of households in much the same way
that negative income shocks do. Private rms can and do offer healthcare expenditure insurance in the form
of medical aid schemes. However, markets for such insurance are prone to both forms of information-
related market failures discussed in Section 9.2.2, namely adverse selection and moral hazard.
Section 9.2.2 explained that adverse selection arises when persons whose exposure to certain risks
makes them particularly likely to seek insurance cover can hide their high-risk status from insurers. It also
pointed out that adverse selection causes allocative inefficiency: rms that cannot distinguish between high-
risk and low-risk clients charge everyone the same premium based on the average risk, and such mispricing
of insurance cover distorts the quantities purchased. In some circumstances – for example, when low-risk
individuals deem the premiums based on average risk excessive and do not buy cover – adverse selection
can make insurance provision impossible by preventing sufficient pooling of high and low risks.
Such inefficiency seems highly likely in markets providing healthcare expenditure insurance, because
persons who already face or are at risk of facing large medical expenses are more likely to seek cover than
those who do not, ceteris paribus. In reality, rms providing medical aid substantially reduce the prevalence
of adverse selection by subjecting prospective clients to various forms of screening such as medical tests and
scrutiny of personal and family health records. Screening enhances efficiency, because it enables medical
aid schemes to charge risk-related premiums. However, it also enables them to increase their pro ts by
limiting healthcare expenditure cover to low-risk individuals, either by charging high-risk persons
unaffordable premiums or by refusing to insure them. Such practices raise important equity questions,
especially if the poor are heavily affected by the risk factors that determine access to and the price of medical
aid. is would be the case, for example, if these risk factors were linked to inadequate diets and poor living
conditions.
Recall from Section 9.2.2 that moral hazard occurs when insured persons can raise the liabilities of the
insurer without the knowledge of the latter by increasing the size of the insured loss or the probability of its
occurrence. Behaviour with both types of effects causes allocative inefficiency, and actions that increase the
likelihood of the occurrence of the loss can even make insurance provision impossible.
Moral hazard is a serious problem in markets for healthcare expenditure insurance. Some fully insured
persons may well take less care to avoid lifestyle-related ailments (for example, those linked to smoking and
unhealthy eating habits) than they would have if they had to pay the full price of medical care. Such
behaviour is the equivalent of actions in other insurance markets that increase the probability of losses to
the insurer. e equivalent of actions that increase the size of the insured loss is also a major issue in
healthcare expenditure insurance markets. is is the so-called third-party payment problem: the reality
that a third party (the medical aid scheme) carries the costs strengthens the incentives of insured persons to
consume healthcare services and that of service providers to supply them. Figure 10.2 shows that the likely
outcome of such incentives is inefficiently large consumption and supply of healthcare services.
e downward-sloping demand curve Dm in Figure 10.2 depicts the marginal bene t of healthcare
services to consumers. In the same way, the upward-sloping supply curve Sm shows the marginal cost of
producing such services. In the absence of medical aid schemes providing healthcare expenditure
insurance, the market would be in equilibrium at a, where the price would be P0 and the quantity consumed
Q0. e total expenditure on healthcare services would be given by the rectangle P0aQ00 (that is, the product
of P0 and Q0).
A person with healthcare expenditure insurance pays premiums to a medical aid scheme, but is not
required to pay service providers in full when receiving treatment or buying medicines. is means that an
insured person does not pay the full marginal cost when using healthcare services. However, medical aid
schemes often require clients to make small payments known as co-payments to providers of medical
services.3 Figure 10.2 illustrates the third-party payment problem in the context of medical aid schemes with
co-payments. Hence, the introduction of medical aid reduces the out-of-pocket cost to consumers of
healthcare services from P0 to P1.4 While a markedly larger quantity of healthcare services is now consumed
– Q1 exceeds Q0 – out-of-pocket spending by consumers on such services is less than before (the area P1bQ10
is smaller than the area P0aQ00). e reason for the drop in out-of-pocket spending by consumers is that
medical aid reduces the price they pay to providers of healthcare services.

Figure 10.2 The third-party payment problem in the market for healthcare expenditure insurance
Source: Adapted from Hyman (2010: 369).

e effects of the third-party payment problem are not necessarily limited to the demand side of the
market for healthcare services. Figure 10.2 shows that the supply side may be affected as well. If the supply
curve has the usual positive slope, the price of healthcare services has to increase to induce providers of
healthcare service to meet the increase in the quantities demanded by consumers. us, the slope of the
supply curve depicted in Figure 10.2 implies that a price of P2 is necessary to elicit the provision of Q1 units of
healthcare services. In this case, total expenditure on healthcare services increases from P0aQ00 to P2cQ10.
e portions of these payments made by consumers of healthcare services and medical aid schemes are
P1bQ10 and P2cbP1, respectively. Note that the marginal cost of providing healthcare services exceeds their
marginal bene ts along the segment Q0Q1 on the horizontal axis. Hence, the triangle abc represents the
allocative inefficiency caused by the third-party payment problem.
Section 9.2.2 pointed out that social insurance schemes can overcome some adverse selection problems
in insurance markets, as well as their negative equity effects. In principle, government-run national health
insurance schemes with compulsory membership can yield similar risk-pooling bene ts and achieve
equitable coverage of all income groups. e case for such schemes is bolstered further by the merit good
argument introduced in Section 3.5, which rests on externality considerations and the (paternalistic) belief
that some persons would not buy enough healthcare expenditure insurance to ensure access to adequate
medical care for themselves and their families.
National health insurance schemes are costly and administratively complex, however, and sufficient
nancing sources and managerial skills are prerequisites for their viability and effective functioning. Most
developing countries lack the payroll tax bases and skills needed to operate such systems. In fact, there is
growing concern in several high-income countries with well-established systems of this nature about the
longer-term scal implications of rapid growth in healthcare spending.5 While population ageing and other
factors have also been at play, this development has con rmed an important conclusion in Section 9.2.2:
moral hazard may well be as common in social insurance schemes as in private insurance markets, because
organisations in the public sector are unlikely to be more successful than private rms are at overcoming the
underlying information-related market failures.
Supplier-induced demand – the tendency of service providers to encourage patients to buy more
healthcare services than they need – has been one of the information-related causes of the worldwide
growth in healthcare expenditure. As was pointed out earlier, third-party payment systems weaken the
incentives of healthcare service providers to economise on the use of resources. Such systems do not
encourage price competition; instead, they incentivise providers of healthcare services to compete on the
real or perceived quality of care. Such competition often revolves around investments in appealing facilities
and the latest diagnostic technologies, as well as cost-raising treatment strategies involving expensive tests,
procedures and medicines. Insured patients’ limited knowledge of health conditions and treatment options
as well as the reality that medical aid schemes foot their bills weaken their incentives to resist supplier-
induced demand.
Public healthcare systems that offer free care at the point of use and are funded from general tax
revenues – such as the British National Health Service – are alternatives to national health insurance
schemes. In principle, such systems can provide healthcare services to all income groups irrespective of
ability to pay. Fixed allocations from the national budget provide control over aggregate healthcare
expenditure, but the enforcement of budget constraints usually involves rationing in the form of waiting
periods for treatment. Like national health insurance schemes, such systems are very complex to run and
require considerable administrative capacity.

10.2 The case for in-kind subsidies

Recall from Section 8.2.2 that government expenditure programmes to change the secondary distribution of
income can take the form of cash transfers, in-kind transfers and the provision of social services. Many
governments levy only nominal fees for at least some education and healthcare services, or provide such
services free of charge. e term ‘in-kind subsidies’ is used to describe social service provision on such
terms. e question arises whether there is economic justi cation for such in-kind subsidies. One reason
why this is a pertinent question is that the costs of subsidies often exceed their bene ts. In such cases,
subsidies give rise to economic inefficiency in the form of excess burdens. Figure 10.3 uses the concept of
consumer surplus to illustrate the excess burden of a subsidy.
Figure 10.3 depicts the market for Good X. e initial equilibrium is at E0, where the demand curve and
the supply curve intersect. e government now introduces a subsidy to reduce the price consumers pay for
Good X. We assume that the cost of producing the good is constant and that the full bene t of the subsidy is
passed on to consumers. e new equilibrium at E1 implies that the subsidy decreased the price from P0 to
P1 and boosted the quantity demanded from Q0 to Q1. Recall that the demand curve is a marginal bene t
curve. It follows that the subsidy increased the consumer surplus (the difference between the bene t a
consumer derives from a good and service and the price he or she pays for it) from the area aP0E0 to the area
aP1E1. us, the area P0E0E1P1 depicts the total bene t of the subsidy to consumers. It has two components:
area P0E0cP1 represents the bene t from paying a lower price for the original quantity and area E0E1c the
bene t from the additional units of Good X purchased because the subsidy made it more affordable.
It is important to ascertain the cost of achieving this increase in the consumer surplus. e budgetary
cost of the subsidy is its unit cost (the difference between P0 and P1) multiplied by the Q1 units consumed,
that is, the area P0bE1P1. Hence, the scal cost of the subsidy (area P0bE1P1) exceeds its bene t to consumers
(area P0E0E1P1) by the area E0bE1. is welfare loss, which is known as the deadweight loss or excess burden
of the subsidy, arises because in-kind subsidies restrict the choice set of consumers. It follows that the above
demonstration of the excess burdens of in-kind subsidies complements the choice-based argument for cash
transfers outlined in Section 9.3.1.

Figure 10.3 The excess burden of a subsidy

ese arguments, however, do not mean that in-kind subsidies are never justi able in economic terms.
Section 3.7.1 and Section 9.3.1 pointed out that a case can be made for subsidising goods or services that
exhibit positive externalities.6 In fact, Section 3.7.1 used education – one of the social services discussed in
this chapter – as an example to illustrate that subsidies can promote allocative efficiency by bringing the
marginal private bene ts of consumption in line with the marginal social bene ts. Healthcare services also
exhibit external effects. Tuberculosis (TB) treatment is a good example. Individuals with TB who fail to seek
treatment or fail to complete the required treatment regime do not only jeopardise their own health; the
contagious nature of the disease means that such behaviour also poses health risks for others. Moreover,
failure to comply fully with the suggested treatment regime has been linked to multidrug resistant TB. is
represents another public health externality, because treatment of this type of TB is more expensive than
that of other strains.
e implication of the positive external bene ts is that the social bene ts of TB treatment compliance
exceed the private bene ts. Figure 10.4 illustrates the potential bene ts of subsidising such treatment. e
gure shows the marginal private bene t (MPB) of treatment to a consumer, the marginal external bene t
(MEB) and the marginal social bene t (MSB) to the community as a whole. Note that the MSB curve is
obtained by vertical summation of the MPB curve and the MEB curve. e marginal cost (MC) is assumed
constant and equal to the competitive market price (P0).
From a social point of view, optimal consumption of TB treatment by the individual consumer is Q1 (this
is the level of consumption associated with E1, where the marginal social bene t MSB equals the marginal
cost MC). e individual will not consider the external bene ts of his or her consumption of TB treatment,
though. Instead, he or she will consume at E0, where the marginal private bene t MPB equals the marginal
cost MC. e resulting quantity consumed (Q0) will be sub-optimal. However, an in-kind subsidy that
reduces the cost to the consumer from P0 to P1 would encourage socially optimal consumption. e shaded
area P0E1E2P1 indicates the size of the required subsidy. In sum: in the presence of a good- or service-speci c
externality, a subsidy for that good or service could induce socially optimal consumption via its effect on the
price.
Would an unconditional cash transfer enhance allocative efficiency to the same extent as an in-kind
subsidy would? One reason why an unconditional cash transfer might not do so is that its consumption-
boosting effects would not necessarily be restricted to goods and services valued by society. Recall, for
example, that South African households’ use of grant money does not differ markedly from their use of other
incomes (cf. Section 9.5.3.1). is implies that the introduction of an unconditional cash transfer aimed at
improving the health of members of such households would increase their expenditure on healthcare. Yet
households would only use a portion of the cash transfer for this purpose, and would spend more on the
other goods and services in their consumption baskets as well. ese goods and services can range from
necessities such as food and transport to ‘sin goods’ such as alcoholic beverages and tobacco products.
Hence, an in-kind subsidy is likely to give rise to a larger increase in the consumption of a good or service
with positive externalities than an unconditional cash transfer with the same monetary value would.
Figure 10.4 Subsidisation and the external benefits of tuberculosis treatment

10.3 Social service delivery

Section 7.2.3 pointed out that government expenditure on education and healthcare is relatively high in
South Africa by international standards, and scal incidence studies (e.g. Van der Berg & Moses, 2012;
Inchauste et al., 2015) have found that these spending programmes are well targeted at the poor.7 It is well
established, however, that the returns to South Africa’s investments in the public education and healthcare
systems have been disappointing. As Van der Berg and Moses (2012: 137) put it: ‘Social spending has often
not produced the desired social outcomes’. is section uses the notion of the ‘service delivery chain’ to
identify some of the factors that in uence social service delivery in South Africa and other Southern African
countries. While these ideas are used in this chapter to comment on the effectiveness of public education
and healthcare systems, they can be applied to other social spending programmes as well.
Service delivery has short-term and long-term effects on citizens’ lives. Access to and the quality of
services such as education, healthcare, housing and sanitation are important aspects of persons’ quality of
life. Moreover, it also creates future possibilities and opportunities, primarily but not exclusively via its
in uence on the labour market prospects of individuals.
e World Bank (2004c) described the process of service delivery as a chain consisting of relationships
between three sets of role players: policymakers, service providers (for example, teachers and doctors), and
citizens. Problems affecting any of the links of such chains undermine service delivery. In primary schooling
for example, outcomes would be compromised if policymakers fail to provide teachers with sufficient
resources, if high rates of teacher absenteeism prevent full coverage of the curriculum in schools, or if
learners do not use the textbooks provided to them. e various reasons why links in service delivery chains
break down range from corruption to weak accountability mechanisms and weaknesses in the capacities to
budget, disburse allocated funds, and monitor programme implementation. Hence, it should not be
assumed that more government spending on social services would always lead to better social outcomes.
Pay hikes for teachers and medical staff are cases in point: such injections of money into the education and
healthcare sectors do not automatically lead to better service delivery and improvements in outcomes.
e distinction between outputs and outcomes is also important. Outputs are deliverables that are
usually easy to observe, count and verify (for example, the number of children enrolled in school or the
proportion of a population that has access to potable water). Such deliverables tend to intermediary aims
that are necessary but not sufficient conditions for meeting the ultimate goals of programmes. Outcomes,
on the other hand, are deliverables that are more difficult to observe, count and verify, but are linked to the
ultimate goals of programmes. Examples of outcomes are whether children enrolled in school are learning at
the expected pace and according to set standards or whether the water provided to communities is actually
clean and safe. e distinction between outputs and outcomes does not matter much when it comes to
services that are simple to administer, such as vaccinations or grants. In education and healthcare, however,
outputs often contribute very little to the effectiveness or success of programmes. In such cases, assessments
of effectiveness should focus almost entirely on outcomes.
Sections 10.4 and 10.5 discuss social service delivery issues in South Africa by outlining weak links in the
delivery chains of public education and healthcare services. ese services are well worth focusing on,
because they absorb large portions of the national budget. Furthermore, the introduction to this chapter
emphasised their importance for economic development.

10.4 Service delivery in education in South Africa

Section 7.2.3 pointed out that education expenditure is the largest item in the functional classi cation of
government spending in South Africa,8 while Section 7.3 showed that the country’s education spending-to-
GDP ratio is high by international standards. As was mentioned in Section 10.3, however, the economic
returns to this large investment in public education are poor. e weak relationship between resources and
results in the South African education sector con rms the importance of effectiveness in the provision of
social services.
South Africa has a large and diversi ed public education system with four components: pre-primary
education (a reception year known as ‘Grade R’), basic education (primary and secondary schooling),
higher education (universities), and technical and vocational education and training provided by TVET
(technical and vocational education and training) colleges and SETAs (sectoral education and training
authorities). is section focuses on public basic education; as such, it does not discuss other components of
the public education system or the signi cant and growing private basic education sector.

10.4.1 Resources and outputs in the South African basic education system
e South African education system was highly fragmented during the apartheid era, when segregated
facilities and administrative structures existed for each race group. In addition, the allocation of funding for
schools, teachers and learning materials was extremely inequitable. While facilities and structures serving
whites were well resourced, those of other race groups were severely underfunded. Such discrimination in
the allocation of resources contributed to large racial disparities in educational attainment (that is, years of
schooling completed) and outcomes.9
e transition to democracy in South Africa was accompanied by a fundamental restructuring of the
basic education sector: an integrated system administered by a national and nine provincial departments of
education replaced the fragmented apartheid structures. is process was accompanied by large shifts in the
allocation of resources. Within a short space of time, the government largely eliminated disparities in public
spending per learner between race groups and income groups by equalising learner–teacher ratios and
teacher pay scales, and by reducing allocations to schools that formerly served only white children while
increasing those to schools that formerly served only children of colour.10 It should be pointed out that these
reforms did not fully eliminate differences in resources among schools in richer and poorer areas. e
infrastructure (buildings and other facilities) of many schools remains inadequate, and schools in richer
communities that charge fees still have lower learner–teacher ratios because they employ and pay additional
teachers. Nonetheless, the post-apartheid shift in resources in basic education was a remarkable
achievement.
Apart from changing the distribution of resources among schools, the government markedly increased
the volume of resources set aside for basic education. is is evident from Figure 10.5, which depicts general
government expenditure on basic education and its two components (pre-primary and primary education,
and secondary education) in the scal years from 1994 to 2016. e gure contains real amounts, that is,
actual expenditure de ated with the consumer price index.
Expressed in 2016 prices, total general expenditure on basic education increased from R98,3 billion in
the 1996 scal year to R165,1 billion in the 2017 scal year, that is, at an average annual growth rate of 2,5%
per annum. Growth in expenditure on pre-primary and primary education as well as secondary education
underpinned this trend (the average annual growth rates in expenditure on both components of basic
education were also 2,5%). It is clear from Figure 10.5 that basic education expenditure grew rapidly from
2008 to 2013. From 2014 onwards, the deterioration in the overall scal situation forced the scal authorities
to limit the growth in most functional categories of expenditure. e restraining effect of this development
on basic education spending was compounded by the need to increase allocations to higher education after
the introduction of fee-free tertiary education at the beginning of 2018. It should be kept in mind when
interpreting these gures that the population of school-going age also grew markedly in this period. In fact,
Spaull (2019: 6) argued that spending per learner decreased in real terms from 2010 to 2017. Nevertheless,
the combined effect of the resource allocation shift and the growth in government expenditure on basic
education was a signi cant increase in the resources available to schools serving poorer learners in South
Africa.

Figure 10.5 General government expenditure on basic education in South Africa in real terms, 1996–2017

Source: Calculated from South African Reserve Bank, Quarterly Bulletin (various issues). Pretoria: South African Reserve
Bank. [Online]. Available: https://www.resbank.co.za/Publications/QuarterlyBulletins/Pages/QuarterlyBulletins-
Home.aspx [Accessed 14 June 2019].

ese shifts and increases in resources were accompanied by further improvements in access to
education, as indicated by enrolments and grade survival rates. For all practical purposes, South Africa now
has universal access to primary and lower secondary schooling. Household surveys show that 98,9% of
South African children aged between seven and fteen years were attending some form of educational
institution in 2016 and that 66% of children in this age group were in schools that did not charge any fees
(Department of Basic Education, 2018: 6, 12). By international standards, the extent of access to primary and
secondary schooling in South Africa is above average, especially in the African context. However, many
learners drop out of school in Grades 10 and 11 before the external matric examination. us, only 85,1% of
16 to 18-year-olds were in school in 2016 (Department of Basic Education, 2018: 6). While this gure
exceeded the 82,6% recorded in 2002, South Africa’s rates of graduation from upper secondary school
continue to lag behind those of peer countries (Gustafsson, 2011: 2). is also explains why comparatively
few young people in South Africa reach and complete post-school education. Fewer than 10% of youths in
South Africa attain fteen years of education (completion of a three-year degree, for example), compared
with at least 15% in Columbia and Peru, and 24% in the Philippines and Egypt (Gustafsson, 2011: 14).

10.4.2 Educational outcomes in South Africa

e question arises how educational outcomes changed amidst these improvements in the availability of
resources to schools as well as the outputs of the basic education sector.11 Unfortunately, outcomes have
remained disappointing. e performance of South African learners in cross-national assessments of
educational achievement has provided particularly compelling evidence of poor learning outcomes that lag
behind those of most developing countries (see Van der Berg, Burger, Burger, De Vos, Du Rand, Gustafsson,
Moses, Shepherd, Spaull, Taylor, Van Broekhuizen & Von Fintel, 2011: 2–3). In fact, the results of such
assessments have indicated that the performance of the country’s education sector lags behind that of some
markedly poorer African countries with smaller economies, lower government expenditure on education
per capita and more severe poverty burdens. Box 10.1 discusses the ndings of such cross-national
assessments in more detail.

Box 10.1 What do South African learners know relative to learners from other countries?

Cross-national assessments, which consist of tests written by a representative sample of students in each
country, make it possible to compare the knowledge of South African learners with that of learners in other
countries on the continent and further afield. The three major cross-national assessments are the Progress in
International Reading and Literacy Study (PIRLS – Grade 4 and 5), the Trends in International Mathematics and
Science Study (TIMSS – Grade 8 and 9) and the Southern and Eastern Africa Consortium for Monitoring
Educational Quality (SACMEQ – Grade 6). This box summarises the performance of South African learners in
each of them.
• PIRLS. The 49 countries that participated in the 2016 round of PIRLS included South Africa, several OECD
countries and other middle-income countries such as Azerbaijan, Bulgaria, Egypt, Georgia, Iran, Kazakhstan,
Morocco and the Russian Federation. The South African students fared worse than their peers from all the
other countries: fully 78% of them failed to achieve the Low International Benchmark that indicates minimal
competence in reading (Mullis, Martin, Foy & Hooper, 2017: 58). Trong (2010: 2) explained the practical
value of this benchmark: ‘Learners who were not able to demonstrate even the basic reading skills of the
Low International Benchmark by the fourth grade were considered at serious risk of not learning how to read.’
The portions of learners from some other countries that failed to achieve the Low International Benchmark
were as follows: Egypt (69%), Morocco (64%), Iran (35%), Georgia (14%) and Bulgaria (5%). The governments
of some countries with significantly better results than those of South Africa spend markedly less per primary
school learner. In Bulgaria, for example, such spending amounted to US$1 363 per learner in 2017 in
purchasing power parity terms, which was 40% lower than South Africa’s spending level of US2 281 per
learner.12
• SACMEQ. South Africa’s performance relative to poorer African countries has confirmed that the availability of
additional financial resources does not guarantee better outcomes. Van der Berg et al. (2011: 4) concluded
after studying the SACMEQ 2000 and SACMEQ 2007 data that ‘South Africa performed slightly below the
average of the other participating African countries in Grade 6 mathematics and reading, despite benefiting
from better access to resources, more qualified teachers and lower pupil-to-teacher ratios’. The relative
performance of South African in SAQMEC 2013 was somewhat better yet far from satisfactory. In that year,
the average reading score of South African learners (558) equalled the average of all the learners from the
14 participating countries, while the average mathematics score of South African learners (587) only
marginally exceeded the average of all participating learners (584) (Portfolio Committee on Basic Education,
2016: 2). It is disconcerting, for example, that South African Grade 6 learners performed significantly worse
than their Kenyan peers, whose average reading and mathematics scores were 601 and 651, respectively.
 This is in spite of the fact that in 2015, public current expenditure per pupil in purchasing power parity terms
was more than seven times higher in South Africa (US$2 271) than it was in Kenya (US$307).13
• TIMSS: South Africa participated in the TIMSS studies in 1995, 1999, 2003, 2011 and 2015. Hence, these
studies of competence in mathematics and science allow the most extensive comparison of the performance
of South African learners since the political transition.
The TIMSS studies showed that there was no improvement in Grade 8 mathematics or science
achievement between 1995 and 2003. Thus, it was decided that the international Grade 8 tests were too
difficult for South African Grade 8 students. Grade 8 and Grade 9 students wrote the Grade 8 test in 2003,
but only Grade 9 students wrote the Grade 8 test in 2011. A comparison of the performance of Grade 9
students in 2003 and 2015 shows that there was an improvement in mathematics and science performance
amounting to approximately two grade levels of learning (Reddy, Visser, Winnaar, Arends, Juan, Prinsloo &
Isdale, 2010: 6). When interpreting this trend, one should keep in mind that South Africa started from an
exceedingly low base in 2003 and that the level of performance remains poor even after the improvement.
Thirty-nine countries participated in TIMMS 2015. In that year (that is, after the improvement), the average
mathematics score of the participating South African learners was the second lowest after that of Saudi
Arabia, while none of the other countries had a lower science score (Reddy et al., 2010: 3).

Although South Africa has achieved relatively high levels of educational attainment (including near-
universal access to primary and lower secondary schooling), the cross-national assessment results
summarised in Box 10.1 suggest that very little learning is taking place in as many as 80% of schools. e
roughly 20% of schools that perform well in such assessments are fee-paying and independent ones that
predominantly serve richer communities and households. Various surveys have also indicated that the
majority of South African learners perform poorly in key learning areas such as Reading, Mathematics and
Science; furthermore, the results of the Department of Education’s Systemic Evaluations have shown that
most children fail to achieve the standards required by the curriculum (Van der Berg et al., 2011: 2–3). In
sum, the large pro-poor shift of resources since the transition has been accompanied by improvements in
educational outputs, but similar improvements in educational outcomes have not been forthcoming.

10.4.3 Service delivery issues in basic education in South Africa

It is now widely accepted that the ability of a country to educate its youth cannot be measured by access to
schooling or enrolment rates alone, but rather by its ability to impart to students the skills, abilities,
knowledge, cultural understandings and values that are necessary to function as full members of their
society, their polity and their economy. is more appropriate de nition of success suggests that the South
African basic education system performs extremely poorly, especially when one considers the substantial
amount of resources that is allocated to this enterprise. Section 10.3 pointed out that failures to turn
additional resources into improved outcomes often re ect problems in the service delivery chain that links
policymakers, social service providers and citizens. A growing body of research has suggested that the South
African basic education system exhibits several problems of this nature. Van der Berg and Hofmeyr (2018:
14–19), for example, discussed four binding constraints that hinder the ability of other interventions (such as
the provision of more resources) to improve educational outcomes:
• Weaknesses in provincial education departments. e bulk of basic education spending in South Africa is
undertaken at the provincial level of government.14 Hence, the lack of capacity in most provincial
education departments is a major constraint in the basic education sector. Its manifestations include
poor ful lment of critical administrative functions (such as the management of learning materials and
other non-personnel resources) and inadequate monitoring of and support to schools and teachers.
• e undue in uence of labour unions. Most teachers belong to labour unions. e power of some unions
extends well beyond issues related to teacher pay and working conditions to include aspects of
education policymaking and implementation. Some strong teacher unions have even distorted post-
provisioning processes to secure nepotistic appointments for their members, and have thwarted several
attempted policy reforms to increase accountability in the basic education sector.
• e weak content knowledge and pedagogical skills of teachers. While most South African teachers are
adequately quali ed, many lack the content knowledge to teach the subjects for which they have been
appointed.15 is is indicative of shortcomings in the pre-service and in-service training of teachers. One
of the outcomes of many teachers’ poor content knowledge is that they cannot accurately judge the
performance of their learners. e quality of teachers and teaching are particularly important aspects of
the relationship between resources and outcomes in schooling. Basic education is a labour-intensive
activity16; hence, large portions of increases in government expenditure on education usually take the
form of salaries and other bene ts paid to teachers. Such expenditure increases will not improve
educational outcomes unless they attract better teachers to the public education sector or motivate
existing ones to work harder and more effectively. In South Africa, the main driver of the rapid growth in
basic education spending from 2008 to 2013 was improvements to the pay scales of teachers. ese
improvements had three aims: to increase the reward to teaching experience, to restore the position of
teachers relative to other civil servants, and to ensure that the teaching profession attracted appropriate
individuals. e adoption of the new pay scales did not lead to immediate improvements in educational
outcomes, and it remains to be seen whether bene ts will realise in the longer-term.
• Wasted learning time. Much learning time is wasted in South African schools.17 is re ects a variety of
reasons, including teacher absenteeism, poor time management, and the lack of a culture of teaching
and learning. A study by the Human Sciences Research Council (Reddy, Prinsloo, Neshitangani,
Moletsane, Juan & Janse van Rensburg, 2010: 84) con rmed that wasted learning time markedly reduces
children’s opportunity to learn in many South African schools.
It is notable that several of these service delivery problems are related to a lack of accountability, that is, the
reality that there are few, if any, consequences for non-performance in the basic education sector.
Education is one of the most important elements of social policy. Hence, it should be a priority function
in the national budget, as is the case in South Africa. However, outcomes do not necessarily improve when
more resources are made available for social service provision, especially when capacity and accountability
are lacking. is disconnect between increased resources and improved outcomes characterises various
areas of government activity in South Africa. It is especially problematic in education, however: compared to
social assistance, which uses taxes and social grants to redistribute a small portion of current total income,
education has much greater potential to in uence the mechanisms that determine the distribution of
income via its effects on labour market earnings.

10.5 Service delivery in healthcare in South Africa

e health status of a country’s population is a critical input into an economy’s growth; furthermore, it plays
a major role in preventing poverty and in ensuring quality of life for the population. Hence, it is important to
ensure that government expenditure on health contributes to good health outcomes.

10.5.1 The South African health system

e South African health system has large public and private components. In 2014, total health expenditure
in South Africa amounted to 9,0% of GDP, of which 49% (4,4% of GDP) owed through the public and 51%
(4,6% of GDP) through the private sector (National Treasury, 2015a: 53). Although their aggregate spending
levels are quite similar, the proportions of the South African population served by the private and public
components of the health system differ markedly (Statistics South Africa, 2019: 26). e 2018 General
Household Survey showed that 16% of South Africans had medical scheme cover and relied on private
sector healthcare. e remaining 84% of the population lacked medical scheme cover and mainly used
public healthcare services. It follows that expenditure in the private healthcare sector is markedly higher on
a per capita basis than that in the public healthcare sector. However, Box 10.2 shows that attempts are
underway to extend health coverage to all South Africans by means of a national health insurance system.
More than 90% of South Africa’s health budget is allocated to the national Department of Health and the
nine provincial departments of health, predominantly to fund a non-contributory public health system.18
e national Department of Health receives only a small portion (some 5%) of the total public healthcare
budget. e bulk of this budget (85% or more) is allocated to the provincial departments of health, which are
the most important providers of healthcare services in the public system.19 is section focuses on total
public health expenditure, which includes all funds set aside for this purpose in the national budget,
whether to the national or provincial departments of health or to other departments or institutions.

Box 10.2 Towards universal health coverage in South Africa

A growing international movement supports the establishment of universal health coverage systems. Definitions
of universal health coverage vary across countries and contexts; in its simplest form, however, it means
‘providing all people with access to needed health services of sufficient quality to be effective without their use
imposing financial hardship’ (Moreno-Serra & Smith, 2012: 917). One of the principles of this movement is that
individual countries should consider their own income constraints when deciding which health services to
include in such systems. One of the outcomes of the Rio+20 United Nations Conference on Sustainable
Development in Rio de Janeiro (Brazil) in June 2012 was a call to action that clearly articulated some of the
envisaged benefits of universal health coverage, namely enhancement of health, social cohesion, and
sustainable human and economic development (see Evans, Marten and Etienne, 2012: 864).
Apart from the obvious health and socio-economic benefits of better access to health services, empirical
evidence shows a strong association between better health status and growth and development. The 2001
Commission on Macroeconomics and Health, for example, found that an increase of 10% in life expectancy at
birth is associated with increases of 0,3% to 0,4% in annual economic growth (World Health Organization,
2001: 24).
South Africa has embarked on a deliberate process to achieve universal health coverage. The aim of this
process is to pool all government health spending in a single insurance scheme by establishing a National
Health Insurance (NHI) system. In an important step in this process, the Government gazetted the NHI White
Paper in June 2017. In addition, the Department of Health released a draft NHI bill. It invited relevant
stakeholders to make submissions on areas of concerns in the bill and produced a revised draft bill, but it has
not been released yet. The most likely sources of funding for an NHI system will be general taxes and
compulsory contributions by all formally employed South Africans whose earnings exceed a certain threshold.
The scheme will contract both public and private healthcare providers to deliver healthcare services to the
citizens of South Africa.
Increased health expenditure via an NHI system may lead to better health outcomes for all South Africans.
Yet cross-country evidence indicates that higher government spending on health does not necessarily lead to
better health outcomes (see Gupta, Verhoeven & Tiongson, 1999). When considering the potential effects of
health spending, policymakers often ignore the possibility that users might not respond to policy changes in the
ways they envisage. The incentives that govern the behaviour of health system staff and other actors in the
public sector may also undermine health outcomes. Filmer, Hammer and Pritchett (2000: 201) put it as follows:
‘Often, health service failures result from a systemic mismatch between the traditional civil service incentive
structure and the tasks required in the health sector.’

e aims of this section are to present data on current public health expenditure levels, describe how these
have changed over time and discuss the problems that arise in efforts to use public funds to improve health
outcomes for the poor. Hence, the section refers to levels of health service access, the quality of health
services provided and the efficiency of the public health system. Although an awareness of private health
expenditure is important for understanding the larger South African health- nancing context and the
inequity in expenditure between the public and private sectors, the public sector provides the bulk of the
health services consumed by poor South Africans. Hence, the remainder of this section focuses on the
public-sector component of the health system.
Public health expenditure in South Africa takes place in the context of a heavy disease burden. e
health system is confronted with a so-called ‘quadruple burden of disease’ that consists of communicable
diseases (for example, tuberculosis and HIV/Aids), a growing burden of non-communicable diseases (for
example, diabetes and cardiovascular disease), injuries (many of which are effects of high levels of
interpersonal violence), and maternal and child health problems. Studies of the South African health system
(for example, Van den Heever, 2012: 2–3) consistently conclude that South Africa produces worse health
outcomes than many low-income countries despite its status as an upper middle-income country. us,
despite the positive effects of the widespread availability of antiretroviral therapy (ART), South Africa
remains amongst the 30 countries in the world with the worst TB burdens (World Health Organization, 2018:
25). In similar vein, South Africa was the country with the highest number of new HIV cases in the world in
2017 (World Health Organization, 2019). e heavy disease burden necessitates high levels of public health
expenditure, but also weakens the effectiveness of such spending.

Levels, growth and composition of government spending on health in

10.5.2
South Africa
Health is the third largest functional category of government spending after education and social protection
(see Table 7.2). Figure 10.6 shows that general government expenditure on health increased in real terms
from R54,1 billion in scal year 1993 to R182,2 billion in scal year 2017.20 is represented an impressive
average annual rate of growth of 5,2%. It is also clear from Figure 10.6 that the real rate of growth in health
spending outstripped the rate of growth in the South African population over this period. Hence, the real
level of general government health expenditure increased in per capita terms from R1 430 in scal year 1993
to R3 206 in scal year 2017.

Figure 10.6 General government expenditure on health in South Africa in real terms, 1996–2017

Sources: Calculated from South African Reserve Bank, Quarterly Bulletin (various issues). Pretoria: South African
Reserve Bank. [Online]. Available:
https://www.resbank.co.za/Publications/QuarterlyBulletins/Pages/QuarterlyBulletins-Home.aspx [Accessed 14 June
2019]; and Statistics South Africa, Mid-year population estimates (various issues). Statistical Release P0302. Pretoria:
Statistics South Africa. [Online]. Available: http://www.statssa.gov.za/?page_id=1866&PPN=P0302&SCH=7362 [Accessed
14 June 2019].

e increase in spending depicted in Figure 10.6 was the result of deliberate efforts by the government,
albeit of varying intensity, to expand healthcare expenditure, equalise it across race groups, and make it
more pro-poor. e emphasis on making public healthcare expenditure more pro-poor was reinforced by
allocation shifts between provinces and expenditure categories.
 Real government expenditure on health grew relatively slowly during the second half of the 1990s, when
the availability of funds was limited by difficult macroeconomic conditions and a strong focus on scal
consolidation guided by the Growth, Employment and Redistribution (GEAR) strategy (see Section 18.6.1).
In fact, such spending even decreased in aggregate and per capita terms in 1999. Aggregate spending grew in
every scal year from 2003 to 2017, however, while per capita spending increased in each of these scal years
except 2014 and 2017. e real growth rates of both aggregates were particularly rapid from scal year 2004
until scal year 2013. Table 7.2 showed that this period of growth also brought a signi cant increase in the
ratio of general government health spending to GDP: following a slight decrease from 2,9% in scal year
1994 to 2,7% in scal year 2005, this ratio rebounded to 4,0% in scal year 2016. However, real public health
expenditure lost some of its upward momentum from 2014 onwards because of the poor growth
performance of the South African economy and the deteriorating scal position. e decreases in the per
capita values of real public health expenditure in 2014 and 2017 underscored this development.
Fiscal shifts towards primary healthcare activities have occurred during the post-apartheid period.21 e
Government introduced free primary care for mothers and young children in 1994 and abolished the
remaining user fees on primary care in 1996. Free primary care has been available to the entire population
since then. e effect of this policy shift has been to raise government expenditure on primary care from 20%
of its spending on hospitals in 2000 to 30% in 2007 (Burger, Bredenkamp, Grobler & Van der Berg, 2012: 682).
is was achieved by simultaneously increasing the share of the budget allocated to primary healthcare and
decreasing the share allocated to hospitals.
Another important trend in public health expenditure in South Africa during the new millennium has
been marked increases in the salaries, wages and other bene ts of workers in the health sector.22 e real
wage bill of the public health sector more than doubled between the 2006 scal year and the 2015 scal year
(Blecher, Davén, Kollipara, Maharaji, Mansvelder & Gaarekwe, 2017: 29). At rst, this trend was driven by
increases in salaries and wages as well as growth in the number of staff employed in the public health sector
(the purpose of the hiring of additional workers was to rectify perceived understaffing, especially compared
to other developing countries). From the 2007 scal year onwards, the main driver was the incremental
implementation of the Occupation Speci c Dispensation (OSD) – a new remuneration structure for health
workers that gave rise to substantially higher salaries.

10.5.3 Access and quality of health services

e distinction between access and quality is more complex in health than in education. School enrolment
rates can be used as indicators of access in education, while the learning outcomes of students can serve as
proxies for quality. In healthcare, utilisation rates and interactions with the health system (for example, visits
to clinics) are useful measures of access. It is very difficult to measure quality, though. Nonetheless, these
concepts remain relevant for assessment of the outcomes of increased healthcare expenditure and shifts in
expenditure allocation in the post-apartheid period.
Did access to public healthcare services improve in the post-1994 period and, if so, how? Did some
groups receive better access or more bene ts than others did? e following evidence exists on four
dimensions of healthcare access:
• Overall utilisation of healthcare. e total number of visits to primary healthcare facilities increased from
82 million in scal year 2001 to 129 million in scal year 2013 before dropping to 126 million in scal year
2016 (Blecher et al., 2011: 39; Blecher et al., 2017: 34). On average, South Africans without healthcare
expenditure insurance paid 2.8 visits to such facilities in scal year 2016.
• Pro-poor access. Not only did expenditure on primary healthcare increase substantially in the post-
apartheid period, there is also evidence that it became more pro-poor. In other words, the share of the
bene ts of public health spending received by the poorest began to exceed their share of the population.
By 2008, 30% of government expenditure on clinics accrued to the poorest 20% of the population, while
50% of such expenditure accrued to the poorest 40% (Burger et al., 2012: 689). e distribution of the
bene ts of changes in expenditure on public hospitals also exhibited a pro-poor pattern, albeit a slightly
weaker one.
• Affordability. An analysis of household survey data on self-reported payments to healthcare providers by
Burger et al. (2012: 692) revealed a large decrease in payments to public clinics and hospitals between
1993 and 2008. is largely re ected the effects of the introduction of free services. It is also worth noting
that the incidence of catastrophic health expenditures amongst South African households has been very
low in the post-apartheid period (Burger et al., 2012: 693).23
• Physical proximity. Since 1994, the government has markedly expanded the health infrastructure
network, especially the number of clinics. As a result, the proportion of the population that report travel
times of more than 30 minutes to the nearest clinic decreased signi cantly between 1993 and 2008
(Burger et al., 2012: 695). Respondents in household surveys now rarely mention the physical distance to
public healthcare facilities as a barrier to accessing healthcare. However, the high public transport costs
associated with accessing such facilities are still frequently cited as an obstacle (see, for example,
Goudge, Gilson, Russel, Gumede & Mills, 2009). ese costs are particularly onerous for people suffering
from chronic diseases such as TB and HIV, who require frequent visits to healthcare facilities.

As was pointed out earlier, it is difficult to capture the quality of healthcare services in aggregate measures.
Hence, we provide evidence on the experiences of health system users, which can be seen as tentative
indications of quality problems in healthcare. An analysis of data from the General Household Surveys from
2002 to 2008 by Burger et al. (2012: 697) identi ed waiting times, insufficient staff and the rudeness of staff as
the main complaints about public healthcare facilities. Another study found that the major supply-side
constraints experienced by chronically ill patients were the unavailability of drugs, weak or inadequate
clinical services at clinics that necessitated referrals of patients to other facilities, and a lack of ambulances
to transport patients (Goudge et al., 2009). Earlier analyses of changes in the demand for speci c categories
of health services in South Africa (for example, Grobler & Stuart, 2007) have found public healthcare to be an
inferior good for which the demand decreases as incomes increase.24
South Africa’s consistent under-performance in critical health areas (for example, maternal and child
health, HIV/Aids and TB) relative to peer countries cannot be explained away by insufficient funding for the
public sector, especially in view of the growth trend in public healthcare expenditure discussed in Section
10.5.2. In 2015, fully 120 of the other 182 countries for which data were available had lower rates of maternal
deaths than South Africa had (World Health Organization, 2018: 24); furthermore, South Africa was one of
only twelve countries in which the child mortality rate was higher in 2008 than it was in 1990 (Chopra,
Daviaud, Pattinson, Fonn & Lawn, 2009: 835). ese and other unsatisfactory health outcomes point towards
the presence of inefficiency or, more speci cally, X-inefficiency, in the public health system.25 While this
issue has not yet been explored in detail in the South African context, there exists some evidence that points
towards the presence of X-inefficiency in the public health sector once the burden of disease and available
resources have been controlled for.
Christian and Crisp (2012) provided several examples of a lack of leadership and decision-making power
in the public health system that are likely to lead to X-inefficiency. In addition, the reality that some public
health districts with scant resources outperform other better resourced ones (see Engelbrecht & Crisp, 2010:
201) probably re ects differences in levels of X-efficiency. Evidence from South Africa also suggested that
levels of efficiency may differ between the public and private sectors: for example, client perceptions
indicated that a sample of low-cost private sector clinics provided higher-quality services at similar cost to
public sector clinics (Palmer, Mills, Wadee, Gilson & Schneider, 2003).
South Africa’s public and total health expenditure levels compare relatively well with those of its upper-
middle income country peers, but the impact of public expenditure is undermined by the heavy disease
burden. e period from 2004 to 2014 was marked by consistent absolute and relative increases in real
public health expenditure, in part because of the anticipated implementation of a national health insurance
system. Rates of growth in public health expenditure have slowed from 2014 onwards, however, because of
poor economic growth and the deterioration in the scal situation. In all likelihood, the effects of X-
inefficiency in the system have been exacerbated by lack of capacity and accountability. While appropriate
levels of health expenditure are important for achieving a healthy population, they should not be valued for
their own sake. e outcomes that should be sought are a healthy population and its attendant bene ts,
which include increased labour-market participation, a reduction in poverty, and more rapid economic
growth and development.
 Key concepts
• catastrophic health expenditures (page 17)
• consumer surplus (page 4)
• co-payment (page 5)
• deadweight loss, or excess burden (page 7)
• human capital (page 2)
• in-kind subsidies (page 5)
• outcomes (page 9)
• outputs (page 9)
• supplier-induced demand (page 6)
• third-party payment problem (page 4)
• universal health coverage (page 14)

SUMMARY
• Education and healthcare services are particularly important for economic development, both as inputs
in the aggregate production function and as ends. Effective service delivery is very important, however,
because the link between the extent of government spending and developmental outcomes in these
areas is often tenuous.
• Education and healthcare are mixed services that can be supplied by the public sector or the private
sector. e existence of market failures underpins strong cases for government intervention in education
and healthcare markets. e case for government intervention in education markets rests on externality
considerations, information problems, capital market failures and equity concerns. Externalities and
imperfect information cause allocative inefficiency in healthcare markets. Equity considerations further
strengthen the case for government intervention in such markets.
• In-kind subsidies often cause excess burdens, because they restrict the choice sets of consumers.
However, such subsidies can be efficiency- and welfare-enhancing when good- or service-speci c
positive consumption externalities exist.
• e process of delivering social and other services can be seen as a chain consisting of relationships
between three sets of role-players: policymakers, service providers and citizens. e reality that links
service delivery chains can break down because of weaknesses in capacity or accountability, among
other things, means that more government spending on social services does not always lead to better
social outcomes. It also emphasises the salience of the distinction between outputs (readily observable
and measurable deliverables that are necessary but not sufficient conditions for meeting the ultimate
objectives of programmes) and outcomes (deliverables that are more difficult to observe and measure
but linked to the ultimate goals of programmes).
• By international standards, government expenditure on education and health is relatively high in South
Africa. Furthermore, there has been signi cant growth in public spending on both sets of services since
1994, accompanied by effective policy reforms that made such spending more pro-poor. Yet the
outcomes of government spending on education and healthcare services have remained disappointing:
various indicators show that these outcomes fall short of those of other middle-income countries and
even those of some low-income countries subject to much tighter resource constraints. Service delivery
problems linked to weak accountability mechanisms and X-inefficiency are major causes of these
disconnects between resources and outcomes. ese problems severely undermine the considerable
developmental potential of government expenditure on education and health.

MULTIPLE-CHOICE QUESTIONS
10.1 Which of the following statements is/are correct?
a. Human capital theory considers the nancial and other bene ts of education.
 b. Capital markets often fail to provide adequate education nancing.
c. Private rms cannot provide healthcare expenditure insurance.
d. Supplier-induced demand is a major problem in markets for healthcare expenditure insurance.
10.2 Which of the following statements is/are correct?
a. In-kind subsidies cannot increase the social welfare, because they generate excess burdens.
b. In-kind subsidies cause excess burdens because they restrict the choice sets of consumers.
c. Social outcomes do not always improve when governments spend more on social service
provision.
d. Outputs are easier to observe and measure than outcomes are.
10.3 Which of the following statements is/are correct?
a. Not all differences between the resources of schools in richer and poorer areas of South Africa have
been eliminated.
b. According to some researchers, real government expenditure on education per learner decreased
in South Africa from 2010 to 2017.
c. Measures of educational attainment are good proxies for educational outcomes.
d. e large pro-poor shift in resources since the end of apartheid has been accompanied by
improvements in educational outcomes, but not by improvements in educational outputs.
10.4 Which of the following statements is/are correct?
a. Resources are equitably distributed between the public and private components of the South
African health system.
b. For a middle-income country, South Africa has a very high disease burden.
c. Real government health expenditure increased in per capita terms from scal year 2003 to scal
year 2013.
d. In South Africa, public healthcare is a normal good.

SHORT-ANSWER QUESTIONS
10.1 Explain the importance of education and healthcare services for economic development.
10.2 Brie y explain the nature and effects of the third-party payment problem in markets for healthcare
expenditure insurance.
10.3 Explain the distinction between outputs and outcomes in social service delivery.
10.4 Brie y discuss four binding constraints in the South African basic education sector.

ESSAY QUESTIONS
10.1 Summarise the case for government intervention in education markets.
10.2 Summarise the case for government intervention in markets for healthcare expenditure insurance.
10.3 ‘Governments should always avoid in-kind subsidies, because they cause allocative inefficiency in the
form of excess burdens.’ Do you agree with this statement? Explain your answer.
10.4 Discuss the nature and manifestations of the service delivery problem in the basic education sector in
South Africa.
10.5 Discuss developments in access to and the quality of public healthcare in South Africa since 1994.

1 Recall that the discussion of trends in the functional composition of government expenditure in South Africa in Section 7.2.3 used
the Government Finance Statistics (GFS) system categories.
2 e income security programmes discussed in Chapter 9 are the main elements of the social protection category of social services.
3 e purpose of such payments is to discourage persons with medical aid from seeking unnecessary or excessively expensive
medical care.
4 e out-of-pocket cost would have been zero if co-payments were not required.
5 Chernew and Newhouse (2011) provided a useful overview of the extent and causes of growth in healthcare expenditure in
developed and developing countries.
6 Recall from Section 3.6 that goods and services exhibit positive externalities when the marginal social bene ts of consumption
exceed the marginal private bene ts because external bene ts accrue to non-users.
7 e only exception is government expenditure on higher education, which mainly bene ts members of higher income groups (Van
der Berg & Moses, 2012: 130–131; Inchauste et al., 2015: 24–25). e main reason for this is that many children from low-income
households attend poorly functioning schools that prevent them from accessing opportunities for tertiary education.
8 Recall from Table 7.2 that expenditure on education by the general government amounted to R285,2 billion in the 2016 scal year.
is constituted 18,8% of total general government expenditure and 6,6% of GDP.
9 Educational attainment gaps began to narrow from the 1960s onwards as the school enrolment and completion rates of all race
groups increased markedly (Van der Berg & Hofmeyr, 2018: 10).
10 Fiscal incidence studies – which were introduced in Section 8.4 – have con rmed the extent of this shift in resources (see, for
example, Inchauste et al., 2015). Van der Berg (2006) and Gustafsson and Patel (2006), among others, discussed the nature and
effects of this shift in more detail.
11 Recall the distinction between outputs and outcomes explained in Section 10.3.
12 e amounts per learner were published by UNESCO (2019: 281, 287).
13 13 e amounts per learner were published by UNESCO (2017: 402).
14 In the 2016 scal year, for example, provinces were responsible for 70,4% of total general government expenditure on education
(Statistics South Africa, 2017: Table 2). e national Department of Basic Education has policymaking and monitoring mandates.
15 To name but one example: fully 79% of the Grade 6 mathematics teachers tested in SACMEQ 2007 had a content knowledge below
Grade 6/7, that is, below the level they were teaching (Venkatakrishnan & Spaull, 2014: 14).
16 Salaries and wages accounted for 87,1% of general government expenditure on basic education in scal year 2017 (Statistics South
Africa, 2018a: 22, 26).
17 In 2008 and 2009, for example, the National School Effectiveness Study (NSES) showed that only 24% of Grade 4 and 5 mathematics
topics were actually covered in a nationally representative sample of schools (Taylor & Reddi, 2013: 193).
18 e remainder of the national health budget is allocated to various other government departments (including Education, Defence,
and Correctional Services) and to extra-budgetary institutions such as the Compensation Fund and the Road Accident Fund. As
was pointed out in Section 9.5, the Compensation Fund and the Road Accident Fund are social insurance elements of the South
African social security system.
19 ree sources determine the total funds receivable by the provincial departments of health: the provincial equitable share,
conditional grants and provincial own revenue. e provincial equitable share is the major funding source of the provinces. e
allocation of this pool of funds among the provinces is based on an equitable share formula that incorporates factors such as their
population sizes and poverty rates. Section 19.7.2.2 discusses the provincial equitable share in more detail. Conditional grants are
typically allocated on a national priority basis, for example, to combat HIV/Aids or to revitalise infrastructure. Such funds are
allocated by the National Department of Health and are disbursed to provinces with lists of conditions. e allocations could be
frozen or withdrawn if the provinces fail to meet these conditions. Provincial own revenue are funds generated by the provincial
departments of health. ese typically form very small shares of total provincial revenues.
20 All amounts in this paragraph are expressed in February 2016 prices. e consumer price index was used to de ate the nominal
amounts.
21 Many de nitions of primary healthcare exist. In South Africa, the budget of the Department of Health contains a programme
entitled ‘Primary health care services’ that manages the district health system and deals with environmental health issues,
communicable and non-communicable disease control, general health promotion and improvement of nutrition. e shift in
scal resources discussed here coincided with a policy shift to increase the focus on these aspects of the public health system at the
expense of that on hospital-based curative services, among other things.
22 Salaries and wages form the bulk of the health budget: in the 2017 scal year, for example, this component accounted for 61,4% of
general government expenditure on health and the next biggest component – purchases of goods and services – for 29,8%
(Statistics South Africa, 2018a: 22, 26).
23 Catastrophic health expenditure refers to health expenditure in excess of a pre-de ned income threshold that may have an
impoverishing effect on households. Various thresholds have been proposed. Two widely used ones are healthcare expenditure
that exceeds 10% of total household expenditure and healthcare expenditure that exceeds 40% of total household non-food
expenditure.
24 Large numbers of South Africans without medical scheme coverage rst seek care at private facilities when they are ill. Even among
the poorest quintile of households, about 20% access private healthcare services (Burger et al., 2012: 682).
25 Christian and Crisp (2012: 726) de ned X-efficiency as ‘an open-ended concept which describes the effectiveness with which a set
of inputs can produce outputs’. While it is closely related to the concept of technical efficiency, it differs from that concept in that
the source of inefficiency is explicitly identi ed as ‘intrinsic to the nature of human behaviour’, for example, aspects of
management and decision-making in organisations (Christian & Crisp, 2012: 727). Section 2.3 also explained the notion of X-
efficiency.
Introduction to taxation and tax equity

Tjaart Steenekamp and Ada Jansen

In Parts 1 and 2, we explained the role and functions of government as well as the nature and patterns of
government expenditure. e need and demand for public goods and services would appear to be varied
and extensive. However, like all other demands (wants) and needs, the demand for public goods and
services is also constrained by limited resources on the supply side. Furthermore, like all other goods and
services, the supply and distribution of public goods and services must be nanced. John Coleman aptly
remarked, ‘e point to remember is that what the government gives, it must rst take away’ (quoted in
Mohr, Fourie & associates, 2008: 339). is chapter focuses on taxation as a source of nance for public
expenditures. Different sources of nance are identi ed in Section 11.1. In the section that follows, tax bases
are identi ed and a distinction is made between different types of taxes, such as general taxes, ad valorem
taxes and direct taxes. Criteria for analysing taxes are proposed in Section 11.3, after which the equity
criterion is described and discussed in depth in Section 11.4 as well as the rest of the chapter. When the
fairness of a tax is considered, the effects of tax shifting are paramount. In Sections 11.5, 11.6 and 11.7, the
incidence of taxes is discussed using both a partial equilibrium and a general equilibrium analysis.

Once you have studied this chapter, you should be able to:
list alternative sources of government revenue
define a tax and describe the structure of tax rates
distinguish between general and selective taxes, specific and ad valorem taxes, and direct and indirect taxes
list the properties of a ‘good’ tax
explain what is meant by an equitable tax
distinguish between the statutory and the economic burden of a tax
analyse the shifting of a tax and its impact on tax incidence using both a partial and a general equilibrium
framework.

11.1 Sources of finance

e dominant source of nance for public expenditure is taxation. In 2017/18, tax revenue constituted
approximately 78,5% of total cash receipts from operating activities of the consolidated general government
(South African Reserve Bank, 2018: S-73). Government expenditure may also be nanced from alternative
sources. In addition to taxation, there are four other important sources of nance: user charges,
administrative fees, borrowing and in ation taxation.
 User charges (also referred to as bene t taxes) are prices charged for the delivery of certain public
goods and services. e role that these charges play in the allocation and distribution of resources is
analogous to the role of prices in the market mechanism. e important difference is that user charges are
set in the political market. User charges can only be levied if exclusion is possible (see Section 3.4 of Chapter
3). In other words, it should be possible to exclude those who do not pay for the consumption of the public
good or service in question. Examples of user charges include toll roads, public swimming pools, ambulance
services and university education (see Section 11.4.1 for a further discussion of bene t taxes).
Administrative fees are similar to user charges, but differ in the sense that the service (or bene t)
received in return for the fee is de ned rather broadly and imprecisely. Fees of this nature include business
licences, television licences, diamond export rights, shing licences and motor vehicle licences. e dreaded
parking ticket and speeding ne can also be added to the list. Administrative fees and nes are insigni cant
sources of revenue.
Government can borrow from its own citizens and from abroad. Borrowing is often used to nance
capital expenditure. Borrowed funds must be repaid at some point and can therefore amount to deferred
taxes. Because lenders have to be adequately compensated by way of interest payment for current
consumption forgone, it is imperative that borrowed money be spent on productive activities. Sometimes
government uses borrowed funds to nance current consumption, a practice that cannot easily be defended
on economic grounds (see Section 17.2 of Chapter 17).
Government-induced in ation can also be regarded as a source of revenue. If public expenditure is
nanced in such a way that increases in the money supply occur, this nancing may eventually raise the
price level. In ation changes the real value of public debt. If government borrows R2 000 from a taxpayer (for
example, if government imposes a loan levy on all taxpayers with incomes in excess of R100 000) and
in ation is 10%, then in a year’s time, the real value of the loan is only R1 800 (R2 000 2 [ 2 000]). If the
value of the loan is not linked to a price index, the real value of government debt decreases. In this case, it
may also be said that government nances its expenditure with an in ation tax.

11.2 Definition and classification of taxes

In 1789, Benjamin Franklin wrote: ‘In this world, nothing can be said to be certain, except death and taxes’
(Cohen & Cohen, 1960). What, then, is so abhorrent about taxes? Taxes are transfers of resources from
persons or economic units to government and are compulsory (or legally enforceable). ere is not
necessarily a direct connection between the resources transferred to government and the goods and services
that it supplies. In fact, government can compel one group of individuals to make payments that are used to
nance activities to the bene t of another group. Taxes are compulsory owing to the free-rider problem. As
no one would pay taxes voluntarily, people have to be compelled to do so. e fact that government has
legally been granted the power to tax, distinguishes government’s con scation of resources through taxation
from other involuntary transfers of resources (for example, theft).
However, government does not have unlimited powers as far as taxation is concerned. e Constitution
of the Republic of South Africa (1996) provides for money bills, that is, bills that provide government with the
legal right to appropriate amounts of money or to impose taxes, levies or duties. All money bills must be
considered in accordance with the procedure established by Section 75 of the Constitution and an act of
parliament must provide for a procedure to amend money bills before parliament. Since the National
Assembly must pass this type of bill, it implies that, in addition to procedural limits, the imposition of taxes
and changes to taxes are subject to the checks and balances of parliament. In a representative democracy,
parliamentarians must take cognisance of the preferences of voters as well as those of other parties (for
example, vested interests) that may in uence scal decisions. e in uence of interest groups in this process
is considered to be particularly important (see Chapter 6) and may even explain why the fear that ‘the
numerous poor will out-vote the rich and middle classes, and tax away much of their wealth’ is more
apparent than real (Becker, 1985).
Taxes can be classi ed in many ways. e classi cation used in South Africa is the one that conforms to
the International Monetary Fund’s Government Finance Statistics Manual, of which the latest version was
published in 2014. Tax revenue is grouped into six categories, each of which is then sub-divided further. e
main tax categories in respect of central government (with the approximate percentage contribution to total
tax revenue net of South African Customs Union payments1 for 2018/19 in brackets) are as follows (National
Treasury, 2019a: 195):
• Taxes on income and pro ts (60,0%)
• Taxes on payroll and workforce (1,4%)
• Taxes on property (1,3%)
• Domestic taxes on goods and services (36,7%)
• Taxes on international trade and transactions (4,5%)
• Stamp duties and fees (0,0%).

Income taxes clearly constituted the most important category of taxes in 2018/19. Taxes on the income of
individuals contributed 66,2% towards the total taxes on income and pro ts, while tax on the income of
corporations contributed 29,1%. In order to do a proper analysis, we will classify taxes according to tax base
in the next section. We will then distinguish between general and selective taxes (Section 11.2.2), speci c
and ad valorem taxes (Section 11.2.3), and direct and indirect taxes (Section 11.2.4).

11.2.1 Tax base and rates of taxation

Taxes can generally be imposed on three tax bases: income, wealth and consumption (sales or
transactions). A tax on people can be added to these three bases. A lump-sum tax per head (or poll tax) is
an example of a tax on people. Tax bases can also be viewed in terms of their ow and stock characteristics.
Flows are associated with a time dimension and are measured over a period. Income and consumption are
ow concepts since both are measured over a period of time (normally a tax year). Stocks have no time
dimension and are measured at a particular point in time, for example, wealth. Which of the three potential
tax bases is the best is a much-debated issue. Most countries have hybrid systems, exploiting three (or four)
bases simultaneously. Tax systems differ according to historical and political circumstances as well as the
stage of economic development of the country. is aspect is discussed in Section 16.5 of Chapter 16.
Once the tax base has been identi ed, the tax rate structure can be set. e tax rate structure describes
the relationship between the tax rate and the tax base. e tax rate refers to the amount of tax levied per unit
of the tax base. ree variants can be distinguished: proportional, progressive and regressive taxes. e rate
structure can be described in at least two ways. One way is to compare the average tax rate (the total rand
amount of taxes collected divided by the rand value of the taxable base) to the size or value of the tax base.
erefore, when the average tax rate remains constant with respect to variations in the tax base, the tax is
proportional (see Box 11.1). When the average tax rate increases as the tax base increases, the tax is
progressive. If the average tax rate decreases as the tax base increases, the tax is regressive.
e rate structure can also be described by focusing on the ratio of taxes paid to income. All taxes are
then evaluated in terms of the income base as a common denominator for purposes of comparison between
different taxpayers, even though, in practice, the tax may be imposed on a different base. In this way, the
distributional consequences of taxes can be determined. Taxes that generate the same proportion of income
as income rises are proportional (for example, corporate income tax). A tax with a proportional rate
structure (e.g. 20% of taxable income) is also called a at-rate tax. Taxes that take an increasing proportion of
income as income increases are progressive taxes (for example, personal income tax). If a tax generates a
decreasing proportion of income as income increases, it is a regressive tax (for example, a value-added tax
without any zero ratings). See Box 11.1.

Box 11.1 Proportional, progressive and regressive taxes

Consider a tax on the consumption base. Suppose a tax of 15% is imposed on the consumption of all essential
expenditure items. Suppose furthermore that there are two households: the Peterson household with a monthly
income of R1 500 and the Chetty household with a monthly income of R250 000. Is this tax progressive,
proportional or regressive? According to a report by the Bureau of Market Research (Bureau of Market
 Research, 2017) on Household Income and Expenditure Trends and Patterns, 2013–2017, the low-income
group (monthly income of less than R1 750) spends approximately 70,0% of its income on essential items,
whereas the high-income group (monthly income in excess of R204 000 per month) spends about 36,0% of its
income on essential items (Bureau of Market Research, 2017). The Petersons thus spend approximately R1
050 on essential items per month and the Chettys spend R90 000 (almost eighty six times more than the
amount of the Petersons). A 15% tax on essential items will therefore cost the Petersons R158 per month in
taxes, while the Chettys will pay R13 500. In absolute terms, the Chettys thus contribute more in tax on food
products to the South African Revenue Services (SARS) than the Petersons. When we divide the tax collected
(R158 or R13 500) by the rand value of the taxable sales or transactions base (R1 050 or R90 000), the
average tax rate is the same . Our first conclusion would
therefore be that the tax is proportional since the average tax rate does not vary with the size of the tax base,
namely consumption. However, if we evaluate the tax in terms of the income base, as is commonly done, the
picture looks rather different. The Petersons have to pay over almost 11,0% of their income ( 3 100) to
SARS, but in the case of the Chettys, it is less than 6,0% of their income .Thus the tax
structure is regressive with respect to income, but proportional with respect to the sales (or transactions) base.

11.2.2 General and selective taxes

Taxes can be classi ed as general or selective taxes. A general tax (also called a broad-based tax) is one that
taxes the entire tax base and allows for no exemptions. A value-added tax (VAT) without any exemptions or
zero-ratings is a general tax on consumption. Similarly, an income tax that taxes all sources of income
(including capital income) without any tax deductions is a general tax. Selective taxes (also called narrow-
based taxes) are imposed on one or a few products, or only wage income (for example, excluding interest
income). e whole tax base is therefore not taxed. An excise tax on cigarettes is another example of a
selective tax. e importance of the distinction between general and selective taxes lies in the fact that under
certain assumptions,2 general taxes are similar to head or lump-sum taxes, which leave relative prices
unchanged. In other words, the behaviour of economic role-players is assumed to be unaffected by these
taxes. In contrast, selective taxes distort relative prices by driving a wedge between the before-tax price and
the after-tax price of a commodity. is violates the Pareto-efficiency condition of optimal welfare in the
sense that one person’s welfare is increased at the cost of another person. is aspect will again receive
attention when we consider the efficiency of different taxes in Section 12.1 of Chapter 12.

11.2.3 Specific and ad valorem taxes

Taxes can also be speci ed according to the size or the value of the tax base. e size of the tax base can be
measured in terms of weight, quantities or units. When a xed amount is imposed per unit of the product,
the tax is called a unit tax or speci c tax. Examples of speci c taxes in South Africa in 2019/20 are excise
duties on the following products: sparkling wine (R13,55 per litre), beer made from malt (R1,74 per average
340 ml can) and cigarettes (R16,66 per twenty cigarettes).
Taxes imposed on the value of products are called ad valorem taxes. Such a tax is usually levied as a rate
(that is, a percentage) of the excisable value (or price) of a commodity. VAT and excise duties are examples of
ad valorem taxes. Ad valorem taxes (sometimes called ad valorem duties) are often imposed on luxury
goods. Ad valorem excise duties were introduced in 1969 for revenue purposes and were levied on certain
locally manufactured goods with a corresponding ad valorem customs duty (at the same rate) levied on
imported goods of the same kind. For example, in South Africa, commodities such as sun protection
products with a sun protection factor of less than fteen, perfumes and toilet waters, golf balls, video
cameras for non-commercial application and cellphones are taxed at rates varying between 7% and 9% of
the price. Motor vehicles are taxed using a graduated formula that distinguishes between vehicle type and
weight, and translates into a higher excise duty on high-priced (luxury) cars.
11.2.4 Direct and indirect taxes
Yet another distinction can be made between direct and indirect taxes. Direct taxes are imposed directly on
individuals and companies (for example, personal income tax and company tax). Indirect taxes are imposed
on commodities (for example, excise taxes and VAT). is distinction fundamentally revolves around the
issue of tax incidence (that is, the question of who really pays the tax). Tax incidence is a complex issue and
is analysed in Section 11.5. For the moment, it suffices to say that we simply cannot tell with certainty in
advance who is going to bear the burden.
From the perspective of tax shifting, direct taxes are de ned as taxes that cannot be shifted readily. ey
are collected from individuals, households or rms and allow for the possibility of adjusting the tax
according to the personal circumstances of the taxpayer (for example, the marital status, gender, size of
household or wealth status). ese taxpayers are the intended bearers of the tax burden and it is assumed
that they pay the tax over to SARS. Nowadays, personal income tax is mostly deducted from employees’
salaries and paid over by employers. Nonetheless, the employer does not have complete information on
non-salary income (for example, donations, interest, rent or capital gains), and the individual is therefore
still responsible for the completeness and correctness of the tax assessment.
Indirect taxes are taxes that are imposed on commodities or market transactions and are likely to be
shifted. Examples are excise duties and fuel levies. It is also more difficult to adjust the tax rate to the
personal circumstances of the consumer. In the case of indirect taxes, it is often possible to shift the burden
of the tax to someone else. VAT is collected from merchants who, in turn, can pass on the tax to consumers
by way of a price increase. e consumer then indirectly bears the burden.
Although there are differences of opinion on the exact distinction between direct and indirect taxes, this
classi cation is widely used. e relative importance of direct versus indirect taxes is much debated,
internationally as well as in South Africa. In 1975/76, the ratio of direct to indirect taxes was 2,7, that is, for
each rand in indirect taxes collected, R2,70 was collected in direct taxes. In 2018/19, this ratio was 1,5, which
shows a clear shift in the direction of indirect taxes. ese trends will be discussed in Section 16.5 of Chapter
16 when the topic of tax reform is considered. e merits and demerits of indirect taxes will receive attention
in Chapter 16.

11.3 Properties of a ‘good’ tax

Tax systems evolve from time to time as new taxes are introduced and others are amended. e question is
whether these changes are good or bad. To evaluate these changes, we can use a list of criteria, some of
which date back to the maxims of taxation proposed by Adam Smith in 1776. A ‘good’ tax system must rst of
all generate sufficient revenue to nance budgeted government expenditure. e other more important
criteria or properties of a ‘good’ tax system can be classi ed under four headings:
• Equity: Taxes should promote an equitable (or ‘fair’) distribution of income. Cognisance has to be taken
of the fact that the burden of taxes can be shifted. To determine fairness, the incidence of taxes, therefore,
has to be examined. ese topics are covered in Section 11.5.
• Economic efficiency: All taxes impose a burden and most taxes affect the behaviour of taxpayers (in
other words, they cause an excess burden). Taxes should be designed in such a way that their distorting
effects on the choices made by taxpayers are minimised. is aspect is discussed in Section 12.1 of
Chapter 12.
• Administrative efficiency: Taxes are levied to yield sufficient revenue, but in order to be efficient,
administration and compliance costs have to be kept low. is calls for tax simplicity and certainty. is
topic is discussed in Section 12.3 of Chapter 12.
• Flexibility: As economic circumstances change, taxes and tax rates need to adjust. Taxes should therefore
be exible enough to facilitate macroeconomic stability and economic development. is aspect is
discussed in Section 12.4 of Chapter 12.
11.4 Taxation and equity: Concepts of fairness
One of the most important judgements in tax analysis is whether or not the distributional impact of a tax is
equitable or fair. However, fairness is a subjective concept and, like beauty, it lies in the eye of the beholder.
Although fairness is a value-laden concept, economists can help to make informed value judgements. In his
rst maxim of taxation, Adam Smith (1776), as quoted in Musgrave and Musgrave (1989: 219), stated:

e subjects of every state ought to contribute towards the support of the government, as nearly as possible,
in proportion to their respective abilities; that is, in proportion to the revenue which they respectively enjoy
under the protection of the state.

is statement contains a tax equity principle that is still used in theory and practice to evaluate the fairness
of the impact (or incidence) of taxes, that is, the ability-to-pay principle. Another principle is based on
bene ts received. We will rst consider the bene t principle.

11.4.1 Benefit principle

e bene t principle stipulates that the tax burden of government expenditure be apportioned to taxpayers
in accordance with the bene ts each receives. Consider the example of a bridge: the more an individual uses
the bridge, the more he or she will have to pay for this bene t. If the bridge is nanced out of general
revenues instead of through a bene t tax (for example, a toll) levied on people using the bridge, those who
do not use the bridge will be in a worse position. It seems unfair to expect non-users also to pay for the
construction and maintenance of the bridge. Here we have an example of forced carrying: while free-riding
refers to someone failing to carry a proper tax burden, forced carrying refers to someone being made to carry
a heavier-than proper burden. It can also be argued that tax morality will be undermined if taxpayers do not
bene t from a tax system and that a democratic society will become intolerant of this type of system in the
long run.
A major advantage of the bene t principle is that it links the expenditure side of the government’s budget
to the revenue side. In this way, it serves to discipline or regulate government expenditure. A further bene t
is that the allocative procedures of market behaviour are approximated. Individuals can adjust their
consumption of services until price is equal to the marginal cost. In a voluntary exchange model, if the price
is higher than the marginal cost, society places a higher value on an additional unit of the service than the
resources required to provide the additional unit. In terms of Pareto-optimality conditions, society’s welfare
can then be improved by allocating more resources to the provision of the service, that is, at least one
person’s position can be improved without causing anyone else to be in a worse position. Bene t taxes
assume the role of prices and can therefore ensure that resources are efficiently allocated.
Unfortunately, the scope for applying the bene t principle to government funding is rather limited.
Governments generally provide goods and services that are public in nature. In other words, the bene ts are
generally non-excludable (see Chapter 3). For example, how would one apportion the bene ts of protection
(for example, policing) to different bene ciaries? ere is no straightforward answer. Some people would
argue that the rich bene t most because they have so much to lose and that they therefore have to contribute
more. Others would argue that the poor are the most vulnerable (in other words, most in need of protection)
and that they thus have to pay more.
Another shortcoming of bene t taxes is that it takes the existing distribution of income and wealth for
granted. e effective demand for public goods and services is often determined by this distribution. If the
distribution is skewed towards the rich, the provision of public services might therefore be tailored largely to
their needs. e bene t approach can ideally only handle a tax-expenditure process that has no
redistributive objectives (that is, where redistribution is not a signi cant justi cation for the service being
provided). It cannot handle a tax-expenditure programme designed for redistributive purposes (for example,
taxes levied to nance transfer payments or expenditure programmes designed to bene t the poor rather
than the rich) or where it is administratively not feasible to exempt poor people from paying for the bene ts
of a particular service (for example, a toll road). It would, for instance, be somewhat ridiculous if pensioners
had to bear the current taxes required to nance the transfers to themselves. e bene t principle can
therefore undermine redistributive objectives of government and result in con icting outcomes.
Bene t taxes (often referred to as user charges) are nevertheless levied in some cases, for example, tolls
for roads and bridges, admission charges to museums and parks, license fees, and (to a certain extent)
university tuition fees and school fees. Note that in the case of user charges, the link between the nancing
of the use of the service and the bene t is reasonably direct.
e bene t approach may therefore be rationalised on the grounds of equity and efficiency. However, a
frequent criticism against user charges is that they may contradict the equity objective by preventing the
poor from using government-supplied services. For example, it is argued that education and publicly
supplied cultural and recreational facilities such as art galleries, museums and parks, often referred to as
merit goods, should be made available free of charge to all people, that is, nanced from general revenue.
is will enable people who are unable to pay for the use of these facilities to enjoy them. A problem with
this approach is that the implicit subsidy that is involved when general nancing is applied accrues to both
the rich and the poor. If the services are heavily used by the rich, the subsidy to the rich may be substantial.
To the extent that the poor contribute to general revenue, a part of the cost of the subsidy is borne by the
poor. In certain circumstances, general fund nancing could even redistribute income from the poor to the
rich. is dilemma may be resolved by imposing a charge for the use of facilities, but offering special
reductions to targeted groups such as children, the elderly and so on.
Sometimes services are nanced in such a way that the link is more indirect. is occurs when charges
are assigned (or dedicated) to special funds or accounts for nancing services that are indirectly related to
the source of the funds. In these cases, the taxes are called earmarked taxes. Examples of earmarked taxes
are levies on the sale of fuel, social security taxes (for example, unemployment insurance contributions) and
skills development levies. Fuel levies at provincial level have been proposed in some provinces (for example,
the Western Cape) and are intended for maintenance of transport infrastructure. At the time of writing, fuel
levies consisted of a general levy of R3,52 per litre (in respect of 93 unleaded octane petrol), the Road
Accident Fund (RAF) levy of R1,98 per litre, a customs and excise levy of 4,0 cents per litre, and a carbon tax
of 9,0 cents per litre (effective from June 2019). Only the RAF has the character of an earmarked tax since a
portion of the general fuel levy nances general government expenditure and accrues to the National
Revenue Fund. Government decided to do this on the recommendation of the Katz Commission (1995) that
a portion of the general fuel levy should be allocated to the National Road Agency for road construction and
maintenance. e Skills Development Act (No. 97 of 1998) as amended (2003) provides for a levy of 1% of
payrolls aimed at, for example, developing the skills of the South African workforce, improving productivity
in the workplace, promoting self-employment, and increasing the levels of investment in education and
training in the labour market.
e shortcomings of the bene t principle generally apply to earmarked taxes as well. In addition,
earmarked taxes affect the procedural fairness of the budgetary process. Since earmarked funds have
assured sources of revenue, these funds need not compete with other departments for nance. In the case of
earmarked funds, accountability and the responsibility for efficient resource allocation are also shifted to the
managers of these funds. Earmarked taxes also complicate scal policy aimed at achieving macroeconomic
objectives. e Katz Commission (1995: 24) therefore recommended against a proliferation of earmarked
taxes (especially in respect of general tax revenue such as VAT, where the link between bene ts and costs is
extremely vague).

11.4.2 Ability-to-pay principle

As already mentioned, the bene t principle cannot be applied to the nancing of public goods and services
that are non-excludable. Public expenditures are predominantly of this type. Total cost of public expenditure
therefore has to be apportioned according to the ability of people to pay. In contrast with the bene t
principle, the application of the ability-to-pay principle implies that the tax problem is viewed
independently from the expenditure aspect.
e ability-to-pay principle calls for people with equal capacity to pay the same amount of tax
(horizontal equity) and for people with greater capacity to pay more (vertical equity). Horizontal equity
requires similar treatment of people in similar economic circumstances for tax purposes, while vertical
equity requires that individuals in different economic circumstances be treated differently.
e implementation of a tax system based on ability to pay requires public consensus on an appropriate
de nition (in other words, measure, indicator, criterion or basis) of ability to pay. It also calls for consensus
on the rate structure. Income is generally regarded as one such measure or criterion. Other possible
measures include consumption, wealth and utility. Much of the discussion concerning the appropriate
measure or base is theoretical, but, in practice, income is commonly used.
However, income is by no means a perfect measure of a person’s ability to pay. Income measures
outcomes and does not necessarily re ect ability or capacity. Consider, for example, two people, Ms
Moleketi and Ms Pienaar, with similar economic circumstances and the same capacity to earn income. e
economic circumstances referred to here include gender, race, religion, marital status, level of education,
disability, number of dependants and so on. Ms Moleketi, however, is much more hardworking and works
ten hours a day at a rate of R50 per hour, compared to the ve hours that Ms Pienaar works for the same
hourly pay. Ms Moleketi therefore earns R500 per day compared to Ms Pienaar’s R250. Using income as the
yardstick of ability to pay implies that Ms Moleketi has a greater ability to pay and that she should therefore
make a greater tax sacri ce. Is this fair? In this case, is the wage rate, instead of total wage income, not a
more appropriate measure of ability? Not surprisingly, the inclusion or exclusion of the various possible
determinants of economic circumstances in the search for an appropriate base for ability to pay has been
hotly debated.3
e appropriate measurement of horizontal and vertical equity is not only a complicated issue, but also
requires subjective evaluations. Economics alone cannot provide unambiguous answers to the
measurement of ability and the nal decision therefore has to be taken via the political process.
Once the appropriate measure of ability to pay has been established, fairness suggests that people with
the same ability should pay the same amount of tax and people with different abilities should pay different
amounts. Suppose income is used to measure ability to pay. is still leaves the question of how to
determine the taxes payable by people with different incomes. For example, if person A’s income is double
that of person B, should person A pay exactly twice as much tax as person B? is relates to vertical equity
and is a rate structure issue. Various concepts involving sacri ce have been developed in an attempt to deal
with the problem of vertical equity.4
Suffice it to say that most countries try to deal with vertical equity by applying the progressive tax rule to
income (in other words, by taxing an increasing proportion of income as income increases). But this does
not solve the dilemma. To determine whether a tax is truly progressive, proportional or regressive, one needs
to know who really pays the tax. In other words, the impact of a tax on the distribution of income has to be
investigated. is is what tax incidence is all about.

11.5 Tax incidence: Partial equilibrium analysis

We noted that in order to understand whether a tax is fair or not, we need to know the ultimate incidence of
the tax. We rst investigate the meaning of tax incidence and the shifting of the tax burden. ereafter, the
impact of tax shifting on the tax burden is analysed using partial equilibrium analysis as well as a general
equilibrium framework.

11.5.1 Concepts of incidence

All taxes reduce the real disposable income of taxpayers. All taxes therefore involve a burden. At the outset, it
should be noted that only people bear the burden. Companies, for example, cannot bear the tax burden.
Only people (for example, shareholders, workers, landlords and consumers) bear the burden of taxation.
Exactly who ultimately bears the burden is a matter for theoretical discourse and empirical evidence. To deal
with this question, we rst have to distinguish between the statutory (or legal) incidence of a tax and its
economic (or effective) incidence.
 e statutory incidence refers to the legal liability to pay the tax over to the revenue authorities.
Economists are not really interested in who is legally obliged to make the tax payment to SARS. Since taxes
affect economic behaviour, economists are more concerned with who ultimately bears the tax burden, that
is, the economic incidence of the tax. For example, customs and excise taxes are levied on cigarettes. At the
time of writing, an importer of cigarettes had to pay a customs duty of R16,66 per twenty cigarettes to SARS.
However, the importer can shift the actual burden forward to retailers who, in turn, can pass on the tax to the
consumers (the smokers) in the form of higher prices. e tax burden can also be shifted backward if, for
example, the importer cuts back on staff or lowers real wages. In addition to tax shifting, the tax can be
avoided by cutting back on the taxed activity, that is, by reducing imports of cigarettes. What we notice here
is that the economic incidence issue is fundamentally about how taxes affect prices (including wages).
Tax incidence studies may apply either a balanced-budget incidence methodology or a differential-
incidence methodology. According to the balanced-budget incidence approach, the overall distributional
effect of a tax and the spending nanced by the tax is considered. Income taxes lower real disposable
income. At the same time, however, the income tax revenue is spent on education and other public services
that raise real disposable income (at least for those with school-going children). e advantage of the
balanced-budget incidence approach is that it relates the cost of spending programmes to those who pay for
it. e disadvantage is that, in reality, government uses a number of different taxes to nance expenditure.
Tax revenue is pooled in the national revenue account. Linking a particular expenditure item to a tax source
is almost impossible.
Since not all taxes are earmarked for a particular expenditure programme, the differential-tax incidence
approach comes in handy. e differential-tax incidence methodology considers the distributional impact
as one tax is substituted for another, holding total revenue and expenditure constant. e benchmark tax
often used for purposes of comparison is a lump-sum tax (for example, head tax), which does not affect
relative prices and hence economic behaviour. For example, suppose government replaced a lump sum tax
on all redheads with an excise tax on beer drinkers that yielded the same tax revenue. Assuming that
redheads were not beer drinkers, they would gain and beer drinkers would lose. Furthermore, the owners of
breweries and their employees would be affected by the tax. By comparing the total impact of the tax change
on the incomes of beer drinkers to those of redheads, we are engaging in differential-tax incidence analysis.
Ultimately, the economic incidence of a tax depends on how the economy reacts to a tax change. e
response of the economy can be determined by tracing the effect of the tax on prices. Taxes change the
relative prices of goods and services. For example, if a R2,00 dairy levy on butter increases the price of butter
from R20,50 to R22,50 per 500 g while other prices remain unchanged, the relative price of butter also
increases. If it is assumed that the initial change in the price of butter does not have signi cant repercussions
on prices in other markets, tax incidence can be analysed in terms of a partial equilibrium framework. In
other words, we can then study the effect of the tax in a single market in isolation. In partial equilibrium
analysis, price and quantity in each market are determined by demand and supply, with the assumption that
other prices remain unchanged (in other words, the ceteris paribus assumption is used).
However, when a tax is imposed on one good, it usually means that the prices of other goods change as
well. For example, a levy on butter will affect the price of margarine, a substitute. After the imposition of the
levy, butter will be relatively more expensive and margarine will be relatively cheaper than before.
Consumers will therefore tend to substitute margarine for butter and the increased demand for margarine
will probably lead to a rise in the (absolute) price of margarine. is will leave margarine consumers with
less real disposable income than before. Once the secondary effects of a tax are taken into account, tax
incidence is studied in a general equilibrium framework.
Both partial and general equilibrium analyses can be used in incidence studies. Partial equilibrium
studies are less complex since all of the rami cations of a tax change are not considered. is form of
analysis is ideally suited to studying taxes levied on goods and services that are characterised by low degrees
of substitutability or complementarity. Where the secondary effects are considered to be small, uncertain or
spread thinly over a number of other markets, economists would argue that these can be ignored and that
the conclusions based on partial equilibrium analysis will not differ signi cantly from those based on
general equilibrium analysis. e general equilibrium framework, however, is more suitable for studying the
incidence of a general sales tax that is levied on a broad base or a levy on an important product (such as fuel)
that has important rami cations for the economy. Since general equilibrium analysis considers relative
price changes in more than one market, it is conceptually superior to partial equilibrium analysis.

11.5.2 Partial equilibrium analysis of tax incidence

In this section, we analyse the incidence effect of taxes within a partial equilibrium framework. e ability to
shift the tax burden under partial equilibrium conditions depends on a number of factors, but we will
consider only two: market structure and price elasticity.
Two types of taxes are considered: a unit tax (in other words, a xed amount per unit of a good or service
sold; see Section 11.2.3) and an ad valorem tax. We start by assuming that there is perfect competition and
that the tax is levied on an increasing-cost industry (in other words, the supply curve slopes upward).

11.5.2.1 Incidence of a unit tax

In Figure 11.1(a), D0 is the demand curve and S0 is the supply curve for bread. e before-tax equilibrium
price is P0 and the equilibrium quantity is Q0. Suppose that an excise tax (a unit tax) of t per unit is imposed
on bread. e excise tax is a xed amount per unit of the product. If the tax is collected from the seller (in
other words, the statutory burden is on the seller), it means that the minimum price at which rms will
supply the equilibrium quantity, Q0, is P0 1 t. Since the sellers are interested in the after-tax price, they will
try to recover the full tax amount at any given quantity. ey will charge a higher price (the original price 1
the tax) at each output level. e supply curve therefore shifts up vertically by the tax amount to S1 (5 S0 1 t).
Note that because it is a unit tax, S1 is parallel to S0. e new market equilibrium is at point A with price Pm
and quantity Q1. e market price, which is the price paid by buyers of bread, increases from P0 to Pm. Sellers
receive the after-tax price, Ps (that is, Pm 2 t).
What is the total tax revenue that accrues to government? e formula for total revenue is price
multiplied by quantity. Total tax revenue is the tax per unit t (5 PmPs 5 AC) multiplied by the number of
units sold, 0Q1 (5 PsC). Geometrically, the total revenue is represented by the rectangle PmACPs. e
important question now is: Who pays what part of the total tax amount? e answer to this question will tell
us what the economic incidence of a unit tax on consumption is.
Before the imposition of the tax, consumers paid a price P0 for the product. After the tax, they pay Pm.
Note that the price that buyers have to pay did not increase by the full amount of the tax (Pm minus P0 is less
than t or AB is less than AC). e total tax amount paid by consumers is equal to the rectangle PmABP0. What
about the sellers? e before-tax price received was P0. e after-tax price received by sellers is Ps, which is
less than P0. Producers therefore also pay part of the tax. e total tax amount paid by sellers is equal to the
rectangle P0BCPs. From this analysis, it is clear that the total tax burden is split between buyers and sellers;
that is, the tax incidence falls on both. Uninformed observers generally assume that sellers simply pass on
the total tax burden to consumers. Partial equilibrium analysis shows that this need not be the case.
Would the result be any different had the statutory burden been on buyers (that is, the consumers)? e
answer is no. To show this, one must remember that the demand curve in our example indicates the
maximum price that consumers are willing to pay for each different quantity of bread. In Figure 11.1(b), the
before-tax equilibrium price is P0 and the equilibrium quantity is Q0. At the equilibrium quantity Q0, the
consumer is only willing to pay P0. If a tax of t [equal to the size of the tax in Figure 11.1(a)] is now collected
from the buyer instead of the seller for quantity Q0, the buyer will still only be willing to pay P0. e price
paid to the seller, however, will be reduced by the tax amount (P0 – t) that is paid over to SARS by the buyer.
is will hold for each point on the demand curve. In this case, therefore, the effective demand curve (in
other words, as perceived by the seller) shifts downward by the tax amount. e new demand curve is D1 (5
D0 2 t). Since a unit tax of t is imposed, D1 is parallel to D0. e after-tax equilibrium is at point C with price
Ps and quantity Q1. While the original demand curve, D0, shows what price buyers are willing to pay for each
quantity, the demand curve D1 shows the price received by sellers (that is, the after-tax price) for each
quantity. us for quantity Q1, buyers are willing to pay Pm and sellers receive Ps (5 Pm – t). e result is
exactly the same as the result obtained when the tax was levied on sellers: the tax incidence is on both
buyers and sellers, and their burdens are exactly the same. e side of the market on which the tax is
imposed does not impact on the distribution of the tax burdens. What matters is the tax wedge, which is the
tax-induced difference between the price paid by the buyer and the price received by the seller.

Figure 11.1 Incidence of a unit tax on consumption imposed on the supply side or the demand side

11.5.2.2 Incidence of an ad valorem tax

An ad valorem tax is levied as a percentage of the price of a good or service (see Section 11.2.3). We now
examine the imposition of an ad valorem tax, collected from the producers of DVD players, for example,
with the aid of Figure 11.2.
Figure 11.2 Incidence of an ad valorem tax

e most important difference between the analysis of an ad valorem tax and a unit tax is that the after-
tax supply curve swivels in the case of an ad valorem tax (compared to the parallel shift in the case of a unit
tax). Because the tax is proportional, the higher the price, the greater the amount of tax to be paid over by
producers (or the higher the after-tax price paid by consumers) will be. For example, at a price of R1 000 per
DVD player and an ad valorem rate of 20%, the absolute amount of tax is R200 (5 3 1000) If the price is
5
R2 000, the absolute amount of tax is R400.
us, when an after-tax supply curve such as S1 in Figure 11.2 is constructed, the vertical distance
between S1 and S0 (the before-tax supply curve) at a relatively high price such as Ph (5 DE) should be greater
than at a low price such as Pl (5 FG). e rest of the incidence analysis of an ad valorem tax is similar to that
of a unit tax. In Figure 11.2, the total tax burden of buyers is the rectangle PmABP0 and the total tax burden of
sellers is P0 BCPs. e ad valorem tax rate, expressed as the ratio of tax to the gross price paid by the buyer,
equals

11.5.2.3 Incidence and pure monopoly

Until now, we have investigated tax incidence under conditions of perfect competition. Pure monopoly is
another extreme form of market structure. It is often taken for granted that a monopolist, being a price-
maker, can always shift taxes forward in full. Under certain circumstances (for example, where a monopolist
is not maximising pro t), this may indeed be the case; however, the conclusion does not necessarily apply.
Let’s consider the tax-shifting capacity of a pro t-maximising monopolist.
A monopolist maximises pro ts where marginal cost (MC) equals marginal revenue (MR). In Figure 11.3,
the before-tax equilibrium of the pure monopolist is at quantity Q0 and price P0. e before-tax pro t is the
area P0 ABC. Assume that a unit tax on output is now imposed on the monopolist.
Figure 11.3 Incidence of a tax on a pure monopoly

A unit tax on the output of the monopolist will raise the average cost (AC) and the marginal cost of the
rm. e reason why both AC and MC increase is that the tax is levied on each unit produced and is
therefore viewed by the rm as a variable cost. In Figure 11.3, the average cost curve shifts from AC0 to AC1
and the marginal cost curve from MC0 to MC1. Pro ts are now maximised at an output of Q1 and at price P1.
If you compare the before-tax equilibrium to the after-tax equilibrium, you will notice the following:
• e after-tax price is higher and the quantity is lower than before the imposition of the tax, but the price
did not increase by the same amount as the tax. e increase in the price from P0 to P1 indicates the
extent of forward shifting.
• e after-tax pro t (the rectangle P1FDE) is less than the before-tax pro t (the rectangle P0ABC). e
pure monopolist therefore does not shift the tax burden fully.

Monopolists can also be taxed (at least in theory) on their economic pro ts. Economic pro t (or excess
pro t) is the difference between total revenue and total costs (in other words, both explicit and implicit
costs). Implicit costs include the opportunity costs of self-owned resources (that is, normal pro t). e cost
curves in Figure 11.3 represent economic costs. In other words, if average revenue exceeds average cost,
economic (or excess) pro ts are earned. In Figure 11.3, the pure monopolist’s before-tax pro t (area P0ABC)
is considered to be economic pro t. A tax at a rate of t per cent on the monopolist’s pro t of P0ABC will
simply reduce its economic pro t by the tax amount. Neither the average nor the marginal cost curves will
be affected. e monopolist will maximise after-tax pro t at the same before-tax level of output and price.
e tax is therefore borne fully by the owners of the rm. In these circumstances, the pro ts tax cannot be
shifted.6

11.5.2.4 Incidence and price elasticities of demand and supply

As mentioned at the beginning of Section 11.5.2, tax incidence is affected by market structure (for example,
perfect competition or pure monopoly) and price elasticities. When a tax is imposed, both the quantity
demanded and the quantity supplied at equilibrium decrease. e magnitude of the decrease depends on
the elasticity of demand and supply. Likewise, the impact on the tax burden and the relative tax shares of
buyers and sellers also depend on the price elasticities.
e importance of price elasticities in tax incidence analysis can be illustrated with the aid of diagrams.
We will rst consider the effect of demand elasticities and then the effect of supply elasticities.
Figure 11.4 shows two demand curves, D0 and D1, and a supply curve, S0, all intersecting at X. e before-
tax equilibrium price (P0) is the same, irrespective of the demand curve used. e intersection of the
demand curves enables us to compare elasticity at this point. e demand curve D1 is more inelastic at any
price than the demand curve D0, that is, quantity demanded is less responsive (or sensitive) to price changes
in the case of D1. Suppose the taxed good is cigarettes. If a unit tax (t) is now imposed on the importer or
seller, the effective supply curve shifts upwards and to the left from S0 to S1 (where S1 5 S0 1 t). Let us
consider the relatively inelastic demand curve D1 rst.
If the demand for cigarettes is represented by D1, the price charged to buyers increases from P0 to P1. e
price received by the seller (or importer) decreases from P0 to P2. e proportional price change for the
buyers therefore exceeds that of the sellers. e tax burden (P1CFP2) is divided between the portion borne by
the buyers (the area P1CDP0) and the portion borne by the sellers (the area P0DFP2). In the case of the more
elastic demand curve D0, the proportional price changes for buyers and sellers are approximately the same
in this example. e total tax burden of P3ABP4 is then divided between the buyers (the area P3AEP0) and the
sellers (the area P0EBP4). us, the more price inelastic the demand for a product is, the greater the relative
portion of the tax borne by buyers, ceteris paribus (sometimes referred to as the inverse elasticity rule).
Conversely, the more price elastic the demand for the product is, the greater the relative portion of the tax
borne by the sellers.

Figure 11.4 Tax incidence and price elasticity of demand

Remember that the ip-side of the tax burden for consumers and producers is government’s tax revenue.
Comparing demand curve D1 to D0, it is evident that the more price inelastic the demand curve is, the
greater the total tax revenue government collects is (P1CFP2 compared to P3ABP4).
 We can also show that, ceteris paribus, the more price inelastic the supply of a product is, the greater the
relative portion of the tax borne by the sellers is. In Figure 11.5, we have two supply curves, S0 and S1,
intersecting a demand curve D0 at point X. e supply curve S1 is relatively more inelastic than the supply
curve S0. To simplify the analysis, suppose that the unit tax (t) on cigarettes is now imposed on buyers (in
other words, the statutory burden is on smokers). is means that the effective (after-tax) demand curve
shifts downwards to the left from D0 to D1 (where D1 5 D0 – t). Let us rst analyse the shifting of the tax
burden in the case of price-inelastic supply S1. e new equilibrium quantity is Q1. e after-tax price paid
by smokers increases from P0 to P2, whereas the after-tax price received by sellers decreases from P0 to P1.
e proportional price change for sellers is greater than for buyers. Put differently, the share of sellers in the
total tax burden P2ACP1 is P0BCP1, which is greater than the share of buyers (that is, P2ABP0). By contrast, in
the case of the elastic supply curve S0, we notice that the proportional change in the after-tax price of sellers
is less than that of buyers. e new equilibrium quantity is Q2. e buyers pay a price (P4) per unit and the
sellers receive a price (P3) per unit. e portion of the total tax burden P4DFP3 borne by the sellers (that is,
P0EFP3) in the case of the elastic supply curve is less than the burden of the buyers (that is, P4DEP0).

Figure 11.5 Tax incidence and price elasticity of supply

From our discussion of price elasticity so far, we can generalise by stating that the more inelastic the demand
and the more elastic the supply are, the easier it is to shift the burden of a tax forward through a higher
selling price. is conclusion is illustrated in Figures 11.4 and 11.5. By rotating the demand curve D0 and the
supply curve S0 around X in each case, we note that the buyer’s portion of the burden increases as the
demand curve becomes steeper (in other words, more inelastic) and the supply curve becomes atter (in
other words, more elastic). In addition to the examples provided here, the extreme possibilities of perfectly
(in)elastic demand and perfectly (in)elastic supply also exist; however, we will not consider these
possibilities here.
Applying the theoretical discussion of tax incidence to, for example, taxing tobacco and alcohol
consumption as well as sugar sweetened beverages (SSBs) (so-called sin products) reveals some interesting
policy perspectives and dilemmas. ere is some empirical evidence that the price elasticity of cigarette
demand ranges between –0,5 and –0,7 in South Africa (see Boshoff, 2008: 128–129). Consumption is thus
relatively price inelastic (it is less than 1 in absolute value) and the quantity demanded is not that responsive
to price (tax) changes (see demand curve D1 in Figure 11.4). A 10% increase in the price of cigarettes will lead
to a reduction in quantity demanded of only 5%. In South Africa a health promotion levy, which taxes sugary
beverages, was implemented from 1 April 2018. In support of the bene cial effects of such a tax, a South
African study used a price elasticity of –1,299 to show that a 10% increase in the price of SSBs would reduce
consumption by 13% and thus impact signi cantly on health outcomes (e.g. obesity) (Manyema, Veerman,
Chola, Tugendhaft, Sartorius, Labadarios & Hofman, 2014: 4). Expectations regarding how high (or higher)
excises on sin products will reduce consumption should not be too optimistic. ere are also other factors
that impact on consumption, such as income (higher disposable income leads to higher consumption) and
people’s health awareness. Policy interventions such as advertising, labelling and physician counselling
improve people’s awareness of health risks and reduce consumption. Higher cigarette prices also impact on
the number of smokers who substitute illegal for legal cigarettes. Similarly, higher taxes on SSBs may lead to
higher consumption of substitutable sweetened drinks such as coffee, tea and hot chocolate. us lower
official consumption gures may not re ect the full picture. To the extent that the calculation did not control
for other factors, the price-elasticity coefficients noted above may thus be overestimating the reduction in
the number of smokers and obese persons caused by higher excises.
Another interesting consideration when analysing excise taxes (sin taxes) imposed on sugar sweetened
beverages, cigarette and alcohol products is how these taxes may affect the welfare of, in particular, those
earning low incomes. In 2014/15 poor households spent approximately 1,5% of their total annual
expenditure on mineral waters, soft drinks, fruit and vegetable juices whereas the upper expenditure group
spent only 0,3% (StatsSA, 2017c: 111). e tax is clearly regressive from this perspective since the poor spend
a higher proportion of their expenditure on these goods.
We have noticed that the incidence of an excise on cigarette consumption is mainly on consumers (the
area P1CDP0 in Figure 11.4). is tax burden is distributed unevenly across households and categories of
users (for example, moderate drinkers versus abusers). Studies for the United States and the United
Kingdom have found that these taxes tend to be regressive. e tobacco and alcohol spending of the poor as
a percentage of total income is much higher than that of the rich. According to the Bureau of Market
Research (BMR), poor households in South Africa in 2017 spent 7,8% of total expenditure on alcoholic
beverages and tobacco, compared to 2,5% by the richest households (Bureau of Market Research, 2017).
Indirect taxes (excises) on these products will therefore be shared approximately in this proportion, implying
that taxes on sin products are regressive. Black (2008: 607–611) and Econex (2010: 7–38) also argue
eloquently that an excise tax on alcohol is a blunt instrument in addressing the negative external effects of
abusive (heavy) drinking. e Econex study (2010: 19) shows that the vast majority of (moderate) drinkers
pay 87% of the tax, yet contribute perhaps 20% to the externality. A better option would be to target heavy
drinkers directly by enforcing stiff penalties for drunken driving as well as abusive and intrusive behaviour
(see Section 3.7 of Chapter 3).
ere is evidence that low- and middle income groups are more sensitive to price increases in tobacco
and alcohol goods than the high-income group. And if – as argued in Chapter 3 – many low-income
household budgets are controlled by males who are also addicted to nicotine and alcohol, an increase in
excise taxes may have perverse effects (see Black & Mohamed, 2006: 131–136 and Econex, 2010: 36–38).
Money may be diverted from the household budget to maintain the user’s level of addiction. He or she may
also substitute lower-quality sin goods (for example, cigarette butts and low-quality wine in foil bags or
home-brewed beer concoctions, which may be damaging to health) for high-quality ones. e negative
externality is actually aggravated. e potential for policy con ict is apparent; applying the inverse elasticity
rule (see Chapter 16, Section 16.2) to tobacco and alcohol consumption implies that tax revenue objectives
must be traded-off against reducing negative externalities, and ensuring that the burden is shared fairly
between the rich and poor, and abusers and non-abusers alike.
11.6 General equilibrium analysis of tax incidence
In Section 11.5, we distinguished between partial equilibrium analysis and general equilibrium analysis.
Partial equilibrium analysis examines a single market in isolation (in other words, on the basis of the ceteris
paribus assumption) and ignores the secondary effects of a price change. When the secondary effects of a
tax are taken into account, we are studying tax incidence within a general equilibrium framework.
Because general equilibrium analysis considers what happens in all markets simultaneously, modelling
tax incidence becomes rather complex. To make it more manageable, certain assumptions have to be made.7
Assume that only two products are produced in the economy: shoes and reed baskets. ere are two
factors of production, labour and capital, which are perfectly mobile between sectors. Both factors of
production are xed in supply (in other words, the supply curves are vertical). Shoes are produced using a
capital–intensive technique (in other words, the capital–labour ratio is high), whereas reed baskets are
produced using a labour-intensive method (in other words, the capital–labour ratio is low). We will analyse
and compare the incidence of a tax on one product (that is, a selective tax) and the incidence of a tax on both
products (that is, a general tax). In Chapter 15, we will discuss a selective tax on one factor of production (for
example, land).

11.6.1 A selective tax on commodities

Suppose that a tax is imposed on shoes. e price of shoes increases relative to that of reed baskets.
Remember that the after-tax price does not necessarily increase by the full amount of the tax (in Figure 11.1,
the tax was equal to AC, but the price increase was only AB). e price increase has two effects: one is on the
consumption (uses) side; the other is on the factors of production (sources) side.
On the uses side, the price increase of shoes (which is, of course, also a relative price increase) causes
consumers to demand fewer shoes (illustrated by a movement along the demand curve for shoes) and
demand more reed baskets. When the demand for reed baskets increases (illustrated by a shift of the
demand curve to the right), the price of reed baskets increases, ceteris paribus. e tax thus causes an
increase in the price of both products; that is, the tax burden (and incidence) is spread to the consumers
(and producers) of the non-taxed product as well.
On the sources side, the fact that the tax on shoes results in fewer shoes being demanded implies that
fewer shoes will need to be produced. When fewer shoes are produced, some of the capital and labour used
in the production process become redundant. But since shoes are produced using capital-intensive
technology, relatively more capital than labour is released into the market. e redundant labour and capital
must now nd employment in the sector manufacturing reed baskets. Since the reed basket sector is a
labour-intensive sector (and technology is assumed to remain unchanged), all of the redundant capital
cannot be absorbed into the sector. e additional capital can only be absorbed if the capital–labour ratio
increases. But with capital in excess supply, its relative price will decrease, so that both sectors will end up
using more capital-intensive production techniques. us, in addition to the rise in the relative price of the
non-taxed commodity (reed baskets), a tax on shoes also causes the relative price of capital (that is, the
return on capital) to decrease. e burden of the tax is therefore also spread to the owners of the factor of
production used most intensively in the production of the taxed commodity.

11.6.2 A general tax on commodities

If a tax is imposed on shoes and reed baskets simultaneously at the same rate, the tax is a general one as
opposed to a selective tax. e general tax is equivalent to a tax on income (for example, a tax on capital and
labour at the same rate). e tax will be borne in proportion to the consumption (or income) of each
member of the economy. A general tax on commodities leaves relative prices unchanged (including the
relative price of leisure).8
11.7 Tax incidence and tax equity revisited
When we introduced the topic of tax shifting and tax incidence, we stated that most taxes alter the
distribution of income. To make inferences about the equity of a tax, it must be ascertained whether the tax
alters the distribution in a progressive, proportional or regressive way.
In Sections 11.5 and 11.6 we examined the question of who bears what part of the tax burden in the case
of consumption taxes in particular. If the economic burden of the tax is on buyers and the expenditure on
this good or service increases as individuals move up the income scale (in other words, if the demand is
relatively income elastic), the tax is progressive. Luxuries tend to fall into this category of goods and services
since individuals tend to spend more on luxuries as their income increases. In contrast, expenditure on
necessities tends to fall as total income rises (in other words, low-income earners spend proportionally more
on these goods and services than high-income earners). A tax on necessities that is shifted to buyers is
therefore regressive. A value-added tax or a general sales tax levied on a broad base will be regressive. e
reason is that consumption as a percentage of income decreases as individuals move up the income ladder.
High-income earners tend to save proportionally more, leaving them with a proportionally lower tax burden
than low-income earners (since saving is exempt from the tax).
When the tax is shifted to the seller, it also has distributional implications. If the seller has to bear the tax,
his or her real factor income (wage, rent, interest or pro t) is reduced. Whether the tax is progressive or
regressive now depends on the factor income shares of the ‘sellers’ and their relative positions on the
income ladder. If the tax is shifted to unskilled workers who nd themselves at the bottom end of the income
scale, the tax could be said to be regressive. On the other hand, if the tax is shifted to highly skilled workers
earning high incomes, the tax will be progressive.
To determine the nal impact of a consumption tax on distribution, the effect on buyers and sellers
should be considered simultaneously. According to Musgrave and Musgrave (1989: 254), there is no
systematic relation between the distribution in consumption of a good and the distribution of the factor
income that the production of a good generates. ey conclude that in the absence of evidence to the
contrary, the distribution of the tax burden is dominated by what happens to consumption (in other words,
the extent of tax shifting to buyers) since the initial impact is on the uses side.
e incidence of alternative taxes will be addressed when they are examined in later chapters. In
addition to the impact of individual taxes on distribution, we would also like to know the overall impact of
taxes on income distribution. is is a daunting task since we need to know the incidence of each tax
beforehand (see Box 11.2). Based on plausible assumptions about the shifting of different taxes, empirical
studies have found the tax structure of countries to be progressive for the low-income and high-income
groups, and regressive for the intermediate income groups.9 Similar results have been obtained for South
Africa.10
It seems that not much redistribution is taking place through the scal system from the tax side. is
conclusion has persuaded many economists to argue that if the objective is to improve the position of poor
people, it is best to pursue the objective through the expenditure side of the national budget rather than
through the tax system (see Section 16.5 of Chapter 16).

Box 11.2 Incidence of a tax on crayfish

There is some evidence that government has shifted its attitude regarding the taxation of luxury goods, as is
evident from the extension of excise duties to aircraft, helicopters, motorboats and sailboats in recent times
(see National Treasury, 2012: 58). In a very unequal society such as ours, a populist approach would be to
stretch the tax net on luxury goods even further. However, actions of this nature may have unintended
consequences if the incidence effects of taxes are not properly traced out. Determining the distributional
impact requires a great deal of information and data on the income and expenditure patterns of socio-economic
groups. For most indirect taxes, the tax incidence is also likely to remain uncertain. Usually researchers and
policy-makers assume that excise taxes are shifted 100% forward to consumers, but this is not the full story.
Let us take a glance at a possible chain of events when a hypothetical ad valorem excise duty is imposed on
crayfish processing establishments (suppliers). This would, of course, be in addition to the application fees,
 permit fees, licence fees and other fees that crayfish harvesting and processing already attract. Crayfish (also
known as West Coast Rock lobster) is usually served as an expensive dish at seafood restaurants. A
reasonable assumption for income and price elasticity of demand for crayfish would be that they are greater
than one (in other words, demand is fairly elastic). Since this delicacy is a luxury good, individuals would be
able to adjust their consumption quite easily as the (tax) price changes. Consumers will nevertheless have to
pay a higher price, while the suppliers will receive a lower after-tax price. From a vertical equity perspective the
higher price paid by consumers would be considered fair. After all, these consumers are mostly in the higher-
income groups. From the consumption side and based on income and expenditure patterns, the tax would thus
be progressive. However, owing to the relatively elastic demand for crayfish, suppliers will pay the larger share
of the tax. How will suppliers respond to the lower after-tax price and the decline in the quantity demanded?
Owners of crayfish processing facilities will surely take a knock (their return on capital decreases) in the short
term. In the long run, they will attempt to shift the burden to their employees by reducing employment numbers,
offering lower wage increases and/or paying lower prices to crayfish permit holders. These factors of production
are mostly unskilled or semi-skilled, earn low average wages and have been historically marginalised; the fish
permit holders (or their employees) are engaged in labour-intensive harvesting techniques and use their
resources to ensure food and livelihood security (see Department of Agriculture, Forestry & Fisheries, 2012;
Mather, Britz, Hecht & Sauer, 2003). Some of the retrenched workers may also try to find employment in other
industries. Depending on the factor intensities in the other industries, the job seekers may force wages down in
these industries. As a result, all labourers will bear part of the burden. It is thus conceivable that on the
production side, the excise tax on crayfish will be regressive. The net redistributional impact can only be
established using quantitative analysis, but the chain of events shows the complexity of designing equitable
taxes.

Key concepts
• ability-to-pay principle (page 227)
• administrative fees (page 222)
• ad valorem tax (page 225)
• balanced-budget incidence approach (page 231)
• benefit principle (page 227)
• differential-tax incidence (page 231)
• direct tax (page 226)
• earmarked tax (page 229)
• economic incidence (page 231)
• forced carrying (page 227)
• general equilibrium framework (page 232)
• general tax or broad-based tax (page 225)
• horizontal equity (page 229)
• indirect tax (page 226)
• inflation tax (page 222)
• inverse elasticity rule (page 237)
• lump-sum tax or poll tax (page 223)
• partial equilibrium framework (page 232)
• progressive tax (page 224)
• proportional tax (page 224)
• regressive tax (page 224)
• selective tax or narrow-based tax (page 225)
• specific tax (page 225)
• statutory incidence (page 231)
• tax base (page 223)
• taxes (page 222)
• tax rate (page 224)
• tax rate structure (page 224)
• tax wedge (page 234)
• unit tax (page 225)
• user charges or benefit taxes (page 222)
• vertical equity (page 229)

SUMMARY
• e most important source of government revenue is taxation. In addition to taxation, there are user
charges (for example, tolls), administrative fees (for example, motor vehicle licences), borrowing and
in ation taxation.
• ere are four tax bases on which taxes can be imposed: income (for example, personal income tax and
company tax), wealth (for example, tax on estates and gifts), consumption (for example, value-added
tax) and persons (for example, poll tax or lump-sum tax). e different bases can be taxed at different
rates. Depending on the relationship between the change in the average tax rate and the change in the
tax base, the tax structure can be described as regressive, progressive or proportional. Taxes can further
be classi ed as general or selective, unit or ad valorem and direct or indirect taxes.
• e properties of a ‘good’ tax are as follows: taxes must be fair (equitable) and non-distorting
(economically efficient), must yield sufficient revenue with low compliance cost (administrative
efficiency) and must adjust to economic circumstances ( exible).
• Taxes are equitable when they meet either the bene t principle of fairness or the ability-to-pay principle
of fairness, or both. e ability-to-pay principle requires that persons with the same capacity (or income)
be treated similarly (horizontal ability) and those with greater capacities should be pay more (vertical
ability).
• Since tax burdens can be shifted (forwards and backwards), the fairness of a tax can only be determined
once the economic incidence of the tax is known. Tax incidence is studied using partial or general
equilibrium analysis.
• Using a partial equilibrium framework, it can be shown that the incidence is determined by the type of
tax (unit tax or ad valorem tax), price elasticity and market structure. A unit tax would cause the after-tax
demand or supply curve to shift parallel to the original demand or supply curve, whereas an ad valorem
tax would cause the demand or supply curve to swivel. Under certain circumstances, a monopolist may
not be able to shift a tax in full. e extent to which consumers (buyers) and producers (sellers) share the
tax burden is dependent on the price elasticity of demand and supply. e more price elastic demand is,
the greater the share of producers; the more price inelastic supply is, the greater the relative share of
producers.
• General equilibrium analysis considers how taxes are shifted when relative prices change and impact on
both the consumption (uses) side of the market and the production (factor source) side of the market. A
selective tax causes a burden not only for the consumers of the taxed good, but also for the consumers of
nontaxed goods. In addition, the factors of production used most intensively in the production of the
taxed good also bear the burden of the tax.

MULTIPLE-CHOICE QUESTIONS
11.1 A 14% tax on the price of cigarettes …
a. is a unit tax.
b. is a direct tax because it taxes the smoker directly.
c. would cause the after-tax supply curve of cigarettes to shift parallel to the original curve.
d. would result in the tax burden being borne mostly by the smokers themselves.
11.2 Which one of the following statements is correct?
a. National Defence should be funded using bene t taxes.
 b. An excise tax of R50 is payable on DVD players priced at R500. e excise tax increases to R120
when the value of DVDs is R1 000. e tax rate structure is therefore progressive.
c. Mr Average earns R5 000 a month and Ms Hasarrived earns R100 000 a month. Each of them
purchases a digital camera valued at R1 000 and pays an import tax of R200. e tax rate structure is
therefore proportional.
d. A tax on luxury goods would be fair from a horizontal equity point of view.
11.3 Consider Figure 11.4. Which one of the following statements is correct?
a. When demand is relatively inelastic, the tax burden can easily be shifted to producers.
b. If the demand curve is D1, the total tax burden is P1CXFP2.
c. If the demand curve is D0, the tax share of the supplier is P0EBP4.
d. Because the tax was imposed on sellers, the demand curve shifted to the left.
e. If the demand curve is D0, the total tax revenue of government is obtained by multiplying the tax
per unit by the tax base, that is, P2P3 3 0Q2.
11.4 Assume our economy produces only two goods: wine and BMWs. Wine is produced using labour-
intensive technology and BMWs are produced using capital-intensive technology. An excise tax is
imposed on BMWs. Which of the following statements is or are correct?
a. e most appropriate analytical framework to use is partial equilibrium analysis.
b. e price of wine will increase.
c. More capital than labour will become redundant.
d. e incidence of the tax will be mainly on capital.
e. If an income tax had been imposed on labour and capital at the same rate, no tax shifting would
have occurred.

SHORT-ANSWER QUESTIONS
11.1 Distinguish between taxes and other sources of government revenue.
11.2 A R1 000 tax payable by all adults could be viewed as both a proportional tax and a regressive tax. Do
you agree? Explain.
11.3 Indicate whether the following taxes or levies are general or selective, speci c or ad valorem, direct or
indirect. Explain your answer in each case.
• Personal income tax
• A 10% tax on DVD players
• Value-added tax
• R1,02 per litre levy on fuel
• A levy of R200 on all economics students.
11.4 ‘When taxes are evaluated, we need only be concerned with fairness.’ Do you agree? Discuss.

ESSAY QUESTIONS
11.1 Explain what is meant by tax equity. Discuss critically how applicable these fairness principles are to
nancing toll roads.
11.2 By using the partial equilibrium framework, explain who bears the burden of an ad valorem tax levied
on buyers (in other words, consumers) if there is perfect competition and demand is relatively inelastic.
11.3 ‘It is a misconception that a monopolist can simply pass on the burden of a tax on its product to
consumers.’ Discuss this statement.
11.4 e Minister of Finance wishes to reduce inequality and alcohol abuse by signi cantly increasing the
excise tax on spirits since it has a high price elasticity (e 5 21,5) and income elasticity (e . 1) of
demand. Using partial and general equilibrium analysis, advise the Minister on the revenue and equity
implications. What alternatives are there to raising the excise tax?
1 e reason for calculating this on a net basis is that South Africa collects trade-related taxes on behalf of the Southern African
Customs Union and thereafter pays over to the other member countries (Botswana, Lesotho, Eswatini and Namibia) their formula-
driven shares of the common customs union pool.
2 e most important assumptions are that the supplies of work effort and of savings remain unchanged (in other words, are
unaffected by the tax).
3 For a comprehensive debate on whether the individual or the household should be the taxable unit as well as the constitutional
implications of discrimination on the basis of gender and marital status in the South African tax system, see Margo Commission
(1987: 106–154) and Katz Commission (1994: 67–84).
4 For a discussion on using the sacri ce of utility as measure for determining ability to pay, see Musgrave (1959: 90–115).
5 e ad valorem tax can be levied as a percentage of the gross price (in other words, the price received by sellers). A 20% tax on the
gross price is equivalent to a 25% tax on the net price. If the buyer pays 20% on R1, the seller receives R0,80 (the net price). e tax
is R0,20. Twenty cents as a percentage of the gross price is 20% (5 3 100). As a ratio of the net price, it is 25% (5 3 100).

6 is conclusion holds only if the rm is maximising pro t prior to the introduction of the tax on economic pro ts. In the case of
unrealised pro ts or non-pro t maximising behaviour, tax shifting is possible.
7 is approach to tax incidence theory was pioneered by Harberger (1962). For a discussion of the extensions to and limitations of
the simpli ed model we are considering, consult Boadway and Wildasin (1984: 368–374).
8 is assumes that leisure can be taxed. e tax therefore cannot be shifted and consumers have to bear the entire burden.
9 For references, see Boadway and Wildasin (1984: 384–385).
10 See McGrath, Janisch and Horner (1997). In the same study, the authors calculated the tax share of the high-income group to be
75,7% of the total tax burden, compared to the tax share of the low-income group of 2,8%.
Tax efficiency, administrative efficiency and flexibility

Tjaart Steenekamp and Ada Jansen

In Chapter 11, we identi ed a number of properties of a ‘good’ tax and considered the equity criterion in
some detail. Most taxes affect relative prices and consequently also the economic choices of participants in
economic activity. is chapter focuses on the impact of taxes on efficiency in the allocation of resources.
Because tax efficiency is handicapped by the fact that people do not like paying taxes, the issue of tax
compliance will be discussed. Since tax systems are continuously being adjusted in response to changing
economic and political in uences, issues surrounding tax reform will be highlighted.
We begin this chapter (in Section 12.1) with an explanation of what the excess burden of a tax is, using
budget lines and indifference curves. e measurement of excess burden is discussed in Section 12.2 by
applying the consumer surplus approach. Once the impact of taxes on economic efficiency has been
analysed, we focus on the two remaining criteria for analysing taxes: administrative efficiency (Section 12.3)
and tax exibility (Section 12.4).

Once you have studied this chapter, you should be able to:
explain what is meant by tax efficiency
compare the excess burden of different taxes using indifference curves
determine the magnitude of excess burden using the consumer surplus concept
explain the meaning of administrative efficiency and how it can be achieved
show what is meant by tax evasion and how to reduce it
define tax flexibility.

12.1 Excess burden of taxation: Indifference curve analysis

All taxes place a burden on consumers, workers or producers. In addition to this direct burden, most taxes
cause a burden that is greater than what is necessary to generate a certain amount of tax revenue. is
additional burden is called the excess burden, welfare cost or deadweight loss of a tax. It measures the loss
in bene ts (well-being) to consumers and producers that results when prices are distorted by a tax, and then
prevents the market for the taxed good from reaching equilibrium at the most efficient level. An example will
help to illustrate this concept.
We know from Chapter 2 that for a given set of relative prices, perfectly competitive markets will allocate
resources in a Pareto-efficient way. is means that the marginal rate of substitution (MRS) of, say, (x) for (y)
will be equal to the marginal rate of transformation (MRT) of x for y (MRSxy = MRTxy = ). Furthermore,
given this set of relative prices, we know that the consumption of x and y yields a certain amount of
consumer satisfaction (or a certain level of welfare). Suppose that a tax (t) is now levied on x. e price of x
becomes (1 + t)Px. Suppose also that the tax is such that the consumption of x becomes zero. is is an
extreme example of excess burden. If this is the case, no tax revenue will be forthcoming and there will thus
be no direct tax burden. However, there must be some burden since consumers can now only enjoy y.
Because the relative price ratio changed consumers reallocated their resources (in other
words, their after-tax income) towards y. e expenditure pattern moved away from what was previously
regarded as optimal and desired, and consumers therefore experience a welfare loss (or loss in utility), even
though there is no direct tax burden. On the production side, production of good x must be discontinued
and the resources used to produce it must be reallocated to producing more of good y. In practice, workers
will probably have to be reskilled and some may even lose their jobs if the technology used in the production
of y is less labour-intensive. In addition, machinery used in the production of x may have to be moved to a
different location and reprogrammed to produce more of y. ese additional costs are (deadweight) losses
that result from the tax-induced price distortion.
Another example that is often cited to illustrate the effect of taxes on the behaviour of taxpayers is the so-
called window tax that was levied in England between 1695 and 1851. e tax was levied in proportion to the
number of windows in a house. It was obviously levied in order to generate revenue, but because the
wealthy, who had the largest houses, paid the most, it can be viewed as a form of wealth tax. However,
people do not like to pay taxes and some owners simply avoided the tax by bricking up some of their
windows. e loss in daylight that they had to endure (that is, the excess burden of the tax) has led some
people to believe that this is the origin of the phrase ‘daylight robbery’!
e theory of excess burden of different taxes can be explained using two approaches: the indifference
curve approach and the consumer surplus approach. e consumer surplus approach is useful for
measuring the excess burden and is discussed in Section 12.2. Here we focus on the indifference curve
approach, which is useful for comparing the price-distorting effects of different taxes. We rst look at the
impact of tax-induced changes in relative prices on the welfare of a particular consumer. We then distinguish
between the impact of a general tax and that of a selective tax. For comparison purposes, we use a lump-sum
tax as our general or benchmark tax. We therefore evaluate the excess burden of a selective tax by comparing
its impact on welfare with that of a lump-sum tax.

12.1.1 Lump-sum taxes and general taxes

A lump-sum tax is a xed amount of tax that an individual would pay in, for instance, a year, and is
independent of his or her income, wealth or consumption. Lump-sum taxes do not distort relative prices
and therefore do not affect people’s choices. For example, a person will not be able to avoid paying a lump-
sum tax by working shorter hours or by reducing his or her consumption of a particular commodity. Put
differently, a lump-sum tax does not cause a substitution effect. It reduces the taxpayer’s disposable income
and thus has an income effect only. Because relative prices are not distorted, lump-sum taxes have no excess
burden. A head tax, levied on each member of society or on all breadwinners, is an example of a lump-sum
tax.
Lump-sum taxes have one major disadvantage. Since the tax as a percentage of income (in other words,
the average tax rate) decreases as income increases, lump-sum taxes are regressive. ey therefore leave the
after-tax distribution of income more unequal than the before-tax distribution. For most policy-makers, the
price tag of this trade-off between equity and efficiency is simply too high. e perceived unfairness of the
poll tax (a prime example of a lump-sum tax), introduced by the atcher government in Britain in 1990, is
widely regarded as one of the factors that led to atcher’s downfall later that year.
A head tax is not an unfamiliar tax in Africa. An early version of a head tax in South Africa can be found in
the Glen Gray Act (No. 25 of 1894). is tax generated insigni cant tax revenue (R9 million, or approximately
0,1% of total tax revenue in 1978/79) and was abolished in 1978/79, the year in which general sales tax (GST)
was introduced.
e effects of a lump-sum tax are illustrated in Figure 12.1. Good X is measured horizontally and Y is
measured vertically. e before-tax budget line is AB. e consumer is initially in equilibrium at point E0,
where the indifference curve U0 is tangent to (in other words, it touches) the budget line. A lump-sum tax is
introduced that lowers the income of the consumer and causes the budget line to shift from AB to CD. e
lump-sum tax yields revenue equal to AC if measured in terms of Y or DB if measured in terms of X. Note
that the after-tax budget line, CD, is parallel to AB since relative prices are unchanged. e consumer is now
in equilibrium at E1, where fewer of X and fewer of Y are consumed than before.
Note that the consumer is in a worse position after the tax since consumption is on a lower indifference
curve, U1. is is as a result of the normal burden of the tax. e condition for Pareto optimality has not been
disturbed (MRSxy = ) and resources are allocated efficiently at the new after-tax income level. A lump-sum
tax thus causes a normal burden, but it has no excess burden.
In Section 11.2.2 of Chapter 11, a general tax was de ned as one that taxes the entire tax base and allows
for no exemptions. If a general tax is imposed at the same rate on Y and X, and it is assumed that these are
the only products produced in the economy (and leisure is ignored), the budget line in Figure 12.1 shifts
from AB to CD. Equilibrium is again at E1. Tax revenue equals AC. A general tax does not distort relative
prices. is means that the aftertax price ratio is the same as the before-tax price ratio . e
condition for a Pareto-efficient allocation of resources in the economy is therefore not distorted [MRSxy =
MRTxy = ].
General taxes resemble lump-sum taxes and do not have excess burdens. ese taxes have the added
advantage that, from a practical point of view, general taxes at uniform rates are easy to administer. However,
as in the case of lump-sum taxes, the disadvantage of a general tax is that it ignores distributional
implications. We will also note later (Section 12.2.2) that a general indirect tax contradicts the inverse
elasticity rule of allocative efficiency.

Figure 12.1 The excess burden of a tax using indifference curves

12.1.2 Selective taxes

In contrast to a lump-sum tax, a selective tax is one that taxes only a part of the general tax base. Suppose a
selective tax is now imposed on X. To analyse its welfare implications, we compare the impact of the
selective tax to that of a lump-sum tax that generates the same tax revenue (that is, AC) in Figure 12.1. We
know that a selective tax on X will increase the price of X and leave the price of Y unchanged. If the
consumer spends his or her entire budget on X, fewer of X can be obtained after the tax than before the tax. If
the consumer spends his or her entire budget on Y, the same amount of Y can still be purchased (because
the price of Y has remained unchanged). e budget line showing combinations of Y and X will, therefore,
swivel inward (or pivot around point A). To obtain the after-tax budget line, which yields the same tax
revenue as the lump-sum tax, we must nd an equilibrium point that is also on budget line CD. A budget
line therefore has to be drawn through A in such a way that it is tangent to an indifference curve at its point
of intersection with CD. In Figure 12.1, budget line AF is a budget line of this nature.
A selective tax on X that yields the same revenue as the lump-sum tax (that is, E2G = AC) will cause the
budget line AB to pivot (or swivel) to AF. Equilibrium is at E2 on indifference curve U2. is is an important
observation. Compared to the lump-sum tax, consumer welfare is lower (indifference curve U2 is lower than
U1). e difference in welfare indicates the welfare cost or excess burden of a selective tax. Selective taxes
distort relative prices and cause an excess burden. In terms of Pareto optimality, the equilibrium condition
for consumers is now where the marginal rate of substitution of X for Y (MRSxy) is equal to the after-tax price
ratio [ ]. Assuming that producers shift the full burden of the tax to consumers, the equilibrium
condition for producers is where the marginal rate of transformation of X for Y (MRTxy) is equal to the
before-tax price ratio ( ). e price ratios for consumers and producers thus differ and, as relative prices are
distorted by the selective tax, resources are allocated inefficiently in the economy (MRSxy ≠ MRTxy).

12.1.3 Tax neutrality

Economic efficiency is considered to be one of the properties of a ‘good’ tax. Efficient taxes are taxes that
minimise the distorting effect on the choices of decision-makers. In other words, taxes are considered
efficient when the excess burden is as small as possible. is brings us to the concept of tax neutrality.
e tax neutrality concept must be understood in its traditional, narrow context and also in its wider,
modern context. e traditional, narrow neutrality concept is mainly concerned with allocation, the idea
being that taxes should not prevent consumers from maximising utility or producers from maximising pro t;
in other words, taxes should be neutral. e tax should have little or no impact on economic decisions about
what to buy, how much to invest or save and how many hours to work. is view of neutrality rests on the
assumption that resources in the economy are allocated optimally, and that non-neutral taxes would result
in a reallocation and therefore a non-optimal allocation. A broader view of tax efficiency, however, also takes
account of market imperfections.1
In reality, the market economy is imperfect (meaning that it is inefficient and inequitable), and
optimality is the exception rather than the rule. Hence non-neutral taxes may be bene cial or harmful,
depending on whether they steer the economy in the direction of optimality or away from optimality. ese
two possibilities are called the positive non-neutral effect and the negative non-neutral effect
respectively.
A selective tax is clearly non-neutral as it disturbs relative prices. Economic choices are affected and,
according to the traditional view, selective taxes are therefore inefficient. According to the modern
approach, however, a selective tax can move the economy in the direction of optimality, which amounts to
positive non-neutral action. In contrast, a general tax such as a head tax is a neutral tax and does not
in uence allocation decisions. From the modern perspective, a general tax may well perpetuate existing
distortions in the economy or perpetuate a socially unacceptable income distribution, which implies that tax
non-neutralities can sometimes improve allocative efficiency. Examples include levying taxes to correct for
negative externalities and taxing sumptuary consumption (for example, excises on cigarettes and liquor).
e term ‘optimal taxation’ is derived from efforts to design tax systems to improve efficiency (minimise
excess burden) and to achieve a more socially equitable distribution of income (maximise social welfare). In
the sections that follow, we will touch upon some of the tax optimisation rules to minimise excess burden
(for example, the inverse elasticity rule; see Section 12.2.2). Related to the theory of optimal taxation is the
theory of second best. We saw in Chapter 2 and subsequent chapters that the market sometimes fails owing
to certain inefficiencies (or distortions). e theory of second best is concerned with designing government
policy (and therefore tax systems) in situations where some inefficiencies cannot be removed. More
accurately, the theory of second best states that whenever there are distortions in several markets, removing
one may not necessarily improve matters. In fact, it may introduce other distortions and, in general, may
result in a lower level of welfare for consumers (see Lipsey & Chrystal, 1995: 414–415).

12.2 Excess burden: Consumer surplus approach

We have made some progress in comparing different taxes in terms of their excess burdens. Selective taxes
cause an excess burden, whereas general taxes do not. Using the indifference curve approach, we have
shown that, compared to a selective tax, the same tax revenue can be obtained by levying a general tax or a
lump-sum tax, which leaves the consumer better off. We now wish to measure the change in consumer
welfare and investigate the factors that determine the magnitude of the excess burden.

12.2.1 The magnitude of excess burden

In this section, we follow the consumer surplus approach in order to measure the excess burden of a tax. To
enable measurement, we employ a partial equilibrium framework and focus on the burden of a unit tax on a
particular commodity or output of an industry. Suppose the commodity is grams of butter. We can simplify
the example by assuming that supply is produced under constant-cost conditions (in other words, the
supply curve is horizontal). Furthermore, we use standard demand curves. Although it is theoretically more
correct to use compensated demand curves2 to calculate the excess burden, we use standard demand
curves. Ignoring this subtlety will not affect our analysis signi cantly.
In Figure 12.2, the demand curve for butter is shown as D0 and the supply curve is S0. e equilibrium
price is Pb and the equilibrium quantity is Q0. We now have to introduce the consumer surplus. You will
recall that the demand curve indicates what consumers are willing to pay for different quantities. However,
once the market price is determined, it applies to all consumers. e difference between what different
consumers are willing to pay for a good and what they actually pay is the consumer surplus. is is
represented by the area under the demand curve and above the price line. e consumer surplus measures
the rand value of consumer welfare at different quantities.3 In Figure 12.2, the consumer surplus is ACE.
Suppose a selective tax (t) is levied on the producers of butter that increases the price of butter to (1 +
t)Pb. e supply curve S0 shifts upwards and parallel to S1. e equilibrium quantity of butter decreases from
Q0 to Q1 and the consumer surplus is reduced from ACE to ABF. e loss in consumer surplus is the
trapezoid FBCE. But this is not the total welfare cost (or loss) to the consumer. e tax revenue is the tax per
unit (tPb) multiplied by the quantity of butter purchased (Q1) (in symbols, FE × 0Q1 = FBDE). e tax on
butter therefore yields revenue equal to the area FBDE. If government, which after all spends the tax
revenue, were to return this amount of tax to consumers as a lump sum, the consumers would be worse off
by the triangle BCD (that is, FBCE – FBDE). e triangle is the welfare loss or excess burden of the selective
tax on butter. In other words, the tax causes a reallocation of resources (less butter and more other goods are
produced than without the tax). e triangle measures the welfare loss caused by this misallocation of
resources.4
Figure 12.2 The magnitude of the excess burden of a tax on a constant-cost industry

e size of the triangle (that is, the excess burden) is determined by the price elasticities of demand and
supply, as well as the tax rate. ese determinants are discussed below.
e magnitude of the excess burden can be measured using simple algebra. We know that the formula
for the area of a triangle is one-half the base multiplied by the height. In Figure 12.2, the excess burden can
therefore be expressed as

where Eb is the excess burden (area BCD), tP is the ad valorem tax (= BD) and ΔQ is the decrease in quantity
demanded (= DC). is formula can be re ned by also considering two other determinants of the excess
burden: price elasticities and the tax rate.

12.2.2 Price elasticities

Price elasticity of demand indicates how sensitive the demanded quantity is to a price change, while price
elasticity of supply indicates the same sensitivity in respect of the supplied quantity. If demand is inelastic,
buyers will tend not to adjust their quantities demanded by much if the price changes. In other words, they
are insensitive to price changes. If demand is elastic, buyers will tend to adjust their quantities demanded
signi cantly if the price changes. In other words, they are sensitive to price changes.
 What is the impact of price elasticity of demand on excess burden? Figure 12.3 is almost the same as
Figure 12.2. e only difference is that in Figure 12.3, we compare the impact of a selective tax on butter for
two cases of demand. In one case, the demand curve (D1) is relatively inelastic. In the other case, the
demand curve (D0) is relatively elastic. e before-tax quantity (Q0) and the price (Pb) are the same in both
cases. A selective tax (t) causes the supply curve to shift upwards from S0 to S1. In the case of demand curve
D0, the equilibrium quantity decreases to Q1. With demand curve D1, the equilibrium quantity decreases to
Q2, indicating that the quantity demanded is less sensitive to the imposition of the tax than in the case of D0.
Comparing excess burdens, we notice that for demand curve D0, the excess burden is BCD (the same as in
Figure 12.2) and for demand curve D1, the excess burden is GCH. is illustrates that the welfare cost of a
given tax is less where the demand for a good is relatively inelastic. In other words, the more elastic the
demand is, the greater the excess burden (or welfare cost).
Figure 12.3 also illustrates how price elasticities impact on the amount of tax revenue that can be
collected from a tax. Consider rst the relatively elastic demand curve D0. e tax revenue is the tax per unit
(tPb) multiplied by the quantity of butter purchased (Q1) (in other words, FE × 0Q1 = FBDE). With the more
inelastic demand curve D1, the tax revenue is clearly much greater (in other words, FE × 0Q2 = FGHE). From
a revenue perspective, it would make perfect sense for tax authorities to levy taxes on commodities for which
demand is relatively inelastic. e implications are that it is more efficient from an economic efficiency point
of view (that is, low excess burden) and tax revenue perspective (that is, high tax revenue) to levy taxes on
price-inelastic commodities than on price-elastic commodities. Commodity taxes should thus be high on
inelastic goods and services, and low on goods and services with high demand elasticities. is tax rule,
commonly referred to as the inverse elasticity rule, is attributed to Frank Ramsey (1927). A further
implication of this tax rule is that uniform tax rates are not necessarily efficient, since the higher the price
elasticity of demand of good X relative to that of good Y is, the lower the tax rate on X should be relative to
that on Y.
Applying the Ramsey rule to the design of taxes leads to interesting results. Luxury goods and services
typically have high price elasticities since the possibilities for product substitution are greater. Consumer
products with high elasticity coefficients include marijuana (–1,50), transatlantic travel (–1,30), restaurant
meals (–2,27), lamb and mutton (–2,65), and motor vehicles (–1,20). Necessities have lower elasticity
coefficients, for example, food (–0,21), bread (–0,15), petrol (–0,60) and medical services (–0,18) (see
Nicholson & Snyder, 2010: 131; McConnell & Brue, 2005: 363 for a list of price and income elasticities for the
United States). e price elasticity of demand for necessities tends to be low compared to luxury goods. In
Section 11.5.2 of Chapter 11, we noticed that the price elasticity of cigarette demand ranges between –0,5
and –0,7 in South Africa. A non-smoker would hardly consider cigarettes necessities, but this would not
surprise a smoker. e low coefficients are inter alia indicative of the addictive nature of nicotine. e
elasticity tax rule implies that a high tax rate on, for example, bread or insulin, would be efficient (the excess
burden is small and the tax revenue would be high). However, distributional implications are ignored in
arriving at these conclusions. Expenditure on bread constitutes a major proportion of the income of poor
people. e tax is therefore regressive and inequitable. Diabetics depend on insulin to keep them alive and
would probably pay anything to obtain it. Few would disagree that it would be very unfair to apply optimal
taxation rules to this product. Tax design often calls for a trade-off between equity and efficiency.
Figure 12.3 The effect of demand elasticity on excess burden

e effect of elasticity on the magnitude of the excess burden can be expressed in mathematical terms. In
Equation [12.1], the term ΔQ depends on the price elasticity of demand (ε), which is de ned as follows:

which can be expressed as

e change in price caused by the tax (ΔP) is equal to the value of the selective tax (that is, tP). e right
side of Equation [12.3] can therefore be rewritten as . By substituting this expression for ΔQ in
Equation [12.1], we obtain the following:

Equation [12.4] tells us three things. Firstly, the value of ε indicates the importance of price elasticities of
demand. If the value is low (demand is price inelastic), it means that the excess burden will be small, and
vice versa. Secondly, PQ is the amount spent on butter before the tax. In our equation, it means that the
higher this original amount spent is, the greater the excess burden. Finally, we notice that the excess burden
is a function of the tax rate. We focus on this nding in the next section.
Figure 12.4 The effect of tax rates on excess burden

12.2.3 The tax rate

e magnitude of the excess burden also depends on the tax rate. As the tax rate increases, the excess
burden increases by a multiple of the tax rate.
In Figure 12.4, the initial equilibrium is at price Pb and quantity Q0. e commodity, butter, is produced
under constant-cost conditions. If a selective tax of t2 is levied on butter, the after-tax price is (1 + t2)Pb and
the equilibrium quantity decreases to Q2. e excess burden is the triangle ACH. e tax revenue is the area
GAHE. Suppose that the tax rate is halved to t1. e new equilibrium price is (1 + t1)Pb and the equilibrium
quantity is Q1. Tax revenue is now equal to the area FBDE. e excess burden is now the triangle BCD. We
notice that although the tax rate was halved, tax revenue did not halve. More importantly, the excess burden
fell by about three-quarters. is can easily be con rmed geometrically by decomposing the triangle ACH
into four smaller triangles of equal size: ABK, KBH, BDH and BCD.
We can conclude that low tax rates on a large number of commodities will produce smaller excess
burdens (and more tax revenue) than high tax rates on a few commodities that yield the same total revenue.
is analysis therefore suggests that broad-based taxes such as VAT and income taxes are more efficient than
narrow-based selective taxes.
Excess burdens are real and, according to some estimates, quite signi cant (see below). Knowing the
excess burdens of different taxes is useful in designing taxes. If, for example, an ad valorem tax of 30% is
levied on a good of which the price elasticity of demand is equal to one and total expenditure on the good is
R200 million, it can be shown (using Equation [12.4]) that the excess burden will be R9 million (that is, 15%
of the tax revenue). Put differently, if the excess burden is expressed per rand of tax revenue, the efficiency–
loss ratio of the tax can be calculated (in other words, excess burden ÷ tax revenue). Given tax revenue of
R60 million from the ad valorem tax, the efficiency–loss ratio is 0,15. is means that for each rand raised in
taxes, the deadweight loss (excess burden) is 15 cents. By comparing the efficiency–loss ratios of different
taxes, tax designers can attempt to minimise the total excess burden of the tax system. Overall economic
efficiency can be improved by reducing taxes on goods and services with high efficiency–loss ratios, and
increasing taxes on goods and services with low efficiency–loss ratios.
Excess burden calculations are subject to theoretical and empirical limitations. When excess burdens are
estimated, compensated demand and supply curves should be used. Using standard demand and supply
curves overestimates the excess burden. e problems associated with attempts to make the concept
operational are illustrated by its absence in public budgets. e excess burden costs of taxes are not recorded
as expenditure items in public budgets. Given an uncompensated demand elasticity of –0,75 for alcohol and
assuming constant costs, Econex (2010: 11) estimates the total current value for the excess burden to be
R881,0 million ( gures for 2009). Of this amount, 87,26%, or R769,0 million, is carried by moderate drinkers.
e estimated excess burden for alcohol may be lower or higher, depending on the size of the income effect
and the possible loss in producer surplus if increasing costs are assumed (the supply curve is upward
sloping). For the United States, it was estimated that for each $1,00 of tax revenue raised in the personal
income tax system in 1994, the additional loss in net bene ts (excess burden or deadweight loss) was $1,65
(Feldstein, 1997). e loss is attributed to the impact of the tax-induced distortion on labour participation
rates and hours as well as the effect on investment in education, occupational choice, effort, location, more
attractive fringe bene ts, nicer working conditions and all of the other aspects of behaviour that affect the
short-run and long-run productivity and income of the individual.

12.3 Administrative efficiency

e excess burden is not the only cost of a tax. Taxes have to be administered and this generates costs. In
2016/17, the estimated expenditure by the South African Revenue Service (SARS) amounted to R10 795
million, which was approximately 0,89% of total tax revenue collected (South African Revenue Service, 2018:
53). In addition to administration costs borne by government and, ultimately, taxpayers, individual
taxpayers incur costs (called compliance costs) in order to meet their tax obligations. ese include the cost
of time spent lling in tax returns as well as the cost of employing tax specialists (accountants and lawyers).
When taxes are designed, these costs must also be considered. ere is some evidence that the compliance
costs are signi cantly greater than the administrative costs incurred by government. For example, Slemrod
and Sorum (1984) estimated compliance cost in the United States at between 5 and 7% of revenue collected
from income taxes. In New Zealand, Sandford and Hasseldine (1992) estimated compliance costs at
approximately 2,5% of GDP.
Administrative efficiency entails minimising both administration costs and compliance costs. Two
phenomena are related to these costs: tax avoidance and tax evasion. Tax avoidance is perfectly legal and
includes the actions by taxpayers to take advantage of special provisions (tax loopholes) in the tax code so
that their tax liability is reduced. Although tax avoidance is legal, it is wasteful in the sense that taxpayers
make choices on the basis of tax considerations rather than economic considerations, that is, it entails high
opportunity costs. Avoidance practices often arise from errors or loosely drafted tax legislation. Exploiting
these loopholes through careful tax planning is therefore not in the ‘spirit of the law’. An example is where a
business is split up into smaller units to take advantage of the graduated company tax rate for small
businesses.
Tax evasion is illegal and consists of actions that contravene tax laws. e most common forms of
evasion are not registering as a taxpayer, underreporting of income and claiming more deductions than
warranted. Tax evasion is quite prevalent in the informal sector (also called the unrecorded sector) and is
characterised by cash transactions that are difficult to trace.
To get a clearer understanding of the actions that can be taken to reduce tax cheating, we need to know
the extent of tax evasion, the reasons why taxpayers engage in evasion practices and what the optimal level
of tax evasion is from the perspective of the cheating taxpayer.
 In some countries, tax evasion is said to have become a national pastime. South Africa is not untouched
by taxpayers’ attempts to reduce their tax burden. e extent of tax evasion can be measured by considering
the tax gap, which is the difference between the tax liability declared to tax authorities (the actual revenue
collected) and the tax base calculated from other sources (the expected tax revenue in accordance with tax
legislation). Knowing the size of the tax gap is important as a performance measure for the revenue
authorities, and helps them to identify problems with tax legislation, national statistics and what the impact
of the informal sector is on tax revenue. Various attempts have been made by countries to estimate the gap
for different taxes and the total tax gap, but it remains a difficult task to quantify the gap accurately (see
McManus & Warren, 2006: 77–79 for examples of tax gap studies). e Margo Commission (1987: 406) and
the Katz Commission (1994: 62) refer to tax gap studies that indicate that the gap is in the order of 10% for
developed countries and an average of 33% for developing countries. e Katz Commission (1994: 66)
guesstimated the tax gap in South Africa to be around 20% of actual collections. If this gure is accurate, it
would have amounted to approximately R232 billion in lost tax revenue in 2017/18. In 2015 the International
Monetary Fund estimated the value added tax (VAT) tax gap for South Africa, a task commissioned by the
Davis Tax Committee5. e main ndings revealed a compliance gap ranging between 5% and 10% of
potential VAT revenues for the period 2007 to 2012 (IMF, 2015a: 5).6 e VAT policy gap was estimated to be
low compared to international standards, due to an efficient VAT policy structure (IMF, 2015a: 6). Another
indicator of tax evasion and avoidance is the difference between persons who have chosen to register for
personal income tax purposes and the number of persons employed (that is, the potential tax base). In 2017,
there were approximately 22,4 million employed and unemployed workers, but only 4,9 million were
assessed as individual taxpayers. We return to this aspect in Chapter 13 when we look at the personal
income tax base.
Various factors contribute to the tax gap, including tax illiteracy, cultural attitudes, the level of tax
morality and de cient tax legislation. Oberholzer (2008) studied the differing perceptions about taxation
that result from people’s cultural, political and social backgrounds. All of the respondents in his study were
of the opinion that waste and corruption in government were high, and almost 80% felt that taxes were used
by government for meaningless purposes. is of course affects taxpayers’ willingness to pay taxes (that is,
their tax morality). At local government level, similar problems apply. Fjeldstad (2004) argues that an
explanation for non-payment is that non-compliance is caused by poverty (in other words, an inability to
pay) and the existence of a so-called culture of entitlement. In addition, the paper argues that non-payment
is related to whether or not citizens trust the local government to act in their interest. Trust has three
dimensions: the local authority will use tax revenues to provide the expected services; the authorities will
apply fair procedures in collecting tax; and other citizens will pay their share. In other words, taxpayers want
value for their money and taxes must be fair from a horizontal and vertical equity point of view. Suffice to say
there are psychological reasons and economic reasons for tax evasion (see Gcabo & Robinson, 2007). In this
section we focus primarily on the economic reasoning behind tax evasion and what the optimal level of tax
evasion is.
Neoclassical economic theory considers decision-makers as rational beings who attempt to maximise
their expected income and utility. Note that deciding how much to evade implies that the taxpayer has
already made up his or her mind about the morality of tax evasion (that is, the taxpayer acts illegally). We can
then employ an ordinary demand-and-supply framework to answer the ‘how much’ question. Tax evasion
generates bene ts and costs. If we assume a progressive income tax structure, the marginal bene ts from
evading an extra rand’s income decline as evasion increases. e individual faces lower tax brackets and
thus lower taxes. For example, if the marginal tax rate is 35%, the tax liability for the extra rand not reported
is reduced by R0,35. is is the marginal bene t received by the taxpayer. If more tax is evaded, the marginal
tax rate may be reduced to, say, 18% (because taxable income becomes less), translating into a tax liability of
only R0,18 for the extra rand evaded. Of course, when no taxable income is reported, the marginal tax rate
becomes zero. We can plot this relationship in Figure 12.5(a) as a downward-sloping marginal bene t (MB)
curve. A marginal bene t curve can also be interpreted as a demand curve for underreported income (D0).
When evading taxes, the taxpayer runs the risk of being caught and penalised. Marginal costs of evasion,
therefore, increase as evasion increases. is can be plotted as a marginal cost (MC) curve or a supply curve
for underreported income that increases from left to right (S0). To complete the picture, we consider both the
marginal bene ts and the marginal costs of evasion in Figure 12.5(a). Equilibrium (E0) is where MC = MB. At
this point, the optimal level of underreported income is Q0.

Figure 12.5 Optimal tax evasion and policy options

To reduce the level of underreporting taxes, the revenue authorities have two clear policy options:
increase the marginal cost for underreporting and/or decrease the marginal bene ts that taxpayers derive
from tax cheating. e marginal cost can be increased by higher penalties if a taxpayer is caught cheating
and by increasing the probability of getting caught. Both would have the effect of shifting the supply curve in
Figure 12.5(b) from S0 to S1. e new equilibrium is at E1, thereby reducing underreported income from Q0 to
Q1. e revenue authorities, however, cannot monitor all taxpayers at all times. In their enforcement
programme, they encourage compliance by randomly auditing taxpayers. e audit coverage ratio of the
South African Revenue Service (SARS) has consistently exceeded the target set. It amounted to 14,47% of
taxpayers in 2017/18 (South African Revenue Service, 2018: 68), which compares very favourably with that of
other countries. Although economic theory predicts that increased sanctions against underreporting of
income or non-payment of service charges should lead to better compliance, this may not necessarily
always be the result. Experience from local authorities in South Africa shows that the more severe the
sanctions observed are, the greater the resistance to paying tax is. Fjeldstad (2004: 552) ascribes this
‘perverse’ relationship to reciprocity considerations: ‘the proposition that the authorities’ unresponsive,
disrespectful and unfair treatment of ratepayers fosters disrespect for and resistance against local authorities
and the service providing agencies …’. A softer approach was followed by SARS at one stage when it
encouraged taxpayers who had failed to meet their obligations to make a disclosure of their sins in return for
a waiving of penalties or additional tax (National Treasury, 2005a: 97). But to ensure fairness in the tax
system and to improve budget revenue, the revenue authorities subsequently revised penalties, requiring
higher penalties on the higher taxable evaders (National Treasury, 2009a: 39). Penalties for income tax
evasion can be up to double the amount evaded (that is, 200%).
e alternative to combating tax evasion through penalties is to decrease marginal tax rates in an effort to
lower the bene ts from underreporting tax. Lower marginal tax rates would cause the demand curve in
Figure 12.5(b) to shift from D0 to D1. Equilibrium is at E2 and the amount of income underreported also
decreases from Q0 to Q1. ere is some evidence that lower marginal tax rates may positively affect
underreporting. In a study measuring the effect of tax rates on tax evasion on Chinese imports, Fisman and
Wei (2004) found that the evasion gap is highly correlated with tax rates and that much more tax revenue is
lost in respect of products with higher tax rates.
 Of course, by applying both compliance options (higher penalties and lower marginal tax rates)
simultaneously, equilibrium E3 can be reached in Figure 12.5(b), resulting in underreporting decreasing
from Q0 to Q3.
us, in addition to taxes having to be designed to minimise the excess burden, administration costs and
compliance costs, good tax administration requires that tax evasion and avoidance be kept to a minimum.
When taxes are evaluated according to the criterion of administrative efficiency, a number of issues should
consequently be considered.
• e golden rule in tax design is simpli cation. Simple tax laws are easy to understand and comply with.
Incentives for tax delinquency should also be minimised. For example, high marginal rates should be
avoided and the poor should not be taxed. Penalties for tax evasion should be high and actively enforced.
High penalties and a high probability of being detected increase the marginal cost of cheating.
• An important consideration is the community’s level of literacy. For example, income taxes require high
levels of skill, whereas a head tax is fairly easy to understand.
• A further consideration is the efficiency and expertise of the tax administration. e importance of this
was recognised in the rst Interim Report of the Katz Commission (1994). Considerable attention was
devoted to tax administration in this report and it was made clear that the South African system required
major structural changes. At that time, the Commission estimated that at least R5 billion in additional
revenue could be netted through administrative reform. e gure turned out to be much higher in
reality.
• To improve tax collection, taxes should be withheld at source. Most taxpayers are subject to the pay-as-
you-earn (PAYE) system, according to which tax is withheld at the source (for example, by the employer)
and paid over directly to the tax authorities. (Note that PAYE is a tax collection system and not a
particular kind of tax.) Under the PAYE system, government receives its revenue before the employee can
lay his or her hands on it. Moreover, there is a regular ow of revenue, which, in turn, reduces the lag in
the operation of the tax multiplier and makes stabilisation policy more effective. On the other hand, the
PAYE system signi cantly increases the compliance costs borne by companies and other institutions in
the private sector on behalf of taxpayers. Clearly, a trade-off between administrative and compliance
costs is required.
• A further consideration is the tax morality of the community. Taxpayers’ willingness to part with their
hard-earned money is linked to perceptions about the vertical and horizontal equity of taxes as well as
the way in which the tax revenue is spent. If taxpayers feel that the tax system is inequitable and that
government is spending their tax money wastefully (or not in their interest), the willingness to pay tax
will be undermined. Government’s proposal to publish the names and particulars of persons who have
been convicted of tax law offences in the Government Gazette ranks as a measure to improve tax
compliance and tax morality (see Box 12.1 for some further insights from behavioural economics).
• Administrative efficiency is also affected by the political will to enforce tax laws.
• For taxes to be administratively efficient, they should be certain and transparent. e tax to be paid
should be certain (predictable) to ensure rational decision making on the part of both taxpayers and the
government. Both the tax collector and the taxpayer should thus be given as little discretion as possible.
ere should also be no uncertainty about who bears the tax burden. Personal income tax satis es this
requirement, but there is still too much uncertainty in respect of company tax as to who actually bears
the burden. Indirect taxes are also characterised by obscurity as to who carries the tax burden, as we saw
when tax shifting and tax incidence of tobacco and alcohol products were explained (see Section 11.5.2
of Chapter 11).
• Transparency means that the government should not take advantage of people’s ignorance.
Transparency also means that the government has a responsibility to subject tax decisions to the political
decision-making process; in other words, tax decisions should be embodied in legislation and be actively
debated. Taxation through in ation is an example of a tax on income that is not legislated for explicitly
(see Chapter 13, Section 13.4.4).

Box 12.1 Insights from behavioural economics in explaining tax evasion

 The standard deterrence model has been criticised for not predicting tax evasion well. In the model the focus is
on financial incentives (such as penalties) to explain tax evasion. New insights have been drawn from
behavioural economics, which is described as a mix of economics and psychology that allows opportunities to
augment existing theories on behaviour, and to develop better models thereof on the empirical side (Thaler,
2016: 1577). These theories provide reasons why taxpayers may be more or less compliant than predicted by
the standard deterrence model. Below are some examples:
• In contrast to the assumption of the neoclassical approach, individuals may face limits in their ability to
compute, that is, they may exhibit bounded rationality (Alm, 2012: 62). Given this, the complexity of the tax
system may affect compliance behaviour if individuals make mistakes in their tax computations (Leicester et
al., 2012: 83).
• It is also possible that the compliance decision is influenced by individuals’ perceptions of the probabilities
they face under tax compliance. For example, taxpayers may overestimate the probability of being audited by
the tax authorities, which increases tax compliance, since their subjective weight of the probability (of an
audit) exceeds that which is implied by its objective level (Alm, 2012: 63).
• Individuals may react more to losses (from penalties in the tax system) than to gains from successfully
evading taxes, and therefore evade less than what the standard model would predict (Leicester et al., 2012:
84).
• Taxpayers’ decisions to comply may depend on the actions of others, reflecting social norms. For example, if
some taxpayers are dishonest and evade taxes, such behaviour can reduce the perceived cost of non-
compliance for others. In the same vein, if others are honest and compliant, the moral cost of evasion may
be high and it can result in increased compliance (Leicester et al., 2012: 85).

12.4 Flexibility
Economic activity is characterised by recurrent recessions and booms, that is, cyclical changes. Moreover,
structural changes occur in the economy as well. Taxes should be exible enough to provide for changing
economic conditions. Taxes can in uence economic activity from both the supply side and the demand side.
On the supply side, economic growth can be in uenced by changing the incentive to work and spend (or
save). For example, if the price elasticity of the supply of female labour is greater (in other words, labour
supply is more sensitive) than that of males, female labour should be taxed at a lower marginal rate
(according to the inverse elasticity rule). is should, theoretically, induce more females to enter the labour
market or to increase their work effort. Hence the supply of labour can be in uenced. is topic is discussed
in more detail in Chapter 13.
In macroeconomics, the use of demand-side measures to smooth out business cycles is a standard topic
of discussion. is is known as stabilisation policy. A distinction is made between automatic and
discretionary stabilisers. e timing of discretionary scal action is decisive. e problem of timing,
however, is not too serious in the case of automatic stabilisers. Automatic stabilisers are characterised by
built-in exibility. An example of an automatic stabiliser is a progressive income tax system. When the
economy is entering a recession, for example, the average income tax rate will automatically begin to
decrease. As individuals’ incomes decline, they are automatically assessed at lower rates. In this way, they
are left with relatively more disposable income than they would have had otherwise. By the same token, tax
revenue for the government declines more rapidly than the national income. In Chapters 13 and 16, we will
note that in ation has largely rendered tax policy’s role in automatic stabilisation ineffective.

Key concepts
• administrative efficiency (page 258)
• compliance costs (page 258)
• consumer surplus approach (page 252)
• efficiency–loss ratio (page 257)
• excess burden (page 248)
• indifference curve approach (page 248)
• inverse elasticity rule (page 254)
• lump-sum tax (page 249)
• optimal taxation (page 251)
• positive and negative non-neutral effect (page 251)
• selective tax (page 250)
• tax avoidance (page 258)
• tax evasion (page 258)
• tax gap (page 258)
• tax morality (page 259)
• tax neutrality (page 251)
• theory of second best (page 252)

SUMMARY
• All taxes (except lump-sum taxes) tend to distort relative prices and generate a welfare loss to producers
and consumers. is is called the deadweight loss or excess burden. is loss is in addition to the normal
burden of a tax. e excess burden can be illustrated with the aid of indifference curves and can also be
measured using the consumer surplus approach.
• e excess burden (or efficiency cost of taxation) is determined by price elasticities, the size of the tax
base and the tax rate.
• From an efficiency and tax revenue perspective, it can be argued that it is better to levy taxes on price-
inelastic commodities than on price-elastic commodities. is tax rule is sometimes referred to as the
inverse elasticity rule or Ramsey rule. It can also be concluded that it is more efficient to levy low tax rates
on a large number of commodities (in other words, a broad base) rather than high rates on a few
commodities.
• To ensure that taxes are administratively efficient, administration costs and compliance costs must be
minimised. Related to these costs are the practices of tax avoidance and tax evasion. Tax evasion is
illegal, but can be reduced by imposing penalties in order to increase the marginal cost of
underreporting. Alternatively, marginal tax rates can be lowered, which would decrease the bene ts
from underreporting.
• Good taxes must be exible enough to adapt to changing economic conditions. Progressive personal
income taxes are characterised by built-in exibility.

MULTIPLE-CHOICE QUESTIONS
12.1 e size of the excess burden …
a. can be measured using indifference curves.
b. is indicated as the difference between the consumer surplus before a tax and after a tax.
c. increases with the square of the tax rate.
d. decreases as price elasticities of demand increase.
e. doubles when the tax rate doubles.
12.2 Consider Figure 12.1. Which one of the following statements is correct?
a. When a selective tax is levied on good x, the relative price ratio changes from
b. e after-tax budget line pivots when a selective tax is levied.
c. e after-tax budget line shifts parallel when a selective tax is levied.
d. e selective tax on good x generates tax revenue equal to Q2E2.
e. e excess burden of a selective tax is indicated by the difference between U0 and U2.
12.3 e more price elastic the demand of the taxed good is, …
a. the greater the excess burden will be.
 b. the greater the tax revenue will be.
c. the more likely it is that the good is a necessity.
d. the higher the tax rate should be, according to Ramsay’s rule.
12.4 Which of the following statements is or are correct?
a. When businesses underreport their pro ts, they are avoiding tax, which is perfectly legitimate.
b. e marginal costs of underreporting can be decreased by naming and shaming tax dodgers.
c. Lower marginal tax rates reduce the bene ts from underreporting.
d. Perceptions about high levels of corruption and the inequity of the tax system may lead to tax
revolts.
e. A head tax has built-in exibility, which makes it a good tax.

SHORT-ANSWER QUESTIONS
12.1 What are the main determinants of the excess burden of a tax?
12.2 Brie y discuss administrative efficiency as a property of a ‘good’ tax.
12.3 Use examples to explain what tax exibility means.

ESSAY QUESTIONS
12.1 Explain why a lump-sum tax is used to compare the excess burden of different taxes.
12.2 ‘A tax on cigarettes is inefficient since it is non-neutral.’ Do you agree? Explain your answer by using
indifference curves.
12.3 Assuming that government needs to raise a certain amount of tax revenue from two goods with different
price elasticities of demand, what would your advice be to government if the excess burden had to be
minimised? How would your answer differ if other tax criteria had to be considered as well?
12.4 ‘Marginal tax rates should be reduced for high-income taxpayers to increase tax compliance.’ Discuss
this statement by making use of a diagram.

1 In addition, the modern approach to neutrality applies a balanced-budget framework. According to this approach, the concept of
scal neutrality relates to both the revenue side and the expenditure side of the budget, in the sense that tax changes are not to
disturb the revenue-expenditure balance.
2 A compensated demand curve shows the change in quantity demanded when price changes and the individual is compensated to
remain on the same indifference curve, thus re ecting the substitution effect of a price change (see Rosen & Gayer, 2014: 331).
3 Assuming that the marginal utility of income is constant, the demand curve can be interpreted as measuring marginal utility. e
area under the demand curve (in other words, the sum of the marginal utilities) represents the total utility (welfare) of consumers.
e difference between the total utility and the actual cost of the bene ts (the market price) is the consumer surplus (Mohr, Fourie
& Associates, 2008: 129).
4 In the case of an increasing-cost industry (in other words, where the supply curve is positively sloped), the excess burden is the
area between the original demand curve and the supply curve, and is restricted by the after-tax equilibrium quantity. e excess
burden triangle therefore contains some consumer and producer surplus.
5 e Davis Tax Committee was constituted in 2013 by the Finance Minister at the time, Mr Pravin Gordhan, to review the tax system
and its contribution to inclusive growth, employment and scal sustainability.
6 Note the IMF’s tax gap analysis encapsulates a broader de nition of the tax gap, which considers the impact of non-compliance
and the effects of the tax policy framework on revenues (see IMF, 2013: 11).
Personal income taxation

Tjaart Steenekamp and Ada Jansen

In Chapters 11 and 12, the properties of a ‘good’ tax were identi ed and explained. We are now ready to
apply these criteria to the analysis of speci c taxes. In this chapter, we focus on income taxation with the aim
of describing and analysing personal income tax in terms of the tax criteria developed in Chapters 11 and 12.
We begin by de ning the income tax base in the rst section. We include the international taxation of
income. In the three sections that follow, we consider personal income tax. We begin this discussion by
de ning concepts such as exclusions, exemptions, deductions and rebates (Section 13.2). In Section 13.3, we
address the important feature of progressiveness in personal income taxation. is is followed (in Section
13.4) by an analysis of the economic effects of personal income tax using the tax criteria we identi ed in
Chapters 11 and 12.

Once you have studied this chapter, you should be able to

define the comprehensive income tax base
discuss the international principles of taxing income
discuss the reasons why the personal income tax base differs from the comprehensive income tax base
explain what a progressive personal income tax is and how progressiveness is achieved
discuss the economic effects of personal income tax.

13.1 The comprehensive income tax base

When we introduced the tax base in Chapter 11, we distinguished between three bases: income, wealth and
consumption. In this chapter, we will focus on personal income. We rst need to know what income is.
We are all familiar with a budget (for example, our own personal budget), which has an expenditure (or
uses) side and a revenue (or sources) side. Income can be de ned from both the sources side and the uses
side of the budget. From the uses side, income is the monetary value of consumption plus any change in the
net worth over a year. Net worth (or the net value of assets) is obtained by subtracting liabilities from assets
(see Chapter 15). Put differently, income is the net increase in the power to consume in a particular period
(for example, a year). It can be expressed as follows: Y = C + S where Y is income, C is consumption and S is
saving (or the change in net worth).
From the sources side of the budget, anything that makes consumption possible (in other words,
anything that is available to nance consumption) is considered as income. Income thus includes salaries,
wages, interest, capital gains, rent, pro ts, royalties, dividends, gifts, employer contributions to pension
funds, unemployment bene ts and income in kind. is de nition of the comprehensive income tax base
is referred to as the Haig-Simons de nition, named after two early twentieth-century economists who
advocated its use. Haig and Simons believed that this de nition of income most accurately re ects the
ability to pay (one of the criteria of fairness) or purchasing power.
For administrative and other reasons, governments tax some of the sources of income separately. In
South Africa and most other countries, income received by individuals is subject to personal income tax. e
income of incorporated businesses (in other words, pro ts) is subject to company tax. In some countries, net
capital gains from increases in the value of assets are subject to capital gains tax. Personal income tax is
discussed in this chapter; company tax and capital gains tax are discussed in Chapter 14. Gifts, which can
also be treated as additions to wealth, are discussed in Chapter 15.
Another dimension of the comprehensive de nition of income is that income is recorded as it accrues
and not only when it is realised. For example, if an asset increases in value during the course of the year, the
capital gain is an addition to net worth. e asset need not be sold (that is, realised) for the increased value
to be regarded as income. e reason is that an increase in the value of an asset represents an increase in the
owner’s purchasing power (that is, ability to pay). In practice, the accrual principle causes considerable
administrative complications (for example, valuation problems) and it may also result in cash ow problems
for those who have to pay tax on accrued amounts not actually received in cash.

International taxation of income: The residence principle versus the

13.1.1
source principle
Income is generated within countries, but also across national borders. In recent years, the economies of
countries have become increasingly internationalised. is has impacted on a tax jurisdiction’s ability to tax
individuals and companies. In an integrated and open economy, the returns of factors of production (for
example, salaries, dividends, pro ts, royalties and interest) ow much more freely within a country and
across national borders. When rates of return differ between countries as a result of taxation, these tax-
induced differentials can be exploited by capital and highly skilled individuals, causing distortions within
countries and across countries. Over the years, tax authorities have dealt with the international taxation of
income using two general principles: the residence of taxpayer principle and the source of income principle.
e residence principle (or worldwide basis principle) is based on the view that the country of
residence of the person or business that receives the income determines the tax liability and collects the tax.
us, a person residing in South Africa would be liable for taxes on his or her total (worldwide) income in
South Africa if a residence system was applied. For example, if the person earns R300 000 from a source in
South Africa and R120 000 from a source in Zimbabwe, the combined income of R420 000 is taxable in South
Africa. For a legal person (for example, a company), residence is determined by where the business is
registered or has a permanent presence. Only income that can be allocated to the activities of the business
(at home and abroad) would be taxable.
According to the source of income principle, income is taxed by the country where the income is
generated. Using the example above, only the R300 000 that originated in South Africa would be taxable in
South Africa if the source principle was applied. In practice, most countries apply a combination of both
systems. is hybrid form of taxing cross-border ows of income could result in double taxation of income of
this nature. If South Africa applies the residence principle and Zimbabwe uses the source principle, then, in
our example above, a person residing in South Africa would be taxed on his or her worldwide income of
R420 000 in South Africa. In addition, the Zimbabwean tax authorities would tax the person on the R120 000
generated in Zimbabwe. To eliminate or reduce the extent of double taxation, countries using the worldwide
basis unilaterally grant tax relief in the form of an income deduction for the income earned in the source
country or a tax credit for the tax paid in the source country. Alternatively, countries enter into bilateral tax
treaties or attempt to harmonise the tax treatment of cross-border income.
On a multilateral basis, however, it is difficult to harmonise tax systems as countries perceive the net
bene ts of each system differently. e debate on the merits of each system is extensive and not clear-cut at
all (see Faria, 1995: 216–221; Tanzi, 1995: 65–89 and Katz Commission, 1997b). e issues that developing
countries have to consider include the following:
• e source basis resembles the bene t principle of taxation. e entity generating the income bene ts
from public expenditures, for example, uses public roads and schools, and should therefore be taxable.
is is not a very convincing argument since a resident who earns foreign-sourced income also bene ts
to some extent from public roads and schools in the home country. e residence basis, on the other
hand, approximates the ability-to-pay principle and enables countries to tax the worldwide income of
residents on a progressive scale. Countries with low levels of foreign income (for example, dividends,
interest and royalties) would have to consider using the source basis on grounds of administrative
expediency. On the other hand, where income from investments abroad is considerable, the residence
basis has to be considered on revenue grounds.
• From a tax neutrality point of view, a tax system (for example, tax rates) should not in uence locational
decisions of businesses. From the perspective of a capital-importing country, a source-based system
would have the advantage of being neutral with regard to capital imports since it does not discriminate
between domestic investment and foreign investment, regardless of where the capital originates.
Developing countries tend to be capital importers. On the other hand, from a capital-exporting
perspective, a residence-based system would be neutral with regard to capital exports. e only concern
to an investor would be the tax rate in his or her country of residence.

In South Africa, the taxation of income was based on the source principle of international taxation in the
past. As a result of the increasing globalisation of the economy and the relaxation of exchange controls, a
residence-based income tax system was introduced as of 1 January 2001. It was argued that by doing this, the
South African income tax base would be broadened, opportunities for tax arbitrage would be limited and the
tax system would be brought in line with accepted norms for taxing international transactions (Department
of Finance, 2000: 84).
is move was contrary to the recommendations of the Katz Commission (1997b). In its Fifth Interim
Report, the Katz Commission (1997b) distinguished between active income (income derived from
operational activities, such as manufacturing and rendering services) and passive income (income derived
from investment, such as interest and royalties). e Commission recommended that active income be
taxed on the source basis and passive income on the worldwide basis. It argued that taxing active income on
a worldwide basis and at the relatively high domestic effective tax rates would encourage South African
multinational companies to relocate to low tax jurisdictions. Changing the tax system to a worldwide basis
would also be administratively complex. e Commission argued that taxing passive income on a worldwide
basis would be necessary to protect the tax base. Passive capital is very mobile when exchange controls are
limited.
ere remain differences of opinion on the relative merits of the change to a worldwide (or residence)
basis in South Africa. e application of the residence principle may enhance equity by taxing the off-shore
income of South African residents. After all, these amounts increase taxpayers’ ability to pay and allow for
progressive rates of taxation in the vertical equity sense of the word. On revenue grounds, it is also sound to
protect the system from undue losses as exchange controls are relaxed. Foreign-employed income (for a
South African resident working more than 183 days a year, for a private-sector company) was exempt from
tax until 2017. e current legislation now only allows exemption if the income earned is taxed in the foreign
destination (National Treasury, 2017a: 138). However, converting to the worldwide basis involves many
administrative problems, including problems of de nition (for example, when is an establishment resident)
and requires the (re)negotiation of various double taxation agreements. e possible impact on net foreign
investment is unpredictable at this early stage.

13.2 The personal income tax base

Total (comprehensive) income is the starting point in calculating personal income tax. Once exclusions such
as unrealised capital gains have been subtracted, gross income is obtained. In South Africa, gross income
consists of all receipts and accruals (for example, wages and salaries, rents, royalties, dividends, capital gains
and interest) of South African residents, irrespective of where in the world it was earned. Exempt income (for
example, interest) is deducted from gross income and the resulting amount constitutes net income. Taxable
income is obtained by deducting all of the amounts allowed as deductions (for example, medical expenses)
from net income. Normal tax is calculated at the applicable rate on taxable income.
e comprehensive de nition of income is much broader than the de nition of taxable income in South
African tax law. e reason is that government provides tax expenditures, sometimes referred to as tax
loopholes. Tax expenditures include exclusions, exemptions, deductions and tax rebates (or credits) that all
affect the size of the tax base. It has been estimated that tax expenditures to the amount of R106,8 billion
were provided to personal income taxpayers in 2016/17 (National Treasury, 2019a: 121). is revenue
forgone represents about 25,2% of the income tax revenue collected from persons in 2016/17. e
calculation of personal income tax liability in South Africa is shown in Box 13.1.

Box 13.1 Calculating personal income tax liability

TOTAL (COMPREHENSIVE) INCOME

minus exclusions (for example, imputed rent and unrealised capital gains)
GROSS (CASH) INCOME
minus exemptions (for example, tax-free portion of interest)
NET INCOME
minus deductions (for example, contributions to retirement funds)
TAXABLE INCOME
Tax according to tables
GROSS TAX LIABILITY
minus rebates (for example, primary rebate and rebate for age 65 and over)
NET TAX LIABILITY

13.2.1 Exclusions
Income tax is generally levied on cash income. Some forms of non-cash income, such as in-kind receipts, are
excluded from the tax base. A tax exclusion can be illustrated as follows: the salary of a person who is
employed as a cook or a child minder is taxable, but if a housewife or mother performs the same functions,
the value (or opportunity cost) of her services is not taxed. If a homeowner lets his or her house, the rent
received is taxable. If, on the other hand, the owner lives in the house, the rent forgone is an opportunity cost
(in other words, imputed rent) that ought to be taxable in terms of the comprehensive de nition of income,
but it is excluded from tax.
In the past, companies could also offer generous fringe bene ts (for example, motor vehicle allowances,
subsidised meals, low interest loans and housing subsidies) that were not taxed to employees. Nowadays,
most fringe bene ts are taxable, but the real value of these bene ts is not always taxed in full. For example,
in the case of (higher education) bursaries for relatives of employees, R60 000 per year is tax-free for
employees earning up to R600 000 per year. e taxation of fringe bene ts is sometimes also difficult to
administer. For example, it is often difficult to distinguish between personal use and business use (for
example, the use of a company cell phone or company stationery for private use). Finally, non-cash
transactions such as barter arrangements are also difficult to detect and frequently go untaxed (for example,
a dentist lling a plumber’s tooth in exchange for the plumber clearing a drain at the dentist’s residence).

13.2.2 Exemptions
A second category of tax expenditures is exempt income (or exemptions). Tax exemptions refer to income
that is exempt from payment of income tax. As a method of providing tax relief to the poor and the aged, an
amount of income is tax exempt. In 2019/20, the exempt amount for persons below the age of 65 was R79
000 and for those over the age of 65, the implied exempt amount was R122 300. A further implied threshold
applies to those aged 75 years and older (R136 750). In practice, the exempt amount is determined by tax
rebates allowed on the amount of tax that is due (see the discussion on rebates in Section 13.2.4). e rst
R23 800 of interest income received by a natural person under the age of 65 from a South African source is
also exempt from income tax in South Africa. In March 2015 the government introduced tax-free savings
accounts, which allows an investor to save up to a maximum of R33 000 per annum1 (up to a lifetime limit of
R500 000). Interest on these accounts is tax exempt.

13.2.3 Deductions
In addition to the exempt categories of income, certain expenditures may be deducted from net income for
tax purposes. ere are various tax deductions (or allowances) that serve different purposes. Some
deductions, such as those in respect of pension fund contributions and contributions to retirement annuity
funds, are intended to serve as incentives for taxpayers to provide for their old age, at which time their
pension may be taxed according to the applicable tax rate. is is particularly important in a country like
South Africa, which does not have a comprehensive social security network. Until a few years ago
deductions were also allowed (within limits) for contributions to medical aid funds and medical
expenditures. It is important to note that a tax deduction of a given amount is worth more to a person with a
high marginal tax rate than a low marginal tax rate. It is for this reason that the monthly deduction for
medical contributions and expenses was converted into tax credits from 1 March 2012 (see rebates below).
Another category of deductions is in respect of expenditures incurred with the purpose of producing income
(for example, travelling and motoring expenses or allowances, or rental income on a second residence).

13.2.4 Rebates
A further tax expenditure, which affects the government revenue from income tax, is tax rebates (or tax
credits). Tax rebates involve the subtraction of a speci ed amount from the amount of tax to be paid (that is,
the taxable liability). In contrast to a tax deduction of a speci ed amount, a tax rebate is a lump-sum bene t
and is independent of the taxpayer’s marginal tax rate. Rebates are primarily aimed at providing tax relief to
the poor and middle-income groups, and to provide for differences in taxpayers’ personal circumstances.
Because it is a lump-sum bene t, its value falls as taxable income increases, thus enhancing tax progressivity
and vertical equity. In 2019/20, our personal income tax system provided for a primary rebate of R14 220,
while persons aged 65 and over (the secondary rebate) were allowed an additional R7 794, and persons aged
75 and over (the third rebate) had a further R2 601 deducted from their tax liability. What does this mean?
e primary rebate determines the de facto tax-exempt income (see exemptions in Section 13.2.2). When the
primary rebate of R14 220 is deducted from the gross tax liability of a person earning a taxable income of R79
000, the net liability is zero. e income level at which the effective tax rate is zero is referred to as the
minimum tax threshold. Any taxable income above this amount would incur a tax liability. It should be
obvious that rebates bene t the poor more than the rich. For example, a rebate of R14 220 is percentage wise
worth more to a person with a taxable income of R100 000 than to a person with a taxable income of R500
000. Without the rebate, a person earning R100 000 would have had to pay tax of R18 000. With the rebate,
the tax liability falls to R3 780. e bene t is 79% of the unadjusted tax liability. For the person earning a
taxable income of R500 000, the bene t (calculated in the same way) is only 11,1%. In addition to the above
rebates, medical tax credits are allowed as a rebate against taxes payable. Standard medical scheme
contributions can be rebated with a tax credit of R310 per month for each of the rst two bene ciaries, and
R209 for the remaining bene ciaries. Qualifying (other) medical expenses can, within limits, also be
converted to tax credits. More generous conversions are applicable to taxpayers over 65 years of age as well
as persons with disabilities. ese taxpayers often have high medical expenses over which they have no
discretion that can signi cantly affect their ability to pay. Equity is thus ensured since the relief does not
increase with higher income levels.
13.3The personal income tax rate structure: Progressiveness in
personal income taxation
In the case of personal income tax, most observers agree that the ability-to-pay principle should apply. is
principle generally takes taxpayers’ income as a yardstick. However, it is not so simple to determine taxable
income and to combine it with the tax rate in such a way that fairness is achieved. For example, one of the
most prominent features of personal income tax is that the structure can be adapted to the personal
circumstances of the taxpayer, but adaptations of this nature complicate the rate structure.
Ability to pay (and thus the rate structure) is affected by the ling status (or unit of taxation) of the
taxpayer, that is, whether he or she is single, married or has children. If two people live together, their living
costs (per person) should be lower than when they live apart (for example, because they share certain
things). eir combined ability to pay should therefore be greater than it would have been if they had lived
apart and they could, therefore, be taxed at a higher rate (on their combined income). But the issue is not
quite so simple. For example, a married couple with both partners working can be at a disadvantage
compared to a couple where there is only one breadwinner. In the case where both spouses work, families
may have to purchase a number of services that would normally be rendered at no additional cost by a non-
working spouse (for example, cleaning services and child care). If imputed income is not taxed, the two-
breadwinner couple should be taxed at a lower rate.
ere is a further complication when children are included in the relationship. Having children involves
costs and affects the disposable income of the parents. Some would argue that these costs are non-
discretionary (for example, where contraception is ruled out owing to religious beliefs) and that tax
deductions or rebates for children should be allowed. Others argue that children are a matter of choice, and
that there is no reason why special provision should be made for expenses involving children and not for
other expenses that are incurred by choice, such as private overseas trips. e particular view taken on these
issues also depends on the values of society in respect of the family. Should mothers be encouraged to
return to work or should the income tax system discriminate against women to promote a stable family life?
Whether or not the personal income tax structure should be used for social engineering purposes is a much-
debated issue. It should be obvious that all kinds of cultural, religious and moral factors can enter the
picture. is brief discussion shows that an individual’s tax rate and tax liability do not necessarily depend
solely on his or her income. In practical terms it also should be clear that there will be limits to any country’s
ability to have its tax system cater for all equity-related matters.
Until the mid-1990s, the tax unit in South Africa was the married couple. Married and single people were
taxed using different schedules. e choice of unit was scrutinised carefully by the Margo Commission
(1987). It was argued that joint taxation of spouses amounted to a marriage penalty for income-earning
wives. is type of taxation discouraged people from marrying, discouraged married women from entering
the labour market and affected the status of married women as individuals. e Margo Commission (1987:
151) successfully recommended that the individual replace the couple as the unit. In the current personal
income tax system in South Africa, the individual is now the unit of taxation. e Katz Commission also
considered the issues of discrimination on the basis of marital status, gender, age, ethnic or social origin,
religion and so on. eir perspective was mainly a constitutional one. e Katz Commission (1994: 73)
recommended that all provisions in the Income Tax Act that were based on gender and marital
discrimination violated the Constitution and should therefore be eliminated. e Katz Commission (1994:
73) also recommended that child rebates no longer be granted, but that discrimination on the basis of age
remains for the time being. ese recommendations were introduced in 1995.
Whether the unit of taxation is the family, a married couple or the individual, the principle that people
should pay tax according to their ability to pay still holds. In Chapter 11, it was said that a tax system based
on ability to pay requires consensus on how ability is to be measured as well as on the rate structure. In the
case of vertical equity, the question is what the respective tax liabilities of rich and poor taxpayers should be.
For this purpose, each taxpayer’s sacri ce of utility as a result of taxation has to be evaluated. Most countries
apply the equal marginal sacri ce principle. According to this principle, the income tax rate should increase
progressively as taxable income increases (in other words, the richer the person, the higher his or her tax
rate should be). e aim is to achieve a more equitable distribution of after-tax income.
Table 13.1 Personal income tax rates in South Africa, 2019/20

Source: National Treasury (2019a: 41).

What is meant by progressivity and how can it be achieved? In Chapter 11, it was explained that the
progressivity of the rate structure can be determined by looking at the average tax rate. When the average tax
rate (that is, the total amount of taxes payable divided by the value of the taxable base) increases as taxable
income increases, the tax is progressive. Progressivity can be obtained by using a combination of income
exemptions (or tax rebates) and by taxing blocks or brackets of income at different rates. A higher tax rate is
speci ed for each higher income bracket, but the higher rate is applicable only to that part of the taxable
amount that falls into the relevant bracket. ese graduated rates are called marginal tax rates since they
represent the change in taxes (in other words, the marginal or extra tax amount) paid with respect to a
change in income (that is, the marginal or extra income).
Table 13.1 shows the tax rates for individuals for 2019/20. e rst column shows the taxable income
brackets. e second column shows the basic amount of tax paid for each bracket. e basic amount in the
schedule is simply the sum of the marginal tax amounts of the preceding brackets and is speci ed to simplify
tax calculations. us a person with taxable income of R220 000 pays R35 253 on the top ‘marginal’ income
in the rst bracket (R195 850 3 18%) plus an extra R6 279 on the ‘marginal’ income in the second bracket
([R220 000 – R195 851] = R24 149 × 26%), which add up to R41 532. Note that this amount is not the effective
tax liability. e primary rebate of R14 220 has to be subtracted from this amount, which leaves a tax liability
of R27 312. is converts to an average tax rate (or effective tax rate) of 12,4% on taxable
income. If we repeat this exercise for an income of R1 500 001, an average tax rate of approximately 34,5% of
taxable income is obtained. Since the average tax rate increases as income increases, the tax structure is
progressive. e importance of the distinction between marginal and average rates will again be emphasised
when the economic impact of personal income taxes is discussed in the next section. A much more
simpli ed personal income tax rate structure applies in some other countries. Take Lesotho, for example. On
the rst M61 080 (one rand equals one Lesotho maloti), a rate of 20% is applied. (A tax credit of M7 260 may
be deducted.) Income in excess of M61 080 is taxed at 30%. is structure almost resembles a at rate tax
(see below). In Botswana, individuals with an income of more than P36 000 are liable for personal income
tax. Five brackets are used. At the threshold income level of P36 000 (one rand equalled 0,74 Botswana pula
in December 2018), a rate of 5% applies. is rate increases to a maximum of 25% for incomes over P144 000.
Eswatini (previously known as Swaziland) taxes persons other than companies using four brackets, with the
highest rate of 33% being levied on taxable income exceeding SZL200 000 (one rand equalled one Swazi
lilangeni in December 2018). A rebate of SZL8 200 is allowed, which means that no tax is payable below an
annual taxable salary of SZL41 000 per annum. e Namibian tax schedule provides for seven tax bands,
with income above N$1 500 000 being taxed at 37%. Income below N$50 000 is taxed at a nil rate (one rand
equalled one Namibian dollar in December 2018). Using GDP per capita as a yardstick, the top personal
income tax rate in Namibia takes effect at approximately 20 times the GDP per capita level. By comparison,
in Botswana, the top rate becomes effective at approximately 2 times GDP per capita, albeit at a lower rate.

13.4 Economic effects of personal income tax

In Chapters 11 and 12, we laid the foundations for analysing different taxes. You will recall that we identi ed
four properties of a ‘good’ tax: economic efficiency, equity, administrative efficiency and exibility. ese
properties or criteria will now be applied to personal income tax.

13.4.1 Economic efficiency

From Chapter 12, we know that when the economic efficiency of any tax is considered, economists try to
establish whether or not the tax has an excess burden. You will recall that the excess burden is a burden in
addition to the normal burden of a tax that reduces the taxpayer’s welfare (in other words, leaves the
taxpayer on a lower indifference curve). To determine whether an income tax has an excess burden, we will
rst consider a general tax on income and then a selective tax on labour income. Since personal income tax
is levied on interest income as well, we will also consider the economic efficiency of this practice separately.

13.4.1.1 General tax on income

e impact of a general tax (a tax that is levied on the entire tax base) was analysed in Chapter 12. We
concluded that a tax of this nature has no excess burden since relative prices are not distorted. An example
of this kind of tax is a head tax. If the entire income base (personal income and company income) is taxed
and leisure is ignored (or it is assumed that leisure can be taxed at the same rate as income), an income tax is
also a general tax. e effect of this type of tax was illustrated using Figure 12.1. e tax on income will
simply shift the after-tax budget line (CD) parallel and downwards to the origin. Relative prices remain
unchanged and there is no excess burden. e tax has a normal burden only, which is illustrated by
consumption occurring on a lower indifference curve (U1) than before. An income tax that taxes the entire
tax base (excluding leisure) is therefore an efficient tax. Unfortunately, leisure cannot be ignored and neither
can leisure be taxed that easily.

13.4.1.2 Selective tax on labour income

Once we include leisure in our analysis, it can be shown that a personal income tax does have an excess
burden. Workers have a choice between income and leisure. Since leisure cannot be taxed easily, a tax on
income only is in fact a selective tax. If income (or the goods that can be purchased with that income) and
leisure are viewed as two commodities, we can argue that an income tax distorts the relative prices of
income and leisure. People may decide to work more or less as a result of the tax (in other words, the supply
of labour is affected). e net result will depend on the relative strengths of the income and the substitution
effects of the price change. We focus on these two effects below and show that it is the substitution effect that
has an adverse impact on the incentive to work. We also show that the substitution effect is determined by
the marginal income tax rate. We start by examining how the supply of labour is determined using budget
lines and indifference curves.
Figure 13.1 The supply of labour and an income tax on personal income

Assume that a person, Peter, has eighteen hours a day (his time endowment), which can be used for two
activities: work and leisure. In Figure 13.1, the daily time endowment is measured on the horizontal axis as
the distance 0L. Note that on this axis, we measure two things. From left to right (in other words, from point 0
to point L), we measure the number of hours spent on leisure activities. From right to left (that is, from point
L to point 0), we measure the number of hours worked. For example, if LQ1 hours are spent working, it
means that 0Q1 hours are available for leisure.
Suppose Peter earns a wage of R10 per hour. If he uses his entire daily time endowment (eighteen hours)
to work, he can earn an income of R180 per day. Peter’s income (or the goods that can be purchased with
that income) is plotted on the vertical axis. If the full-time endowment is used to earn income, one
combination of income and leisure or work is obtained (in other words, point Y). If, instead, he wastes all his
time doing nothing, zero income is earned and we have another combination of income and leisure/work
that we can plot (that is, point L). We can continue in this way and trace out the different combinations of
income and leisure or work for each number of hours worked. is is shown as line YL, that is, Peter’s budget
line. e two ‘goods’ in this case are income and leisure. By adding indifference curves (which indicate
Peter’s preferences or tastes), we can determine the combination of income and leisure or work most
preferred by Peter. Suppose that the highest indifference curve that Peter can attain is U0. He is then in
equilibrium at E0, where 0Q0 hours are spent on leisure and LQ0 hours are used or supplied for work.
Suppose a proportional tax or at rate tax (for example, 14%) is now levied on income only. A
proportional tax lowers the after-tax wage. is means that Peter’s income (his after-tax wage multiplied by
the number of hours worked) is now less at each number of hours worked. e after-tax budget line pivots
(or swivels) from YL to BL. e after-tax equilibrium is at E1, where indifference curve U1 is tangent to the
new budget line, BL. Peter’s welfare has declined as indicated by the fact that he nds himself on a lower
indifference curve. At the new equilibrium, Peter has a tax liability of E1G (in other words, the difference
between his before-tax income and his after-tax income earned by working LQ1 hours). e quantity of
labour supplied has increased from LQ0 to LQ1 hours per day. is, however, need not always be the case.
e movement from E0 to E1 is the result of the combined effect of the income effect and the substitution
effect of the tax change. Whether the number of labour hours supplied increases or decreases will depend
on which effect is stronger.2
What is the income effect? e tax reduces Peter’s after-tax income. Peter is worse off as a result of the
tax and he will tend to work more to offset partly this loss in income. Put differently, Peter has a lower after-
tax income and can therefore not afford the same amount of leisure than before. Less leisure means more
hours of work. e income effect will therefore cause the quantity of labour supplied to increase. What about
the substitution effect? To understand the substitution effect, it is important to note that leisure has a price.
e price of one hour of leisure is the hourly wage sacri ced by not working. us, the opportunity cost of
leisure is the wage foregone. Since the introduction of an income tax reduces the after-tax wage, the
opportunity cost of leisure decreases; that is, leisure becomes cheaper. In other words, consuming leisure
involves a smaller sacri ce in income than before the introduction of the tax. Because leisure is now
relatively cheaper, it is substituted for work. e substitution effect increases leisure, which means that it
reduces the quantity of labour supplied.
e income and substitution effects can also be explained using Figure 13.1. After the introduction of the
proportional tax, equilibrium changes from E0 to the new equilibrium at E1. e movement from E0 to E1 can
be decomposed into the income effect and the substitution effect. If the individual were (hypothetically)
compensated with an amount just enough to make him as well off as before the tax, the budget line BL
would shift parallel outwards to HJ, which is tangent to indifference curve U0 at E2. e movement from E2 to
E1 is, therefore, the income effect of the tax. e income effect shows that the individual increases his work
effort in response to the imposition of the proportional tax. e movement from E0 to E2 depicts the
substitution effect, which is a consequence of the change in the relative price of labour alone. e movement
shows how the individual substitutes more leisure for less work if a proportional tax on labour is imposed or
increased.
As stated earlier, the income and the substitution effects combine to determine whether a person will
work more or less. If the income effect dominates, the after-tax quantity of labour supplied will increase. In
contrast, if the substitution effect dominates, the number of hours worked will decrease and the quantity of
leisure will increase. On the basis of this analysis alone, it is impossible to say what the outcome will be.
Empirical evidence, however, suggests that the labour-supply elasticities for prime age men (men aged
between about twenty and sixty years old) are generally close to zero (see Brown & Jackson, 1990: 449). is
means that the supply curve is almost vertical or, put differently, that the quantity of labour supplied is very
insensitive to changes in the net (after-tax) wage. On the other hand, the estimated elasticities are generally
positive and high for married women. In their case, the quantity of labour supplied is thus quite sensitive to
changes in the net (after-tax) wage. For example, there is evidence in the United Kingdom that the hours
men work are not very responsive to changes in work incentives; however, participation by those with low
levels of income is responsive. In contrast, the hours of work and the participation of women with young
children are quite sensitive. e income of men with high levels of education is responsive to tax, but in the
sense that they shift their income to non-taxable forms (see Meghir & Phillips, 2010: 204 and 252).
e analysis used for a proportional tax can be repeated for a progressive tax on income. One difference
would be the shape of the after-tax budget line. In the case of a progressive tax, the budget lines will be
kinked or curved (see Rosen & Gayer, 2010: 418). is does not change the conclusion with regard to the
impact of the tax on the supply of labour fundamentally, but there is one important difference between
progressive and proportional taxes: for a proportional tax, the average tax rate is equal to the marginal tax
rate, but for a progressive income tax, the marginal tax rate is greater than the average tax rate. is means
that the adverse incentive effects owing to the substitution effect are likely to be more important for a
progressive tax than for a proportional tax (see Box 13.2).

Box 13.2 Marginal tax rates and the substitution effect

In the case of a proportional tax, every taxpayer faces the same marginal tax rate, but in the case of a
progressive tax, the marginal tax rate varies with income (for example, in Table 13.1, the marginal tax rate
 exceeds the average rate for each income tax bracket). What does this mean?
The size of the income effect is determined by the average tax rate, whereas the size of the substitution
effect is determined by the marginal tax rate.
The average tax rate indicates how much of his or her income the taxpayer sacrifices to the revenue
authorities for his or her total work effort. To recover some of the income lost as a result of tax, the taxpayer
will work more. You will recall that this is the income effect of the tax change. The income effect is therefore
related to how much tax is paid and, for a given income, this is determined by the average tax rate. The
marginal tax rate, on the other hand, indicates the additional income an individual sacrifices to the revenue
authorities for additional work effort (in other words, changes in work effort) and is therefore related to the
substitution effect. The higher the marginal tax rate, the more of the additional income a taxpayer must
sacrifice. As the marginal tax rate increases, the incentive to substitute leisure for income becomes greater. It
is therefore the size of the marginal tax rate that determines the incentive to work (the substitution effect).

Figure 13.2 The excess burden of a proportional tax on personal income

We know at this stage that a tax that selectively taxes income (and not leisure as well) has income and
substitution effects that together may cause the quantity of labour supplied to increase or decrease. We also
know that the taxpayer’s welfare is reduced by the tax (as illustrated by the movement to a lower indifference
curve). However, we still have to determine the economic efficiency of the tax, that is, whether or not it has
an excess burden. To arrive at an answer, we have to compare the results of the proportional tax to a lump-
sum tax of equal revenue. Remember that a lump-sum tax does not distort the relative prices of leisure or
work and income, and therefore has no excess burden. e impact of the personal income tax on economic
efficiency (the excess burden) is explained in Figure 13.2. is gure is similar to Figure 13.1, except that the
income and substitution effects are not illustrated.
 If a lump-sum tax (for example, head tax) is introduced to raise the same tax revenue as the proportional
tax in Figure 13.1 and Figure 13.2 (in other words, E1G at Q1 hours), the after-tax budget line shifts parallel to
the original budget line (YL) to DF, which intersects BL at point E1. Peter will now be in equilibrium at E2,
where the highest indifference curve, U2, is tangent to the (new) budget line. Peter is working LQ2 hours a
day, which is more than the LQ1 hours worked under the proportional tax. Peter’s welfare is also higher
under the lump-sum tax than under the equal-yield proportional income tax, as illustrated by the
attainment of a higher indifference curve (U2) than before (U1). e difference between U2 and U1 can be
ascribed to the excess burden of a proportional income tax that selectively taxes income. Note that
compared to lump-sum taxes, selective income taxes lower the incentive to work, even though they may lead
to increased work effort (labour supply increases from LQ0 without any tax to LQ1 with the tax). A lump-sum
tax of equal revenue yield thus results in an even greater quantity of labour being supplied (the quantity of
labour supplied increases from LQ0 to LQ2). We can therefore conclude that income taxes are economically
inefficient, the reason being that relative prices are distorted by the tax.
Our conclusion that income taxes are distortional rests on the premise that it is difficult or impossible to
tax leisure. If it were possible to tax both labour and leisure hours, we would have had a tax that generates no
excess burden. One option put forward, which became known as the Corlett-Hague rule, was to tax goods
and services complementary to leisure (Corlett & Hague, 1953). Consumers use certain goods, such as golf
clubs, DVDs, television sets and romance novels, along with leisure time. On their own, these goods have
little value. ey only become useful if used in combination with leisure time. erefore, if you cannot tax
leisure explicitly, the policy solution would be to tax goods complementary to leisure at a rate that would
reduce demand for leisure indirectly. In this way, the excess burden of the income tax system could possibly
be reduced.
From this analysis, it follows that taxing women who are not the main breadwinners at lower marginal
tax rates than men may, for example, increase the supply of labour. is would induce more women to enter
the labour market. is option is not available, however, since the South African Constitution does not
permit discrimination on the basis of gender or marriage.
Based in part on the theoretical arguments that lower marginal tax rates will increase work effort and
improve tax compliance (the incentive to cheat is lower if tax rates are lower), there has been a worldwide
trend towards lower marginal tax rates (see Section 15.5.2 of Chapter 15). South Africa has followed suit. In
the early 1970s, the maximum marginal tax rate was 72%, compared to the present rate of 45%.3

13.4.1.3 Selective tax on interest income

Personal income tax not only impacts on the supply of labour, but also affects private savings if interest
income is taxed. South African households have a poor savings record. Savings as a percentage of disposable
income of households declined from 2,6% in 1998 to 0,3% in 2017 and was in negative territory for the period
2006 to 2016. Because taxation may affect savings behaviour, governments attempt to encourage individuals
to save more. In South Africa, domestic interest income is exempted up to a certain threshold, depending on
the age of the individual (above or below 65 years). Contributions to retirement annuities, pension and
provident funds are also tax deductible up to certain levels to provide incentives for individuals to save for
their retirement. A comprehensive framework for social security and retirement reform was proposed in
2007. is framework provides consistent tax incentives for mandatory contributions and supplementary
voluntary contributions. To analyse the impact of taxation on savings, we will show that the supply of savings
is a function of lifetime income, the interest rate and the preferences (tastes) of individuals. We then
introduce a tax on interest income and show that because the tax changes the opportunity cost (or ‘price’) of
present consumption, it has income and substitution effects. e supply of individual savings may increase
or decrease.
We know that income (Y) is equal to the sum of consumption (C) and saving (S). Less saving implies
more consumption. We can understand the impact of a tax on a person’s savings choices by analysing
consumption in two periods: the present and the future. In deciding how much to save, the individual will
have to consider his or her lifetime income, that is, the income that is now received and what is expected to
be earned in the future. Because individuals prefer present consumption to future consumption, our
individual (let’s call her Grace) would require some form of compensation for postponing consumption to
the future. e compensation for giving up one rand of present consumption (that is, saving) can be seen as
the interest that can be earned on her savings. Put differently, the opportunity cost of one rand of present
consumption is (1 + r) rand in the future, where r is the interest rate. If Grace has R2 000 available now and
the interest rate is 10%, she would be able to postpone her consumption of this amount (the principal) in
return for future consumption of R2 200, that is, (R2 000 + [0,10]2 000). is can also be analysed graphically.

Figure 13.3 Effect of an income tax on savings

In Figure 13.3, present consumption is measured on the horizontal axis and future consumption on the
vertical axis. Suppose Grace has an initial endowment of present income (Y0) and future income (Y1). If she
neither saves nor borrows, this income and consumption bundle would correspond to point E0 (the
endowment point). She now decides to save S. is can be shown as a movement to the left of E0, and a
reduction in present consumption equal to the distance Y0C0. In the next period, Grace will be able to enjoy
increased consumption equal to (1 + r)S. is represents a movement to a point above and to the left of E0,
such as D, entailing an increase in future consumption equal to the distance Y1C1. But Grace can also decide
to borrow; that is, her present consumption exceeds her present income, leaving her with less future
consumption. In Figure 13.3, the net effect is represented as point F, a movement to the right and below the
initial endowment point E0. By linking points D and F through E0 and by considering all of the other possible
values of savings and borrowing, the lifetime budget constraint AB is derived. is budget line shows the
different combinations of consumption over the two time periods and has a slope of –(1 + r). But which
combination will Grace choose? Each person has a set of indifference curves. Each curve gives those
combinations of present and future consumption that leaves the person indifferent, that is, at the same level
of utility. Whatever the person’s preference, each person will attempt to maximise utility subject to his or her
lifetime budget constraint. If Grace prefers to consume less today in return for more in the future, a point
such as D can be attained, where her indifference curve U0 is tangent to the lifetime budget line AB. She
prefers present consumption equal to 0C0 and future consumption of 0C1.
Let us now consider the effect of a proportional tax, at rate t, on interest income. We will not analyse the
case of the borrower who has to pay interest and who may, in some cases, deduct interest payments from
income tax. Deductible interest payments are permissible in countries such as the United States. Some
taxpayers in South Africa may also claim part of the interest payments on their housing bonds as a personal
income tax deduction. e tax reduction on interest payments on housing bonds is the exception and we
will focus on the more important practice where interest income (in other words, the return on savings) is
taxed.
A tax at a rate of t reduces the rate of interest received to (1 – t)r. e opportunity cost of present
consumption changes to [1 + (1 – t)r] compared to the before-tax opportunity cost of (1 + r). How does this
affect the budget line? First of all, the after-tax budget line must include the initial endowment (Y0, Y1), since
the individual may still opt neither to save nor to borrow at that income. If Grace decides to save (in other
words, move above and to the left of point E0 in Figure 13.3), she will be able to increase future consumption
by less than before. e new life-time budget line becomes the atter segment GE0. Put differently, the slope
of the budget line now has an absolute value of –(1 + [1 – t] r). e cost of borrowing is not affected by the tax
on interest income. erefore, to the right of point E0, the individual can still consume combinations on
segment E0A of the budget line. e after-tax lifetime budget constraint becomes AE0G and is kinked at point
E0. Again, the question is: Which combination of present and future consumption will Grace choose after the
tax on interest?
Before the tax, Grace consumed at D where indifference curve U0 touches the lifetime budget line AB.
After the tax consumption is at E0*, where the lower indifference curve U1 is tangent to the after-tax budget
line AE0G. Present consumption is now at 0C0* and future consumption is equal to 0C1*. e important
observation is that saving has declined from Y0C0 to Y0C0*. But depending on her taste for future and present
consumption (the shape of the indifference curves), the outcome could have been different. Saving could
have increased. e net result is ambiguous. Why?
e tax has changed the relative ‘price’ of present consumption, and a price change has a substitution
effect and an income effect. Let us consider the substitution effect rst. Since the opportunity cost of present
consumption has decreased (the product has become ‘cheaper’), Grace would increase present
consumption. is implies that savings is reduced, resulting in less future consumption. e income effect
can be explained by noting that the reduction in the interest rate received caused after-tax income in the
future to become less. e only way in which Grace could maintain her consumption level in the future is to
reduce present consumption. e income effect implies more saving. One could also understand the
income effect by considering that many people aim at reaching a certain retirement goal or that they are
target savers. If the return on their savings declines, they have to compensate by saving more (reducing
present consumption). To summarise: the substitution effect causes saving to decrease; the income effect
causes saving to increase. If the substitution effect dominates, the net result would be lower saving. If the
income effect dominates, the net result would be increased saving. e end result is therefore theoretically
ambiguous and remains a matter for empirical research.
We have established that an income tax levied on interest income causes the relative price of present
consumption in terms of future consumption to change. e impact on the supply of savings remains
uncertain. How is economic efficiency affected? To determine whether the tax on interest income has an
excess burden, we could repeat the procedure we employed when we analysed a selective tax on labour
income. We simply contrast the effect of the interest income tax with a lump-sum tax of equal revenue. A
lump-sum tax does not have an effect on the consumption choices of the individual in the two periods (in
other words, the relative price ratio between present and future consumption remains unchanged). We can
conclude that an income tax on interest income is economically inefficient since it causes the individual to
substitute between present and future consumption (an excess burden is created). e magnitude of the
burden may also be compounded when different forms of savings are taxed differently, thereby changing the
rates of return on savings vehicles. is would imply a further distortion of relative prices.
Empirical results of the effect of taxation on saving are inconclusive. Some economists estimated a low
interest elasticity of supply of savings, whereas others came to different conclusions. e consensus opinion
is that the rate of return effects on saving is at best small (Jappelli & Pistaferri, 2002). e evidence on
whether taxation may have an impact on pension savings is also not clear. People do not necessarily increase
their net savings; they shift their savings in response to tax incentives (Gale & Scholz, 1994). e Katz
Commission (1994) came to similar conclusions, noting that the impact of non-tax factors is much more
important and that it is only the composition of savings that is affected by personal income tax. In this
regard, by stimulating personal saving, the loss in revenue for government could result in an increased
budget de cit (in other words, greater government dissaving). Promoting savings through mandatory
contributions to pension funds also crowds out voluntary savings. However, there may be other reasons for
introducing compulsory savings programmes, including equity reasons. Low-income earners below the
income tax threshold have no tax incentive to save, while high-income earners have the resources and
nancial information needed to exploit tax incentives. To leave saving for retirement entirely in the hands of
individuals could lead to free-riding behaviour, which could cause a burden for future generations. e
reform of the social security system was addressed in Chapter 9.

13.4.2 Equity
As mentioned, personal income taxation lends itself to the application of the ability-to-pay principle.
rough a system of exemptions, deductions, tax rebates and marginal tax rates, the rate structure can be
made to conform to society’s notion of fairness. However, before we can make nal conclusions in respect of
tax equity, we need to know who ultimately bears the burden of the tax. If the tax can be shifted quite easily,
the achievement of equity objectives could be compromised.
Even though the statutory burden of the tax is on the individual, it may be possible for individuals to shift
the burden. e critical issue is how sensitive the supply of labour is to price and tax changes. As already
noted, the net effect on work effort will be determined by the relative strengths of the income and
substitution effects. Empirical evidence points to an insensitive or inelastic supply of labour for men (in
other words, the supply curve is almost vertical). If the supply curve is relatively inelastic, the burden is on
the supplier, that is, the employee in this case (see Chapter 11, Figure 11.5). If the supply curve is less
inelastic, the employer and the employee will share the burden. is scenario is possible in the case of
married women and high-income professionals who are internationally mobile. us governments should
take due cognisance of tax shifting possibilities when personal income taxes are levied with a view to
changing the income distribution. Intentions and actual policy outcomes do not necessarily point in the
same direction. e unintended consequences of attempts at changing the distribution of income using
populist approaches are considered in Box 13.3.

Box 13.3 Taxing the rich4

The debate on taxing the rich was invigorated by a New York Times article in which billionaire investor Warren
Buffett alleged that the super-rich do not pay enough taxes (The Guardian, 2011). United States president
Barack Obama was also calling for ‘millionaires and billionaires’ to ‘pay their fair share’ and the new French
president was arguing for a 75% tax rate on household incomes above $1,3 million (Economist, 2012a,
2012b). In the United Kingdom, it was announced in 2008 that a 45% tax on incomes above £150 000 would
take effect in 2011, but this rate was increased to 50% and the date was brought forward to April 2010
(Brewer, Browne & Johnson, 2012: 181).
South Africa is characterised by large income and taxable income inequality. In 2010, the taxable income
share of the top 10% was 47% and that of the top 1% of taxpayers was approximately 18% (see Steenekamp,
2012: 3; Alvaredo & Atkinson, 2010: 14). In line with international trends, marginal tax rates applicable to the
rich were similarly reduced in South Africa in steps from a high of 75,5% in 1948 to the current level of 45%.
Using a micro-simulation model, Van Heerden and Schoeman (2010: 13) confirm that taxpayers at the higher
end of the income tax scale have benefited relatively more from tax reforms in recent years.
It is obvious that the rich are important to economic development, for that is where the concentration of
money is as well as wealth, net savings and talent. A progressive income tax system taxes success, and may
discourage entrepreneurial entry and innovation. The very rich receive relatively more compensation in the form
 of cash, bonuses, share options and capital income than other income groups, which makes them more
responsive to demand conditions and the business cycle. It also enables them to shift the form in which they
receive income in the short run. The hours of work of high-income taxpayers are sometimes assumed to be
more responsive to high marginal tax rates. High tax rates may furthermore encourage evasion and emigration.
Because of the economic importance of the rich, it is imperative to study how sensitive their behaviour is to
changes in income, and what the implications are for public policy and taxation in particular. All the mentioned
behavioural responses are important because they impact on tax revenue and economic efficiency (the
deadweight loss). In principle, the elasticity of taxable income (ETI) can capture all of these responses.
The ETI concept is central in the analysis to estimate the impact of changes in marginal tax rates. It
measures the responsiveness of top reported incomes (as a share of all reported incomes) with respect to the
net of tax rate.5 The higher the elasticity coefficient, the more responsive tax earnings are to the net of tax rate.
If government increases the tax rate by a small amount, it will have two effects on tax revenue: a mechanical
effect and a behavioural effect. The mechanical effect causes tax revenue to increase when the higher tax rate
increases, whereas the behavioural effect causes a reduction in tax revenue. The behavioural response is also
equal to the marginal excess burden (deadweight loss) created by the tax increase.
Using published tax data by income group contained in Tax Statistics 2011, it was determined that the
approximate top 1% of taxpayers (approximately 100 000 taxpayers) in 2010 earned taxable income starting at
R750 000. Because tax policy results are so sensitive to different estimates of the ETI, the results for an ETI of
0,20, 0,40 and a much higher ETI of 0,80 were used to investigate the revenue and efficiency implications of a
marginal tax increase for the top 1%.6 Using these coefficients, it was estimated that a 10% increase in the top
marginal tax rate (an increase from 40% to 44%) would result in taxable income ranging from a gain of
approximately R2 billion to losses of R340 million. Although these results are tentative, they show that taxing
the rich at higher rates may not produce the revenue windfall expected. The marginal excess burden increases
from R0,39 to R3,16, depending on the ETI and the effective marginal tax rate (society is worse off by this
amount for each extra rand of tax revenue raised). Using a micro-simulation model, Van Heerden (2013) also
concludes that from an efficiency perspective, increases in marginal tax rates will be detrimental to the
economy. Kemp (2017: 31) estimates the elasticity of taxable income in South Africa at approximately 0,3
(using bracket creep in the absence of large tax reforms), and for gross income the estimate is smaller at
around 0,2. He also finds that higher income groups are more behaviourally responsive (i.e. the elasticity
estimate for the top 10% of income earners is 0,37). He concludes that profound increases (such as the
increase of the maximum marginal rate to 45%) could dampen anticipated revenue gains. Empirical evidence on
progressivity and the impact of taxation in particular on redistribution in South Africa since 1994 is also not
very encouraging (see Nyamongo & Schoeman, 2007; Van der Berg, 2009; StatsSA, 2008).

13.4.3 Administrative efficiency and tax revenue

In the discussion of administrative efficiency in Chapter 12, a number of references were made to income
tax (Section 12.3). We will highlight some of the issues here. Income taxes are complex, and administrative
efficiency requires relatively sophisticated taxpayers and administrators. In countries where these
requirements are generally lacking, the income tax code should be as simple as possible. To simplify tax
administration and ease the compliance burden for some taxpayers, SARS introduced an electronic ling
facility for individuals.
Personal income tax is the largest source of government revenue. In 2017/18, it contributed 39,7% of total
tax revenue collections. It must, however, be recognised that the tax base is rather small. For the 2017 tax
year, there were 20,0 million individuals registered as taxpayers, but approximately 32% of them were
expected to submit tax returns and only 4,9 million were assessed (National Treasury and South African
Revenue Services, 2018: 30). is must be compared to a labour force of approximately 22,4 million
employed and unemployed workers in the rst quarter of 2017 (StatsSA, 2018b: 1). us although the labour
force is indicative of the potential income tax base, many of the total number of potential taxpayers are not
liable for income tax owing to the tax threshold and tax rebates. e tax base can be increased by lowering
tax rebates or lowering the minimum tax threshold. is would cause hardship for the poorer taxpayers and
increase tax administration. Alternatively, efforts could be made to capture people outside the tax net, such
as those in the informal sector and other groups that are difficult to tax. ese groups could be taxed using
presumptive taxes. Presumptive taxation involves the use of certain indicators (for example, ownership of
certain assets, personal servants, average pro t margins or average gross turnover) to determine tax liability.
We will return to this concept when we discuss the company tax on small businesses in Chapter 14 (Section
14.3.2).
Another method that has been proposed to increase tax revenue (or maintain the same level of tax
revenue) is to reduce tax rates. is recommendation is based on the alleged tax-rate–tax-revenue
relationship that has become known as the Laffer curve, named after the economist Arthur B. Laffer, who
popularised the idea (see Laffer, 2004). e logic of his argument is that higher tax rates will not necessarily
produce more tax revenue, since the tax base will shrink as taxpayers reduce their work effort in response to
the higher rates. Fundamentally, the debate is again about how sensitive labour supply is to tax changes.
Without discussing this issue again, it should be recognised that the higher the tax rate is, the more likely it is
that the substitution effect will dominate. For example, if income is taxed at increasingly higher marginal
rates and ultimately at 100%, work effort will not only decline, but a person will eventually not be prepared to
do any additional work whatsoever. Tax revenue is the product of the tax rate and the tax base, where the tax
base depends on work effort (the number of hours worked). At low tax rates, an increase in the tax rate will
tend to increase tax revenues (at low rates, people will still work more; in other words, the income effect
dominates). However, this will only continue up to a point, beyond which further tax rate increases will
reduce tax revenues (people will eventually reduce their work effort; in other words, the substitution effect
will dominate).
e Laffer curve is illustrated in Figure 13.4. Total tax revenue is plotted on the vertical axis and the tax
rate on the horizontal axis. If the tax rate is at A, government can still increase tax revenue by raising the rate.
Once the tax rate is at B, however, a further increase in the rate will cause tax revenue to decline (for
example, from R3 to R2). e supporters of the Laffer hypothesis use the mechanism to show that if the tax
rate is at C, the authorities should lower the rate since this will result in higher tax revenue, as illustrated by a
movement from R2 to R3. e critical question, of course, is to determine where countries are on the curve.
e proponents argue that some countries may well be beyond point B. e debate can only be settled
empirically, but it does appear that labour supply elasticities are such that it is unlikely that rate reductions
would be fully counteracted by increased work effort.
Figure 13.4 The Laffer curve

13.4.4 Flexibility
In Section 12.4 of Chapter 12, it was pointed out that one of the ‘good’ characteristics of a personal income
tax is its built-in exibility to counter cyclical economic behaviour. Personal income tax is thus considered to
be an automatic stabiliser. On the negative side, in ation has serious implications for a progressive personal
income tax, to the extent that it has rendered the automatic stabilising effect meaningless (see Chapter 18).
In ation erodes the value of tax thresholds and deductions, and leads to bracket creep; this is a process
whereby a person is pushed into a higher income tax bracket as his or her nominal income increases
(irrespective of what happens to the person’s real income). is could be bene cial to government (it raises
more and more revenue without the legislative process needed to increase tax rates), but if tax brackets and
the value of tax preferences are left uncorrected, all taxpayers could, in the extreme, end up paying the
maximum marginal tax rate. is disguised way of increasing the tax burden is, of course, not conducive to
transparency, an important requirement of a good tax system. Sometimes the phenomenon of bracket creep
is erroneously referred to as ‘ scal drag’. e latter concept really refers to the dampening effect on the
economy of the higher tax revenues (caused by bracket creep); that is, rising tax revenues automatically push
the government’s budget into a surplus, creating a restrictive scal policy. is could have very adverse
implications if substantial in ation occurs in recessionary times (when a de cit on the budget would be
more appropriate in Keynesian terms) unless tax brackets are adjusted for in ation.
 To understand the phenomenon of bracket creep, it is essential to distinguish between nominal income
and real income. Real income refers to the purchasing power of income (in other words, income after
adjustment for in ation), whereas nominal income is the actual monetary value or rand (or home-
currency) value of income.
e personal income tax schedule taxes a person on his or her nominal income. In in ationary times,
employers often compensate workers for the decline in purchasing power by increasing their wages or
salaries. If the rate of increase is equal to the in ation rate, workers’ nominal income increases, but their real
income stays the same. However, if the personal income tax schedule is not corrected for in ation,
individuals are pushed into higher tax brackets with higher marginal tax rates because their nominal
incomes have increased. Consider the numerical example below using the tax schedule in Table 13.1.
Suppose that Ms Shope received an income of R180 000 in 2018/19. Now assume that the general price
level increases by 10% in 2019/20 and that Ms Shope’s employer raises her salary by 10%. In real terms, her
income stays the same, but in nominal terms, her income increases to R198 000. We can calculate Ms
Shope’s tax liability in 2018/19 and 2019/20, assuming that the tax schedule in Table 13.1 remains
unchanged (inclusive of the tax rebate). In 2018/19, her tax liability (based on her income of R180 000) was
R18 180 (remember to take the primary rebate of R14 220 into account). In 2019/20, her nominal income of
R198 000 pushed her into the higher (R195 851 and more) tax bracket and the last R2 150 of her income is
taxed at a marginal tax rate of 26%. In 2019/20, her tax liability increases by 18,8% to R21 592, much more
than the in ation rate of 10% with which her salary increases. Ms Shope therefore pays R3 412 more tax,
even though her real income has not increased. Her average tax rate increased from 10,1%
to . is increase in her average tax rate (while her real pre-tax income remains unchanged)
is what bracket creep is all about. If left unchecked, bracket creep undermines vertical equity. Even poor
taxpayers will be drawn into the tax net and will eventually be taxed at ever higher marginal tax rates.
e tax authorities can temper or eliminate the effects of bracket creep by linking the rate structure to a
price index (in other words, by raising the income brackets each year in line with the increase in the general
price level). Alternatively, government can adjust the brackets and rebates on an ad hoc basis. is has been
the preferred method of the South African tax authorities. Following the recommendation of the Margo
Commission (1987: 81), government has also reduced the number of tax brackets to slow down the
in ationary creep to increasingly higher marginal rates. For example, in 1991/92, the income tax schedule
for married couples provided for fteen brackets compared to the seven brackets in 2019/20.

Key concepts
• bracket creep (page 287)
• comprehensive income tax base (page 267)
• Corlett-Hague rule (page 280)
• elasticity of taxable income (page 285)
• fiscal drag (page 288)
• income effect (page 277)
• Laffer curve (page 286)
• lifetime budget constraint (page 282)
• lifetime income (page 281)
• marginal tax rates (page 273)
• minimum tax threshold (page 272)
• nominal income (page 288)
• presumptive taxation (page 286)
• real income (page 288)
• residence principle or worldwide basis principle (page 268)
• source of income principle (page 268)
• substitution effect (page 277)
• tax deductions (page 271)
• tax exclusions (page 270)
• tax exemptions (page 271)
• tax expenditures (page 270)
• tax rebates (page 271)
• unit of taxation (page 272)

SUMMARY
• Income is anything that is available to nance consumption and falls within the de nition of the
comprehensive income tax base, also known as the Haig-Simons de nition of income. ree important
forms of income are personal income, company income and capital gains. For administrative and other
reasons, these forms of income are taxed separately.
• In calculating personal income tax, gross income is rst determined. It consists of all receipts and
accruals of South African residents irrespective of where in the world the income was earned. Taxable
income is obtained after tax expenditures (that is, exemptions and deductions) have been provided for.
e gross tax liability is calculated using tax tables. Once rebates (for example, primary tax rebate) and
credits (for example, medical tax credit) have been deducted, the net tax liability is obtained.
• A tax on income distorts relative prices and therefore impacts on the choice that workers have between
work and leisure. A tax on labour income has an income effect and a substitution effect. Depending on
the relative strengths of these effects, it will result in work effort increasing or decreasing. Empirical
evidence suggests that the supply of working-age men is not sensitive to tax changes, whereas that of
married women is.
• Individuals have a choice between current and future consumption (savings). A tax on interest income
will have an income effect and a substitution effect. e substitution effect causes saving to decrease; the
income effect causes saving to increase. e empirical evidence on whether taxation may have an impact
on savings is inconclusive, but the consensus opinion is that the effect is small.
• Because a tax on personal income or interest income is a selective tax, it distorts relative prices. By
comparing the effect of the selective tax to a lump-sum tax of equal revenue, we can show that a tax on
labour income or interest income has an excess burden. e personal income tax is therefore allocatively
inefficient from this perspective.
• Personal income taxation is suited to address equity concerns of society. rough a system of
exemptions, deductions, tax rebates and marginal tax rates, the rate structure can be made to conform to
the ability-to-pay principle. But because some groups of income tax payers (mobile professionals and
married women) can shift their tax burden, the equity outcome of high tax rates may not necessarily be
as intended.

MULTIPLE-CHOICE QUESTIONS
13.1 In South Africa, gross tax liability …
a. exceeds comprehensive income.
b. minus the medical tax credit equals net tax liability.
c. is calculated by applying personal income tax tables to taxable income.
d. equals taxable income.
13.2 Personal income tax …
a. has a substitution effect when leisure can be taxed at the same rate.
b. levied at progressive marginal tax rates decreases the substitution effect.
c. increases the opportunity cost of leisure.
d. generally decreases work effort according to available empirical evidence.
e. causes labour supply to increase when the income effect dominates.
13.3 Personal income tax has an excess burden …
a. because leisure and income are taxed at the same rate.
b. since there is a choice between current consumption and saving.
 c. that can be illustrated as the difference between U0 and U1 in Figure 13.2.
d. but can be reduced by taxing goods and services, which are substitutes for leisure.
e. meaning that the tax is not only inefficient, but also inequitable.
13.4 A selective tax on interest income …
a. only causes an income effect.
b. increases the slope of the after-tax budget line, for the segment to the left of the endowment point
(assuming that future consumption is measured on the vertical axis).
c. increases the opportunity cost of present consumption.
d. has no excess burden.
e. decreases the opportunity cost of present consumption.

SHORT-ANSWER QUESTIONS
13.1 De ne and explain the comprehensive income tax base.
13.2 Differentiate between the following concepts:
• Gross, net and taxable income
• Nominal and real income
• Accrued and realised income
• Average tax rate and marginal tax rate
• Tax exclusions, tax deductions and tax credits.
13.3 Why are policy-makers so concerned about the impact of in ation on personal income tax?

ESSAY QUESTIONS
13.1 Critically evaluate the introduction of a worldwide basis of taxation in South Africa.
13.2 Explain the rationale behind the progressive nature of individual income tax structures.
13.3 With the aid of a graph, explain the likely effects of an increase in individual income tax on the supply of
labour.
13.4 ‘Personal savings in developing countries are at low levels. By exempting interest income from taxation,
governments can stimulate higher levels of savings.’ Do you agree? Make use of a diagram and empirical
ndings in your discussion.
13.5 Discuss the economic effects of personal income tax by referring to economic efficiency, equity,
administrative efficiency and exibility.
13.6 Discuss the relevance of the so-called Laffer effect.

1 In 2015 the maximum annual tax-free amount was R30 000. It increased to R33 000 as of 1 March 2017.
2 e decision to work less or more is also affected by noneconomic factors (for example, the status of work in a society as well as the
need to avoid boredom and to get out of the home environment). We concentrate only on the economic effects.
3 Since the 1970s the top marginal income tax rate decreased to 40%, but it was subsequently rasied to 41% in 2015/16, and then to
45% in 2017/18 (for equity and revenue purposes).
4 is section is based on Steenekamp (2012).
5 For a comprehensive description of the ETI concept and the model framework for analysing behavioural responses, see Saez (2004:
11–15), Saez, Slemrod & Giertz (2012: 5–10) and Giertz (2009: 115–117).
6 Van Heerden (2013: 78) estimated the ETI for the high-income taxable group to be 0,79.
Company income tax, capital gains tax and income tax
reform

Tjaart Steenekamp and Ada Jansen

In Chapter 13 we applied the properties of a ‘good’ tax as identi ed and explained in Chapters 11 and 12 to
personal income tax. In this chapter we continue our application of these properties to other income taxes,
namely company income tax and capital gains tax.
We begin by attempting to answer the question of why companies are taxed in the rst section. In Section
14.2, we discuss the structure and calculation of company tax. Section 14.3 analyses the economic effect of
company tax. In Section 14.4 we turn our attention to the taxation of capital gains. We conclude this chapter
with a section on income tax reforms that have implications for both individual and company income tax
and that include the topical issues of the at rate tax and dual income tax.

Once you have studied this chapter, you should be able to

Explain why companies are taxed separately
Define the company tax base
Describe the classical and fully integrated company tax systems
Describe company tax in South Africa
Explain the efficiency of company tax
Discuss the importance of company tax in the investment decision
Discuss the merits and disadvantages of tax incentives for investment
Explain the equity implications of company tax
Discuss the arguments for and against capital gains taxation
Explain a flat rate income tax
Explain the dual income tax.

14.1 Why company taxation?

Owing to certain unique properties of company income, most countries tax this source of tax separately.
ese are the main reasons for taxing companies:
• From a legal point of view, companies are separate entities (legal persons). ey function as institutions
with their own identity and are independent from their shareholders, who are taxed in their own right.
• Companies receive bene ts from government and should be taxed for these privileges according to the
bene t principle. e bene ts include companies’ limited legal liability to shareholders, the creation of
an orderly environment by government, which is necessary for conducting business, the use of
infrastructure and so on.
• If companies are not taxed, it is possible for shareholders to limit their personal income tax liability by
retaining pro ts in the company. is increases the capital value of the shareholders’ investment in the
company. If capital gains are also not taxed, the integrity of the whole system of income taxation is
jeopardised. Tax avoidance by individuals is therefore limited by taxing company pro ts. Since it is
generally members of the higher-income group who are shareholders in companies, taxing this income
is fair from an ability-to-pay perspective.
• By taxing the excess pro ts of imperfectly competitive rms (for example, monopolies and oligopolies),
market failures are addressed. Even though this may contradict the tax neutrality arguments against tax
expenditures, the government may attempt to achieve other economic policy objectives, such as
promoting foreign and local investment and achieving regional development aims, by manipulating
company tax rates and providing tax incentives (for example, liberal depreciation allowances, training
allowances and tax credits).
• Company taxation is administratively simple and generates signi cant revenue as a separate tax,
particularly in developing countries (see the discussion of tax reform in Section 16.6).
• By taxing companies, revenue is derived that would otherwise accrue to foreign investors and their home
governments. Foreign investors are usually taxed on their investment income at company tax rates in
their home countries, that is, according to the residence basis of taxation discussed in Chapter 13. For
example, when South Africa (the host country) levies company tax, the pro ts repatriated to the home
country (for example, the United States) will also be taxed at prevailing American company tax rates. To
avoid double taxation, countries usually enter into double taxation agreements, whereby the home
country credits taxpayers with taxes paid on business income in the host country. In other words, only
the difference between the home country tax liability and the host country liability is payable in the
home country. is amounts to a sharing of company tax revenue between the home and source
countries, thus deviating by mutual agreement from the residence basis of taxation.

Company tax is a signi cant but declining source of tax revenue. In 1975/76, income tax on companies
amounted to R1 969 million, or about 41% of total tax revenue (net collections) in South Africa. In 2018/19,
company tax came to R218 436 million, or about 17,4% of total tax revenue (excluding Southern African
Customs Union (SACU) payments). ere are various reasons for this decline, one of which concerns the
nature of economic development. As a country develops economically, the consumption tax base broadens
and the personal income tax base also becomes more important. Another set of possible reasons includes
factors that reduce company pro t margins, such as rising wage costs, import costs and debt nance
charges. Tax exemptions, tax evasion and tax avoidance are further possible causes, since they lead to
effective tax rates that turn out to be much lower than the nominal or statutory rates. Moreover, the company
tax is a relatively unstable source of government revenue, in the sense that changing domestic and global
economic conditions cause volatility in the contribution of company income tax to total tax revenue as
pro ts track upswings and contractions in economic activity. e company tax rate was also reduced from
50% in the 1980s to the current 28%.
Although the contribution of company tax revenue shows a declining trend, South Africa uses company
taxation very intensively compared to other low- and middle-income economies. When countries are
ranked according to their company income tax efforts, South Africa ranks second out of 27 countries studied
(Steenekamp, 2007).
Mining and quarrying in South Africa has for many years been an important contributor to income tax
revenue, but its role has changed signi cantly over the last decades. Provisional tax payments by the mining
and quarrying sector vary greatly because of global changes in commodity prices, the rand exchange rate
and other reasons. In 1975/76, income tax revenue from mining companies was 10,8% of total tax revenue
(excluding SACU payments). In that same period, tax revenue from gold mining was approximately 10% of
total tax revenue, but it declined to only 0,1% in 1998/99 (the last year in which revenue from gold mining
companies was reported separately from other companies). Gold and other mining companies contributed
approximately 1,2% to total tax revenue in 1998/99. In 2017/18, tax payments by this sector remained low at
only 1,9% to tax revenue collections (R21,9 billion) (National Treasury & South African Revenue Service,
2018: 170).

14.2 The company tax structure

Before company tax can be analysed, it is necessary to discuss the company tax structure. is can be done
by describing the tax base and by distinguishing between different types of company tax as well as different
sizes of enterprises.

14.2.1 The company income tax base

Businesses can be classi ed as incorporated or unincorporated. Incorporated businesses or companies
include private and public companies, state-owned companies and non-pro t companies. Unincorporated
businesses include sole proprietorships and partnerships. In South Africa, mining companies are also
distinguished from non-mining companies. e nature of business differs so much between these two
categories that they are taxed differently. e South African government introduced a mineral and
petroleum royalty regime in 2010/11. e mineral royalty generated R7,6 billion in 2017/18. e taxing of
mining enterprises is still receiving the attention of government. e terms of reference for the South African
Tax Review Committee (the Davis Tax Committee (DTC)), appointed in 2013, speci cally required the
Committee to review whether the royalty regime is sufficiently robust and to assess the appropriateness of
the current mining tax regime. e DTC (2016a) released a nal report on hard-rock mining. Below are some
of their recommendations on the mining tax system and the royalty regime.1
• Retain the mineral royalty regime as its design balances responsiveness to uctuating economic
circumstances (obtaining rents when mining companies earn high pro ts and ensuring minimum
revenues for the scus during downturns). Some aspects of the regime, though, require attention (see
DTC (2016a)).
• Move towards an alignment of the current mining tax system with that of the other taxpaying sectors, and
retain the royalty regime in response to the non-renewable nature of mineral resources.
• e existing capital allowance that makes provision for the upfront expensing of capital expenditure
should be replaced by an accelerated depreciation allowance similar to that of the manufacturing sector
(40/20/20/20 basis). is will also allow for the removal of ring fences (which was put in place to prevent
deductions of future capital expenditure against the tax base of other mines, or non-mining income).
• Existing gold mines should continue to be taxed according to the gold mining formula. e motivation
for this recommendation is the possible loss in employment if the formula is not retained for these
mining companies.
• e mining industry has experienced profound changes over the past two decades, with a resultant
environment conducive to smaller companies making a meaningful contribution in the industry. Given
that their involvement is often based on a contract mining business model, it is recommended that an
appropriate template contract be designed to guide such contract mining arrangements.
• In the rest of this chapter, we focus on the taxation of non-mining companies only.

Taxable income is de ned as total receipts (or revenue) of the company minus certain allowable expenses
(or costs). Expenses comprise costs incurred to run the business. Examples include any labour costs
incurred, interest payments or materials bought. e de nition seems simple enough, but a number of
issues in respect of expenses complicate company taxation. ese are the two most important
complications:
• Interest payments can sometimes be deducted from taxable income, but dividends cannot. is has
implications for company nance. Companies have three basic sources of nance: share capital,
borrowing (debt nancing) and retained earnings. In the case of borrowing, the company can write off
the cost of the loans (interest) against income for tax purposes. When share capital is used in South
 Africa, the cost of shares (dividends) cannot be written off against income. erefore, there is a built-in
bias against the use of share capital as a source of nancing. In other words, the tax system encourages
the use of loan nance, but excessive use of loans increases indebtedness and may lead to bankruptcy.
• Companies are allowed to deduct depreciation on assets when taxable income is calculated. e
rationale for a depreciation allowance is that assets (for example, machinery) lose some of their value
each year and wear out over their lifetime; this constitutes a cost of production. But it is difficult to
determine the true rate of depreciation precisely (that is, economic depreciation). Sometimes the
allowance exceeds economic depreciation. From time to time, governments also allow companies to
depreciate assets at an accelerated rate. is is sometimes used, for example, as an incentive to attract
foreign investors or stimulate investment in selected industries. Practices of this nature have the effect of
reducing the effective company tax rate below the statutory tax rate (see Section 14.3.2.1). Assets are also
classi ed into different categories and depreciation allowances differ accordingly. For example, 5% per
annum is allowed on commercial buildings (in other words, a write-off period of twenty years), but 40%
of the cost of manufacturing machinery will be deducted in the rst year and 20% of the cost for the
subsequent three years.

In 2004, a special depreciation allowance for investment undertaken for construction or refurbishment of
buildings in underutilised designated urban areas was introduced. Taxpayers refurbishing a building within
this type of zone receive a 20% straight-line depreciation allowance over a ve-year period. If a new
commercial or residential building is constructed within a zone of this nature, taxpayers receive an eleven-
year write-off period with a 20% write-off in the rst year and 8% thereafter. e incentive is aimed at
encouraging investment in areas with high population-carrying capacity and central business districts. is
depreciation allowance applies to sixteen municipalities, including areas such as Johannesburg, Cape Town,
Polokwane, Sol Plaatje Metropolitan (Kimberley) and Nelson Mandela Metro-politan (Port Elizabeth).
e problem with differentiated depreciation rates is that this practice tends to open up tax arbitrage
opportunities. Investors, for example, will invest in assets that they can write off quickly or immediately for
tax purposes, but that actually depreciate much slower or even appreciate over time. e tax authorities
usually only become aware of these ‘tax shelters’ after a while, and it then takes some time to eliminate the
loopholes. In South Africa, investments in forests, ships, motion pictures and aircraft were, and still are,
examples of this type of tax shelter. Needless to say, these tax-driven schemes are not models of efficient
resource allocation (see Section 14.3.2.3 on tax incentives).

14.2.2 Types of company tax

ere are two extreme types of company tax. At one extreme, companies are regarded as entities separate
from their shareholders and both are taxed on the same income source (referred to as the classical system
or partnership approach). Company income is thus taxed twice, rst as company tax, and then also as
personal income when distributed as dividends. is is commonly known as double taxation of dividends.
At the other extreme, the company is simply seen as a conduit for all company income to the shareholders.
All company income (retained as well as distributed) is taxed in full in the hands of the shareholders at their
marginal income tax rates. e company tax then only serves as a withholding tax, which is credited in full at
shareholder level. is type of company taxation is referred to as the full integration system.
Prior to 1990, South Africa had a modi ed classical company tax system incorporating a tax on company
income and a tax on dividends, with a certain portion of dividends being excluded from taxation in the
hands of the shareholders. On the recommendation of the Margo Commission (1987: 204), the double
taxation of dividends was discontinued in 1990, when dividends became tax-exempt in the hands of resident
individuals and close corporations. e company tax system was modi ed again in 1993 in an effort to
reduce the company tax rate without undue revenue loss to the scus, and to encourage companies to
nance themselves. A dual tax rate was introduced, that is, a basic rate and a secondary rate. Non-mining
companies are taxed at a xed percentage on their taxable income. is is the basic tax on companies.
Company tax is therefore a proportional tax. e basic company tax rate was lowered from 48% to 40% in
1993 and to 30% in 1999. In 2005, the rate was further reduced to 29% and again in 2008 to 28%. e basic
rate applies to all retained pro ts, that is, pro ts that are used to nance company investment or build
company reserves, and that are therefore not paid out as dividends to shareholders. In addition to the basic
rate, a secondary tax on companies (STC) was introduced in 1993. STC was levied on all pro ts distributed
to the company’s shareholders (in other words, dividends). STC was levied at company level and paid by
companies. e rate was originally set at 15% in 1993, but it was raised to 25% in 1994, reduced to 12,5% in
1996 and reduced again to 10% in 2007. Both rates had to be considered in order to determine the company
tax rate.
e Katz Commission (1994: 175–177 and 226–227) thoroughly investigated the advantages and
shortcomings of STC. e Commission concluded that STC had served its purpose and that ‘it has become
desirable to consider better ways to achieve its objectives’. Starting in 2007, STC was phased out and replaced
with a dividend tax, which became effective on 1 April 2012. e tax rate is set at 20%.2 It applies to all
distributions and is administratively enforced as a withholding tax at company level; legally it is not a tax on
companies. e withholding tax only applies to dividends paid to individuals and non-resident shareholders
(dividends to all South African tax resident companies are exempted). No deductions may be claimed
against the dividends received and these dividends are not included in the shareholder’s taxable income.
Company tax rates differ somewhat between the BLSN countries. In Botswana, resident companies pay
company tax at 22%. Where a business falls within the de nition of ‘manufacturing’, it is entitled to a reduced
corporate tax of 15%. is relief is conditional on the approval by the Minister of Finance on consideration
of, for example, the number of citizens employed, training, replacement of non-resident employees with
skilled citizens and citizen management participation. e Kingdom of Lesotho taxes non-manufacturing
companies at a rate of 25%. Income derived from manufacturing activity sold within the common SACU
market is taxed at a rate of 10%. Manufacturing companies selling outside the SACU market attract no
company tax. In Namibia, corporate income is taxed at 32%. Special rebates may be granted to approved
manufacturing companies. Taxable income (after the special deductions) derived from manufacturing is
taxed at 18% for a total period of ten years. A tax-free regime applies in Namibian export-processing zones
(EPZs). All companies generating income in Eswatini (Swaziland) are taxed at a at rate of 27,5%.

14.2.3 Taxation of small and medium-sized enterprises

It is generally recognised that small and medium-sized enterprises are important for economic growth and
creating job opportunities in the economy. It is also recognised, however, that the tax system severely affects
these types of businesses insofar as tax compliance is concerned3. In addition, small businesses are highly
dependent on working capital provided by the owners. When their pro ts are taxed, owners often have to
use short-term debt, which increases their risk exposure. Various options can be pursued to assist small
enterprises through the tax system, including taxing these businesses on their cash ow, taxing enterprises
at differentiated rates (for example, a progressive corporate rate structure) and using tax incentives (for
example, tax holidays or accelerated depreciation allowances) (see Steenekamp, 1996).
e real and nancial cash- ow tax base equals [sales + borrowing + interest received] − [real purchases
+ interest payments + debt repayment]. e main advantages of a cash- ow tax (Mintz & Seade, 1991: 177–
190; Shome & Shutte, 1993; Sunley, 1989) are as follows:
• e de nition of the tax base is simple as the tax does away with problems of de ning depreciation,
measuring capital gains, costing inventories and accounting for in ation.
• e tax is neutral with respect to the use of capital. Immediate expensing implies that the tax does not
discriminate between debt and equity.

e cash- ow tax has a number of shortcomings, however:

• During the transition from an accrual to a cash- ow system, the tax creates windfall tax revenue gains or
losses, depending on whether government allows or denies companies depreciation on earlier
investment.
• Full and immediate expensing seems to reduce the tax base compared to the accrual income tax base.
However, this conclusion is disputed by some, who argue that it has to be considered that the corporate
income tax base is also eroded through all kinds of tax expenditures.
• e incentive for base shifting is high (the tax base can be shifted from a high-tax entity to a low-tax
entity). is could be done by purchasing inputs from the low-tax entity at in ated prices, by low-rate
leasing and expensing by a high-tax party to a low-tax party, and by the sale of expensed assets at
understated prices to the low-tax entity. Monitoring these practices is difficult and increases the
administrative burden.
• e cash- ow tax raises serious problems in international taxation, as it may not be creditable in home
countries.
• Since no country has implemented the tax as yet, administrative problems are difficult to predict.

e Katz Commission (1994: 157–158) considered cash- ow accounting and recommended that it be
introduced as an option for small enterprises. e South African government accepted this recommendation
in principle and considered implementing cash- ow taxation in a particular sub-sector of the economy,
namely the micro and small business sector. Limiting cash- ow taxation to a sub-sector has the advantage of
improving its cash ow and capital requirements. Although the theoretical case for a cash- ow tax system is
quite strong, one serious problem concerns the international consequences. On the negative side, cash- ow
taxation increases the incentive for tax evasion and avoidance. e application of this form of taxation to a
sub-sector of the economy will probably add to opportunities for tax arbitrage.
In 2000, the South African government opted for a graduated tax rate structure and generous
depreciation allowances for small business corporations. e tax rate structure is set out in Table 14.1.
Small businesses may write off investment in manufacturing assets in full in the tax year in which the
assets are brought into use. An accelerated allowance for machinery, plant, implements, utensils, articles,
aircraft or ships (other than plant or machinery used in a manufacturing or similar process) acquired by
small business corporations can be depreciated at 50% of the cost of the asset in the tax year during which it
was rst brought into use, 30% in the second year and 20% in the third year (a 50/30/20 rate structure over a
three-year period). At the time of writing, these bene ts were limited to businesses with an annual turnover
of less than R20 million. Businesses in which more than 20% of gross income consists collectively of
investment income are excluded. As a means of reducing compliance cost, small businesses with a payroll of
less than R500 000 are also not required to pay the skills development levy. ese tax reliefs are thus mainly
intended for manufacturing businesses, and are aimed at promoting job creation and improving the cash
ow of small businesses. Small businesses engaged in the provision of personal services qualify for relief as
long as they maintain at least four full-time employees for core activities.
To further reduce compliance costs (income tax and VAT paperwork) for very small businesses, the
government proposed a simpli ed tax regime in 2008. e regime is a turnover-based presumptive tax
system. Businesses with a turnover of less than R1 million per year will have the option to elect this form of
tax. Tax liability is calculated using a graduated tax rate structure based not on taxable income, but on
turnover (sales) (see Table 14.2). In addition, these micro businesses are not required to register for VAT4.

Table 14.1 Graduated tax rate structure for small business corporations

INCOME TAX: SMALL BUSINESS CORPORATIONS

Financial years ending on any date between 1 April 2019 and 31 March 2020

Taxable income (R) Rate of tax (R)

0–79 000 0% of taxable income

79 001–365 000 7% of taxable income above 79 000

365 001–550 000 20 020 + 21% of taxable income above 365 000

550 001 and above 58 870 + 28% of taxable income above 550 000

Source: South African Revenue Services (SARS). 2019. Budget 2019 Tax Guide. [Online]. Available:
http://www.treasury.gov.za/documents/national%20budget/2019/sars/Budget%202019%20Tax%20Guide.pdf [Accessed
4March 2019].

e use of presumptive taxes is widespread globally. is approach uses criteria such as turnover, sales,
employment, assets or size of the rm to determine tax liability administratively. It is intended to capture a
minimum tax amount from hard-to-tax groups of taxpayers such as those operating in the informal (or
unrecorded) sector. e aim is eventually to draw these taxpayers into the regular tax net once they have
reached a certain maximum threshold (for example, R1 000 000 of turnover in Table 14.2). Above this limit,
rms do not qualify for presumptive tax status and must migrate to the regular company tax structure.
Presumptive taxation has some advantages, including simpli cation. It not only reduces compliance cost
for the business, but makes it easy for the revenue authorities to administer. ey can use readily available
data to benchmark and verify tax liability, and it also releases them from spending too much time on large
numbers of small tax entities, freeing them to concentrate on the ‘big sh’. On the negative side, presumptive
taxes may cause inefficiencies in the tax system as rms try to take advantage of the regime by restructuring
their activities to qualify for presumptive tax status. Alternatively, if the maximum threshold is set too high
(resulting in a tax liability that is low compared to the regular tax liability), rms will attempt to remain in the
presumptive tax net. is may compromise horizontal fairness in the company tax system.

Table 14.2 Presumptive turnover tax for micro businesses

Turnover tax for micro businesses

Financial years ending on any date between 1 April 2019 and 31 March 2020

Taxable turnover (R) Rate of Tax (R)

0–335 000 0% of taxable turnover

335 001–500 000 1% of taxable turnover above 335 000

500 001–750 000 1 650 + 2% of taxable turnover above 500 000

750 001 and above 6 650 + 3% of taxable turnover above 750 000

Source: South African Revenue Services (SARS). 2019. Budget 2019 Tax Guide. [Online]. Available:
http://www.treasury.gov.za/documents/national%20budget/2019/sars/Budget%202019%20Tax%20Guide.pdf [Accessed
4 March 2019].

As part of its mandate, the DTC investigated the tax system of the small and medium size business sector.
In 2014 the DTC released a rst interim report on small and medium enterprises, and this was followed by
the nal report in April 2016 – see DTC (2016c). ree focal issues were the turnover tax system for micro
businesses, the small business corporation tax system (in particular the graduated tax structure and other
incentives), and the VAT registration threshold for small businesses. Below is a brief synopsis of these issues,
as reported in the nal report of the DTC (2016c):
• e turnover tax thresholds were adjusted in the 2015/16 Budget (which were part of the
recommendations in the DTC’s rst interim report). is was done to encourage small business
participation in the economy and form part of the tax system (National Treasury, 2015c: 49). In its nal
report, the DTC proposed retaining the current turnover tax package. Furthermore, taxpayers should be
given the option to exit the system on an annual basis. e National Treasury and SARS should also
explore measures that would require micro businesses to register for tax before allowing them to use
electronic payment facilities.
• e DTC found that the small business corporation (SBC) tax incentive system was not effective. A large
proportion of the bene ts of the system accrued to service-related businesses (such as nancial,
medical, and veterinary services), who were not the intended bene ciaries. Furthermore, since the
turnover tax system is in place, the SBC tax system does not have to account for small businesses with a
turnover of less than R1 million. e DTC proposed the following options for consideration: retain the
current system with its accompanying problems; remove service-related small businesses from the SBC
 de nition; consider extending the SBC system with a compliance rebate system to all tax compliant
SBCs; and nally, remove the SBC incentives and direct more targeted interventions through the office of
the Department of Trade and Industry, or the recently formed Department of Small Business
Development.5
• e DTC received submissions suggesting that the compulsory VAT registration threshold should be
increased. In response, the DTC indicated there was no justi cation for increasing this threshold, as it
aligns favourably with international standards.

14.3 The economic effects of company tax

Company tax impacts on economic efficiency, the decision to invest and the fairness of the tax system. ese
topics will be the focus of the next subsections.

14.3.1 Economic efficiency

In analysing the economic efficiency of company tax, we again have to consider the excess burden of the tax.
Does company tax cause an excess burden? In other words, are relative prices (or returns) distorted,
resulting in resources not being optimally allocated? e answer depends on whether company tax is seen
as a tax on excess pro ts (or economic pro ts) or as a selective tax on capital.
If company tax is regarded as a tax on economic pro ts only, then company tax has no excess burden.
e economic pro ts of rms are simply reduced by the tax and rms still maximise pro ts. e tax does not
cause any other changes in behaviour (in other words, marginal costs are unchanged) and relative prices are
not affected. However, companies are taxed on their accounting pro ts (that is, economic pro t plus normal
pro t).6 In the short term, normal pro ts are a xed cost. erefore, marginal costs, which change only when
variable costs change, are not altered by the tax. However, in the long term, there are no xed inputs (all
become variable) and taxing normal pro ts as part of accounting pro ts may then affect marginal cost. us,
in the long term, a tax on accounting pro ts may cause a change in the behaviour of owners and prices,
which could result in an excess burden.
e return on capital of shareholders invested in companies (in other words, dividends) is a non-
deductible expense. Company tax can therefore be regarded as a tax on capital. However, all forms of capital
are not taxed at the same effective rate. In South Africa, incorporated businesses (for example, public
companies) are taxed at a rate of 28%. Non-incorporated businesses (for example, sole proprietorships) are
taxed at marginal personal income tax rates, which vary between 18 and 45%. Small business corporations
are taxed at graduated rates that vary between 0 and 28%. e important point about differential taxes on
capital is that if companies are taxed at different rates, the net (after-tax) returns differ from sector to sector.
Capital will then move from the sectors with lower net (after-tax) returns to sectors where a higher net (after-
tax) return can be obtained. In theory this process will continue until net returns are equal in all of the
sectors. is migration of capital is purely in response to tax-induced differences in returns and not because
capital can be used more productively elsewhere. Resources are therefore misallocated and an excess
burden results. In the United States, this loss in welfare has been estimated at between 12% and 24% of total
tax revenue (see Shoven, 1976: 1261–1283; Jorgenson & Yun, 2001: 302).

14.3.2 Company taxation and investment

When decisions are made to invest in a country, investors consider taxes. However, non-tax factors and tax
incentives are just as important.

14.3.2.1 The decision to invest

Another issue concerning the economic efficiency of company taxation is its impact on local and foreign
xed investment. e company tax structure can lower or raise the user cost of capital. e user cost of
capital includes, rstly, the opportunity cost of holding an asset in a company. By investing capital in a
company instead of, for example, saving it, an opportunity cost is incurred (in other words, the interest
forgone). Secondly, the user cost of capital includes depreciation costs. Capital invested in, for example,
machines, decreases in value. Finally, the user cost of capital is affected by company tax. e returns on
share capital (realised pro ts and dividends) are taxed and constitute a cost to the investor. By lowering
nominal company tax rates, providing depreciation allowances or giving tax incentives (for example, tax
holidays and tax rebates), the user cost of capital can thus be lowered and vice versa.
In 2014, the DTC commissioned the World Bank to update a 2006 sectoral study7 on the effectiveness of
investment incentives by determining the marginal effective tax rate (METR) of capital investment. e
METR is a summative measure8 that captures the impact of the tax system on the marginal investment of a
pro t-maximising rm, and it determines the scale of the investment project (World Bank, 2015: 8). In the
case of company tax, we can determine the METR on income from physical capital by determining the
impact of the tax system on the user cost of capital. Hence, the METR is simply the difference between the
before-tax and after-tax rate of return on a marginal investment, divided by the before-tax rate of return
(Fullerton, 1999: 270). e effective tax rate typically considers the statutory tax rate, and any tax incentives
related to the investment (such as investment tax credit, accelerate depreciation allowances, and investment
allowances). ere are two main approaches to determine the METR, namely, the forward-looking approach
and the backward-looking approach (see Sørensen, 2004: 4). e former applies the parameters of the
current tax legislation and determines what the expected tax burden will be on a hypothetical investment.
e latter uses (historical) data on capital taxes paid, and compares it to before-tax capital income estimates.
e main ndings of the World Bank study revealed that the METR on capital is lower than the statutory
company tax rate (of 28% at the time of the study) for almost all sectors investigated. is implies that tax
incentives (see Section 14.3.2.3 below) have a signi cant effect on the tax burden of the marginal investment
(World Bank: 2015: 9). In particular, the METR for the manufacturing sector was 19,6% (using a neutral debt-
equity scenario, i.e. a ratio of 0,5). Interestingly, even for sectors where there are no speci c tax incentives,
the allowance of deducting interest payments from taxable income has reduced their METR. Hence, where
xed assets were nanced through debt, the METR was lowered considerably (World Bank, 2015: 10).
e question in developing countries is whether taxes impact on foreign direct investment (FDI); that is,
can developing countries attract new capital to meet various economic, social and political objectives?
Empirical studies on the impact of taxes on FDI in developed countries generally conclude that taxes have a
strong effect. Limited evidence for developing countries also shows some impact of tax rates on location
decisions of investors (see Zee, Stotsky & Ley, 2002: 1509–1510; Keen & Mansour, 2009: 22; Norregaard &
Kahn, 2007: 18–19). Using dynamic panel data econometrics, Klemm and Van Parys (2010) nd evidence
that lower company income tax (CIT) rates and longer tax holidays are effective in promoting foreign direct
investment in Latin America and the Caribbean, but not in Africa. is is a surprising result. It means that
investment incentives are ineffective in Africa and simply lead to a loss of revenues. is ineffectiveness of
tax incentives is presumably attributed to a generally poor investment climate in Africa, thus indicating that
non-tax factors are important considerations when investors make locational decisions.
Based on the evidence brought before the Katz Commission (1994: 213), the Commission observed that
‘while tax is an important investment consideration, it ranks well down the list of priorities unless it poses a
speci c and actual inhibition’. However, corporate taxation might be a major policy issue for the following
reasons:
• A punitive corporate tax regime constrains investment, while a liberal regime could result in an
unnecessary loss of revenue to the treasury, thus allowing foreign treasuries to capture bonus revenue.
• An unfavourable corporate tax regime might lead to other forms of pro t-taking (for example,
repatriation of pro ts through transfer pricing and setting up branches instead of subsidiaries). e
willingness of multinational enterprises (MNEs) and others to lower their effective tax rates through
transfer pricing indicates their sensitivity to tax rates. Globalisation has opened opportunities for MNEs
to minimise their tax burden, which impacts on tax equity (see Organisation for Economic Co-operation
and Development, 2013).
 When non-tax factors are approximately the same in countries competing for investment, taxation
• becomes all-important, as is evident from the intense tax competition in the East Asian region.

14.3.2.2 Non-tax factors

One should guard against overemphasising the impact of the company tax structure and rates on the
decision to invest. Overall, it appears that there is much more empirical evidence indicating that the effect of
taxes is minor compared to the impact of other factors that affect investment decisions (see Norregaard &
Kahn, 2007: 18; Zee, Stotsky & Ley, 2002: 1509–1510). e following non-tax factors are important to
investors:
• Business opportunities: Investors look for market opportunities to maximise pro ts, for example, by
increasing market size or vertically integrating their production processes across national borders.
Portfolio investors look for speculative pro ts (such as those offered by exchange rate differentials) or are
interested in spreading risk.
• Political stability and good governance: Political instability disrupts investors’ calculations of the
expected rate of return on their investment. Conditions in badly governed countries impact negatively on
the business environment.
• Economic stability: is includes price and exchange rate stability as well as scal stability (in other
words, a sustainable scal de cit).
• Labour stability: Rigidities in the labour market can result in labour costs becoming internationally
uncompetitive. Work ethic is important for certain types of investors. For a business person interested in
manufacturing goods that use labour-intensive technology (for example, clothing), a reliable workforce
is essential.
• Security and respect for property rights: Investors are sensitive to policy in respect of nationalisation and
privatisation, and to the honouring of international commitments such as double taxation agreements
and debt repayment obligations.
• Transportation costs and communication infrastructure: In principle, it is always desirable to
manufacture goods as close as possible to the market or markets that they serve. Poor communication
infrastructure can raise the cost of doing business.

14.3.2.3 Tax incentives

Tax incentives (for example, tax holidays, capital allowances, incentives for research and development, and
deductions for training expenses) are used in many developing countries in an effort to attract investment to
particular regions, sectors and industries. ey are used for speci c purposes such as export promotion,
employment creation, local participation, local sourcing of materials, improved infrastructure, the
promotion of technological transfers, skills development and the development of geographical regions.
e reasons for and arguments in favour of the use of tax incentives (see Zee, Stotsky & Ley 2002: 1499–
1501; Boadway & Shah, 1995; Chia & Whalley, 1995) are as follows:
• In the case of market failures (for example, positive externalities), incentives may correct these. To
reduce congestion and pollution in urban areas, tax incentives could be used to promote economic
activity in rural areas. Similarly, projects that use new technology or require much research and
development (R&D) input can be targeted.
• Incentives are regarded as a means of offsetting the effects of regulations and controls in developing
countries, such as exchange control and licensing restrictions. Incentives compensate for these
distortions.
• Likewise, labour market distortions may be corrected by using investment incentives as a second-best
policy. In this way, employment can be created.
• Capital markets are often imperfect in respect of small rms. By targeting these rms selectively,
incentives can be used to overcome these constraints.
• Temporary investment incentives may be effective devices for assisting infant industries.
• To the extent that they are effective in encouraging investment, incentives generate external bene ts for
the economy over and above those accruing to the investor, for example, innovation and training.
• Tax incentives can be used to signal to investors that a country is open for business. ey announce to
investors that government has certain priorities and that investors will be compensated for their
commitment. Countries remain vulnerable to competitive incentives by other countries, whereby the
signalled bene ts of incentives are neutralised. Niche markets can also be promoted through income tax
incentives.

Various analysts have questioned the use of incentives to promote investment and other objectives. e
arguments raised against incentives (Zee, Stotsky & Ley, 2002: 1501–1502; Easson, 1992: 387–439; Shah, 1995;
World Bank, 1991) include the following:
• Incentives complicate tax administration and increase administration costs.
• Tax concessions erode the revenue base and necessitate higher tax rates. In some cases, pro table
investments would have taken place without the tax incentive. e incentives then become a free gift
from the government to the investor or even the treasury of the investor’s home country if the latter uses
the residence principle.
• Incentive schemes are often the result of pressure group action rather than an analysis of the economy as
a whole, thus favouring certain sectors over others (socially unproductive rent-seeking). If the incentive
is not aimed at correcting a market failure, it may actually create distortions and economic costs
(inefficiencies).
• Tax incentives result in an uneven tax burden among taxpayers, thus violating horizon equity.
• Tax incentives are less likely to succeed in attracting foreign investment than a general reduction in
corporate income tax.
• Investment incentives often lead to windfall gains in respect of investments that would have taken place
anyway. ey have little impact on new investment.
• If there are obstacles to investment (for example, untrained labour), it is better to address these problems
directly instead of using tax incentives, since such incentives simply add further distortions to the
economy.

In developing countries, tax incentives for investment are now much more widely used than in the 1980s.
Low-income countries use incentives more extensively than do middle-income countries. It appears that tax
holidays are very popular. At the same time, the use of free zones (for example, export processing zones, free
trade zones and economic development zones) has increased to attract footloose industries9 (see Keen &
Mansour, 2009: 20). e number of countries in Sub-Saharan Africa with free zones has increased from 17 in
2005, to 27 in 2014 (James, 2016: 159). Research by Klemm and Van Parys (2010) con rms that countries in
Africa, Latin America and the Caribbean compete for FDI by lowering company income tax rates and
offering tax holidays. ere is no evidence of spatial interaction (competition) using investment allowances
and tax credits.
e popularity of tax holidays runs counter to best practice tax advice. Tax holidays exempt (for a
period) investment projects from company income tax (CIT) or offer a rate lower than the regular rate. e
advantage of tax holidays and preferential CIT rates is ease of administration. e disadvantages are many:
these incentives favour investors expecting high pro ts and who would have undertaken the investment
without the incentive, it encourages shifting of pro ts from non-exempt to exempt rms, it attracts short-run
projects rather than sustainable ones and the revenue losses are not transparent. Some forms of incentives,
for example, investment allowances or tax credits and accelerated depreciation, are preferred to tax holidays
(see Zee, Stotsky & Ley, 2002: 1504–1505). ese incentives (for example, shorter write-off periods for
investment costs) provide for investment in plant and equipment, thereby raising capital intensity, however.
ey are more targeted and transparent in addressing, for example, human resource training needs and
cash- ow problems associated with new investments, and are also less prone to abuse than tax holidays. To
ensure that incentives achieve their bene cial outcomes, well-designed incentives must follow three basic
rules (Bird 2008: 9):
• Keep them simple.
• Keep records on who receives them and what the costs are in revenue sacri ced.
• Evaluate the numbers at regular intervals to check the results.

To this should be added that all new tax incentives should be accompanied by a legislative sunset clause,
whereby the incentive automatically expires unless re-legislated following a proper assessment of their net
bene ts.
e Katz Commission (1994: 204) thoroughly reviewed tax incentives in South Africa and concluded that
‘… the range of incentives should be narrowed as far as possible and that those which exist should all be
justi ed in terms of the objectives in the Reconstruction and Development Programme.’ is, however, is a
very vague guideline. e Commission also endorsed the principle that tax incentives be subject to thorough
cost-bene t analysis. Incentives can be economically justi ed mainly where markets fail (for example, when
investments generate positive externalities). When granted, incentives of this nature should be transparent
and discretion should be minimised.
For some years, South Africa has used a range of incentives to encourage foreign and domestic
investment. Investment support is provided by the Department of Trade and Industry, and consists of an
extensive grant system. Tax incentives (see Box 14.1) are also provided to promote investment. ey consist
of accelerated depreciation allowances, graduated tax rates for small businesses (see Section 14.2.3), and
incentives for capital expenditure on research and development (R&D). As indicated earlier, the World Bank
(2015) determined the METR for several economic sectors in South Africa, and they concluded that most
sectors bene ted from a reduced effective tax rate because of the tax incentives. In particular, the effective
tax rate for the mining sector was calculated at -1.2% (weighted average across minerals), and a signi cant
factor for this result is the full expensing in the year of purchase of capital expenditure (an incentive that the
DTC has recommended should be adjusted – see Section 14.2.1). In the late 1990s, South Africa also
experimented with a tax holiday scheme for certain companies, but it applied only until September 1999.
Much emphasis is now placed on grants and depreciation allowances. is augers well for transparency and
better targeting of projects.
In the 2019 Budget Review, a statement concerning tax expenditures was provided in an effort to quantify
revenues forgone as a result of tax incentives. It was estimated that incentives amounting to approximately
R10,0 billion were provided to the corporate sector in 2016/17 (National Treasury, 2019a: 121). Compared to
the approximate R204 billion tax revenue collected from companies in the same year, the revenue forgone
owing to tax incentives is insigni cant. Nevertheless, with expenditures now quanti ed, analysts will be able
to attempt a proper cost-bene t analysis of tax incentives. One such attempt is the second study completed
by the World Bank (2016), in which the analyst rst investigated whether tax incentives reduce the cost of
capital, and whether the reduced cost of capital resulted in additional investments. Using individual rm
data for the period 2006 to 2012, the study determined that although tax incentives reduced the cost of
capital to between 3% and 6,5% for all sectors, investment increased (by R2,1 billion each year over this
period) in only the following sectors: manufacturing, trade and services, agriculture, and construction
(World Bank, 2016: 6). However, revenue foregone (due to tax incentives) amounted to approximately R4,5
billion per year over the seven-year period (World Bank, 2016: 7).
Additional employment opportunities to the tune of 340 000 were created, but the costs thereof were
relatively expensive (on average an additional job cost around R116 000 in terms of revenue foregone). Two
subsequent studies followed in 2017 and 2018 in which the World Bank analysed the efficiency of speci c tax
incentives (research and development and incentives for small business corporations and property
investment). e DTC (2018a: 51–53) reported these results: in the case of the research and development
incentive, rms who received the incentive increased their research and development spending, compared
to those who did not receive it. e results showed that the additional investment in research and
development (approximately R13 billion) outweighed the revenue foregone of around R7 billion (between
2008 and 2015). On the tax incentives for the small business corporation sector (i.e. graduated rate structure,
and investment incentives), the DTC (2018a: 53) reported the World Bank study’s nding that an additional
job created by the incentives in this sector cost the government on average lost revenue to the tune of
around R120 000.
 Box 14.1 Grants and tax incentives for investment

A generous system of taxable and non-taxable cash grants is provided by the Department of Trade and Industry
to promote new investment in South Africa. Programmes benefiting from this grant system include:
• The Clothing and Textile Competitiveness Improvement Programme
• The Automotive Investment Scheme
• Manufacturing Competitiveness Enhancement Programme
• Black Industrialists Scheme
• Critical Infrastructure Programme
• Sector-Specific Assistance Scheme
• The South African Film and Television Production Incentives
• Special Economic Zones.

Tax incentives to support investment are provided in terms of the Income Tax Act, No. 58 of 1962, and include
accelerated depreciation allowances, graduated tax rates for small businesses, and incentives for research and
development (R&D) capital expenditure. Accelerated depreciation allowances permit investors to use a
40/20/20/20 regime for manufacturing assets (plant and machinery). Investments in renewable energy and
biofuels production as well as capital expenditure on R&D can be depreciated over a three-year period on a
50/30/20 basis. Furthermore, an accelerated depreciation regime applies to buildings within Special Economic
Zones. The special allowance makes provision for a 10% per annum depreciation over ten years.
Other tax incentives include the following:
• A tonnage tax regime for South African shipping companies according to which companies are taxed not on
their business income, but on the size of the ship, thus lowering the effective tax rate.
• Investments in energy-efficient equipment get an additional tax allowance of up to 15%.
• The government’s industrial policy strategy provides for investment and training allowances for large
manufacturing projects. It is designed to support Greenfield investments (in other words, new industrial
projects that utilise only new and unused manufacturing assets) as well as Brownfield investments (that is,
expansions or upgrades of existing industrial projects). A points system is used to determine qualifying
status (criteria include improved energy efficiency, skills development, innovation, business linkages, location
in Special Economic Zone, and SMME procurement). Between 35 and 55% of the cost of fixed capital
investment and employee training is deductible from taxable income.
• Expenditure in respect of scientific and technological research and development can be deducted at a rate of
150% of expenditure.
• Special economic zones exist where a 15% company tax rate applies together with an employment incentive
allowing tax deductions for workers earning less than R6 500 per month.

14.3.3 Fairness
When we considered tax incidence in Chapter 11, we explicitly stated that only people could bear the
burden of a tax. In the case of companies, the people in question are shareholders, workers and consumers.
We will now analyse the incidence of company tax on each of these groups.
In Section 14.3.1, we noted that when companies are taxed at different rates, capital will move from high-
taxed sectors to low-taxed sectors so that a higher net return can be obtained. eoretically, this process will
continue until net (after-tax) returns are equal in all of the sectors. All owners of capital therefore bear the
burden of the tax, not only the capital owners in the taxed sector. e real controversy regarding the
incidence of company tax is whether capital owners bear the full tax burden. Empirical studies indicate that
capital bears almost the entire burden, but it is still worth examining how the tax can theoretically be shifted
to other groups.
When the net (after-tax) return of taxed businesses declines, rms will attempt to recover the tax by
increasing their prices to consumers. e extent to which the taxed businesses can shift the tax will, of
course, depend on the price elasticities of demand and supply of the goods and services produced by these
businesses (see Chapter 11).
How can the company tax be shifted to workers? Just as rms may attempt to pass on the tax burden to
consumers in the form of higher prices, so they may try to shift the burden to workers in the form of lower
wages or lower levels of employment. Some observers argue, for example, that company tax affects savings
and investment over time. Since capital is internationally mobile, a tax on capital will cause capital to move
to lower tax jurisdictions. is lowers the amount of capital available per worker in the country where capital
is taxed and causes the marginal physical product of labour (MPPL or labour productivity) to fall. A decline
in labour productivity may lead to a decrease in wages or lower employment. us labour may bear a part of
the company tax burden. Details about how company tax can be shifted to labour fall beyond the scope of
this chapter. However, the incidence of company tax on consumers and labour remains controversial.
Ultimately, equity (or fairness) depends on who the owners of capital are. Capital is usually owned by the
higher-income group. If the burden of company tax falls mainly on capital owners, vertical equity is served.10
Horizontal equity depends on the type of company tax system in operation and will not be discussed here.
e fully integrated income tax system (Section 14.2.2) appears to serve both horizontal and vertical equity
objectives best, but this type of tax system is unfortunately administratively complex.

14.4 Capital gains tax

Capital gains can be de ned as increases in the net value of assets over a period of time (for example, an
accounting period or scal year). According to the Haig-Simons de nition of comprehensive income,
anything that makes consumption possible without diminishing wealth at the beginning of a scal year is
considered to be income. Capital gains are, accordingly, often classi ed as a form of income and are taxed as
such. Capital gains can be taxed as they accrue (an unrealised gain) or when they are realised. An unrealised
capital gain occurs when an asset increases in value in a given scal year and the asset is not sold, for
example, an increase in the rand value of a Krugerrand when the rand depreciates against the dollar. A
capital gain is realised when an asset has increased in value and is sold for cash.
Capital gains tax is currently levied in a number of developed countries, such as Canada, the United
States, the United Kingdom, Australia and Japan, as well as in some developing countries, including
Argentina, Brazil, India, Nigeria and Zimbabwe. In South Africa, capital gains were generally not taxable in
the past. Where assets were kept as an investment and then sold, the yield on realisation of the asset was
regarded as a receipt of a capital nature: one asset (capital) was simply converted into another (cash) and the
yield was therefore not taxable. But where assets were sold in the course of normal business (that is, to make
a pro t), this pro t was regarded as income and taxed as such. In many cases, the courts had to rule on the
application of this principle, which caused a great deal of uncertainty about the taxability or otherwise of
capital gains. On 1 October 2001, South Africa implemented a capital gains tax (CGT).
Capital gains tax comes into play when there is a change in the ownership of an asset, that is, when it is
sold, given away, scrapped, swapped, lost or destroyed. It is thus a ‘realisation’ or transaction-based tax and,
when capital gains are realised or deemed to be realised, these gains form part of the income tax base. e
capital gain (or loss) is determined as the difference between the realised proceeds from the sale of the asset
and the total base cost of the asset. e base cost of an affected capital asset includes the original acquisition
costs and related transaction costs (for example, legal fees and brokerage), the costs of any improvements
and VAT.
Capital losses may only be deducted against capital gains. ere is no such thing as a negative CGT.
Capital losses incurred on assets not used for business purposes cannot be subtracted from realised gains
for tax purposes. ese include assets used for personal consumption, such as sailboats, second vehicles and
aircraft.
e rst R40 000 of net capital gains of a natural person during a tax year is excluded from CGT. Although
capital gains in excess of R40 000 are included in taxable income, some relief is granted to individuals and
other legal persons. In the case of an individual, only 40% of the net capital gain (the inclusion rate) is
included in taxable income. A company has to include 80% of its net capital gain. is means that the
effective capital gains tax rate in respect of individuals varies between 0% and 18%, depending on the
marginal tax rate, and for companies it is 22,4%. e effective capital gains tax rate for small businesses
ranges from 0% to 22,4%.
Any individual or legal person (for example, a company, close corporation or trust) resident in South
Africa is liable for CGT in respect of the disposal or deemed disposal of capital assets held both inside and
outside the country. Where an individual or legal person is not resident, a liability will only arise in the event
of the disposal of immovable property inside South Africa or the sale of the assets of a local branch,
permanent establishment, xed base or agency through which a trade, profession or vocation is being
carried out.
Capital assets liable for CGT are property of any kind, whether movable or immovable, tangible or
intangible, and include land, mineral rights, office blocks, plant and machinery, motor vehicles, boats,
caravans, trademarks, goodwill, shares, bonds and Krugerrands. Some assets, such as trading stock and
mining assets qualifying for income tax deductions as capital expenditure, are exempted from CGT. In the
case of individuals, principal owner-occupied residences (gain/loss of less than R2 million), private motor
vehicles and personal belongings (for example, clothing, stamps, works of art, antiques, medallions, foreign
exchange and coins not minted in gold or silver), are exempted from CGT. Small business assets (businesses
with a market value of assets of less than R10 million) realised by individuals over 55 who use the proceeds
for retirement purposes are also exempted from CGT, provided that the assets had been held for at least ve
years. is provision is limited to a once-off exemption of R1,8 million per taxpayer.
In order to safeguard the reinvestment of pro ts, a capital gains tax liability may in certain cases be
deferred until a subsequent CGT event. Deferral (rollover) relief applies to asset disposals such as certain
transfers of property to establish or reorganise a business, transfers of property from a deceased estate,
donations of property and transfers between spouses.
e Franzsen Commission (1968) already recommended the introduction of a separate capital gains tax
50 years ago and more than 30 years before it was established. e Commission regarded pro ts arising from
the sale of shares and xed property (with the exception of property that the person liable for tax uses for
residential purposes) as the principal components of the capital gains tax base. Two subsequent tax
commissions held contradictory views. e Margo Commission (1987) opposed a capital gains tax primarily
because of the administrative problems involved. e Katz Commission (1995: 49), in its ird Interim
Report, also recommended that ‘… by reason of the lack of capacity on the part of the tax administration,
there should not be capital gains tax in South Africa at this stage’. e low revenue potential of this tax
reinforced the Katz Commission’s conclusion. Capital gains are not really taxed for the sake of the revenue
they yield, but rather for other reasons. At the time of introduction, it was estimated that CGT could raise
about 1% of total tax revenue a year directly (in other words, around R5 to R6 billion). In 2017/18, capital
gains tax raised R17,6 billion (1,5 % of total tax revenue), of which companies paid R7,6 billion and
individuals paid approximately R10,0 billion (National Treasury & SARS, 2018: 274).
e most important reasons for capital gains taxation are as follows:
• To protect the integrity of the personal and corporate income tax base. If capital gains are not taxed,
taxpayers have an incentive to convert income into capital gains in order to avoid taxation. Consider the
example of a sole proprietor who reinvests his or her pro t instead of taking it as a salary (which is taxed
at marginal income tax rates). e reinvested income increases the value of the business. When the
business is eventually sold, the bene ts are reaped in the form of long-term capital gains (regarded as
non-taxable capital income in the absence of capital gains taxation).
• To ensure horizontal equity. A capital gain represents an increase in economic power, and it increases
the individual’s ability to earn income and to be taxed. Consider two persons with the same net additions
to wealth (income plus net assets) Person A’s net additions consist of salary income and capital gains;
person B’s net additions consist of salary income only. Both have the same horizontal ability to pay in
terms of the comprehensive de nition of income (that is, the Haig-Simons de nition). If capital gains are
not taxed, person B is taxed unfairly on his or her income.
• To ensure vertical equity. Capital gains accrue mostly to higher-income tax-payers. If they are not taxed
on these gains, the vertical ability-to-pay principle is jeopardised.
• To improve economic efficiency. If investments are chosen on the basis of tax considerations, the
allocation of investment funds is distorted and an excess burden results.
ere are also several arguments against capital gains tax, of which the most important include the
following:
• Capital gains taxation is subject to numerous administrative problems. Assets have to be valued, and
there is a need for accurate and up-to-date deeds registers in the case of, for example, works of art and
real property. e valuation problem is more acute in a situation where an accrual base is used (that is,
where unrealised capital gains are also taxed). e problem of valuation is less severe when a realisation
base is used (in other words, when the selling price is compared to the purchase price when the asset is
sold and tax payment is only due when the asset is sold).
• If nominal pro ts (instead of real pro ts) are taxed, equity is at risk. In ation causes imaginary capital
gains (that is, increases in the nominal value of assets) and it may be unfair to tax someone just because
in ation has increased the nominal value of an asset. Nominal capital gains should therefore be de ated
by an appropriate price index. e choice of a suitable index is a further complication.
• Capital gains are usually once-off events. Taxpayers tend to lock in rather than realise investments in
order to avoid the tax. is lock-in effect can affect investment negatively. Concessions are therefore
usually made either in the form of lower personal income tax rates on capital gains or by not taxing
capital gains once a certain period has elapsed. To compensate for the effects of in ation and the lock-in
effect, the South African tax authorities opted for low effective capital gains rates.

14.5 Income tax reform

In the previous chapter and preceding sections of this chapter, attention was devoted respectively to
personal income taxation and company taxation within the broad framework of a comprehensive income
tax system (the Haig-Simons de nition of income). is tax base includes labour and capital income, and
ideally applies the same rate structure to all forms of income. In practice, tax systems are far removed from
the intended comprehensive income tax. Capital income and labour income are taxed at different rates, and
some forms of income are not taxed. Certain activities are encouraged using tax incentives, deductions and
credits, all leading to a smaller tax base. Because of the international mobility of capital and, to some extent,
labour, countries try to outcompete their rivals using the tax system. But distortions are created that impact
on the work effort as well as saving choices (that is, tax efficiency). e comprehensive income tax does not
deal adequately with irregular income earned over the lifetime of the taxpayer and results in excessive
progressivity in this case. Tax systems also include taxes other than income taxes, such as consumption taxes
(for example, VAT and excises) and wealth taxes (for example, inheritance tax). is mix of taxes has
different distributional effects, and spoils both horizontal and vertical equity. e problem becomes even
bigger considering the administrative complexity and compliance issues resulting from the comprehensive
income tax system. In addition to the tax preferences and expenditures, the integration of the personal
income tax and company income tax adds to the administrative difficulties. What are the alternatives? In this
section, we consider two reform options that have received attention in recent times: the at rate income tax
and the dual income tax.

14.5.1 The flat rate income tax

e at rate income tax option is mainly a response to the administrative complexity of the comprehensive
progressive income tax and the inefficiencies created by taxing labour income and capital income at
different rates (see Barreix & Roca, 2007: 127). Two broad variants of the at rate income tax can be
considered. e system proposed by Hall and Rabushka (1983 and 1995) received particular interest in the
United States. eir proposal provides for a combination of a cash- ow tax on business income (with full
allowance for capital expenditure, but not labour expenditure) and a tax on labour income (with a non-
taxable personal exemption) at a single (low) at rate. In principle, it is none other than an origin-based VAT
collected by the subtraction method and supplemented by a (non-refundable) tax credit against labour
income (see Chapter 16 for a description and discussion of the destination-based VAT system). is design,
therefore, taxes consumption, but has some progressivity built in (the personal income tax exemption). is
system has not been implemented in this form anywhere.
e second variant has been adopted in a number of Eastern European countries (including Russia) and
has attracted much attention. It deviates from the Hall-Rabushka form in that only labour income is taxed at
a single positive rate (with some income allowance based on the taxpayer’s circumstances). Sometimes
company income is also taxed at the same rate, but not necessarily. erefore, when we refer to a at rate
income tax, we mean a single low rate on personal income (see Keen, Kim & Varsano, 2008: 714 for a
comprehensive discussion and analysis of the at rate income tax).
Supporters of the at rate tax base their arguments on various advantages (see Keen et al., 2008: 712–751;
Browning & Browning, 1994: 378–379; Boskin, 1996):
• A at rate tax is simple to administer, and complexity is reduced by the elimination of all kinds of
exemptions and exceptions. Tax liability can be determined on a single-page tax return where all forms
of income are entered, and then multiplied by the tax rate. Tax compliance should improve because of
the lower marginal tax rate. Improved compliance may also lead to higher tax revenue (see the Laffer
curve analysis in Section 13.4.3).
• To get some perspective on the size of the tax rate required to maintain tax revenues, it should be
considered that in 2017, current taxes on income and wealth of households in South Africa were R465
800 million (South African Reserve Bank, 2018b: S-134). is amount was raised with average tax rates
ranging from zero % to in excess of 34,7% (taxpayers with taxable income of R1 500 001 in 2017/18). If the
gross balance of primary income of households as de ned in the national accounts is used as the tax
base (R3 061 637 million), the same amount of tax could have been obtained by levying a tax of
approximately 15,2% on all income (with zero personal income exemptions).
• Lower marginal tax rates increase productivity and the incentive to work.
• Horizontal equity is promoted since different special provisions are not available to people with the same
taxable income.
• Bracket creep (explained in Section 13.4.4) during in ation is eliminated.
• A political economy argument favours the adoption of a at rate. e public choice school would reason
that a single low tax rate will limit the amount of tax revenue government can extract from voters. e
size of a government pursuing its own interest and wasting resources is consequently restricted. It should
be noted that there are other ways of constraining government, such as scal rules (see Section 18.5).

One disadvantage of the at rate tax regime is that the built-in exibility of income taxation is reduced if the
aggregate marginal tax rate falls; although this will be less the higher the basic allowance accompanying the
at tax is. Another problem is that capital income and labour income are in most cases still taxed at different
rates, resulting in tax arbitrage and competition between different countries. e most important
shortcoming of the at rate tax proposal (and probably its nemesis) lies in its redistributive impact. e tax
burden will be redistributed. e burden of the high-income group will be reduced, while the burden of the
low-income group will increase (although some adjustments may be possible to counteract these effects). Its
impact on the tax burden of the low-income and even the middle-income group can, therefore, be curtailed
by applying the tax rate to all income above a minimum income. It would thus introduce a degree of
progressivity in the tax structure.
Another serious equity problem with the at rate tax is the treatment of taxpayers with income around
the threshold. For example, if a at rate of, say, 14% were levied on incomes equal to or in excess of R30 000,
a taxpayer with an income of R30 001 would be taxed at an inordinately high rate of 14%, whereas a person
with an income of R29 999 would have no tax liability.
e distributional and work incentive effects of the at rate tax are not obvious and Keen et al. (2008:
722–732 and 741) show they are somewhat complex. e reason is that the progressivity (or lack of equity) of
the at rate tax must be compared to that of the system it replaces. In addition, the tax revenue yield and
implications for compliance must be considered. Empirical evidence and the practice in at rate income tax
countries of allowing for a personal income tax allowance show that the equity impact is not unambiguously
adverse for the low-income group. e impact on work incentives is also not clear-cut and there is no
evidence that high-income taxpayers improve their work effort signi cantly. Complexity is somewhat
reduced and the potential for arbitrage is lessened, but this is not necessarily owing to the rate structure
itself; it is the result of fewer exemptions and special treatments. Finally, the sustainability of the at rate tax
movement is unclear: the low (or lower) rate imposes constraints on revenue and the framework does not
provide adequately for internationally mobile capital income (see Keen et al., 2008: 742).

14.5.2 The dual income tax

Another alternative to the progressive comprehensive income tax system is the dual income tax system
(DIT). Variants of the DIT were introduced in the 1990s in the Nordic countries and this system is now also
being suggested as an option for developing countries.11 e potential for introducing the dual income tax in
developing countries is discussed in great detail in Bird and Zolt (2010).
In contrast to the comprehensive income tax system, which includes all income minus the costs to
acquire and maintain that income in the tax base, the dual income tax is a ‘schedular’ tax system. is
means that different types of income are taxed separately and at different rates. e DIT has the following
characteristics:
• Personal capital income is taxed separately from labour income and is levied at a at or single rate.
Under a pure dual tax system, the single proportional rate is set equal to the lowest income tax rate (the
rst income tax bracket) on personal (wage) income. Personal capital income includes interest,
dividends, capital gains, rental income, royalties, imputed rent on owner-occupied housing, accrued
returns on pension savings and pro ts from personal businesses (small enterprises). Pro ts from small
enterprises is the aggregate return on capital invested and labour provided by the owners of sole
proprietorships, closely held companies and partnerships. An important element of the DIT is the
taxation of capital income on a broad basis and at a low at rate to ensure uniformity and neutrality in
capital income taxation.
• Labour income (wages, salaries and transfers from government) or non-capital income is subject to a
progressive rate structure. It provides for tax deductions and exemptions to achieve equity objectives.
Overall, the taxation of labour income is therefore at higher rates than the taxation on capital income.
• Company income (incorporated businesses) is taxed at the same rate that applies to personal capital
income. In a pure system, the company income tax is fully integrated with the personal income tax to
eliminate double taxation. is can, for example, be achieved using the imputation method, whereby
shareholders receive credit for company tax paid. e company tax thus becomes a simple withholding
tax at source (see Section 14.2.2).

e DIT system was implemented in Sweden, Norway, Denmark and Finland in the early nineties, and a
number of other industrialised countries have subsequently adopted forms of the dual income tax system.
Uruguay is one of the few developing countries that have moved to a dual income tax system. e reasons in
favour of the DIT were speci c to Nordic countries at that time, but still have relevance. ese are listed
below. In addition, there are convincing economic and administrative considerations.
Firstly, the Nordic countries were feeling the impact of globalisation and the international mobility of
some forms of capital (for example, portfolio investment). Because of their comparatively high tax rates,
investors were moving their capital to lower tax jurisdictions. High in ation also meant that effective tax
rates on capital were high, which led to arbitrage actions. High-income taxpayers could exploit tax
preferences for capital investment using, for example, interest rate deductions to lower their earned income.
is eroded the labour income tax base. Because of high unemployment, political support was mobilised by
decision-makers who were in favour of a dual tax arguing that lower tax burdens on capital would increase
economic activity and thus reduce unemployment. A dual tax left marginal tax rates on labour income
untouched, which would not have been possible to achieve from an equity point of view. e combination of
a lower at tax rate on capital income and a progressive tax on labour income was therefore considered to be
a pragmatic way of dealing with problems of capital ight, tax arbitrage activities and in ation.
Secondly, the economic argument for lower tax rates on capital income and progressive rates on labour
income needs to be considered (see Boadway, 2005: 913–922). e basic efficiency argument is that the
more elastic the source of income is, the more appropriate a lower tax rate would be. Capital income has the
characteristics of being more elastic than labour income. In addition, it is mobile and is used for riskier
investments. High capital income tax rates, therefore, will lead to evasion and discourage investment. Other
efficiency arguments claim that investment generates externalities that should be encouraged by having
lower tax rates on savings. e broadening of the capital income tax base and the imposition of a low at rate
reduce the distortions associated with the differential treatment of different sources of income (for example,
avoidance through tax planning, shifting of assets or selecting a business form on tax grounds).
Levying a single (proportional) tax on people who choose to save reduces discrimination against savings.
Taxing capital income (for example, interest) at progressive rates (as under a comprehensive income tax
system) can cause horizontal inequities because some individuals use savings to smooth their income over
their life cycle. An individual who saves for a rainy day or retirement will end up paying a higher lifetime tax
bill than a person with similar earnings who does not save. Taxing labour income at a progressive rate is
easier to justify than a corresponding tax on capital income.
Taxing labour income causes some distortions as it affects the choice between labour supply or effort
and leisure. e variability of labour income is, however, more predictable (for example, seasonable work)
and the impact can be determined based on knowledge of the worker pro le and the income distribution.
e responsiveness to progressive taxation of individuals to work and how much to work is reasonably well
known (see Section 13.4.1). e personal income tax system can be designed to compensate for individuals
who experience income variability thatis beyond control. In addition, a social security network can play a
supporting role (see Boadway, 2005: 917).12 In short, progressiveness of the tax system as a whole is
important, but not all taxes need to be progressive. It can be argued that the most efficient way to achieve
progressiveness is to tax labour income using a progressive rate schedule, a provision that continues to
characterise a dual tax system. e DIT thus provides the exibility that developing countries need in order
to meet the international competition for capital and to maintain the progressiveness generated by the
personal income tax system. However, personal income tax rate schedules have become atter globally,
which means that the advantage of exibility of the DIT is somewhat dissipated (see Bird & Zolt, 2010: 201).
irdly, from an administrative perspective, the dual income tax system has the advantage that capital
taxes are rationalised by treating all income from capital uniformly. e base is broadened, and the
accompanying removal of tax preferences and exclusions simpli es the personal and company tax system,
in addition to reducing tax compliance costs.
Although much can be said in favour of the dual income tax system, it also has a number of
disadvantages. One of the most important is that the DIT requires the splitting of income of active owners of
small rms into a capital and labour component. is administrative challenge is the Achilles heel of the
system (see Sørensen, 2007: 566; Sørensen 2005: 780–781). As noted above, pro ts from small enterprises are
the aggregate return on capital invested and labour provided by the owners of sole proprietorships, closely
held companies and partnerships. If capital income is to be taxed at a different rate from labour income,
pro ts must be split into the two components. If it is left to the owners to decide, arbitrage opportunities
arise or are sought; that is, attempts are made to reduce the tax liability. For example, if the tax rate on capital
income is lower than the marginal tax rate of the taxpayer, the owner will attempt to have pro ts declared as
capital income (for example, dividends) or capital gains on shares. e problem becomes even more acute if
the small enterprise makes losses: is the loss to be treated as labour income or capital income? In the Nordic
countries, income-splitting rules are applied. is is done by imputing a rate of return to business assets
(classi ed as capital income) and treating the residual pro ts as labour income. e imputed rate of return
provides for a ‘normal’ return on capital (for example, the interest rate on government bonds) as well as a
risk premium to compensate investors for their exposure to risk and a pure (excess) pro t part (referred to as
rent). One problem related to income splitting is the practice by entrepreneurs of including low-yielding
assets such as motor cars and real estate (used for private consumption) in the asset base. In effect, this
increases the part of the business pro t imputed as capital income (the asset base to which the imputed
return is applied increases) and is thus taxed at the lower capital income tax rate. e income-splitting
mechanism for self-employed and other small entrepreneurs, closely held companies and small companies
remains imperfect, and requires all kinds of special tax rules and anti-avoidance measures (see Sørensen,
2005: 777–801), thus increasing the cost of tax administration.
A second disadvantage of the dual income tax system is the unresponsiveness of the company tax rate to
international competition. Remember that the company tax rate is set equal to the capital income tax rate,
which is pitched at the lowest personal income tax bracket. If this rate is higher than global rates,
investments and pro ts may be shifted between jurisdictions. International mobility of capital requires a
large degree of cooperation between nancial institutions and governments to police these movements. In
developing countries, the alternatives for a company rate equal to the lowest income tax bracket would be to
set the company tax rate either higher or lower than the personal income tax rate for income from capital.
e choice would depend on whether the business is, for example, in resource extraction, which generates
location-speci c excess pro ts. In this case, the company tax could be higher. On the other hand, if small
rms are to be encouraged to incorporate or formalise their activities (developing countries often have a
large informal sector), the rate must be set lower. In the latter case, incentives are unfortunately again
created for small entrepreneurs to reduce their tax liability (see Bird & Zolt, 2010: 203–204).
irdly, from a vertical equity perspective, lower tax rates on capital may be viewed as unfair since
capital income is earned mainly by the rich. Boadway (2005: 915–916 and 924) argues that this is a awed
argument, since capital income must eventually be consumed and will then be taxed using the consumption
tax system. On the other hand, the negative vertical equity effect can be mitigated by higher personal income
tax rates on the labour income of the well-to-do and by taxing the inheritances of the rich.
Measuring capital income remains a problem. Included are accrued returns on human capital
investment, imputed returns on consumer durables (for example, owner-occupied housing), the return on
investment in small businesses and accrued capital gains. If capital income is taxed at a low uniform rate but
excludes these hard-to-measure incomes, there are still opportunities for avoidance and evasion, since
returns between assets are distorted.
In designing tax systems, it is almost inevitable that governments will be faced with policy con icts and
goal choices. Tax fairness, efficiency, neutrality and administrative simplicity cannot always be achieved
simultaneously, and must sometimes be traded off. e dual income tax system appears to sacri ce
neutrality; income from capital is taxed differently from income from labour. Vertical equity is sacri ced for
horizontal equity, but is preserved for tax on labour income. Efficiency gains can be achieved by taxing all
capital at the same low rate. Countries face different policy objectives, depending on their economic and
political considerations.

Key concepts
• basic tax on companies (page 296)
• capital gains (page 308)
• classical system or partnership approach (page 296)
• dividend tax (page 297)
• dual income tax (page 314)
• flat rate income tax (page 312)
• full integration system (page 296)
• income-splitting rules (page 316)
• labour income (page 314)
• marginal effective tax rate (page 302)
• personal capital income (page 314)
• secondary tax on companies (STC) (page 297)
• tax holidays (page 305)
• tax incentives (page 304)
• user cost of capital (page 302)

SUMMARY
• e company income tax base is simply total revenue minus certain allowable expenses. Interest costs
are deductible, but not dividends, which biases nancing of companies through loans instead of share
capital. Companies may also deduct depreciation. Special depreciation allowances often differ from the
true (economic) rate and lead to arbitrage opportunities.
• A modi ed classical company tax system is used in South Africa. It means that the company is taxed as a
separate unit on its pro ts. e pro ts distributed (dividends) are taxed again in the form of a dividend-
withholding tax. e dividend tax is paid by the company to SARS, but the individual shareholder retains
the ultimate responsibility for the tax liability.
• e basic tax on companies is 28%. Small and medium-sized businesses are taxed at a graduated tax rate.
Micro businesses have the option of being taxed on their presumed income (based on turnover).
Because companies are taxed at different rates, the after-tax returns differ from sector to sector. e tax-
induced distortion implies that the tax causes an excess burden (allocative inefficiency).
• Company tax raises the user cost of capital and is thus a cost to the investor. Developing countries have
attempted to reduce the user cost by lowering tax rates and providing other incentives, such as tax
holidays and generous depreciation allowances, to attract foreign direct investment (FDI). Reasons for
and against tax incentives abound, but they remain popular with the governments of developing
countries.
• Capital gains tax is classi ed as a form of income tax. If capital gains are not taxed, taxpayers have an
incentive to convert income into capital gains in order to avoid taxation. In South Africa, capital gain (or
loss) is determined as the difference between the realised proceeds from the sale of the asset and the
total base cost of the asset. Only 40% of the net gain is taxed at the individual’s marginal tax rate.
• Globally, income tax systems do not conform to the ideals of a comprehensive income tax system and are
not always fair, efficient or easy to administer. ere are two major tax reform alternatives: the at rate
income tax and the dual income tax. e at rate income tax provides for a single low rate on income
with a non-taxable personal exemption. is alternative is administratively simple, but the equity
implications and impact on work effort are not that clear. e dual income tax taxes labour income at a
progressive rate and capital income at a low at rate. e progressive rate on labour income is based on
equity grounds, whereas the lower rate on capital is based on efficiency grounds (promoting savings and
investment).

MULTIPLE-CHOICE QUESTIONS
14.1 Company income tax …
a. is based on turnover for all companies in South Africa.
b. rates are differentiated depending on the size of companies.
c. can be reduced by nancing new ventures using share capital instead of borrowed capital.
d. is levied at a xed rate to which a tax on dividends is added.
14.2 Which of the following statements regarding the economic effects of company tax is or are correct?
a. If all forms of businesses were taxed at the same proportional tax rate, it would be an efficient tax.
b. A company tax rate below 30% would attract investors to a country such as the Central African
Republic.
c. Industries are likely to be attracted to regions with a surplus supply of labour by providing a tax
credit for youth employment.
d. e incidence of the company tax is likely to be borne by shareholders in full.
14.3 A capital gains tax …
a. conforms to all of the requirements of the comprehensive de nition of an income tax.
b. is levied at the top marginal income tax rate.
c. is intended to reduce personal income tax avoidance.
d. is levied primarily for revenue purposes.

SHORT-ANSWER QUESTIONS
14.1 Why is company income taxed separately from other forms of income? Explain.
14.2 Why are the economic consequences of company tax regarded as controversial?
14.3 Explain the arguments for the integration of the individual and the company income taxes.
ESSAY QUESTIONS
14.1 e combined effect of income tax on companies and individuals implies the double taxation of
dividends. Explain the efficiency and equity aspects of this interaction.
14.2 If special provision is made in income tax law for small business or the agricultural sector, it means that
the tax is not neutral. What is your opinion on this issue?
14.3 How will income tax on companies affect the decision of an entrepreneur to invest?
14.4 ‘Tax incentives are not effective instruments for attracting investment.’ Do you agree? Discuss.
14.5 Analyse the following statement: ‘Capital gains tax is an inappropriate form of taxation for South Africa.’
14.6 What is a at rate tax and what are the reasons for the interest in this type of tax?
14.7 Compare a progressive comprehensive income tax to a dual income tax. Is either of these two taxes
appropriate for a developing economy? Why or why not?

1 See DTC (2016a) for a discussion of these recommendations and DTC (2016b) for the report on the oil and gas sector.
2 In 2017 the dividend tax rate increased from 15% to 20%, to ensure alignment with the increase in the top marginal income tax rate
to prevent arbitrage opportunities (National Treasury, 2017a: 39).
3 Rankin (2006: 11) investigated the regulatory environment of SMMEs in South Africa, and on the basis of the survey data analysed,
he found that smaller rms were more likely to mention tax regulation as a constraint.
4 e DTC (2014: 32), in their rst interim report on small and medium enterprises, indicated that 50% of VAT vendors registered had
a turnover below R1 million, even though compulsory registration was not required.
5 e Department of Small Business Development was established in May 2014 (DTC, 2016c: 5).
6 See Mohr & associates (2015: 146–148) for an explanation of these pro t concepts.
7 See World Bank (2006).
8 Another important measure is the average effective tax rate (AETR), which determines the average tax burden on an investment
(World Bank, 2015: 8). e AETR applies when rms earn a return above normal pro t on an investment, and it is important for
locational decisions (i.e. where to invest).
9 Footloose industries tend to have few constraints in deciding its location. ey have high value-to-weight ratios, are highly mobile
and locate where the availability of inputs leads to the lowest overall manufacturing costs. See Decker & Crompton (1993: 69) and
Salvatore (1998: 175).
10 It should, however, be recognised that there are people with low incomes (for example, pensioners) who derive their income
mainly from capital. is could affect our equity conclusion with regard to company taxation.
11 For a comprehensive analysis of the dual income tax system, see Boadway (2005: 910–927), Sørensen (2005: 777–801), Sørensen
(2007: 557–602), and Sørensen and Johnson (2010: 179–235).
12 Also see Boadway, Chamberlain and Emmerson (2010: 760–770) for a brief outline of the welfarist, equality-of-opportunity and
paternalism criteria for evaluating a tax system.
Taxation of wealth

Tjaart Steenekamp and Ada Jansen

In Chapter 11, we noted that taxes could be imposed on four bases: income, wealth, people and
consumption. In this chapter, we consider the wealth base. Wealth is the product of accumulated savings,
assets that have gained in value and the free gifts of nature. Richard Bird (1992: 134) argues that the existing
distribution of wealth in a country is ‘… largely the outcome of historical accident, as condoned by the state
and frozen in law. e result of this pattern of distribution of initial wealth is that many of those successful in
life stand not on their own feet but on the shoulders of their fathers.’ Wealth holdings therefore contain an
element of personal effort (self-accumulated wealth) and an element of luck (inherited wealth). ese
characteristics make wealth taxation a much-debated topic and also an emotive one, especially in countries
with vast inequalities of wealth. e aim of this chapter is to analyse the economic impact of the personal net
wealth tax, property taxes and capital transfer taxes. We begin this chapter by distinguishing between
different types of wealth (Section 15.1). In Section 15.2, we investigate the reasons why wealth is taxed. e
taxation of real property, which is an important type of wealth, is studied in Section 15.3. e chapter
concludes with a discussion of capital transfer taxes such as estate duty and gift taxes.

Once you have studied this chapter, you should be able to:
define the wealth tax base
explain the merits and shortcomings of taxing personal net wealth
define the property tax base
describe property tax rating and assessment
explain the effect of property tax on equity using the benefit principle
analyse property tax incidence using partial and general equilibrium analysis
discuss the efficiency effects of a property tax
define a capital transfer tax
discuss the economic effects of capital transfer taxes.

15.1 Wealth and types of wealth taxes

Income and consumption are ow concepts since both are measured over a period of time. Income consists
of wages, rental income from property, interest on savings, dividends on shares and so on. In contrast to
income, wealth is a stock concept that is measured at a particular point in time.
 Wealth is the value of accumulated savings, investment, gifts and inheritances. If a person does not save
or receive inheritances or gifts, he or she will never accumulate wealth. A person’s wealth consists of the net
monetary value of assets owned. Another and technically more correct de nition is that personal wealth is
the present value of a person’s expected real income. Personal wealth includes tangible things such as
houses, durable goods (for example, motor cars, jewellery and valuable paintings) and land. In addition,
individuals hold nancial assets such as cash, deposits in bank accounts, shares in businesses and
government bonds. ese are all assets that can be traded in the market. We can also identify other items,
such as insurance policies and pension rights, although these types of assets are difficult to trade in the
market and to value. Human capital acquired through investment in education and training should also be
included as (intangible) forms of personal wealth, although valuing human capital is obviously difficult. A
person’s wealth must also take account of any liabilities, since assets are often acquired through incurring
debt. By subtracting liabilities from assets, we obtain the net value of assets (in other words, personal
wealth), which is also called net personal worth.
e wealth tax base is not restricted to personal wealth; company wealth should also be considered.
Company wealth includes different forms of capital such as xed capital (premises, plant and machinery),
oating capital (raw materials and inventories) and nancial capital (stocks and shares, cash and bank
deposits). It is not difficult to see why the term ‘capital’ is often used synonymously with company wealth. To
these assets, we should, in principle at least, add intangibles such as goodwill, brand name and market
power and intellectual property rights. As in the case of human capital, it is difficult to value these assets. To
arrive at net company wealth, we must subtract liabilities from the gross value of assets.
e taxation of the wealth base has a long history dating back to a form of property tax that was
introduced in ancient Rome. e most important types of wealth taxes today include:
• Annual wealth taxes (for example, on persons and/or companies)
• Property tax (for example, tax on land and improvements)
• Capital transfer tax (for example, tax on estates and gifts).

15.2 Why tax wealth?

e case for a wealth tax is best presented by considering the arguments in favour of an annual wealth tax on
individuals (also called a personal net wealth tax or net worth tax).1 e arguments for a net or gross assets
tax on businesses are somewhat different and will not be addressed here.2

15.2.1 Equity considerations

When we discussed fairness as a criterion of a ‘good’ tax in Chapter 11, two principles emerged: the bene t
principle and the ability-to-pay principle. A wealth tax is often justi ed in terms of the bene t principle.
Governments provide public services that increase the value of real assets (for example, property).
Governments also provide protection for property (for example, law enforcement and legislation).
Consequently, owners of assets should pay for the bene ts (expenditures) of the protection they receive. It
may, however, be difficult to determine the extent of the bene ts received. Some properties, for example,
need more protection than others.
As far as the ability-to-pay principle is concerned, monetary income, which may include income arising
from realised or unrealised capital gains, is normally subject to income tax (see Chapter 14). From the
horizontal equity perspective, however, wealth confers an additional ability to pay on the owners of wealth
(over and above the monetary income derived from wealth itself and from work). Consider the example of a
poor person with almost no income and a wealthy person who keeps his or her wealth in the form of
Krugerrands and Persian carpets. Surely, even if the wealthy person has no income, he or she is bound to
have greater taxable capacity on account of the potential income-generating capacity of the assets than the
poor person. Besides, because wealth owners have the ability to realise assets, they have economic security
that enables them to exercise purchasing power (for example, by obtaining loans). Wealth also provides
them with the power to exploit economic opportunities.
 From a vertical equity point of view, people with different taxable capacities should be taxed differently.
Wealth is generally distributed very unevenly and it is argued that wealth taxes could reduce this skewness.
e distribution of personal wealth in South Africa is also very uneven. Until very recently, there have been
no reliable estimates of the distribution of wealth (assets) for the whole of South Africa since 1994. At a
regional level, the Gini coefficient (see Section 8.1 of Chapter 8) in respect of the distribution of wealth in the
former Transvaal in 1985 was estimated at 0,67 (Van Heerden, 1996).3 is distribution, however, was not
more skew than in other countries (for example 0,81 for the United States in 1983, 0,68 for the United
Kingdom in 1968 and 0,52 for Australia in 1968). What makes South Africa’s distribution different is its racial
dimension. In 1985, whites comprised approximately 36% of the population of the Transvaal, but owned
approximately 91% of the wealth. In 2015, Orthofer (2016) used two data sources4 to estimate the
distribution of wealth in South Africa. Her main ndings were that wealth is more unequally distributed
than incomes: ten per cent of the population owns about 90–95% of all wealth in South Africa. e estimated
Gini coefficient is close to unity at 0,95. ese results are supported by the research of Mbewe and Woolard
(2016), who conducted a similar analysis5 and found wealth inequality to be extremely high with an
estimated Gini coefficient above 0,90.
According to these equity principles, it seems fair to tax wealth; however, we need to consider the
incidence of a wealth tax as well. Remember that wealth is mostly accumulated savings. A tax on wealth is
therefore largely a tax on savings. e incidence of a wealth tax depends on the price elasticity of the supply
and demand for savings. If the supply is perfectly inelastic, the tax is borne by the owners of capital (wealth)
and, since wealth is concentrated in the hands of the middle- and upper-income groups, the tax is
progressive. If, however, the supply of savings is not perfectly inelastic, the incidence becomes complex, but
it is probable that wealth owners would then be able to pass on some of the tax to consumers and workers.

15.2.2 Efficiency considerations

Wealth taxes may affect economic efficiency positively or negatively. On the positive side, a net wealth tax
has a smaller disincentive effect on work effort than an income tax. A tax on wealth is levied on past effort,
whereas income tax is levied on current effort. Furthermore, in the absence of a wealth tax, a person may
invest in assets that do not yield income to avoid paying income tax. A wealth tax would counteract this type
of tax avoidance to some extent. A wealth tax might therefore serve as an incentive to wealth holders to put
their assets to productive use. We will address this issue when land taxation is considered in Section 15.3.5.
On the negative side, wealth taxes may affect saving because the tax base (wealth) is mainly accumulated
savings. A tax on saving will also translate into a tax on labour, that is, insofar as the purpose of work is to
save (for future consumption). When the tax becomes a disincentive to work, efficiency is lost. In addition,
although wealth taxes may encourage a more productive use of assets, they may also have the opposite
effect. High yield and efficiency are not necessarily the same. Assets are sometimes invested in ventures that
yield little or no pro t over the short term (for example, investments in plantations). A wealth tax on these
assets would cause an undue liability on its owners and ultimately on the community or economy, especially
if taxes have to be paid when wealth increases, even though there is no cash ow.
A discussion of the efficiency of wealth taxes would be incomplete without reference to the excess
burden of taxes of this nature. Wealth taxes reduce the net return on saving (that is, the after-tax interest
rate). Since wealth taxes distort the relative prices of goods that can be consumed presently and goods that
can be consumed in the future, an excess burden results. e size of the excess burden will depend on how
sensitive saving is to changes in interest rates.

15.2.3 Revenue and administrative considerations

Similar to all taxes, wealth taxes can be introduced to generate revenue, although this is not a particularly
important consideration. In fact, in most countries where this type of tax is in force, the net wealth tax
revenue rarely exceeds 1% of total tax revenue.
Wealth taxes do have administrative advantages. e database for these taxes can be used to cross-check
income tax returns. Since the income tax base is often eroded by tax avoidance and evasion schemes, wealth
taxes may complement income tax systems by curtailing these schemes. On the other hand, wealth taxes
also have administrative shortcomings. In the past, wealth was relatively easy to tax since it was held in very
visible forms (for example, immovable property). Nowadays, wealth holdings and their valuation have
become very complex (for example, derivatives such as options and swaps) and are often held in foreign
countries. is calls for a sophisticated and costly tax administration, which deters many countries from
introducing taxes of this nature.
It appears as if the trend is to move away from taxes on wealth. In 1990 twelve Organisation for Economic
Co-operation and Development (OECD) countries levied tax on individual net wealth. is number dropped
to three OECD countries by 2008. After the nancial crisis Iceland and Spain re-introduced net wealth taxes
on a temporary basis as scal consolidation measures; in 2017 only four countries (France, Norway, Spain6
and Switzerland) still levied these taxes (DTC, 2018b: 37). Other developed countries, such as Japan and
Ireland, also abandoned the net wealth tax for reasons of equity, complex administration and impeding
economic growth. Colombia introduced a net wealth tax in 1935, but recently repealed it. In South Africa,
the Margo Commission (1987: 310–322) and the Katz Commission (1995: 50–55) both considered an annual
wealth tax, but concluded that this type of tax (including an inheritance tax) should be avoided owing to the
administrative and compliance burdens involved. e Katz Commission (1995: 50) argued that
redistribution is better achieved by other means, particularly through the expenditure side of the budget.
More recently, the Davis Tax Committee (DTC) considered the feasibility of a recurrent net wealth tax. ey
recommended that more work be done to ensure a well-designed tax, one that will raise more revenue than
what it would cost to administer the tax (DTC, 2018b: 7). In a survey of the issues on the economics of taxing
net wealth in the OECD, Schnellenbach (2012) concludes that the equity, efficiency and political arguments
for taxing wealth are generally quite weak. ere are also hints of a negative effect on economic growth.

15.3 Property taxation

e property tax base can be de ned very broadly to include real property (realty) and personal property
(for example, furniture, motor vehicles, shares, bonds and bank deposits). We will focus exclusively on real
property, partly because it is the most common one. A tax of this nature is an impersonal (in rem) tax.
Property tax can be levied at the national level and the provincial government level or local authority
level. In South Africa, it is collected mainly by (urban) local authorities. National government taxes on
property are primarily in the form of transfer duties (payable by the person who acquires a property), and tax
on donations and estates. In this type of case, the tax is levied when immovable property is alienated or
acquired in terms of a donation or an inheritance. However, the property taxes levied by local authorities are
by far the most important form of property tax in South Africa. Property tax is a major source of revenue for
local authorities.
Property tax revenue to the tune of R63,1 billion was generated in 2017/18. Property tax represented
approximately 16,7% of the cash receipts from operating activities of municipalities (South African Reserve
Bank, 2018b: S-72). Important as this source may be for local governments, its relative insigni cance as a
national revenue source is illustrated by the fact that it would contribute only about 4,4% to the total tax
receipts of the consolidated general government.

15.3.1 The tax base

e real property tax base includes land (farm, residential, commercial and forest land) and capital
invested in improvements (farm buildings, homes, business buildings, fences and so on). e importance of
the distinction between the two tax bases will become clear once we analyse the incidence effects of a
property tax.
In South Africa, all land and improvements in urban areas (from metropolitan local councils to small
local councils) are rateable. In the past, property tax (collected by local authorities) was not levied on rural
land. e Katz Commission (1995 and 1998) considered the possibility of a national tax on agricultural land,
but found that a tax of this nature was not a viable option at the national level of government, either as a
means of raising revenue or as an instrument of a land reform programme. It did, however, support a rural
land tax at local authority level. It also acknowledged that a rural land tax may have certain social
advantages, such as encouraging the economic use of land.
A land tax on farmland was implemented in Namibia in 2004. e purposes of this tax are to fund the
government’s acquisition of agricultural land for resettlement of previously disadvantaged Namibians as
well as to compel absentee landlords to sell unused land to the government.
As far as the urban property tax is concerned, municipalities in South Africa had the option of choosing
between a site value rating (in other words, rating the value of the land only), a at rating (that is, rating the
value of improvements), or a composite rating or differential rating (in other words, rating both land and
improvements, but at different tax rates). In four provinces, the majority of municipalities used the site value
rating system. Of the remaining ve provinces, KwaZulu-Natal opted for composite rating, while many
councils in the other four provinces used at rating (Franzsen, 1997).
In 2004, the Local Government: Municipal Property Rates Act, 2004 (hereafter the Property Rates Act
2004) was enacted, to be implemented over a four-year period. It empowers the eight metropolitan
municipalities (for example, Tshwane, Johannesburg and Cape Town), 205 local municipalities and 44
district municipalities to levy a rate on all rateable property. Property generally means immovable property
and includes immovable improvements (excluding underground mining structures). Different categories of
rateable properties may be determined, including residential properties, industrial properties, business and
commercial properties, farm properties, smallholdings, and formal and informal settlements.

15.3.2 Tax rates

Property tax is usually levied as a percentage of the taxable assessed value of the property. e assessed
value for tax purposes is often less than the market value of the property. Owing to this difference between
the assessed value and the market value, the nominal tax rate differs from the effective rate (the effective rate
being the tax liability expressed as a percentage of the estimated market value). Tax rates generally differ
between jurisdictions as a result of the selectivity of the property tax. Since the property tax is considered
burdensome to certain categories of taxpayers (for example, elderly homeowners), these taxpayers usually
receive some kind of tax relief.
In the case of the Namibian land tax, a rate of 0,4% is levied on the unimproved site value per hectare
owned by commercial farmers. It increases by 0,25% for each additional Namibian-owned farm. If the land
is owned by foreigners, the rate starts at 1,4% of the site value for the rst farm and increases by 0,25% for
each additional property. e proceeds of the tax go into a land reform programme.
In the past, at (uniform) rates were levied in South Africa, but these rates differed among provinces and
among municipalities within provinces. e taxes were levied annually, but were generally collected in
monthly instalments. A maximum rate was prescribed in some provinces. Before 1994, a number of
categories of properties were exempted from paying property tax (for example, properties used for religious
and educational purposes, hospitals and national theatres). Since 1994, however, almost all exemptions have
been repealed and replaced by a system of grants-in-aid that could not exceed the equivalent rateable
amount. Rebates were granted for certain classes of properties (for example, residential properties) and
categories of ratepayers (for example, the handicapped, pensioners and other ratepayers with annual
incomes below a speci ed minimum).
In terms of the new Property Rates Act (2004), a rate levied by a municipality must be an amount in rand
based on the market value of the property. e valuation roll remains valid for a period of four years (which
can be extended to ve years) in respect of a metropolitan municipality and ve years (which can be
extended to seven years) in the case of a local municipality. Differential rates for different categories of
rateable property may be set. e categories may be determined according to the use of the property or
geographical area in which the property is situated. Certain categories of owners, such as pensioners,
owners temporarily without income and indigent owners, may be exempted or may be granted a rebate. e
ratio for residential property to public bene t organisation property (for example, churches and schools),
public service infrastructure property and agricultural property should not exceed 1:0,25. e rate for non-
residential property is the same as that for residential property. e right to levy rates is limited by a number
of provisions. One of these provisions stipulates that the Minister of Finance may set upper limits on the
percentage by which rates may be increased annually. Furthermore, the rate may not:
• materially prejudice national economic interest and economic activities across its boundaries
• be levied on the rst R15 000 of the market value of a residential property
• be levied on places of worship (excluding a structure used for educational instruction in which secular or
religious education is the primary instructive medium)
• be levied on property belonging to a land reform bene ciary, or his or her heirs, for a period of ten years.

15.3.3 Assessment
Probably the most controversial part of property taxation is the valuation of property. ere are a number of
different approaches to assessing the value of property, such as the capital value of land and improvements,
the site value system, the rental value of premises and the comparable sales method (see Bahl, 1998). e
capital value system attempts to determine the full market value of land and improvements (as a bundle)
on a willing buyer and willing seller basis (in other words, the value they would agree to in an open market).
In practice, the capital value is determined by assessing land independent of improvements. Land is valued
using data based on a judgemental approach, and the expert opinions of professional valuers and real estate
agents are enlisted. Improvements are valued with the aid of schedules of value per square metre (based on
building costs).
e site value system assesses the value of land only. e value of land can be based on sales of
comparable vacant land or estimated by means of a residual method (for example, by determining the
bundled value of the property, and then deducting the value of improvements).
e British tradition is to assess the value of property on the basis of some estimate of the rental value of
the property or the rental income derived from the asset. e sales value of a property tax and the rental
value are supposed to yield equal results.7 However, in developing countries, property markets may not be
adequately developed to generate plausible results, property is often underutilised and an assessment based
on the rental value would thus understate the value of the property. Idle land also generates no net income
and rental value will therefore have to be imputed in such a case.
In South Africa, the general basis of valuation will be the market value of a property, that is, the amount
the property would have realised if sold on the date of valuation in the open market by a willing seller to a
willing buyer (Republic of South Africa, 2004). e market value would be inclusive of the land value and the
value of improvements to the property. In the case of agricultural land, the values of any annual crops or
growing timber that have not yet been harvested are disregarded for valuation purposes.
All rateable property must be valued during a general valuation. Physical inspection of the property to be
valued is optional, and comparative, analytical and other systems or techniques may be used to determine
the value, including aerial photography and computer-assisted mass appraisal systems. e valuation roll
must remain valid in total for not more than four nancial years.
e taxing of farm properties could be considered a new category of property that is now rateable by
municipalities. e Property Rates Act (Republic of South Africa, 2004) provides that newly rateable property
had to be phased in over a period of three nancial years. When rating properties used for agricultural
purposes, the extent of services provided by the municipality in respect of these properties, the contribution
of agriculture to the local economy, and the impact of agriculture on the social and economic welfare of
farm workers must be considered. ese factors con rm that the property tax is not a national land tax on
agriculture.
Using market values instead of use value for property, especially in respect of farm land, is a much-
debated issue. e Katz Commission (1998) was in favour of use value as a valuation base. Use value can be
determined using the income capitalisation method, land resource quality index method and lease value or
rental value method (see Katz Commission [1995: 1–36] for a description of use value methods and their
merits). According to the income capitalisation method, use value is determined by dividing net farm
income by an appropriate capitalisation rate. A major shortcoming of this method is that income is volatile.
e land resource quality index method compiles an index using information on the quality of the land
obtained from farmers, population censuses and the Weather Bureau. Farm land is then assessed on the
quality of the land only. In terms of the rental value method, the value of land is assessed with reference to
the potential market rent that could be obtained for a unit of land. e usefulness of this method depends on
the size of the rental market. If properties in a geographical area are mainly owner-occupied, it is difficult to
determine rental value.
ere is a tendency for use values to be lower than market values, since market values are in uenced by
non-farm factors such as location and investment value. is discrepancy is most noticeable where farm
land is close to urban concentrations. e market value of farm land is determined under conditions of
supply and demand. If the farm is not actually going to be sold, it is difficult to determine market value
accurately. e comparative sales method then has to be employed. According to this method, the market
value of the property is obtained by comparing it to similar properties that have recently been sold. Another
shortcoming of the market value method is that rates based on the market value could put severe strain on
the cash ow of property owners during periods of property price booms. e proponents of use value
methods argue that these methods are more equitable from an ability-to-pay point of view compared to the
market value method, which would tax wealth. When the capacity of municipalities to appraise farm land
was considered, the Katz Commission (1998: 51) noted that municipal property valuers have more
experience in market valuation methods. In addition, not all rural land is used for agriculture. In the end, we
have to consider that, from an efficiency perspective, market values should re ect the value of the property
when used in its most productive use (or best forgone opportunity). In other words, when valuing a
property, the assessor should consider the best use of the property. If farm land close to an urban area could
best be used for residential purposes, the market value should re ect this condition. It should also be clear
that determining the highest or best use has a subjective element. For this reason, valuation rolls are
normally open for inspection and objections, and there is a right of appeal.

15.3.4 Equity effects

e fairness of a property tax can be analysed by considering the bene t principle of taxation. Further
analysis of its fairness is possible by using a partial equilibrium framework and a general equilibrium
framework to determine its incidence.

15.3.4.1 Benefit principle of taxation

According to the bene t principle of taxation, people should pay tax in relation to the bene ts (or cost of
expenditure) of public services received. At the level of local authorities, this rule suggests that owners
should pay for services that raise the value of their properties, such as pavements, tarred roads and street
lighting. A property tax on site value can then be viewed as a comprehensive bene t tax and may be
regarded as a fair tax from this perspective.
Although the link between bene ts received and the tax paid is not always clear, empirical studies
indicate that property taxes and the value of local public services are capitalised into house prices (see Box
15.1 for the meaning of the term ‘capitalisation’). e implication is that we can expect property taxes to
depress property values, but that this effect could be countered by the positive effect of public services
nanced by these taxes (Rosen & Gayer, 2008: 525–526).

Box 15.1 The capitalisation of a property tax

The capitalisation principle can be illustrated by means of a simple example. Suppose a piece of land in Cape
Town has an expected yield of R10 000 per annum and that the current interest rate (the opportunity cost of
capital) is 20% per annum. The capitalisation value is determined by answering the following question: What
amount needs to be invested to earn this return? This amount can be determined by using the formula for
capitalised value. The capitalised value (CV) is income from the property (R) divided by the interest (i):
. Thus, the capitalised value of the land will be R50 000 . Let a property tax of R1 000
now be levied. This would be equivalent to a tax on the income from the land at a rate of 10% (R1 000 as a
 percentage of R10 000 = 10%). The capitalised value of the land changes to , where t is the tax
rate (equal to 10%) or CV R45 000. The value of the Cape Town property has been reduced
by R5 000. When the owner wants to sell the property, he or she will have to absorb the loss since the new
buyer will want to obtain a net return (after the R1 000 property tax) of 20% on the investment (= R9 000).

Property taxes based on land values have until now not been considered as bene t taxes (or user
charges) because bene t taxes are usually linked to speci c services (for example, access to a public park or
bus fees for public transport). However, there are taxpayers who are eligible for, but do not use, these
services. Consider the example of a childless couple: although an activity centre in a public park may affect
the value of their property, they derive no further bene t from it. Taxing them at the same (uniform) rate as
other property owners would be unfair. erefore, it can only be said that property tax is a bene t tax if the
revenue from property tax is spent on infrastructure and activities that genuinely bene t property owners as
a group (individually and collectively).

15.3.4.2 Incidence of a property tax

By now, we know that information on the statutory (or direct) burden of a tax is insufficient to allow us to
arrive at conclusions about the fairness and distributional implications of that tax. We also need to identify
the economic incidence of the tax. e incidence of property tax can be studied using partial equilibrium
analysis and general equilibrium analysis.

Partial equilibrium analysis

When property tax is analysed within a partial equilibrium framework, it is regarded as an excise tax that
falls on land and on the capital invested in improvements.

Figure 15.1 Incidence of an ad valorem property tax on land

• Property tax on land: Since the quantity of land in a country is xed, the supply of land is in effect
perfectly inelast ic. is means that landowners cannot increase the quantity of land if prices increase.
All they can do is to change the use of land and, to a very limited extent, reclaim parcels of land (for
example, swamp land). In Figure 15.1, the supply curve (S0) is vertical and the demand curve is D0. e
equilibrium rental before tax is P0 and the equilibrium quantity is Q0. Suppose that an ad valorem
property tax (t) is now introduced. e before-tax demand curve D0 swivels downwards to D1, which is
the new effective (after-tax) demand curve (the demand curve as perceived by landowners). e rent
paid by the users of land remains unchanged at P0. However, the rent received by the owners of land is
reduced by the tax to P1. e total tax revenue is equal to the area ABCD and is borne entirely by the
owners of land. us, the incidence of a tax on land is on the landowner. Since income from land
ownership increases as income increases, the tax falls relatively more on taxpayers with higher incomes.
is makes a land tax progressive and, from the ability-to-pay point of view, the tax is therefore fair. e
lower rent on the land will cause the price of land to fall in order to account for the future tax burden;
taxes are capitalised into the value of land (see Box 15.1). When subsequent buyers make property tax
payments, these payments should not be seen as a burden in the true sense of the word, since the
payments have already been capitalised into a lower purchase price. In other words, the incidence of the
property tax is only on the initial owner of the land (that is, the owner at the time when the tax was
imposed). Should property taxes be increased, the increase would again be capitalised. e extra tax
burden would fall on whoever owns the property at that time. e determination of the incidence thus
becomes quite complicated, since the identities of the owners at the time when the tax was imposed or
increased must be known.
• Property tax on improvements: In contrast to land, which is xed in supply, the capital invested in
improvements is elastic in the long run. e argument is that in the long term, investors can nd all of the
capital they require at the market price. When market rents are taxed, it is possible for property owners to
shift the burden to the users of property (tenants and, indirectly, consumers and workers). is is
illustrated in Figure 15.2. e supply of capital for improvements is the perfectly elastic curve S0. e
before-tax equilibrium is at Q0 and P0. If an ad valorem property tax is now levied on property, the
effective (after-tax) demand curve becomes D1. is is the demand curve as perceived by property
owners. At the after-tax equilibrium, property owners still receive P0, but the users (for example, tenants)
have to pay P1, which includes the property tax. Note that the after-tax equilibrium quantity of property
supplied is lower. Property owners are therefore able to shift the full tax burden (= ABCD) of a tax on
improvements to users. e distributional implications depend on who the users are. If the users are
lower- to middle-income tenants or owner-occupiers of residential property, the distribution tends to be
regressive. e reason is that housing expenditures decrease with increases in income; that is, low-
income taxpayers spend relatively more on housing than high-income taxpayers. If the users are tenants
of commercial property, the higher after-tax rents will be re ected in higher prices for goods and
services. Since consumption as a whole decreases as income increases, the regressive effect of a tax on
the improvements part of property is again emphasised. However, since these effects disregard dynamic
general equilibrium effects, the question of who bears the burden of the property tax is not fully resolved.

General equilibrium analysis

In a general equilibrium framework, the full effect of the tax is traced throughout the economy. Viewed from
this perspective, a property tax levied at different rates is equivalent to a selective tax on capital. In the short
term, the tax will be capitalised and the value of property will be lowered. Since land and the capital invested
in improvements are immobile in the short term, the incidence of the tax is on the initial owners of the
assets. e burden is distributed progressively because property is not only heavily concentrated in the
hands of the wealthy (particularly in developing countries), but income from ownership of property tends to
be an increasing proportion of income.
Figure 15.2 Incidence of an ad valorem property tax on improvements

In the long term, the capital invested in improvements is more mobile. Capital then migrates from high
tax rate areas to low tax rate areas. e stock of capital is reduced in the high tax rate area (increasing the net
return on capital in this area), while the stock of capital in the low tax rate area increases (thereby reducing
the net return of capital in this area). eoretically this process continues until net (that is, after-tax) rates of
return are equal throughout the economy or region. erefore, even though the tax was imposed on capital
in one area, the burden is shared by capital owners in other tax jurisdictions. e process does not stop here.
When capital moves out of an area in which labour is immobile (often the case with unskilled labour), the
labour in that area must now be combined with the reduced stock of capital in production. e capital–
labour ratio decreases and this adversely affects the productivity of labour, with the result that real wages
decrease. us labour also bears part of the burden of the property tax, which adds another regressive
element to the equity equation.

15.3.5 Efficiency effects

Note that the property tax is a selective tax on wealth. Elements of selectivity include the following:
• Only immovable property is taxed.
• e tax base can include land or improvements.
• Tax bases and rates can differ between tax jurisdictions (for example, between urban and rural areas and
between different urban areas and rural areas).

Selective taxes have non-neutral allocation effects. As discussed in Chapter 12, tax non-neutralities can have
positive or negative in uences on resource allocation. ese are some of the effects of selective property
taxation:
• e industrial and residential locational decisions of entrepreneurs and individuals are in uenced by tax
differentials. ese differentials cause a reallocation of resources away from jurisdictions with high
 property tax rates to those with low property tax rates. is kind of regional competition based on tax
differentials can promote efficiency if, for example, agglomeration bene ts8 are created. However, it can
also be very disruptive in areas where labour is not very mobile. A lower capital–labour ratio could
depress already low wages even further or could result in unemployment.
• High property tax rates on improvements may discourage investment in the taxed property and lead to
speculative purchases of low-taxed unimproved land. Large pockets of unimproved land in developed
urban areas are inefficient since infrastructure (such as roads and telephone lines) has to be provided for
the greater area.
• High property tax rates on land may encourage more efficient use of land. A tax on land, for example,
could serve as an incentive to owners of land to develop it to its most pro table use. Badly managed land
or idle land that is taxed may generate no net (after-tax) income. e owner will have to improve the
productivity of the land or sell the land to someone able to do so, or else the owner may go bankrupt.
• Property taxes on improvements may have perverse effects. As mentioned in Section 12.1, in 1695, a
window tax based on the number of windows in a house was introduced in England. Of course, large
houses tend to have more windows. e tax proved to be very unpopular and many households simply
bricked up their windows to avoid payment; the tax was nally abolished in 1851 (Richardson, 1986).

15.3.6 The unpopularity of property taxation

Property taxation is known to be particularly unpopular (Rosen & Gayer, 2008: 527–528; Skinner, 1991).
Why?
• Property taxation is a tax on unrealised capital income, which may cause solvency and liquidity
problems for some taxpayers.
• e burden of the land part of a property tax is on the owner of the property (when the tax is rst
introduced). Since land ownership is mostly concentrated in the hands of a few property owners and
ownership is often coupled with political power, the tax is ercely resisted.
• e land part of the property tax increases the riskiness of farming, which is already subject to external
risks and seasonal uctuations in income.
• e administration of property taxes has shortcomings. Properties and the liable taxpayers must be
accurately identi ed. e taxation of, for example, communal land is problematic. Each parcel of land
must be valued. is process is difficult and sometimes subjective. Since valuation is not done regularly,
there is often outrage when values are updated. Property taxes are generally viewed as regressive and
unfair. In addition, property taxes are highly visible forms of taxation and relatively easy targets for
political action.

15.4 Capital transfer taxes

Capital transfer taxes are taxes on inheritances, estates and gifts. Inheritance tax and estate duties are
called death duties. e estate duty is donor-based. In other words, it is imposed on the estate out of which
the legacy has been made. e liability is determined on the basis of the size of the total estate. Inheritance
tax is donee-based (in other words, it is levied on the receiver of the legacy). e tax liability is determined
by the size of the wealth transfer. Donations (gift) tax is levied on gifts. e reason for this tax is that estate
duties can be avoided if a tax of this nature does not exist. Individuals would simply transfer wealth before
their deaths. e donations tax is levied on the donor and certain exemptions are usually allowed.

15.4.1 Economic effects of capital transfer taxes

e economic effects of capital transfer taxes can be analysed using the now familiar properties of a ‘good’
tax introduced in Chapter 11.
15.4.1.1 Equity issues
Capital transfer taxes are viewed as important tax instruments in reducing wealth inequalities, particularly
vertical inequalities. Inheritances tend to perpetuate wealth inequalities. In this respect, the inheritance tax
has an advantage over estate duties since it targets the source of inequality. Inheritance tax is also fair to the
extent that it is levied according to the donee’s ability to pay. Taxing the donee may also serve as an incentive
to the donor to spread his or her wealth in order to minimise the tax burden of donees.
To comprehend the impact of death duties on wealth distribution fully, it must be realised that the act of
leaving inheritances in itself can reduce wealth inequalities. Consider the example of a father leaving his
worldly possessions to four low-income earning children with no accumulated wealth. Inter-generational
inequality is reduced in this manner. erefore, by taxing each child on his or her inheritance, the promotion
of intergenerational equality of wealth is thwarted.

15.4.1.2 Efficiency issues

Capital transfer taxes affect an individual’s choice between current consumption and saving (or future
consumption). If bequests (accumulated savings) are taxed, the individual must give up more current
consumption (in other words, save more) to maintain a given estate size. is is the familiar income effect of
a price change. However, in addition to the income effect, there is also a substitution effect. Capital transfer
taxes increase the opportunity cost of accumulated wealth (that is, savings), causing an increase in current
consumption (that is, lower saving). e nal impact of capital transfer taxes therefore depends on the
relative strengths of the two effects, causing the outcome to be uncertain.

15.4.1.3 Administrative issues

Capital transfer taxes are probably easier to administer than an annual wealth tax as the estate is valued
once only. By the same token, estate duties are easier to administer than an inheritance tax. In the case of
estate duties, there is only one tax identity, whereas there could be a number of tax identities under an
inheritance tax. Possibly the most serious problem associated with death taxes is the ability of taxpayers to
minimise tax liability through various forms of generation-skipping devices, such as trusts and interest-free
loans. ese devices not only erode the tax base, but are also mainly utilised by wealthy individuals, thereby
perpetuating wealth inequalities. Tax avoidance activities are also unproductive (no new wealth is created).
Other problems with death taxes relate to the hardships they may cause surviving spouses and children.
Special provision has to be made for these categories of donees, which often complicates tax administration.
Many forms of wealth (for example, jewellery) are easy to hide and lend themselves to evasion under death
duties. Transfers made in kind are also difficult to tax under a donations tax. It is no coincidence, therefore,
that capital transfer taxes in the form of death and gift taxes are an insigni cant source of revenue in
developing countries. In 2016 these taxes comprised less than 0,1% of total tax revenues in Africa and Latin
American countries (see OECD.Stat. 2019).

15.4.2 Capital transfer taxation in South Africa

Capital transfer taxation in South Africa includes taxes on estates and gifts. A rate of 20% applies to
donations and estates up to R30 million. e rate increases to 25% when the donations and estate exceed
R30 million. A rebate of R100 000 is applicable to donations, and any donations between spouses are exempt
from tax. An abatement of R3,5 million in respect of estate duty applies and may be carried over, in some
instances, to the surviving spouse, with a combined limit of R7,0 million. If the same asset is taxable due to a
second death occurring within 10 years of a preceding death, relief applies on a sliding scale.
Estate duty is not only decreasing in importance, but is also an insigni cant source of total tax revenue.
From 1984/85 to 2018/19, estate duty as a percentage of total tax revenue (net of SACU payments) declined
from 0,42% to 0,15%. In 2018/19, approximately R2 435 million was raised in donations tax and estate duty.
 Both the Margo Commission (1987) and the Katz Commission investigated transfer taxes. In its Fourth
Interim Report, the Katz Commission (1997a) con rmed the viewpoint that a donations tax and a system of
estate duty should be retained rather than be replaced by an inheritance tax. It was argued that:
• An inheritance tax is more complicated.
• Estate duty has been in place over many years and is well documented.
• Estate duty has been the subject matter of numerous judicial decisions.
• e administrative systems are geared to the system of estate duty. e Katz Commission (1997a)
recognised that the transfer tax system had de ciencies. It recommended that there should be provisions
to deal with generation-skipping trusts that would tax capital transfers in a trust. It was suggested that net
assets of a trust be valued at intervals of 25 to 30 years (more or less re ecting a single generation).

More recently, amendments to capital transfer taxation was based on recommendations made by the DTC
(2016d). e proposals the DTC submitted to the Minister of Finance included the additional rate (of 25%)
applicable to donations and estates valued above R30 million, discussed above. In the DTC’s rst interim
report on estate duties (DTC, 2015) the following recommendation was accepted: the retirement fund
exemption from estate duties was retained, but additional retirement fund contributions made from 1 March
2015 above the annual income tax allowance are not to be taken into account when the general retirement
abatement is computed for estate duty purposes (DTC, 2016d).

Key concepts
• capitalised value (page 328)
• capital transfer tax (page 333)
• capital value system (page 326)
• composite rating or differential rating (page 325)
• donations (gift) tax (page 333)
• estate duty (page 333)
• flat rating (page 325)
• impersonal (in rem) tax (page 324)
• inheritance tax (page 333)
• market value (page 327)
• net wealth tax or net worth tax (page 321)
• personal property (page 324)
• real property tax base (page 324)
• site value rating (page 325)
• site value system (page 326)
• use value (page 327)
• wealth (page 321)

SUMMARY
• Wealth is a stock concept and includes accumulated assets (tangible and nancial). e net value of
assets is obtained by subtracting liabilities from assets. ree types of taxes often levied on wealth are the
personal net wealth tax, property tax and capital transfer tax.
• e taxation of wealth is fair in terms of both the bene t principle and the ability-to- pay principle. A tax
on wealth is a tax on savings. Consequently, the incidence effect is determined by the price elasticity of
demand and supply of savings. If the supply of savings is relatively inelastic, the incidence is mostly on
the owners of capital, who tend to be middle- and upper-income individuals. e tax is therefore
progressive.
• Although wealth taxes may encourage the more productive use of assets, they may also reduce efficiency
when individuals reduce savings and work effort. ey also have an excess burden, since relative prices
 of goods consumed presently and in the future are distorted.
• Property (realty) tax is an important revenue source for local authorities. In South Africa, property tax is
levied as a rand-based amount on the market value of the property. e market value is inclusive of the
land value and the value of improvements.
• It can be shown using a partial equilibrium framework that the incidence of property tax on land is borne
by the landowner (the tax is progressive). e incidence of property tax on improvements can be shifted
to tenants and tends to be regressive. General equilibrium analysis indicates that a tax on property will
be borne by all capital owners in the long run as well as by labour as the productivity of labour is
adversely affected by a lower capital–labour ratio.
• Property taxes distort relative prices and cause an excess burden. Although the tax may serve as an
incentive to use land more productively, it may also discourage investment in xed property.
• Capital transfer taxes include death duties (tax on inheritances and estates) and gift taxes (donations).
Taxes of this nature may reduce vertical imbalances, but impact adversely on savings, and tend to be
avoided by taxpayers by using trusts and interest-free loans.

MULTIPLE-CHOICE QUESTIONS
15.1 Wealth is …
a. income from capital gains.
b. gross personal worth.
c. interest on government bonds.
d. net value of assets.
15.2 Which of the following statements in respect of a net wealth tax is or are correct?
a. e tax is fair because the rich bene t more from public services than the poor.
b. e tax is equitable because the rich have greater taxable capacity.
c. Relative prices of savings and other goods are distorted by the tax and an excess burden results.
d. Asset values are easily determined, thus simplifying tax administration.
15.3 e municipal property tax in South Africa …
a. rates land and improvements at different rates.
b. is based on market values of the assets of property owners.
c. may reduce the value of properties when the tax is capitalised.
d. is unfair to land owners because they cannot shift the tax.
15.4 Capital transfer taxes …
a. are paid by the donor when levied as an estate duty.
b. have only an income effect and are therefore efficient taxes.
c. are signi cant sources of revenue.
d. include donations between spouses.

SHORT-ANSWER QUESTIONS
15.1 De ne a net wealth tax.
15.2 Illustrate the tax incidence of a tax on the land part of property. Who bears the burden?
15.3 Distinguish between an estate duty and an inheritance tax.

ESSAY QUESTIONS
15.1 ‘Wealth taxes tend to generate limited revenues for government and should therefore be eliminated.’
Discuss this statement.
15.2 ‘Municipal rates boycotts in South Africa would con rm the notion that a property tax is a bene t tax.’
Discuss.
15.3 ‘Property taxes on land are not only efficient, but are also equitable.’ Do you agree? Discuss.
15.4 Who really pays the property tax on improvements? Explain by using partial and general equilibrium
analysis.
15.5 Discuss the possibility that high property rates in Sandton in Johannesburg can be shifted to other
forms of capital and even to low income earners in Soweto.
15.6 A factory located in Epping, an industrial area near Cape Town, is valued at R1,2 million. e
opportunity cost of capital (or current interest rate) is 14%. Owing to the high incidence of crime, the
Cape Town Metropolitan Council introduces a once-off ‘protection levy’ of R1 200 payable by all local
property owners. Assuming full capitalisation, what will be the impact on the value of the property in
Epping?
15.7 Discuss the economic effects of capital transfer taxes in South Africa.

1 e personal net wealth tax as implemented in a number of developed countries is discussed in detail in Organisation for
Economic Co-operation and Development (OECD) (1988: 30–75).
2 For a discussion of a gross assets tax on businesses, see Sadka and Tanzi (1993: 66–73).
3 e land area largely corresponds to the present-day South African provinces of Gauteng, Mpumalanga, Limpopo and North West
province.
4 e study used the National Income Dynamics Study (NIDS) Wave 2 data conducted in 2010–2011, and an unpublished dataset of
Personal Income Tax records for the 2010/11 tax year (see Orthofer (2016) for more details on the nature of these data sources).
5 Mbewe and Woolard (2016) also used the NIDS Wave 2 data, but included the more recent data from NIDS Wave 4 (2014–2015).
6 Spain has continued to extend the wealth tax since its re-introduction in 2011 (DTC, 2018b: 37).
7 In theory, the discounted stream of net rent payments is equivalent to the capital value of the property.
8 As rms concentrate (or agglomerate) in an area, bene ts such as better access to a larger market and larger pools of managerial
talent result.
Taxes on goods and services, and tax reform

Tjaart Steenekamp and Ada Jansen

In Chapter 11, we identi ed four major tax bases: income, wealth, people and consumption. We have
already studied taxes on income (Chapters 13 and 14) and wealth (Chapter 15). In this chapter, we examine
taxes on goods and services (in other words, the consumption base). When we introduced tax equity and
efficiency in Chapters 11 and 12, we explained most of these concepts using excise taxes as examples. In this
chapter, we begin by identifying the types of indirect taxes in Section 16.1. e debate on the relative
importance of indirect taxes versus direct taxes is an ongoing one, and the advantages and disadvantages of
indirect taxes are discussed in Section 16.2. Value-added taxes have increased in importance worldwide.
is type of indirect tax is described and its economic effects are analysed in Section 16.3. In Section 16.4, we
consider the personal consumption tax, which has until now only received theoretical attention by tax
analysts. International tax reform is discussed in Section 16.5. e chapter concludes with a brief reference
to tax reform in South Africa.

Once you have studied this chapter, you should be able to:
distinguish between different indirect taxes and indicate their relative importance as sources of revenue
discuss the merits of indirect taxation
describe the consumption-type VAT
explain the economic effects of VAT
describe the personal consumption tax base
discuss the rationale for a personal consumption tax
discuss the shortcomings of a personal consumption tax
distinguish between patterns of taxation in advanced countries and developing countries
identify the direction of international tax reform
compare tax competition with tax harmonisation
contrast international tax reforms with the major tax reforms in South Africa since the late 1960s.

16.1 Types of indirect taxes

Indirect taxes are taxes that are imposed on commodities or market transactions (see Section 11.2.4 of
Chapter 11). e burden of an indirect tax is likely to be shifted. Consider, for example, an excise tax on
locally produced washing machines. Although the statutory burden is on the supplier, the consumer usually
bears the burden indirectly. Indirect taxes can be imposed at different stages of the production process: the
resource (mining or farming) stage, the manufacturing stage, the wholesale stage or the retail stage. If the tax
is collected at one stage only, it is called a single-stage commodity tax. If it is collected at more than one
stage, it is called a multi-stage commodity tax. VAT is an example of a multi-stage tax.
We can distinguish between selective (narrow-based) taxes (for example, speci c excise duties) and
general (broad-based) indirect taxes (for example, turnover tax, general sales tax and value-added tax) (see
Section 11.2.2 of Chapter 11). Excise duties are selective taxes levied on certain goods or transactions.
Excise duties can be speci c (unit) taxes or ad valorem (percentage of value) taxes (see Section 11.2.3 of
Chapter 11). Excise taxes are collected on both domestically produced and imported goods. When levied on
imported goods, they are generally known as customs duties (or tariffs). e personal consumption tax (see
Section 16.4) is also included under indirect taxes.
From Table 16.1, it is clear that VAT is by far the most important indirect tax source in South Africa
(contributing 70,8% of the revenue from domestic taxes on goods and services), followed by the fuel levy
(16,4%), which is an excise tax. Speci c excise duties are levied mainly on alcoholic beverages and cigarettes
(the so-called sin taxes), whereas ad valorem excise duties are levied on a number of luxury goods. Excises
imposed to reduce consumption of certain goods are known as sumptuary taxes (or sin taxes). Speci c
excise duties (8,8%) generate much more revenue than ad valorem duties (0,9%). Other levies included in
the total of domestic taxes on goods and service include various environmental levies, such as the plastic
shopping bag levy (twelve cents per bag), a levy on incandescent light bulbs (R8,00 per bulb), a carbon
dioxide (CO2) emissions tax on new motor vehicles, a levy applied to electricity generated from non-
renewable and nuclear energy sources of 5,5c/kWh and the international air passenger departure tax (R190
per passenger). ese levies serve as (negative) incentives to consumers to reduce greenhouse gas emissions
and to improve the environment.

16.2 Indirect taxes: A general critical assessment

Indirect taxes have a number of advantages.
• Indirect taxation is a practical way of raising revenue from people who have small incomes and those
who are not captured by the income tax net. is advantage hinges on the proposition that all citizens
should contribute to some extent to the upkeep of government (in other words, it is based on the bene t
principle).
• Taxes on goods and services are often invisible. Consumers hardly know that they are paying excise taxes,
while the inclusion of VAT in prices is mostly noted only after the goods and services have been paid for.
e advantage of scal illusion makes these taxes less susceptible to tax resistance. e downside is that
it does not conform to transparency requirements of good governance.
• e tax liability is largely determined by how much is purchased of the taxed good. In the case of
consumption goods (excluding certain necessities), consumers sometimes have a choice between
different goods and services (in other words, substitution possibilities exist). For example, if an excise tax
is imposed on golf equipment (for example, golf balls), individuals may decide rather to take up jogging
(an untaxed leisure activity). Within certain limits, taxpayers themselves can thus determine their tax
liability. To the extent that the shifting of the tax burden is possible, the tax liability would be reduced
further. In respect of direct taxes (income and company tax), the liability cannot be avoided that easily
and the substitution possibilities are also fewer (for example, in the case of income tax, breadwinners
have to earn an income). Government enforces the tax liability much more strongly in the case of income
tax.
Table 16.1 Main domestic taxes on goods and services

Sources: Budget Review 2019, National Treasury (2019a), Available: http://www.treasury.gov.za [Accessed 6 March
2019]; and SARS (2019), Schedules to the Customs and Excise Act, 1964 (Tariff Book), Available:
http://www.sars.gov.za/Legal/Primary-Legislation/Pages/Schedules-to-the-Customs-and-Excise-Act.aspx [Accessed 7
March 2019].
• Consumption taxes can be used to achieve multiple objectives. Excise taxes can be used to correct
market failures such as externalities. By levying a lower tax on unleaded petrol, environmental objectives
are promoted. High sumptuary taxes on liquor and cigarettes are in part aimed at reducing alcohol abuse
and smoking, and thus help to improve general health levels. Similarly the tax on sugar sweetened
beverages (SSBs) is intended to reduce obesity and related health problems among adults and children.
More broadly, it could be argued that it may be worth levying the highest excises on luxury goods
produced by means of capital-intensive technology and the lowest on necessities produced by labour-
intensive means in developing countries.
• Taxes on goods and services are often levied on the value of the commodity, that is, on an ad valorem
basis (for example, ad valorem excises and VAT). Tax revenue from this source automatically increases as
the price of the commodity increases and is therefore effectively indexed to in ation.
• Indirect taxes are relatively simple to administer. Consumers also have limited scope for evading taxes on
goods and services, and compliance is accordingly easier to enforce.

Analysts have also pointed to a number of disadvantages of indirect taxes.

• According to the ability-to-pay principle, broad-based taxes on goods and services tend to be regressive.
e reason is that consumption declines as a percentage of income as income increases. is conclusion
has led to the exemption or zero-rating of certain basic consumption goods and services from indirect
taxation, in particular where a broad-based VAT applies (see Section 16.3). Even excise taxes, which are
generally levied on sumptuary goods, can be regressive. According to the Bureau of Market Research
(BMR, 2017), poor households in South Africa spent approximately 2,1% of their expenditure on
cigarettes and narcotics, in comparison to the 0,3% spent by the affluent in 2017. Even if it is assumed
that cigarette prices are the same for both income groups (the poor often pay higher prices since they
buy cigarettes at in ated prices in rural areas or townships and in units, for example, a single cigarette),
the tax is regressive. e same conclusion applies to excise taxes on beer and ‘luxuries’ such as skin care,
hair and shaving preparations (Steenekamp, 1994; Katz Commission, 1994: 124).
• Indirect excise taxes are selective and lead to inefficiencies. For example, in developing countries, the
opportunity to purchase luxury goods may act as an incentive to work harder and save more. However, if
high rates of indirect taxation are levied, this may result in a substitution effect in favour of leisure,
thereby adversely affecting work effort (see Cnossen, 1990). In addition, high indirect tax rates may lead
to smuggling and black market activities. To counter smuggling, the Swedish government lowered taxes
on cigarettes and tobacco in 1998. A recent study by Ipsos (2018) in South Africa revealed that
approximately three of four cigarettes sold in small and medium businesses (such as spaza shops and
cafes) were sold at prices below the minimum tax due to SARS (which is typically taken to indicate an
illicit cigarette).
• Since indirect taxes can be levied for various purposes, there is often a policy con ict (too many goals
and only one instrument). When we considered economic efficiency as one of the properties of a ‘good’
tax in Chapter 12, we concluded that, to minimise the excess burden, commodity taxes should be high on
goods and services with an inelastic demand, and low on goods and services with high demand and
supply elasticities. is tax rule was referred to as the inverse elasticity rule. In practice, excise taxes are
imposed on luxuries for equity reasons. e demand for luxury goods, however, tends to be both price
and income elastic. Taxes on these commodities will signi cantly decrease the quantity demanded.
Application of the inverse elasticity rule means that, from an efficiency point of view, these goods should
bear low tax rates, which is contradictory to the equity requirement. us there is a policy con ict that
requires a trade-off between equity and efficiency objectives.
• Indirect taxes may represent an in ationary impulse if wages are raised in response to tax increases.
Some multi-stage commodity taxes also have a cascading effect on prices. If the tax at each stage of
production is based on the gross price up to that stage, including tax levied at earlier stages, tax is in
effect levied on tax. is may encourage vertical integration of production processes and thereby reduce
the degree of competition in the markets concerned.
16.3 Value-added tax
Value-added tax is a multi-stage sales tax levied on the value added at the different stages of production.
Roughly speaking, value added is the difference between sales and purchases of intermediate goods and
services over a certain period (normally a month). If a retailer purchased goods to the value of R150 000 from
suppliers in a month and had sales worth R300 000 in that month, the value added by the retailer would be
R150 000. e tax would then be applied to this value. e value added consists of wages, rent, interest,
depreciation and pro t. e calculation of value-added tax is illustrated in Box 16.1. South Africa’s
neighbouring countries, Namibia, Botswana, Lesotho and Eswatini (previously known as Swaziland), have
all recently introduced VAT. eir systems are broadly similar to that of South Africa, but differ in some
respects with regard to rates, exemptions, and goods and services that are zero-rated.
VAT comes in many forms. A tax authority wishing to introduce VAT therefore faces a number of choices.
We now discuss these choices and the South African practice in respect of each.1
ere are three broad types of VAT: a consumption-type VAT, an income-type VAT and a VAT on gross
product. When South Africa introduced VAT in 1991, the universal practice of a consumption type VAT was
chosen. In a closed economy with no government,

GNP = C + I = W + P + D

where GNP is gross national product, C is consumption, I is gross investment, W is wages, P is net pro t after
depreciation and D is depreciation. e consumption VAT base is then

C=W+P+D–I

e regime for international trade can be based on the origin principle (exports taxable, imports zero-rated)
or the destination principle (exports zero-rated, imports taxable). Again, South Africa decided to follow the
popular route of applying the destination principle. is is perceived to be a fair practice (domestic and
imported goods are treated the same) and one that does not affect the competitiveness of exports.
Since destination-based VAT taxes imports but not exports, tax rate differentials between countries are
normally offset at the border. is implies that production takes place in the least-cost location, or put
differently, global allocative efficiency is achieved. e destination principle, therefore, promotes neutrality
in the taxation of internationally traded commodities. Within a customs union or union such as the
European Union (EU), where border controls on trades among members do not exist or are not effective, the
destination principle breaks down. Cross-border shopping is, for example, encouraged if rate differentials
between countries are signi cant. is has relevance for the SACU (Southern African Customs Union; see
Section 16.5.4) should there be a liberalisation of border controls in this union. Alternatives to the
destination-based VAT include applying the restricted-origin principle to intra-union trades. is option
provides for a clearing house allowing for exports of union members to be taxed at origin. Importers would
be entitled to a tax credit for the VAT paid on imports. Tax-rate harmonisation could also lessen the impact
somewhat. e incentive for cross-border shopping diminishes with the distance consumers have to travel
to buy goods. e overall impact may not be that signi cant and reduces the need for rates to be
approximately the same.2
Tax liability can be computed using three methods: subtraction, tax credit (or ‘invoice’) or addition. e
tax credit (invoice) method is generally used, also in South Africa. is is the type of VAT illustrated in Box
16.1.
Two techniques can be used to free goods and sectors from VAT (see Section 16.3.1): outright exemption
(the rm need not le a VAT return and does not, therefore, levy VAT, but it also cannot claim refunds for any
VAT included in the price of purchased goods and services), and zero-rating (the rm les a return, but pays
zero tax on sales and receives a refund in respect of VAT payments made at earlier stages in the production
and distribution chain). South Africa uses both techniques. Education and health services as well as the
services of various non-governmental organisations are exempt. In addition to goods and services destined
for export, a list of basic foodstuffs is zero-rated. is list includes brown bread, maize meal, samp, mealie
rice, dried beans, lentils, pilchards, milk powder, dairy powder blend, rice, vegetables, fruit, vegetable oil,
milk, cultured milk, brown wheat, eggs and edible legumes. After the one percentage point VAT rate increase
(to 15%) in the 2018 budget, the Minister of Finance commissioned a panel to review the current list of zero-
rated items, and to make recommendations on additional items to be zero-rated. e panel recommended
that the following items be considered: white bread, cake our, bread our, sanitary products, nappies, and
school uniforms (with the proviso that a uniform be clearly demarcated)3. e 2019 budget con rmed that
three additional items would be added to the zero-rated list: cake our, white bread four, and sanitary pads
(National Treasury, 2019a: 43).
VAT can be levied at a single rate or at multiple rates (rates in addition to the zero rate). Namibia taxes
most goods and services at a standard rate of 15%. Lesotho levies VAT on most goods and services at a
standard rate of 15%. Alcohol and tobacco products are taxed at a rate of 15%, 8% on electricity, and
telecommunications at 9%. South Africa, Eswatini and Botswana have a single-rate VAT (apart from the zero
rate that is applied for equity reasons – see Section 16.3.1.3). In contrast to the 15% VAT levied in South
Africa and Eswatini, Botswana levies VAT at only 12%.

Box 16.1 Value-added tax: An illustrative example

Consider the transactions of three firms for the month of September. Agent A is an importer of bicycle
components who, for argument’s sake, is assumed to add no further value to the value of imports. Firm B is an
assembler of bicycles and Firm C is a bicycle shop. Agent A imports bicycle components to the amount of R100
000 and sells them for R100 000 to Firm B, who, after assembling the bicycles, sells them for R150 000 to
Firm C. Firm C sells the bicycles for R300 000 to the cycling public (consumers). The tax trail is shown in the
table below.

Table 16.2 The VAT tax trail

VAT is collected at different stages of the production process and, at the end of the distribution channel, the
tax is passed on to the consumer. Assuming for simplicity reasons that consumption is perfectly price inelastic,
the incidence (the burden) of the tax is therefore on consumers, but the sellers make the tax payments to
SARS. VAT is collected by the seller at the point of sale (this is referred to as the output tax). The seller may
then deduct taxes paid on intermediate products (this is referred to as the input tax). For example, Firm B
purchases components from Agent A. Included in the price is input tax of R15 000 (this amount is collected by
 Agent A and paid over to SARS). Firm B may deduct the input tax from the VAT of R22 500 (the output tax)
payable on his or her selling price. At the end of the month, the total tax liability of Firm B is R7 500 (or R22
500 minus R15 000 on inputs). Firm C’s tax liability is R22 500 (R45 000 on the selling price less input tax of
R22 500). The total VAT collected is R45 000 (R15 000 from Agent A, R7 500 from Firm B and R22 500 from
Firm C). Note that the same result can be obtained by levying 15% VAT on the value added at each stage of the
production process (remember to include the value added included in imports).

16.3.1 The economic effects of VAT

e economic effects of VAT can be analysed by considering its revenue-generating potential as well as the
familiar and important properties of a ‘good’ tax: efficiency, equity and administrative efficiency.

16.3.1.1 Revenue
Value-added tax (VAT) has a worldwide reputation of being a ‘money machine’ and in developing countries
this has indeed proven to be the case. In South Africa, VAT was introduced in October 1991 at a rate of 10%,
but the rate was increased to 14% in April 1993. e revenue importance of VAT cannot be contested. In
1992/93, collections amounted to R17,5 billion, or approximately 21,7% of total net tax revenue. In 2017/18,
the budgetary outcome amounted to R298 billion, or 25,7% of total tax revenue (net of SACU payments).
Likewise, Namibia expects 24,3% of the estimated 2018/19 tax revenue (including customs and excise) in the
form of VAT (Republic of Namibia 2018: 3). When countries are ranked according to their efforts at raising
VAT revenue, South Africa ranks 21st out of 27 countries studied (Steenekamp, 2007). More recently, an IMF
study on the South African VAT gap estimated the average C-efficiency ratio between 2007 and 2013 to be
approximately 64% (IMF, 2015: 13). e C-efficiency4 ratio is calculated by dividing the actual VAT revenue
collections by what could be collected if VAT is imposed on all consumption at a uniform rate without
exemptions (Keen, 2012: 427). It is a measure of how efficiently VAT revenue is collected. In comparative
terms, the estimated value of South Africa’s C-efficiency ratio was the third highest 5 in Sub-Saharan
countries over the same period. A further breakdown of the ratio allows the identi cation of the policy gap
(which gives an estimate of the tax expenditures associated with VAT), and the compliance gap (which
provides an indicator of the efficacy of VAT collection and enforcement, given the tax base)6. e policy gap
estimated was relatively low compared to international standards, whereas the compliance gap was
estimated at between 5% and 10% of potential VAT revenues (IMF, 2015a: 16).

16.3.1.2 Efficiency and the tax rate

In our discussion of the efficiency effects of general taxes in Chapter 12, we concluded that taxes imposed on
a broad base and at a uniform rate resemble lump-sum taxes and are efficient. In arriving at this
conclusion, we ignored equity considerations. Whether efficiency requires uniform rates is a much debated
issue. e theory of optimal taxation (see Section 12.1.3 of Chapter 12) provides convincing arguments on
efficiency and equity grounds, which refute the notion of uniformity (Newbery & Stern, 1987). e inverse
elasticity rule is one example of this line of thinking (see Section 12.2.2 of Chapter 12).
You will recall that the inverse elasticity rule states that the excess burden of selective taxes can be
minimised by taxing price inelastic goods and services at higher rates, that is, inversely proportional to the
product’s price elasticity of demand. Put more eloquently, the rule states that the excess burden will be
minimised if the proportional reduction in compensated demand that results when a set of selective taxes is
imposed is the same for all goods. When we move away from the one-person assumption of the model and
also include equity considerations, the rule calls for higher taxes on goods with low distributional
characteristics and lower taxes on goods with high distributional characteristics (goods where the share of
the poor in its total consumption is high). e important conclusion from optimal tax theory is that, from an
equity perspective, uniform taxation is not desirable. In other words, to minimise inefficiency, different tax
rates should be applied to different commodities. A case for uniform rates can only be made under certain
strict conditions. One condition is that governments should make optimal lump-sum transfers to
households. In other words, when taxes are designed, one has to consider what government does with the
tax revenue.
Using optimal tax theory in formulating policy is, however, severely restricted by data limitations. For
example, to design separate rates for each commodity requires information on elasticities and patterns of
complements and substitutes that are difficult to come by. It is also doubtful whether the efficiency gain
from designing a great variety of tax rates would outstrip the costs involved in administering a system of this
nature. One solution could be to lump together large categories of commodities and to subject them to
uniform ad valorem taxes. Another option is to achieve the desired equity objectives with a combination of
differentiated or uniform excises on luxuries and a uniform VAT rate. e lumping-together process is,
however, still analytically and empirically problematic. In the end, there appears to be some agreement that
the loss of economic efficiency owing to VAT is likely to be minimised when uniform rates or a few rates
(three or four) are applied to the broadest possible base. Moreover, if a system of income and expenditure
supports for the poor is in place (see Chapters 8 and 9), the case for uniformity in rates is strengthened.

16.3.1.3 Equity
ere is no question that a broad-based (comprehensive) VAT with no exemptions or zero-rating is
regressive. To lessen the impact on low-income households, VAT rates can, for example, be restructured
using different tax rates for commodities that are important to poor consumers (Go, Kearney, Robinson &
ierfelder, 2005). is would again affect the economic efficiency and the administrative complexity of the
system. To nd a balance between revenue, equity and efficiency in taxation, for example, pursuing equity
would imply trading off distributional and economic efficiency objectives. As mentioned, tax relief includes
exemption from VAT and zero-rating. Exemption of a good or service from VAT means that the rm or
supplier need not levy VAT on sales, but at the same time, that rm may not claim refunds of the VAT already
collected at earlier stages of the production process. e buyer of the service therefore pays VAT levied on all
but the nal stage in the production chain. Because the invoice trail breaks down when a retailer is exempt
from VAT (see also Section 16.3.1.4), there is also no way of knowing or checking – especially in remote rural
areas or even in urban areas with little competition – whether or not the exempt retailer actually adds a
margin to the price that resembles a tax at this nal stage of the supply chain. Zero-rating of a good or
service means that the rm charges a zero rate of tax on sales of the commodity and is also allowed to deduct
VAT collected at earlier stages. e buyer of a zero-rated product does not pay any VAT. None of the stages of
production is thus subject to VAT.
A major shortcoming of zero-rating is that the tax base is eroded, perhaps necessitating a higher VAT rate,
given the total revenue that the government requires. For example, the estimated revenue loss due to zero-
rating in South Africa in 1994/95 was R2 600 million. It was calculated that by abolishing zero-rating on
foodstuffs, the standard rate could have been reduced by about 1,25 percentage points without affecting the
yield from VAT. Furthermore, since zero-rated goods and services are consumed by the rich as well, they also
bene t from the zero rate. Affluent households spend substantially more in absolute terms on zero-rated
goods than less affluent households. It was estimated that of the above-mentioned R2 600 million loss in VAT
revenue, more than two-thirds of the bene t accrued to households in the top half of the income
distribution (Katz Commission, 1994: 113). Zero-rating may also lead to over taxation of suppliers who
cannot credit VAT collected at earlier stages. In South Africa, this is of particular concern to unregistered
vendors who operate in the informal sector.
Another method of reducing the regressivity of VAT is to levy multiple rates (for example, higher rates on
luxuries). is is similar to the option mentioned earlier of combining uniform VAT rates with differentiated
excises on luxuries (as is the case in Namibia, for example). is option, however, is subject to various
administrative and efficiency complications.
In its First Interim Report, the Katz Commission (1994: 133) recommended that the further erosion of the
VAT base through zero-rating or exemptions should not be considered, and that targeted poverty relief and
development programmes should receive priority. In addition, the Katz Commission (1994: 133)
recommended against higher VAT rates on luxury goods or a multiple VAT rate system. e Commission
argued that a system of this nature would make an insigni cant contribution to reducing regressivity, would
have high administration and compliance costs, and would not have much additional revenue potential. e
South African government accepted these recommendations.
In their nal report on VAT, the DTC (2018c) evaluated the structural features of VAT in South Africa.
Some of their recommendations to the Minister of Finance were as follows:
• Zero-rating of basic foodstuffs, if considered in isolation, mitigates the regressivity7 of VAT to some
extent. However, this may not be the most efficient policy tool given the gain (in absolute terms) to rich
households. e DTC (2018c) recognised the difficulties of removing zero-rating and acknowledged that
the best scenario would be to retain those items that explicitly bene t poor households, whilst removing
others disproportionately consumed by the rich. ey recommended that no additional food items
should be zero-rated.8
• e DTC (2018c) also considered the option of multiple rates (particularly on luxury items). eir key
recommendation was that there is scant evidence that higher rates improve the equity of the VAT system;
rather, it adds further administrative complexities. Furthermore, excise duties are already imposed on
selective (more luxurious) items. Given this, they recommended that multiple rates should not be
considered.
• With reference to exemptions, the DTC (2018c) acknowledged the difficulties associated with the
taxation of the nancial sector in the VAT system and in particular where services are not explicitly
charged. South Africa charges VAT on explicit services. When there is no explicit charge for the supply of
the nancial service, it renders the service VAT exempt. e problem of VAT cascading consequently
arises. Since the nancial institution that supplies the service cannot on-charge the VAT paid on inputs
to produce the service, it becomes a cost to the business that purchases the service as it cannot claim
input tax credit. Various approaches were considered to alleviate VAT cascading and reduce the incentive
of vertical integration, and the DTC recommended that National Treasury and SARS give urgent
consideration to approaches adopted by other countries in addressing this problem.

To reduce the regressive impact of VAT, tax relief could be given to the poor or transfer payments could be
directed at them. Van Oordt (2018) investigated the plausibility of taxing foodstuffs and providing cash
transfers from the additional revenues as more welfare enhancing than the zero-rating policy. He concludes
that under conditions of tax earmarking 9 and an efficient government (that increases the value of social
grants by the total additional tax revenue, i.e. a revenue-neutral policy), zero-rating could be removed.
However, Van Oordt (2018) indicates it is unlikely that both assumptions will hold (i.e. full tax earmarking
and government being perfectly efficient in raising social grants) in a developing country context. Hence, he
recommends that under these circumstances, zero-rating can be considered for food items that are not
disproportionately consumed by affluent households.

16.3.1.4 Administration
e credit-type VAT system has the reputation of being effective against tax evasion. e anti-evasion
features are its self-policing attributes, the possibilities for the cross-checking of invoices and the fact that a
large portion of tax revenue is collected before the retail stage. e self-policing feature reveals itself in the
lack of incentives for sellers and buyers to collude in making underpayments of VAT. Sellers would prefer to
understate the output tax, whereas all buyers who are not nal consumers would like to overstate the input
tax since they can reclaim it. erefore, if the seller does not pay the full VAT, it increases the VAT liability of
the buyer, who will certainly complain about it. Since VAT requires the maintenance of records of both
purchases and sales, the revenue authorities have a basis for cross-checking returns. e bene t from
collecting VAT at the different stages of the production process can be seen from our example in Box 16.1. If
the retailer (Firm C) is not a registered VAT vendor and therefore does not charge VAT on sales or claim an
input tax credit, SARS will still collect R22 500 from Firms A and B.
Opportunities for fake claims increase when goods are zero-rated, exempt or taxed at different rates. A
retailer can, for example, understate output tax by understating sales of higher-rated goods. Multiple rates
not only open up avenues for tax evasion, but also complicate administration for the tax authorities and
taxpayers alike.

16.4 Personal consumption tax

In Section 16.3.1 we saw that one of the disadvantages of a value-added tax is its regressivity. An alternative
tax on consumption that can address this shortcoming is the personal consumption tax (or expenditure
tax). e base of the personal consumption tax is income less net saving (saving minus dissaving). is tax is
collected directly from the consumer, similar to the personal income tax, and can be made progressive by
applying a rate schedule and allowing for exemptions on certain consumption items (for example, medical
expenditures). Although it looks simple enough, it is more complicated to design and implement than
income tax or VAT. Nonetheless, there is a large body of support among economists for a tax of this nature.

16.4.1 The rationale for a personal consumption tax

e proponents of a consumption tax argue that it is more equitable to tax what an individual takes out of
the economic system (as re ected in consumption) than what an individual contributes to society (as
measured by income). From this perspective, it would be considered fair that a millionaire who lives like a
miser ends up with a low current tax liability. is con icts with the ability-to-pay principle, which views
potential consumption (the power to consume) as the yardstick. e counter-argument is that the
millionaire is simply postponing the tax until he or she consumes the funds accumulated. e tax liability of
the individual must therefore be viewed over a longer period; that is, a lifetime equity perspective is
required.
It is further argued that a personal consumption tax is more efficient than an income tax. is conclusion
rests on two assumptions:
• at income tax affects saving
• at the supply of labour is xed.

A tax on saving (for example, an income tax) distorts the choice between present and future consumption.
An income tax therefore causes an excess burden, that is, in addition to the burden associated with the
choice between income (work) and leisure. In contrast, a tax on consumption does not create an excess
burden, since saving is not taxed. If the supply of labour (or work effort) is not affected by a tax on personal
consumption, it has no excess burden. If, however, a tax on consumption induces a consumer to work less
(in other words, enjoy more leisure time), it entails an excess burden. Nevertheless, there is some empirical
evidence that a consumption tax is on balance more economic efficient than an income tax.
If consumption is taxed (as leisure is too difficult to tax), the price of consumption goods increases
relative to leisure. It means that one hour of leisure (or labour sacri ced) is now equivalent to less
consumption than before; that is, the opportunity cost (or relative price) of leisure has decreased. Put
differently, a tax on consumption decreases the return to work effort. Leisure hours will increase and work
effort decreases. erefore, a tax on consumption does cause an excess burden. Since the consumption tax
base is smaller than the income tax base, the tax rate on consumption would have to be higher in order to
yield the same tax revenue. Because the excess burden of a tax increases with the square of the tax rate, the
excess burden of the consumption tax is higher than the equal-yield income tax. is efficiency loss must be
subtracted from the efficiency gain derived from not taxing savings. e net effect must then be compared to
the excess burden caused by an income tax. Whether the excess burden of a personal consumption tax is
less than that of an income tax ultimately depends on empirical evidence. Some studies show that a
consumption tax creates a smaller excess burden than an income tax and this has advanced the case of the
proponents of the personal consumption tax (Rosen & Gayer, 2008: 479).
It is also argued that a consumption tax would be bene cial to developing countries. ese countries
have a critical shortage of saving. Since the personal consumption tax is neutral in respect of the choice
between present and future consumption (saving), consumption would be a good tax base. Furthermore,
consumption (like income) tends to be distributed highly unequally in these countries. A progressive
expenditure tax could tap this base effectively and equitably.
e personal consumption tax is usually considered too complex to administer. It is argued, for example,
that to arrive at the taxpayer’s annual consumption, a list would have to be made of all expenditures, and
then added up. is, together with the required record-keeping, would be a mammoth task. Proponents
argue, however, that these problems can be overcome by observing the individual’s cash ow in quali ed
bank accounts. In addition, certain problems normally associated with income tax, such as valuing
unrealised capital gains and depreciation, are also avoided when consumption is taxed. Under a
consumption tax, capital gains are taxed when they are realised. Capital purchases are immediately
expensed (written off when purchased), making allowances for depreciation unnecessary.

16.4.2 The disadvantages of a personal consumption tax

e personal consumption tax has not been implemented successfully anywhere in the world. India and Sri
Lanka, for example, experimented with this tax, but abandoned it. e problem areas in designing a
personal consumption tax are administration, treatment of bequests and gifts, and the problems of
transition.
Critics are concerned about the risks of implementing a tax of this nature because we know too little
about the practical administrative problems to be encountered. In contrast, the problems with the current
income tax system are known and can be addressed. Furthermore, proponents of a consumption tax tend to
compare an ideal consumption tax to the current income tax, with all its impurities introduced over years.
is is not really a fair comparison since there is no guarantee that a personal consumption tax would not
follow the same route, and become progressively more impure and complicated.
A personal consumption tax creates a host of speci c administration problems. For example, under an
income tax system, taxes are withheld at source for administrative and compliance purposes (for example,
the PAYE system mentioned in Chapter 12). is would be difficult under a consumption tax. How would an
employer estimate the consumption and saving of each employee? A presumptive consumption–income
ratio may have to be applied. As mentioned earlier, extensive record-keeping would also be required in
respect of bank balances, expenditures and assets.
It would be necessary (and difficult) to distinguish between consumption and investment. Consider
expenditures such as housing and education. e purchase of a house, for example, should be regarded as
investment and subtracted from consumption to determine the tax base. Owner-occupied housing,
however, generates a service that should be classi ed as consumption. An imputed rent value would have to
be determined for this purpose. An alternative would be to exclude housing altogether from the tax base, but
this would erode the tax base. Education also has both an investment and a consumption component.
Another source of base erosion is the consumption of goods and services in kind. e consumption tax
system is not necessarily superior to the income tax system in detecting consumption. Under the cash ow
system, consumption is calculated as a residual (income minus saving). e de nitional problems related to
income all still apply and are compounded by problems relating to the de nition of saving. Would it not be
simpler to use an income tax system where only income needs to be determined?
Another major problem is the treatment of bequests and gifts. Should bequests and gifts be considered
as consumption by the donor or as income of the donee (that is, the recipient of the donation)? According to
one view, a gift (for example, cash) by a parent to a child is no different from any other form of expenditure
and should be treated as consumption. From another point of view, it is argued that consumption only
occurs when the child spends the cash. Exempting bequests and gifts would solve the administrative
problems, but could lead to large concentrations of wealth. Some form of wealth taxation, however, could
address this problem.
Finally, introducing a personal consumption tax will cause transition problems. One dilemma is the
treatment of savings once the new system comes into effect. Under the income tax system, saving comes
from after-tax income. If an existing asset is now realised or previous saving is spent on consumption goods,
the same base will be taxed again under a consumption tax, which appears to be unfair.
16.5 Tax reform: International experience
e existing tax systems of countries evolved over time. Changes to tax systems are implemented through ad
hoc reforms or comprehensive tax reform programmes. e goals of these reforms are varied and differ from
country to country. e driving force is often a desire for more revenue. In addition, non-revenue goals, such
as redistribution or equity, promotion of growth, tax simpli cation and a more efficient allocation of
resources, are also pursued.
Tax reform can be triggered by various factors, one of which is political change. A change in government
often implies different voter preferences. For example, if the new constituency is, on average, composed of
relatively more low-income voters (such as in Southern Africa), tax reforms that redistribute income to them
may be forthcoming (see Section 7.4.3 of Chapter 7). Political change may also be accompanied by
ideological shifts (for example, a shift away from centralised planning to a more devolved or federal-type
structure or a market-based system). Developments in tax theory often provide new analytical insights, and
attempts are then made to operationalise these through tax reforms. Some reforms are undertaken in
response to international trends. For example, there is a global trend towards VAT and lower marginal
income tax rates, and in the current integrated world economy, it is difficult for a small open economy to
ignore this trend. Tax reforms have also often resulted from a scal crisis (for example, short-term budget
de cits) or a concerted effort to prevent future scal crises (for example, chronic de cits and in ation).
Several African countries have recently introduced VAT, partly for these reasons.

16.5.1 Patterns of taxation in industrialised and developing countries

A cross-country comparison of tax systems indicates that there are vast differences in the composition of
taxes between countries. As each country’s tax system has been established over time by many – often
unique – forces, one should be careful of overgeneralising best tax practices and reforms on the basis of
international comparisons alone. Nevertheless, interesting patterns do emerge when countries are grouped
together, for example, according to the level of economic development, as shown in Table 16.3.
Table 16.3 Tax revenue of total government by type, 2016

Note: Total government comprises the following categories: supranational authorities, federal or central government,
state or regional government, local government, and Social Security Funds.
Source: Compiled from the OECD Global Revenue Statistics Database. Available: http://www.oecd.org/tax/tax-
policy/global-revenue-statistics-database.htm [Accessed 6 March 2019].

Table 16.3 contains a comparison of total government taxes for a group of OECD countries (advanced
countries), two regional groups (Africa, and Latin America and the Caribbean) and South Africa. When total
tax revenue as a percentage of GDP is considered, we notice that the tax burden is much higher in the
advanced countries (34,0% of GDP) compared to the other regions. South Africa comes closest to the OECD
countries with a tax–GDP ratio of 28,6%. A number of observations can be made in respect of the
composition of taxes (the percentages are unweighted average shares of tax types to total tax revenue):
• Taxes on income, pro ts and capital gains constitute one of the predominant sources of revenue in
OECD countries (33,6%), whereas taxes on goods and services (mostly value-added taxes) are the major
sources of revenue in Africa (54,6%) and Latin America and the Caribbean (50,5%).
• In the OECD countries, the average income tax on individuals (23,8%) is much more important than
income tax on companies (9,0%). In contrast, in Latin America and the Caribbean for example, the tax
contribution of companies exceeds that of individuals by 5,7 percentage points.
• Taxes on payroll and workforce as well as property taxes are insigni cant tax sources in both advanced
countries and developing countries.
• Trade taxes are insigni cant sources of tax revenue in OECD countries (0,5%) compared to developing
countries in Africa (11,6%).

Explaining differences in levels of taxation (the tax burden) and the composition of tax revenue is rather
tricky. A few generalisations will suffice. e high tax burden and reliance on income taxes in advanced
countries can probably be attributed to their level of development. Not only does the level of development
determine the size of the tax base, but it also has an effect on a country’s capacity to administer taxes. In
addition, taxpayers are more sophisticated in advanced countries, enabling tax authorities to levy relatively
complex taxes, thereby broadening the tax base even further. e greater reliance on company tax and taxes
on international trade in developing countries can be explained by the administrative ease with which a
relatively few companies and points of import and export can be targeted.
When South Africa’s tax composition is compared to the tax compositions of the OECD countries and
countries in the other two regions, it is evident that in most respects, the South African pattern is quite
similar to that of the advanced countries. An obvious deviation from most developing countries and some of
the advanced countries included in the sample is the relatively high total tax burden that South Africans face
(28,6% of GDP). It has to be considered that the percentage contributions are averages and the values for
different countries in each group show considerable variation. Another difference between the two groups of
countries is the importance of social security contributions to government revenue. In the past, these
contributions (compulsory social security payments by employees and employers) were included in the tax
revenues of government, but are now treated separately. e social security contributions add up to
approximately 26,2% of GDP for the OECD countries, but merely 8,4% for countries in Africa. In South Africa,
social security contributions constitute only 1,4% of GDP. It should be recognised that social security
contributions are negligible in South Africa, since most individuals provide for old age through voluntary
private retirement schemes. Furthermore, South Africa has a large non-contributory, means-tested old-age
grant system that caters for lower income elderly people in the form of transfers nanced by government tax
revenue.

16.5.2 International tax reform

A number of countries have reformed their tax systems in recent years. Given South Africa’s status as an
emerging market developing country, we are particularly (although not exclusively) interested in reforms
undertaken in similar countries. A detailed discussion of tax reform in developing countries falls beyond the
scope of this chapter, however. We will therefore focus on a selection of prominent tax issues in tax reform
debates. e lessons for reform of speci c taxes were discussed in the chapters on speci c taxes (see, for
example, Section 14.5 of Chapter 14 on income tax reform). In the discussion below, the emphasis is on the
direction of tax reform.10
ere has been a reappraisal of the redistributive role of taxes. e importance of using the tax system for
redistributive purposes has been reduced, partly because both vertical and horizontal equity have proved to
be elusive goals. Nonetheless, the contention is that the tax burden on the poor should at least be reduced or
removed altogether. In this way, a levelling-up process can take place. Bird and De Wulf (quoted in Brown &
Jackson, 1990: 175) summarised the position as follows:

Taxes cannot, of course, make poor people rich … If the principal aim of redistributive policy is to level up –
make poor people better off – the main role the tax system has to play is thus the limited and essentially
negative one of not making them poorer.

Consequently, more emphasis has been placed on public expenditure policies as instruments of
redistribution (see Chapters 8 and 9). More attention is also given to efficiency considerations when raising
revenues. is has resulted in a larger role for consumption taxes and less reliance on the principle of
comprehensive income taxation (see Norregaard & Khan, 2007: 5).
Concerted efforts have been made to broaden the base of the tax system. Tax systems in developing
countries are known to be allocatively non-neutral; that is, they cause distortions in the goods and factor
markets. Various reforms have been introduced to reduce the excess burden of taxation in these countries.
e general direction has been towards a broadening of the tax base accompanied by reductions in tax rates.
e concern with base broadening stems from the following:
• e narrower the base is, the higher the rate required to generate a given income.
• e higher the rate is, the greater the incentive for avoiding or evading the tax.
• Resources used for evading taxes are socially unproductive.
• High tax rates cause changes in relative prices that may lead to a reallocation of resources away from
taxed activity.

e objective, therefore, should be lower rates on a broader base. e base-broadening policy debate
focuses on the merits of a broad-based value-added tax, a at tax on consumption and the reduction or
removal of tax expenditures (for example, tax incentives to promote economic activity).
Major efforts are being made to improve tax administration. Tax simpli cation, which is one of the
mainstays of better administration, requires a rationalisation of the number of taxes. Taxes that provide little
revenue and have high administrative costs should be done away with. Tax rates should also be streamlined
(for example, there should be fewer personal income tax brackets). ere is growing recognition that less
complex taxes are easier to administer and will improve tax compliance. In addition, steps are being taken to
improve information systems and to limit political interference in tax administration.
Lower tax rates and a movement to more uniform tax rates (for example, less differentiation in VAT rates)
have been a worldwide phenomenon in the last decade. Lower tax rates are aimed at reducing the
disincentive effects of taxation. Examples include lower import tax rates, lower marginal rates of personal
income tax and lower effective company tax rates. Lower tax rates as well as the transnational convergence
of tax bases and rates are also the result of the increased tax competition that accompanies economic
globalisation and the regional integration of countries in geographic proximity.
A popular development in tax reforms is the levying of ‘green’ and ‘carbon’ taxes to address
environmental externalities. ese taxes are aimed at reducing pollution and greenhouse gas emissions, and
moderating climate change.
e recent nancial and economic crisis caused industrialised and developing countries to review their
tax systems. For most of the last decade, the countries of the EU and others have experienced high economic
growth and a cyclical scal dividend. is led to tax rate reductions and a weakening of automatic stabilisers.
An aging population as well as the increasing burden of nancing social pensions and healthcare systems
further exposed structural scal balances. e crisis in the nancial sector added an additional burden to
the scus and the extent of bonuses received by executives raised the ire of taxpayers. Tax reforms, which are
intended to meet these challenges, include a tax on nancial transactions, shifting the tax structure towards
taxes on property, consumption and environmental taxes, and higher tax rates for the rich and super rich.
Some of these reforms are short term, but it is clear that public nances must be put on a sustainable path:
governments need to nance the cost of the crisis and the increasing future cost of an aging population.
Many of the above-mentioned reforms underpin the ndings of tax trends in a very recent report by the
Organisation for Economic Development (OECD, 2017). e report highlights the following, amongst others:
orientation towards economic growth – particularly reductions in corporate income tax; enhancing fairness,
which entailed numerous personal income tax cuts in the range of middle- and low-income tax payers; and
tax measures to deter harmful consumption and behaviours, such as reforms of excise duties and
environmentally-related taxes and increasing popularity of health-related taxes (e.g. on sugar).

16.5.3 Globalisation and tax reform

Economic globalisation may be de ned as the integration of economies throughout the world through
trade, nancial ows, the exchange of technology and information, and the movement of people (Ouattara,
1997: 107). It is a process whereby economic interdependence among nations has increased since World
War II and has gained particular momentum since the fall of the command economies in Eastern Europe
towards the end of the 1980s.11 is increasing interdependence has led to increasing competition and is
re ected in cross-border economic integration between politically sovereign countries. In a globalised
economy, the effect of policy measures of one country spill over to other countries.
When the tax systems of the world came into being, it was at a time when economies were by and large
closed economies. Much of the economic activity was highly regulated and controlled, and the tax policies of
other countries could be disregarded. Globalisation has, however, altered the behaviour of economic actors
in ways that required tax redesign. Firstly, cross-border shopping has increased in many regions in the
world, including Southern Africa and Europe. is enables some countries to lower their excise taxes on
high-value and easily transportable commodities to attract foreign consumers. In this way, parts of the tax
base of some countries is in actual fact ‘exported’ to other countries and tax revenue is generated in
neighbouring countries instead. Secondly, transfer pricing has resulted from the expansion of the
multinational rm. rough transfer pricing, pro ts can be repatriated from high-tax jurisdictions to low-tax
jurisdictions by over-invoicing imports and under-invoicing exports. Multinational enterprises are in the
advantageous position of being able to minimise their global tax liability by shifting pro ts (through transfer
pricing) from high-tax jurisdictions to low-tax jurisdictions. To entice multinational and other rms to locate
in their countries, tax authorities compete by offering lower tax rates and other tax incentives. irdly, tax
evasion and tax avoidance by individuals has become possible on a global scale, as personal savings have
become more mobile. e proliferation of tax-haven areas and even countries as well as new nancial
market instruments have made it extremely difficult for tax authorities to monitor the non-reporting of
personal income from savings invested12.
How will globalisation impact on the future of tax systems? In his analysis of the globalisation
phenomenon, Tanzi (1996 and 2004) identi ed a number of possible trends. Globalisation tends to put
pressure on developing countries to lower the level of taxation. As far as taxes on consumption are
concerned, it is expected that as borders are effectively removed, countries that have high initial tax rates will
be under pressure to reduce their rates in the face of competition. ese reductions would in particular
affect excises on luxury products. e opening-up of economies requires that taxes on international trade be
reduced or eliminated. International tax competition leads to reductions in marginal tax rates for personal
income. In as much as taxation is a locational factor, tax competition will tend to drive down effective
company tax rates. In an environment where multinational enterprises dominate, it is even conceivable that
income tax on pro ts may be replaced by another tax base, such as a tax on net assets or gross assets. Owing
to the mobility of incomes from capital sources (for example, interest and dividends), these sources could
become taxed separately from wage income. is would enable tax authorities either to exempt these forms
of income entirely or to tax them using a separate schedule – see in this regard our discussion of the dual
income tax (Chapter 14, Section 14.5.2). Lastly, taxes on property will probably increase, as the tax base is
reasonably immobile.
e downward pressure on levels of taxation implies that developing countries have to rely increasingly
on reforming personal income taxation and VAT. Personal income tax serves the purpose of raising revenue
and ensuring that equity objectives are reached. Since it is mostly the high-income earners who bene t most
from globalisation, the personal income tax system is ideally suited to capture revenue from these income
groups for redistributive purposes. At the same time, if the tax base is broad enough, tax rates need not be set
that high. Value-added tax should become the most productive source of revenue. Tanzi (2004) notes that it
would be much better to have a low single rate on a broad base to generate sufficient revenue to deal with
poverty and equity issues as well as the pressures of globalisation on spending. According to Keen and
Mansour (2009: 38), most sub-Saharan African countries have been successful in recovering much of the
revenue losses owing to trade liberalisation by implementing VAT systems in the 1990s.

16.5.4 International tax competition and tax harmonisation

Economic integration within some regions and between countries bordering each other has become more
important in recent times. e economies of the European Union (EU) member states now form an
integrated market. e same applies to the economies of bordering countries such as Canada and the United
States, and Mexico and the United States. In Southern Africa, the economies of South Africa, Botswana,
Lesotho, Eswatini and Namibia (the BLSN countries) have been interlinked historically for many decades.
e Southern African Customs Union (SACU) agreement between the BLSN countries aims to have goods
interchanged freely between these countries. e Union provides for a common external tariff and excise
tariff. All customs and excise collected in the common customs area are pooled and the revenue is shared
among members according to a revenue-sharing formula. In the BNLS countries, revenue from the common
revenue pool (customs, excise and additional duties) is by far the most important source of revenue. In
2017/18, revenue from this source as a percentage of total tax revenue ranged from approximately 36,0%
(Namibia) to 45,4% (Eswatini). e economies of South Africa and other Southern African countries, such as
Zimbabwe, Malawi and Mozambique, as well as other African countries, are similarly interlinked through
their membership of the Southern African Development Community (SADC). e aim of the SADC is to
create a community providing for regional peace and security, and an integrated regional economy. A
number of protocols have been signed promoting economic cooperation, planning and assistance in areas
such as trade, mining, tourism and education. e SADC also provides the building blocks for the African
Union (AU), which has the promotion of accelerated socio-economic integration of the continent as its
vision. e most recent regional integration development has been the signing by early July 2018 of the
African Continental Free Trade Area (AfCTFA) agreement, by more than 50 African countries. It is aimed at
facilitating a single market for goods and services on the continent.
In a global and regional context each country remains an independent nation state with political
autonomy, but their economies have become much more integrated, historically and by agreement. In this
integrated regional setting, the monetary and exchange rate policies as well as the tax policies of these
countries impact directly on each other. Added to the policy issues are the increased international mobility
of consumers and factors of production, such as capital and labour (for example, professionals). e
increased economic integration, coupled with the mobility of factors of production and consumers, has led
governments to compete for investment capital and the purchasing power of neighbouring consumers by
lowering tax rates. is may be welfare reducing in that public services are underprovided. Others argue that
tax competition is welfare-enhancing.
In this section, we will approach the theory of tax competition (that is, when countries lower their tax
burden relative to competitor countries to attract investment) and the need for tax harmonisation (that is,
when countries make their tax burden identical or effect marginal differences between jurisdictions) from
the international tax perspective. (In other words, we will consider competition between independent
countries.) e theoretical discourse also applies broadly to competition between independent
governmental jurisdictions (or sub-national governments) such as local governments or federal states. e
federal states perspective is provided in Chapter 19. In fact, the theory on tax competition has its roots in the
literature of local public nance (or scal federalism).13
One of the earlier contributors to the debate was Oates (1972: 143), who noted that ‘in an attempt to keep
taxes low to attract business investment, local officials may hold spending below those levels for which
marginal bene ts equal marginal costs’. Put differently, there is a perception that high taxes on capital will
drive mobile capital out of the country, and even cause a lowering of wages, employment and land rents.
Governments then lower their public expenditure below efficient levels to reduce their dependence on tax.
What is worrisome, of course, is if governments do not adjust expenditure to lower tax revenues and the
prospective investments and their associated future tax revenue do not materialise. In such cases rising
budget de cits and public debt may be the unfortunate result. In addition, tax competition leads to scal
externalities or scal spill-overs. Some public services generate positive bene ts for neighbouring countries,
for example, malaria control efforts. Countries engaging in tax competition would ignore these positive
externalities when setting levels of public expenditure. ese externalities also come in the form of negative
costs, for example, tax base externalities, which can be described as the result of a country lowering the tax
rate on mobile capital to increase its welfare. e country gains the capital in ow, but it is at the expense of
another country in that the tax base of the latter is reduced and tax revenues decline. e same negative
scal externality can be caused by a reduction in a commodity tax to attract cross-border shoppers. Other
adverse effects of tax competition include weaker environmental standards to attract investment and
reductions in welfare payments by countries trying to make it unattractive for poor households from other
countries to migrate or immigrate to their home countries. Another interesting negative outcome of tax
competition between countries is that it may lead to government failure in the provision of public goods and
services. e argument is that government provides some goods and services when competitive private
markets fail to do so, but when countries compete for the provision of these goods, governments might fail to
provide the necessary public goods and services.
From an efficiency point of view, the potential adverse effects of tax competition suggest a role for tax
rate harmonisation as well as tax base harmonisation. at is, if all countries were to agree to raise their
capital taxes or value-added taxes simultaneously, to set minimum rates or to unify tax bases, they would all
bene t from increased levels of public expenditure. e interaction between countries then pertains to
comparative and competitive advantages, rather than tax-driven opportunities. e welfare-enhancing
effects of tax rate harmonisation can, in simple terms, be illustrated with the aid of Figure 16.1. Assume two
countries with identical demand curves and a world price given by Pw. e high-tax country sets its tax rate
at (th) and the low-tax country sets its rate at (tl). e taxes create an excess burden in each country,
measured by ABC in the high-tax country and ADE in the low-tax country. By harmonising at a common rate
equal, for example, to the average of the two rates, 0,5 (t1 + th), the excess burden is reduced by BCFG in the
high-tax country and increased by GFED in the low-tax country. e net effect is a reduction of the global
welfare loss (excess burden) equal to GHB, implying a potential Pareto improvement14 under general
conditions (for example, assuming that tax revenues can be returned to consumers without additional
distortions being created).
Tax harmonisation may, however, be undesirable, since tax competition may generate bene ts that could
offset the efficiency costs of the under-provision of public services.
e Tiebout model is discussed in Chapter 19. Modern formulations of this model argue that tax
competition between local governments or states within a single nation will tend to limit overexpansion of
the government at the local level. A market-like solution to the efficient provision of public goods is reached
as a result of the fact that local governments are closer to the people and therefore more accountable, and
that individuals tend to vote with their feet. is conclusion can be extended to the between-country
situation. It is believed that tax competition is an important argument against tax harmonisation in unions
such as the EU, where labour is fairly mobile over national borders. Furthermore, when labour mobility and
population economies of scale are considered together, the under-provision of public goods associated with
tax competition becomes even less of a problem. From a public choice perspective, we know that politicians
and bureaucrats often attempt to maximise their self-interests and budgets, and it may thus follow that tax
harmonisation bene ts government officials. Tax competition may curtail undesirable attempts at raising
budgets.
Figure 16.1 Efficiency gain from harmonising tax rates

e net effect of the positive and negative factors is very difficult to determine. Zodrow (2003) concludes
that the argument for corporate tax rate harmonisation in the EU is not yet compelling. Economic models of
the welfare costs of tax competition also suggest that its costs may not be excessive. Zodrow recommends
that a cautious approach to tax coordination (harmonisation) is appropriate. In designing or analysing tax
regimes, tax autonomy must be traded off against tax efficiency and the minimisation of tax administration.
In this regard, Cnossen (2003) emphasises the importance of subsidiarity (or jurisdiction). Subsidiarity
implies that the power to tax (or tax autonomy) rests with countries. With reference to the EU experience,
Cnossen (2003) is of the opinion that all the proposals for VAT coordination require greater involvement of a
central authority, and thus a reduction in country autonomy and sovereignty. Ade, Rossouw and Gwatidzo
(2017) analysed tax harmonisation in the SADC, with a speci c focus on corporate taxes and VAT. Based on
their empirical analysis, they conclude that tax harmonisation is feasible in the SADC, and that member
countries must expand their corporate tax base to allow for the adoption of a low optimal corporate tax rate.
ey also recommend that if SADC members adopt an optimum VAT rate, this can lower the usage of
different VAT rates that may be politically driven (Ade et al., 2017: 12).

16.6 Tax reform in South Africa

e government of South Africa has appointed three commissions and one committee of inquiry reporting
on aspects of the tax structure since the late 1960s: the Franzsen Commission (1968), the Margo
Commission (1987), the Katz Commission (2013 to 2018), and the Davis Tax Committee (from 2013). e
work of these commissions resulted in comprehensive reforms of the South African tax system. In addition
to these comprehensive tax reforms, several major ad hoc tax reforms have also been introduced.
Instrumental in these initiatives were the Standing Commission of Inquiry with regard to Taxation Policy of
the Republic (Standing Tax Commission) and its successor, the Tax Advisory Committee (TAC), which was
an advisory body to the Minister of Finance. Presently the organisational structure of the National Treasury
provides for a Tax Policy Unit, which is responsible for advising the Minister of Finance on tax policy issues
that arise at all three levels of government. In designing tax policy, this unit cooperates with the South
African Revenue Service (SARS), and interacts with the corporate sector and the general tax-paying public.
In 2013, the Minister of Finance announced the formation of a South African Tax Review Committee
chaired by Judge Dennis Davis. e Davis Committee (DTC) was tasked with inquiring and advising on the
role of the tax system in the promotion of inclusive economic growth, employment creation, development
and scal sustainability. e terms of reference of the Committee were comprehensive, and some of the
aspects on their agenda included the following:
• An examination of the overall tax base and tax burden, including the appropriate tax mix between direct
taxes, indirect taxes, provincial and local taxes
• e impact of the tax system on the promotion of small and medium-sized businesses
• A review of the corporate tax system, with special reference to the efficiency of the tax structure, tax
avoidance, tax incentives and effective company income tax rates
• e advisability and effectiveness of VAT dual rates, zero-rating and exemptions
• e progressivity of the tax system and the relevance of estate duty in particular
• e evaluation of proposals to fund the proposed National Health Insurance system.

In the paragraphs below, we highlight some of the main recommendations and reforms proposed by the
different tax inquiries.
e Franzsen Commission (1968) concluded in 1968 that the tax structure at that time was increasingly
inhibiting economic growth. e focus of taxation was shifting from indirect to direct taxes and from direct
taxes on companies to direct taxes on individuals. e Commission therefore believed that structural
changes were required in the form of the following:
• Reduced progression in direct taxes
• A shift towards indirect taxes by broadening the base
• A broadening of the scal concept of income by including capital gains

In its rst report, the Commission consequently recommended that the maximum marginal income tax rate
on individuals be reduced from 66% to 60%, that selective sales duties on a number of items (to be collected
from manufacturers and importers) be introduced and that capital gains tax of 20% on net realised gains be
introduced. With the exception of the recommendation in respect of capital gains tax, all of the other
proposals were accepted by government and duly implemented.
e next major tax reform occurred in 1978, when sales duties was replaced by a general sales tax (GST)
at a rate of 4%. e sales duties had inherent disadvantages (for example, narrow base and high rates). e
major aim of introducing GST was to broaden the tax base and eliminate tax non-neutralities. GST was
followed by the introduction of Regional Services Council (RSC) levies in 1985. e RSC levies were
implemented in 1987, and consisted of a levy based on the remuneration of employees (0,25%) and a levy
based on the turnover of enterprises (0,10%).
e report of the Margo Commission (1987) was released in 1987. e Commission reported at a time
when in ation was rampant, the business cycle was in an upswing and foreign disinvestment was a
threatening factor. e Commission took the view that tax reform should not be driven by short-term
economic problems, but rather by aspects of the existing tax structure that could hinder economic
development. e Commission’s general approach (1987: par 1.28) was founded in a base-broadening
philosophy:

e ideal, both for direct and indirect imposts, is a broad-base, widely distributed, low-rate, high-yield tax,
conforming to these other requirements (equity, neutrality, simplicity, certainty and so on) as far as
possible.
A tax system of this nature would reduce the ‘brain drain’, encourage immigration, improve standards of tax
morality and compliance, promote entrepreneurship and capital formation, and create job opportunities.
e following major recommendations of the Commission were accepted by government:
• e taxation of fringe bene ts
• Lower personal income tax rates with fewer brackets
• Accepting the individual as the unit of taxation and phasing in marriage neutrality, and the equal
treatment of men and women
• e rejection of capital gains tax
• e scrapping of certain tax expenditures and allowances
• e modi cation of GST and the reduction of the rate; if the recommendation was not accepted, GST
would be replaced by an invoice VAT system
• e imposition of a capital transfer tax to replace estate duty and donations tax.

Between 1987 and 1994, two of the most important tax reforms were the introduction of value-added tax
(VAT) and the lowering of the company tax rate (along with the introduction of the secondary tax on
companies [STC]). VAT was introduced in 1991 to eliminate the distorting effects of tax cascading inherent in
GST and to reduce tax evasion. Initially, the VAT rate was to be 12%, with very few exemptions and zero rates.
After much political lobbying by the trade union movement in particular, VAT was eventually introduced at
10%, with allowance for a number of zero-rated items. Secondary tax on companies (STC) was introduced in
1993. is was a tax on distributed pro ts, levied on rms. e aim was to encourage rms to reinvest their
pro ts and thereby promote economic development. In a sense, STC was also an astute way of reintroducing
tax on dividends, since government had exempted the taxation of dividends in 1990.15
In 1994, the rst interim report of the Katz Commission (1994) was released. It was supplemented by
another nine reports between 1994 and 1999. e Commission conducted its investigations at a time when
South Africa had just entered a new political and constitutional era. e major thrust of the rst and third
interim reports was to improve tax administration and collection, and to reappraise the equity aspects of
certain taxes.
e DTC has completed several tasks and a summary of the work done for the period 2013 to 2018 can be
found in its closing report (DTC, 2018d). In short, this report highlights that the DTC submitted 25 reports to
the Minister of Finance, all of which have been published on their website16. e following topics were
covered in these reports (refer to the respective chapters and sections of this book (in parenthesis below) for
some of the recommendations of the DTC):
• Small business (see Section 14.2.3)
• Macro analysis (with concurrent reports on tax incentives – see Section 14.3.2.3)
• Base erosion and pro t shifting
• Mining (see Section 14.2.1)
• Value-added tax (see Section 16.3.1)
• Estate duty and CGT implications (see Section 15.4.2)
• Wealth tax (see Section 15.2.3)
• Carbon tax
• Public Bene t Organisations
• Tax administration
• Funding National Health Insurance and Tertiary Education
• Corporate income tax (see Section 14.3.2.3)

We have studied a number of taxes in the previous chapters. Tax reforms that were speci c to these taxes
were touched on and analysed. e following relatively important tax reforms were introduced and
proposed between 1994 and 2019:
• e status and independence of the revenue authorities were enhanced by the establishment of the
South African Revenue Service (SARS) as a separate department. Various changes and modernisation
 measures were introduced by SARS to improve tax collection as well as to simplify tax administration and
compliance. An electronic ling facility was introduced, improving the quality of returns and timeliness.
• A single rate structure with six brackets for personal income tax was introduced. To enhance the
progressive nature of the tax, marginal tax rates were increased by one percentage point for all personal
income tax brackets except the lowest.
• All gambling and fee-based nancial services were subjected to VAT. Gambling winnings above a
threshold are subject to a nal withholding tax.
• Mandatory (tax) contributions to a national social security fund and incentives for additional savings to
promote retirement savings were proposed. Contributions to retirement funds are deductible, but
became capped at an annual maximum.
• Capital gains became taxable.
• e source-of-income base was replaced by a residence-based income tax.
• e company tax rate was lowered for small businesses with turnover below a certain threshold.
Furthermore, a turnover-based presumptive tax system was introduced as an elective system.
• Tax incentives to promote direct investment were introduced, including an accelerated depreciation
allowance for investment in underdeveloped designated urban areas. A company income tax rate of 15%
will be authorised in special economic zones.
• Secondary tax on companies (STC) was phased out and replaced with a dividend tax on shareholders.
e dividend tax is enforced through a withholding tax at company level.
• Regional Service Council (RSC) levies and Joint Services Board levies were abolished.
• A mineral and petroleum royalty regime was introduced.
• Various environmental charges and incentives were introduced in response to climate change. A carbon
tax was proposed as an appropriate mechanism to reduce greenhouse gas emissions in South Africa.
• A national health insurance (NHI) scheme is to be phased in over the next few years (see Section 10.5.1
of Chapter 10). is will have a major funding impact, and suggested options under consideration
include a payroll tax payable by employers, an increase in the VAT rate and a surcharge on individuals’
taxable income.
• Medical tax deductions were converted into tax credits per bene ciary.
• A youth employment subsidy in the form of a tax credit was introduced.
• e personal income tax rate structure was amended in 2017/18 and a further tax bracket was added to
tax the top income bracket of incomes above R1 500 000 at a marginal tax rate of 45%.
• A health promotion levy was implemented to reduce the consumption of sugar-sweetened beverages.
• e VAT rate was increased from 14% to 15%.
• ree additional food items were added to the zero-rated list.
• Estate duties increased from 20% to 25% for estates greater than R30 million.
• Approval of six special economic zones that will bene t from additional tax incentives was granted.

Key concepts
• base broadening (page 354)
• consumption VAT (page 342)
• customs duties (or tariffs) (page 339)
• destination principle (page 342)
• economic globalisation (page 355)
• excise duties (page 339)
• exemption (page 341)
• input tax (page 344)
• multiple rates (page 343)
• multi-stage commodity tax (page 339)
• output tax (page 344)
• personal consumption tax (or expenditure tax) (page 348)
• restricted-origin principle (page 343)
• single-stage commodity tax (page 339)
• sumptuary taxes (or sin taxes) (page 339)
• tax competition (page 357)
• tax harmonisation (page 357)
• uniform rate (page 345)
• value-added tax (page 342)
• zero-rating (page 341)

SUMMARY
• Indirect taxes are taxes that are imposed on commodities or market transactions. We distinguish
between selective taxes such as excise duties and general taxes (for example, VAT). ese taxes are called
indirect taxes since the tax burden is likely to be shifted.
• Indirect taxes have advantages such as being useful in addressing market failures (for example,
externalities) and raising revenue not captured by the income tax net. e liability is to some extent
determined by how much is consumed. e disadvantages are that indirect taxes tend to be regressive
and may cause excess burdens when goods and services are taxed selectively.
• Value-added tax (VAT) is a major revenue source for the revenue authorities. e VAT system in South
Africa and the BLNS countries is of the consumption type and taxes international trade using the
destination principle. e tax credit method is used to determine the liability and only a few services are
exempted. A range of goods is zero-rated, with all other goods being taxed at a single rate.
• e VAT systems used in Southern Africa are considered to be ‘clean’ and efficient systems, given that
there are few exemptions, some goods are zero-rated and there is a uniform tax rate. Fairness is
somewhat sacri ced in that the system taxes the poor disproportionally.
• An alternative to the personal income tax is a consumption (or expenditure) tax regime. e
consumption tax base is de ned as the difference between income and savings. Taxing consumption is
bene cial considering that the individual is taxed on what is taken out of the economy, in contrast to an
income tax, which taxes effort and savings. e personal consumption tax has not been implemented
successfully anywhere in the world and faces a host of administrative challenges.
• Tax systems differ between developing countries and advanced countries. Developing countries tend to
rely more on consumption taxes, whereas income taxes are the most important sources in advanced
countries. Tax systems change over time. ese changes are often triggered by international tax reforms,
globalisation and the harmonisation of taxes between countries. Some of the important elements of
global tax reform are as follows:
- Income taxes: Lower tax rates, fewer tax brackets and fewer exemptions
- Sales taxes: e introduction of broad-based VAT-type taxes at a single rate
- Tax administration: Simpli cation of tax codes.
• Tax reform in South Africa was much in uenced by government-appointed commissions and
committees of enquiry into the tax system. Important tax reforms since 1994 include the introduction of
a capital gains tax, a mineral and petroleum royalty regime, green taxes and a withholding tax on
dividends as well as lower personal and company income tax rates.

MULTIPLE-CHOICE QUESTIONS
16.1 In South Africa, the tax on cigarettes and tobacco products …
a. contributes more to total tax revenue than the fuel levy.
b. is an ad valorem tax.
c. cannot be shifted.
d. is a sumptuary tax.
16.2 Which of the following statements about a value-added tax is or are correct?
a. Exempted rms may claim an input credit from SARS.
b. Because VAT is levied at each stage of production, tax is levied on tax, causing a cascading effect.
c. ere is some consensus that taxing consumption at a uniform rate on a broad base is efficient.
d. Zero-rating certain goods and services is fairer to the poor, but erodes the tax base.
16.3 A personal consumption tax …
a. has no excess burden.
b. taxes savings, but not work effort.
c. would encourage individuals to save.
d. requires taxpayers to report their annual income and savings.
e. Both c. and d. are correct.
16.4 Which of the following statements on tax competition and harmonisation is or are correct?
a. Tax competition between countries or regions is neutral in respect of the impact on public
expenditure levels.
b. Tax harmonisation impacts on tax autonomy.
c. Tax harmonisation reduces tax efficiency in the low tax country, but increases efficiency in the high
tax country.
d. e global excess burden after tax harmonisation increases by BHG in Figure 16.1.

SHORT-ANSWER QUESTIONS
16.1 Distinguish between the following indirect taxes:
• Single-stage and multi-stage sales taxes
• Excise tax and customs duty
• VAT and a personal consumption tax.
16.2 A uniform VAT rate is preferable to multiple rates. Discuss.
16.3 Identify the major tax reforms implemented globally.

ESSAY QUESTIONS
16.1 Explain why the government should levy indirect taxes.
16.2 ‘Indirect taxes are not transparent enough and inhibit informed choices by taxpayers. e direct tax
base should therefore be the major basis of government tax revenue.’ Discuss this statement.
16.3 What are the characteristics of value-added tax in South Africa? Why is it said that VAT is inequitable
and what can be done to correct the inequity?
16.4 In designing VAT and other indirect taxes, there is always a con ict between equity and efficiency. Do
you agree? Explain your answer.
16.5 Personal consumption is a better tax base than income. Discuss.
16.6 ‘Tax competition is preferable to tax harmonisation.’ Discuss critically.
16.7 Evaluate tax reform in South Africa since the late 1960s in light of the patterns and directions of
international tax reform.

1 For a comprehensive discussion of VAT, see Gillis, Shoup and Sicat (Eds) (1990: 3–16; 219–233) and Katz Commission (1994: 101–
147).
2 See Cnossen (2003) in this regard and also for alternatives to the restricted-origin principle.
3 See the full report at http://www.treasury.gov.za/comm_media/press/2018/2018081001%20VAT%20Panel%20Final%20Report.pdf.
4 C-efficiency refers to collection efficiencies (Cnossen, 2015: 1080).
5 Although relatively high compared to other countries, the IMF (2015: 13) revealed some uctation in the ratio over this time period,
due to changes in compliance and the economy.
6 See Keen (2012) for more details on these indicators.
7 For empirical evidence on this nding, see Inchauste et al. (2016).
8 After the increase in the VAT rate (from 14% to 15%) in the 2018 SA Budget (see National Treasury (2018a)), the Minister of Finance
appointed a panel to review the zero-rating policy (see earlier comments on the panel recommendations and further zero-rating in
Section 16.3).
9 Tax earmarking of course creates problems of its own.
10 For a more comprehensive discussion of international tax reform, see Bernardi, Barreix, Marenzi and Profeta (2008), Norregaard
and Khan (2007), Tanzi and Zee (2000), Stotsky and WoldeMariam (2002), Boskin (1996), World Bank (1991), and Khalilzadeh-
Shirazi and Shah (1991).
11 e remarks made in this section are attributed to Tanzi (1996). For a more comprehensive discussion of the effects of globalisation
on tax policy, see Tanzi (1995).
12 In order to establish the nature and extent of so-called tax base erosion and pro t shifting, the OECD initiated a major project in
2013 that has since involved more than 100 countries and which is aimed at collaboration on how to combat these phenomena in
the interest of all countries. For more information, the following website can be visited: http://www.oecd.org/tax/beps/.
13 For an excellent survey of theories of tax competition and formal models of tax competition, see Hau er (2001) as well as Zodrow
(2003) and Wilson (1999). e discussion here is a summary of these surveys.
14 See Keen (1989) for a detailed discussion.
15 Some argued that since dividends are paid out of after-tax income, they were previously double-taxed.
16 See the DTC website for full details on all reports released (https://www.taxcom.org.za/library.html).
Public debt and debt management

Estian Calitz and Ian Stuart1

e aim of this chapter is to study the essence of, and rationale for, public debt as well as its impact on the
economy. Public debt arises from the borrowings of government, as re ected primarily in the annual
budget. A systematic study of public debt and its management provides important insights into the
relationship between the various components of public nance as well as the interaction between scal and
monetary policy in their impact on the economy.
Our study begins with a discussion of the concept of public debt, followed by an overview of the size and
composition of public debt in South Africa. Next we discuss various theories on the rationale for public debt,
followed by an investigation of this question: Should government tax or borrow? Our nal section deals with
public debt management.

Once you have studied this chapter, you should be able to:
define public debt
describe salient characteristics of the size, composition and nature of public debt in South Africa
explain and critically compare different theories of public debt
explain the rationale for borrowing versus taxation when funding government expenditure
define public debt management
identify and describe the different types of public debt cost
identify the goals of public debt management and discuss their pursuance, with special reference to South
Africa.

17.1 The concept of public debt

Public debt may be de ned as the sum of all the outstanding nancial liabilities of the public sector in
respect of which there is a primary legal responsibility to repay the original amount borrowed (sometimes
called the principal of debt) and to pay interest (sometimes called debt servicing). Most of the time,
especially when considering the macroeconomic implications, the term ‘public debt’ is used to refer to the
debt of the national government only (see Figure 1.1, Chapter 1). We use this narrower de nition in this
chapter. When we are referring to debt of other public sector entities, this will be speci ed.
Public debt arises primarily from the government’s annual budget de cits, which are one of the
consequences of scal policy, to be discussed in Chapter 18. e majority of public debt is incurred through
the sale of government bonds (also called stock) that have a maturity of more than three years. Most of the
time, these are xed-interest bearing securities issued by the national government, that represent a charge
on the revenues and assets of the Republic. Government can also borrow in the short-term in the form of
treasury bills (debt normally issued for a 91-day period) or bank overdrafts for bridging nance.
For many countries, the size and quality of the public debt is a politically charged issue. Debates focus on
whether government borrowing is counter- or pro-cyclical, being used effectively (e.g. nancing
infrastructure or consumption spending) and, most importantly, whether it can be repaid without drastic
scal measures. Despite this, there is limited consensus on the macroeconomic effects of public debt, and
even the de nition of debt sustainability.
Government’s public debt management can have implications for the wider economy. For example, the
interest rate on the R186 – a 10-year xed-rate bond – is a benchmark on which the private sector and state-
owned companies base their own interest rates when issuing bonds. Treasury bills form part of the liquid
asset base of the private banking sector, which can in uence the stock of money. is means that an increase
in treasury bills (that is, an increase in short-term government loans), immediately translates into an
increase in the supply of money, with in ationary implications. From a broader institutional perspective,
public debt management has been a major driving force behind the deepening and maturation of nancial
markets in South Africa.
Other varieties of government bonds are variable-interest bonds (of which in ation-indexed bonds are
an example) and zero-coupon bonds. Zero-coupon bonds are bonds that do not pay interest during the
lifetime of the bonds. Instead, investors buy zero-coupon bonds at a deep discount from their face value,
which is the amount a bond will be worth when it ‘matures’ or becomes due. When a zero-coupon bond
matures, the investor will receive one lump sum equal to the initial investment plus interest that has
accrued. As nancial markets develop and investor sophistication increases, the variety of bonds increases.
All government bonds, irrespective of their maturities, are regarded as liquid assets in the hands of banks.
Occasionally, debt is incurred outside the budget (off-budget debt) and is not re ected in the budget
de cit. For example, in the early 1990s, the government transferred bonds directly to the public employees’
pension funds to improve their funding levels, instead of budgeting for the expense in the normal fashion.
Details of the national government debt are published in the Quarterly Bulletin of the South African
Reserve Bank. is excludes the debt of extra-budgetary institutions (such as universities), provincial and
local governments, and the non- nancial state-owned enterprises (such as Eskom and Transnet). Of course,
if the national government were to take over the debt of any of these institutions, the legal responsibility to
service and repay the debt would be transferred to the national government. From that moment, the debt
would be counted as part of the public (in other words, national government) debt. But this cannot happen
without a speci c policy decision. e rising debt position and increasing nancial predicament of Eskom,
South Africa’s energy utility, is a case in point: in his 2019 Budget Speech Finance Minister Mboweni
explicitly stated that Eskom’s debt will not be taken over by Government.
Our de nition of public debt also excludes contingent liabilities (see Section 17.5.5), that is, the
outstanding nancial liabilities of public entities (such as public enterprises) and private entities whose debt
carries an explicit guarantee by the national government. Only when a guarantee of this nature is called up
will the payment obligation be transferred to the national government (as guarantor) and the amount
involved will be added to the national debt. It is, of course, prudent to keep track of the government’s
contingent liabilities (inclusive of formal and implied guarantees) (see also Section 17.5.5) when considering
the total debt risk to which the government is potentially exposed. Implied guarantees arise when an
investor (for example, a bank) regards a loan to a public sector entity (for example, a local government) as
implicitly carrying a government guarantee on the assumption that the local government will not be allowed
to go bankrupt for political reputation reasons. We return to this possibility in Section 19.5 of Chapter 19
when explaining the occurrence of moral hazard in inter-governmental nancial relations.
We begin with a discussion of the concepts of public debt and debt sustainability, followed by an
overview of the size and composition of public debt in South Africa. Next we discuss various theories on the
rationale for public debt. Our nal section deals with public debt management.
17.2 Public debt sustainability
In Chapter 18 we will discuss the issue of scal sustainability and consolidation in the context of scal
policy. As a precursor it is appropriate to say something about debt sustainability. is has been de ned ‘as a
situation in which a borrower is expected to be able to continue servicing its debts without an unrealistically
large correction to the balance of income and expenditure’ (IMF, 2002: 4). When deciding on the probability
of scal distress or default, government bondholders and ratings agencies consider both qualitative and
quantitative information, including the economic growth outlook, and the institutional strength of the scal
authorities. A key measure of scal vulnerability is the public debt–GDP ratio. Typically, countries with high
debt–GDP ratios and large scal de cits are at higher risk of default. In addition to debt repayments, the high
debt ratio makes a country less resilient to scal shocks (e.g. a global recession). At the same time, however,
a country with a low debt–GDP ratio can nd itself in difficulties on account of the structure of its debt. For
example, a high share of foreign-currency or short-term debt exposes the sovereign to adverse exchange rate
movements, and rollover risk. Furthermore, a country with a moderate but rapidly increasing debt–GDP
ratio may be more likely to have the sustainability of its scal position called into question than a country
with a high, but stable or declining debt level.
Ultimately a country’s debt position is no longer sustainable when markets deem it so. A debt stock can
quickly become unsustainable if global risk appetite and the availability of nancing changes, especially if
accompanied by doubts about the government’s ability to push through contentious scal measures.

17.3 Size and composition of the public debt

On 31 December 2017 the total debt of the South African government (public debt for short) amounted to
R2 467 billion. is was 53% of the GDP, or roughly R43 600 per head of the population. is tells us that if all
of the public debt were to be repaid immediately, the government would on average have to impose a once-
off tax of R43 600 on each citizen, which amounts to a rather high 54% of the gross national income per head
of the population. ose citizens who are government bond holders as well will of course also be on the
receiving side when this debt is repaid. e impact of the international nancial crisis of 2007–2009 and the
subsequent recessionary conditions show how quickly the state of public nance can change in a country
(see Figure 17.1).
e average size of public debt as a percentage of GDP declined during the 1970s and 1980s, and then
rose substantially during the early 1990s until just after the political change of 1994. (See Figure 17.1 for the
period 1960–2017 and Table 17.1 for average data and highest and lowest debt–GDP ratios for indicated
times.) e surge led various economists to warn against the dangers of a debt trap, a term used to signify
the inability of a government to repay and service its debt. Part of this increased share was because the
government borrowed money to fund the government employees’ pension funds and took over the debt of
the independent homelands under the apartheid system. From the beginning of the next decade,
government debt as a percentage of GDP fell quite dramatically, averaging 35,3% in 2000–2007. e lower
debt–GDP ratio was mainly the result of the systematic reduction of the annual budget de cit as a
percentage of GDP and the use of privatisation income to reduce government debt. In 2008, the debt–GDP
percentage reached its lowest level (of 26,5%) during the 57 years covered by Table 17.1, more or less just
when the international nancial crisis erupted. South Africa incurred a substantial Keynesian-type debt
increase during the next few years. e debt–GDP ratio subsequently reached levels of rising concern.
Together with political and economic uncertainties, this led to the downgrade of South Africa’s sovereign
debt in the international nancial markets to sub-investment status. e ratio of 53% registered at the end of
2017 was the highest in the period 1960–2017. In the 2019/20 Budget the ratio was expected to be
signi cantly higher and to only stabilise at 60,2% in 2023/24.
Since 2012 the debt–GDP ratio exceeded the average of 39% during the rst twenty years since the
constitutional change in 1994. Macroeconomically, South Africa had more scal manoeuvrability before the
international nancial crisis than many emerging market economies. However, at the end of 2013, the
country’s debt–GDP ratio was higher than that of peer countries such as Chile and Mexico, and was rising
faster after 2008 than that of Hungary, Brazil, Malaysia and Mexico.
 e South African government has traditionally made relatively little use of foreign nancing, so that
most of the issued public debt is domestic debt. During the period 1960 to 2000, foreign public debt as a
percentage of total public debt uctuated between 10,9% (1976) and 1,6% (1992). During this period, foreign
debt never exceeded 4,3% of GDP. In 1985, 1986 and 1987, foreign loans were used to counter private capital
out ows, but access to the international nancial markets subsequently became increasingly difficult owing
to international nancial sanctions. Access to international nancial markets was normalised in 1994.
Although this, together with the gradual phasing out of exchange control, has provided the scal authorities
with an increased array of foreign nancing options, the rise in the share of issued foreign debt in recent
years has remained modest. A substantial jump occurred in, and was maintained after 2001, resulting in an
average of 10,4% of total debt for 2001 to 2008. At the end of 2017 the ratio was 11,8%. It should be
remembered that non-South Africans can buy rand-denominated debt in the South African secondary
market (in other words, the market in which bond issues are bought from the original or subsequent
owners). Between 2009 and 2013 foreign holdings of domestic (rand-denominated) government bonds
increased signi cantly from 12,8% to 36,4%, subsequently reported to have risen further to 41% at the end of
June 2018, the highest equivalent ratio among emerging market economies. is exposed the government to
the risk of a sudden out ow of capital in the form of the sudden sale of bonds by foreigners who may be
concerned about the relative attractiveness of their South African investments.

Figure 17.1 Public debt in South Africa, 1960–2018 (current prices as on 31 December, % of GDP)

Note: For the last two years, gures are preliminary.

Source: South African Reserve Bank. Various issues. Quarterly Bulletin. Pretoria: South African Reserve Bank.
Table 17.1 Public debt–GDP ratio in South Africa by sub-period, 1960–2017

Note: a For the last two years, gures are preliminary.

Source: South African Reserve Bank. Various issues. Quarterly Bulletin. Pretoria: South African Reserve Bank.

An analysis of the ownership distribution of public debt shows that the majority of public debt is in the
form of long-term bonds held by pension funds (including the Public Investment Corporation [PIC]) and
long-term insurers. At the end of 1992, the PIC alone owned about 37% of the long- and short-term domestic
marketable bonds of the national government. In line with the strategy of bigger investment freedom,
however, this ratio declined to the lowest recent gure of 16% at the end of 2012. At the end of 2017 the ratio
was still at a low level of 16,6%.
e biggest investor in government bonds nonetheless remains the Government Employees’ Pension
Fund. is fund’s investment in government bonds, along with investable funds of other government
pension funds and other public bodies, is channelled via the PIC. Until 1989, insurance companies and
private pension funds were compelled by law to hold 53% of their untaxed liabilities and 33% of their tax
liabilities in xed-interest-bearing public sector securities (Abedian & Biggs, 1998: 261). is provided the
government with a captive loans market. To the extent that the interest on these bonds was lower than would
have applied if government had to compete for these funds in a competitive market, these prescribed
investments constituted a hidden tax on the relevant institutions. For the government, the cost of debt was
therefore below the market rate. is implicit tax, which impacted negatively on savings, was criticised for its
unfairness and adverse in uence on investment performance, and was abolished in 1989.
For some time, the PIC continued to be subject to strict investment requirements. However, this has also
changed and the PIC was increasingly allowed to make market-related investment decisions during the
1990s. is was owing to the fact that the PIC, as the investment arm of the government’s pension funds, was
responsible for the investment yield of these funds. Public employees contribute to a de ned bene t fund. It
means that the weaker the investment performance of the PIC is, the higher the government’s future
obligation to ensure the solvency of these funds will be. For this reason, the government transferred bonds to
the pension funds at various occasions in the early 1990s; it reduced its contingent liability by increasing its
actual debt to ensure future solvency. When the PIC was transformed into a public corporation in 2004, its
investment freedom was formalised further, although this remains subject to political in uences because
the organisation is wholly owned by the government on behalf of its employees. In recent times more doubts
were being expressed about the riskiness of PIC investments. At the time of writing the PIC was under
investigation by a Commission of Inquiry into allegations of impropriety, with possible nancial risk for the
government.
 An intriguing question, which we will address in Section 17.4, is whether, and on what basis, public debt
is justi ed. When attempting to understand the nature and causes of public debt, an important issue that
has to be considered revolves around the purpose for which debt is incurred. For example, are the borrowed
funds to be used to nance current or capital expenditure? Spending on goods and services that are used up
within a speci ed, usually short period is called current expenditure or consumption expenditure
(Bannock, Baxter & Rees, 1971: 82). In scal terms, these goods and services are normally associated with tax
rather than debt nancing. Capital expenditure refers to expenditure on durable items that yield services or
revenue over a long period, such as roads, school buildings, hospitals, irrigation dams and electricity
networks. is kind of expenditure is normally nanced through loans (public debt).
e inverse of the question about the justi cation for debt is whether public debt is something that
should be repaid. Most people would argue that a government is not like a business, the health of which is
determined by factors such as the value of its shares, its pro t, its debt–equity ratio, and measures of liquidity
and solvency. ese criteria are important determinants of whether a business is bankrupt or thriving. If the
business is to be sold, one needs to know its value or net worth to determine the price. Net worth may be
de ned as the difference between the value of all assets and liabilities, that is, the shareholders’ (or, in the
case of government, the taxpayers’) ‘interest’. In the case of government, the mirror image is the net
indebtedness, which is the difference between the value of all liabilities and assets.2 Many would argue that
this kind of information is irrelevant when analysing the nancial state of a government because a
government allegedly cannot become bankrupt or is unlikely to be put up for sale. In fact, one type of
government bond, known as a consul, is a perpetual bond, that is, a bond with an inde nite maturity, never
to be repaid. is is well-known in the UK, but no bond of this nature has been issued in South Africa to
date. e international nancial crisis in Europe has shown the vulnerability of countries with poor scal
track records. Countries like Greece found their government stock trading at very large premiums, to the
extent that they could not nance their budget de cits without nancial support from the European Union,
which came with rather strict scal conditions.
In recent times, the balance-sheet view of public nances has gained much more prominence. e
government is not only the supplier of public goods and services. It is also the custodian of public assets
owned collectively by the citizens (taxpayers) of the country. Informed and enquiring citizens have tended
to become interested in the way in which the government is managing their (public) assets and liabilities.
e net worth of government has become important, not as an indication of the potential selling price of the
government, but as a measure of the quality of scal management and of the impact on the next generation
(in other words, the inter-temporal burden and its implications for inter-generational equity). Attention to
the balance sheet of government has become a feature of public economics. Privatisation, for example, has
raised questions such as whether society is becoming poorer if public assets are sold and whether the
revenue from privatisation should be used to repay public debt or to acquire new assets. Of course, when the
government sells high-worth assets to pay off the debt of state-owned enterprises, the net worth of the
public sector is reduced.
Incidentally, the balance-sheet accounting implied above is also required to answer many of the
questions raised by public auditors who have over the past two to three decades advocated the importance
of value for money in a number of countries such as Canada and New Zealand. In South Africa, we focus
increasingly on commercially oriented questions such as the value of public assets, the (opportunity) cost of
non-earning or badly managed assets and the cost of excessive stockpiling (as was common in the defence
force in the past, for example).

17.4 Theory of public debt

Having now considered the concept of public debt and its size and composition in South Africa, we are in a
position to explore different theories of public debt.

17.4.1 Introduction
In Section 17.3, it was stated that loan nance (debt) is an acceptable method of nancing capital
expenditure, while current expenditure may be a better candidate for tax nance. Is this necessarily true,
especially as the distinction between current and capital expenditure can be ambiguous? What is the
rationale for debt nance, and what are the criteria for choosing between tax and debt nancing of
government expenditure? ese are some of the oldest questions in economics and there are a number of
divergent views on the issue. e answer depends in part on who actually pays the public debt, which is a
question about the nature of the distribution or incidence of the public debt burden.
We begin our analysis of the rationale for debt nance by introducing the concept of the inter-temporal
burden. is refers to the shifting of the burden of the public debt from one generation to the next over time.
e burden of the debt refers to the responsibility for the actual payment of the principal and interest.
e American President Herbert Hoover once remarked, ‘Blessed are our children, for they shall inherit
the public debt.’ Is this necessarily true? ere are different views on this topic,3 and before investigating
them, a few preliminary observations are necessary. When debt is incurred, bene ts accrue to the present
generation since the proceeds of the loan are used to supply public goods and services. If capital goods such
as infrastructure are provided, the future generation also stands to bene t. e debt furthermore establishes
a responsibility to pay interest and to repay the loan or to re nance it. e requirement to repay or re nance
the loan means that a statutory burden is conferred on the next generation. e next generation will pay
interest to bond holders until the bonds expire. In the case of bonds that expire and are not replaced by the
issuance of new bonds (re nancing), the next generation will transfer income to bond holders. In the case of
the re nancing of maturing bonds – a common practice in public nance – the next generation will pay
interest to the new bond holders.
In our study of tax incidence (Section 11.5 of Chapter 11), we saw that there may be a marked difference
between statutory and effective (or economic) tax incidence. e same applies to public debt. e chain of
events set in motion by government borrowing may lead to an economic incidence that may differ
substantially from the statutory or legal incidence. e actual incidence depends on the assumptions about
economic behaviour and the concomitant relationships between economic agents. e answer obviously
also depends on the balance of the scal bene t and burden, that is, the net scal burden.
Two kinds of public debt have to be distinguished. Domestic or internal debt is the debt incurred by
government when borrowing from domestic residents or institutions, that is, when selling bonds in the
domestic primary capital market. e value of the debt is expressed in terms of the home currency. e sale
of the bond does not involve an in ow of foreign capital, and the repayment of principal and payment of
interest do not cause an out ow of funds from the country, provided the transaction is between two
residents. e balance of payments is therefore unaffected by the issuance of these bonds.4 Foreign or
external debt is the debt incurred by government when borrowing from foreign governments, residents or
institutions. e value of the debt is normally expressed in a foreign currency, but can also be denominated
in the home currency. e sale, repayment and servicing of the bond all affect the balance of payments.
ere is an impact on the balance of payments in a number of cases as explained below.
• When bonds are sold to foreigners by the government (in the primary market) and bought by foreigners
from resident bond holders (in the secondary market), an in ow of foreign capital results (affecting the
nancial account). In Figure 1.3 (Chapter 1), this will re ect as an increase in Sf .
• When foreigners sell the bonds to South Africans before the expiry date (in other words, before maturity).
is results in an out ow of capital ( nancial account). In Figure 1.3, this will re ect as a decrease in Sf .
• When bonds are repaid in the hands of foreign investors, this also amounts to a capital out ow ( nancial
account). In Figure 1.3, this will again re ect as a decrease in Sf .
• When interest is paid to foreign bond holders, this is a payment for a foreign service that affects the
current account. In Figure 1.3, this will re ect as an increase in M (a factor service).

Since South Africa started to liberalise its nancial markets, foreign investors increasingly started buying
government bonds in the secondary capital market (in other words, the Bond Exchange of South Africa on
which government bonds are traded after the date of issue), as was mentioned in Section 17.3. is means
that the bond may often change hands so that the debt associated with a bond may, during the ‘lifetime’ of
the bond, be counted as foreign or domestic debt at different times, depending on the nationality of the
registered bond holder at the time.
In nancially integrated markets, the distinction between domestic and foreign debt becomes blurred,
since a foreigner can also purchase bonds issued domestically, just as a South African may buy foreign
issued bonds of the South African government. It is thus more appropriate to distinguish between bonds
issued in the domestic money and capital markets (domestically issued bonds), and bonds issued in foreign
money and capital markets. e distinction between domestic and foreign thus becomes one of source
rather than residence.

17.4.2 Internal versus external debt and the burden on future generations
A view that characterised economic thinking in the 1940s and 1950s was that internal debt does not create a
burden for the future generation. is view is attributed, inter alia, to the American economist, Abba Lerner
(1903–1982). e argument is that certain members of the future generation will inherit a debt repayment or
debt-servicing obligation, but other members of the same generation will be the recipients of these
payments. In other words, members of the future generation owe the debt to themselves. On repayment of
the debt, income is transferred from one group of citizens (those who do not hold bonds) to another group
of citizens (the bond holders). As a whole, the future generation is therefore not in a worse position since it is
capable of the same aggregate level of consumption that would otherwise have been the case. e
repayment of the debt results in an intra-generational transfer or redistribution of resources (in other words,
a transfer within the same generation) rather than an inter-generational transfer (that is, a transfer between
two generations). In terms of this line of thinking, therefore, internal debt is neutral with regard to inter-
generational equity.
e situation is different when external debt is used. In this case, the distribution of the burden depends
on the way in which the funds are used, the fact that interest payments constitute a net transfer of funds to
the rest of the world and the fact that bond holders are now external to the economy. If the borrowed money
is used to nance current consumption expenditure, the future generation has to repay the loan without
enjoying the bene ts. In fact, their income will be reduced by the amount of the loan and/or the accrued
interest that needs to be paid to the foreign lender. Aggregate consumption will be lower than in the case of
domestic debt repayment. Should the money be used to nance capital accumulation (for example, a
railway line), the project’s productivity is crucial. If it is a long-term asset with a real investment return in
excess of the marginal cost of funds obtained abroad (which, in a perfect market, will be the real interest rate
on the loan), the combination of the debt and the performing asset actually makes the next generation better
off. e opposite applies when the investment return is less than the marginal cost of funds.
Today it is realised that the incidence of the burden of public debt is a more complicated matter.
Generations overlap, and the inter-generational incidence of the debt burden may differ according to the
income distribution, tax burdens and in ation rates experienced by future generations, information that is
unknown at the time of the decision to borrow.
It should be noted that, apart from the equity considerations, external debt could alleviate pressures on
the domestic nancial markets (in other words, on the domestic supply of saving) during an economic
upswing. Should the government decide to borrow abroad during these times, external debt may therefore
also ful l a macroeconomic stabilisation role.

17.4.3 Inter-temporal burden

Lerner regarded a generation as consisting of everyone who is alive at a given time. However, if we de ne a
generation as everyone who was born at the same time, several generations exist at any particular point in
time and the burden of the debt may in fact be transferred across generations. An inter-temporal burden is
said to exist.
Suppose the government introduces a new programme that bene ts everyone who is alive at a speci c
point in time and debt nancing is used. Suppose further that a special tax has to be instituted once the debt
repayment commences. It may well be that older citizens who bene ted from the programme are no longer
alive. is implies that they escape the burden of the tax, which has to be carried by members of younger
generations.
e burden of the debt can thus be transferred to future generations. e distinction between external
and internal debt is of no relevance in this case. Even if the debt is completely internal, a burden is created
for the future generation. It is, of course, possible for the next generation to shift the burden forward once
again.
From the point of view of inter-generational income distribution, it is very difficult to anticipate whether
the next generation will be richer than the present generation. It is not inconceivable, of course, that if this is
to be known or suspected, the present generation may deliberately vote for an inter-generational
redistributive scal policy such as using debt nance, much the same as they may vote for intra-generational
redistribution from the wealthy to the poor. Of course, the wealth of the next generation cannot be known,
but it could be that the present generation’s actions add to the wealth of the future generation. If, for
example, there is a legacy of high-quality public infrastructure for the next generation, the present
generation may judge that the next generation will have a much better prospect of generating wealth than
they had. In the eyes of the present generation, debt nancing would then be an appropriate policy
instrument of inter-generational redistribution.
e inter-generational debt burden is more complicated in a country experiencing a major change
(discontinuity) in its political dispensation, such as that experienced during the last quarter of the twentieth
century by countries converting to democracy in Eastern Europe, Asia, South America and Africa (including
South Africa during the 1990s). Often there are strong populist pressures on the new government to renege
on repayment of inherited debt or on foreign countries (or institutions) for debt exoneration. e argument
is that the present generation cannot be held ransom by creditors of the rejected previous regime. However,
the discontinuity in the political system does not imply a discontinuity in debt commitments, which are legal
obligations and represent investments in the hands of bond holders. Moreover, the new generation cannot
claim debt forgiveness while enjoying the bene ts of the assets that were accumulated in the process.
e inter-temporal scal burden reminds us that the distribution of scal bene ts and burdens between
generations cannot be measured in terms of what happens in a particular scal year. A lifetime perspective is
required. In generational accounting, the present value of lifetime taxes to be paid by a representative
person of each generation is compared to the present value of lifetime bene ts to be enjoyed from
government services. e difference is the net tax (at present value), or the net scal burden. A comparison
of the net tax of different generations provides a sense of how scal policy distributes income or purchasing
power across generations (Rosen & Gayer 2008: 468–470).

17.5 Should the government tax or borrow?

On the basis of the foregoing discussion of the rationale for public debt and our analysis of taxation, we
consider the following question: which is the best nancing instrument for public expenditure: taxation or
loans (debt)? Remember that debt is nothing but postponed taxation; in a sense, our question is therefore
one of when to tax, rather than whether or not to tax. We explore the question by referring to the allocation
(efficiency) and distribution (equity) of resources, and by considering the implications for macroeconomic
stability.

17.5.1 Allocation (efficiency)

Is taxation more efficient than debt? e answer lies in its impact on allocative efficiency, that is, the amount
of excess burden created. Earlier we noted that the difference between tax and debt is one of timing. In order
to compare efficiency, therefore, we must inquire as to the impact of the time difference on efficiency.
Consider a particular project that is completed in one year and that may be nanced through either tax or
debt. If tax nance is used, the full cost of the project is nanced by levying a once-off tax. If loan nance is
chosen, the loan will have to be repaid over a number of years, for the purposes of which an amount of tax
will have to be levied every year. If lump-sum taxes are used in each case, there is no excess burden,
efficiency is ensured, and there is thus no difference between tax and debt on efficiency grounds.
Not all taxes are efficient, however. In Chapter 12 (Section 12.2.3), we saw that, in the case of an ad
valorem tax on a commodity, the amount of excess burden increases exponentially with increases in the tax
rate. is means that the excess burden of a single, relatively high tax is bigger than the sum total of the
excess burdens of a series of small taxes that generate the same revenue as the single tax. When a speci c tax
is collected by levying a once-off, relatively high tax rate, it will therefore be less efficient than if the same
amount of tax is obtained by levying a succession of low tax rates. A succession of low tax rates occurs in the
case of debt nancing. e debt-servicing cost is spread over a number of years and requires a
corresponding series of tax revenues. In this case, debt nancing will be more efficient than a once-off tax to
nance the entire project up front. e choice between tax and debt on efficiency grounds, therefore,
depends on the type of tax used.
e efficiency issue has further implications, though. It also extends to the source of the funds. Generally
speaking, taxation reduces private consumption and savings. Debt nance – that is, if the debt is not
monetised – represents a direct use of savings, reducing the amount of savings available to nance private
investment. To the extent that debt nance represents a larger reduction in the country’s savings than tax, it
will be less efficient than tax from the point of view of investment decisions. is is the crowding-out
argument discussed in Chapter 18, Section 18.3.2.
Combined, it is impossible to know a priori which of the above efficiency effects will dominate, that is,
whether tax or debt nance is more efficient. e net effect can only be established empirically.

17.5.2 Distribution (equity)

Intuitively debt nance constitutes the one method whereby more than one generation could contribute to
the nancing cost of activities that confer an inter-generational bene t, something that once-off tax nance
cannot effect. All debt imposes a burden on the future generation. e difficulty lies in knowing in advance
what the concomitant bene t to the future generation will be. If we were to know, for example, that the
future generation would be poorer than the present one, it would make sense to transfer income from the
present to the future generation, for example, by tax- nancing an infrastructural project with long-term
bene ts. e opposite would, of course, apply in the case of a more wealthy future generation.
Ricardian equivalence theory (see Chapter 18, Section 18.3.2) suggests that the government needs not
concern itself with inter-generational equity, since society will voluntarily effect this equity as is preferred.
By increasing their bequests in the face of debt nance, individuals will ensure that the impact of tax and
loan nance on the next generation is equated. From an inter-generational equity point of view, it is thus
immaterial whether tax or debt nance is used.
e nal equity consideration is linked to the bene t approach to taxation (see Section 11.4.1 of Chapter
11). According to this approach, it is fair (and efficient) for a particular group to pay for a particular
programme if members of that group bene t from it. ere is no reason why the future generation should
not pay for programmes that bene t them. To the extent that a programme bene ts a future generation, they
should carry part of the nancing burden, that is, pay for the bene t that will accrue to them. is could be
achieved by means of debt nance.

17.5.3 Macroeconomic stability

From a macroeconomic perspective, the choice between tax and debt arises at the margin; in other words,
should additional expenditure be nanced by taxes or by loans, thus incurring a de cit or increasing the size
of an existing de cit? No one has suggested that the full budget be loan- nanced. Keynesian economists
argue that when unemployment is high, for example, debt- nanced scal expansion is warranted in order to
stimulate aggregate demand until it equals aggregate supply at the full employment level of income. A scal
expansion of this nature may occur with or without a reduction in tax. De cit nance is a choice in favour of
new debt rather than tax and may also entail substituting new debt for existing tax. On the other hand, when
unemployment is low in the Keynesian world, de cit nancing may be in ationary and tax increases (a
lower budget de cit or a higher budget surplus) will be necessary to constrain private spending.
e Keynesian consensus started to break down in the early 1970s, when periods of high unemployment
and high in ation were experienced and when it was no longer possible to increase employment in a non-
in ationary manner (see Chapter 18, Section 18.3.2.4). Lower budget de cits, preferably achieved by
keeping government expenditure in check, became the consensus view of Keynesians and monetarists
(Dornbusch et al., 1994: 397), albeit for quite different reasons. e monetarist argument revolves around
crowding out, explained in Chapter 18, Section 18.3.2. Debt nance has to be reduced to ‘crowd in’ private
investment. e positive impact of reduced budget de cits on investor con dence may outstrip the
depressing effect on income in the short term and may, on balance, be ‘good for growth’. In addition, lower
budget de cits reduce the risk of in ationary nancing. e Keynesian argument is that a reduction in the
budget de cit should be effected by curtailing government expenditure in order to avoid the cost-push
effects of higher taxation; that is, to avoid the cost of using tax rather than debt. Of course, the international
nancial crisis of 2007–2009 threw the debate about appropriate stabilisation policies wide open.
In the era of globalisation, the freedom of choice that countries have to change budget de cits
independently by large margins in order to pursue macroeconomic stability has been reduced substantially.
e strong drive for macroeconomic policy coordination subsequent to the nancial crisis of 2007–2009 is a
case in point. e structural choice in industrial and developing countries has become one of how much and
how fast to reduce excessive public debt and budget de cits, rather than of how high the budget de cit
should be (in other words, how much debt instead of tax).

17.5.4 Summary of views on the impact of debt

We now summarise our ndings, while also incorporating some of the macro issues contained in our
chapter on scal policy (see Chapter 18, Section 18.3). Table 17.2 consists of a matrix of the possibilities
associated with the various viewpoints discussed or to be discussed (the rows) and the basic economic
considerations of efficiency, equity and macroeconomic stability (the columns). Where possible, the name
of the major exponent or school of thought is also indicated.
Table 17.2 Summary of views on the impact of public debt

17.6 Public debt management

Having explored theories of public debt and considered the choice between tax and loan nance, we now
focus on the practical issue of public debt management.

17.6.1 Introduction
us far, our discussion has dealt with the economic justi cation (or not) of borrowing or incurring public
debt. Once a government has decided to use debt nance, another important set of questions arises: when
to borrow, for how long and at what cost, from whom to borrow, where to borrow, which debt instrument to
use and so on. ese questions relate to debt management and, owing to their economic impact, are
important in their own right.
Given the existing debt and debt structure at any point in time, we de ne public debt management5 as
decisions regarding the timing of borrowing, the term structure of the existing debt, the desired future
maturity structure, the nancial instruments, the cost of borrowing and the markets in which new debt is to
be issued. A somewhat more general de nition is used by the International Monetary Fund and World Bank
(IMF and World Bank, 2001): sovereign debt management is the process of establishing and executing a
strategy for managing the government’s debt in order to raise the required amount of funding, achieve its
risk and cost objectives, and meet any other sovereign debt management goals that the government may
have set, such as developing and maintaining an efficient market for government securities. e approach
by the South African government closely resembles this de nition.6
We discuss the questions raised above with reference to the following objectives of public debt
management (which may at times be in con ict):
• Minimisation of state debt cost
• Macroeconomic stability
• Development of domestic nancial markets
• Financial credibility (ensuring access to nancial markets, both domestic and foreign).

Before exploring debt management in terms of these objectives, we explain a few basic concepts and
operational issues.

17.6.2 Bonds and the cost of borrowing

You will recall that the price of a government security (P) is inversely related to its yield (i). If the coupon (the
xed amount of earnings) on a consul (a perpetual bond) is denoted as E, we can write:

Assuming efficient markets so that the long-term interest rate is equal to the yield rate on all bonds, higher
interest rates imply lower bond prices (value) and vice versa. In the case of a security with a nite maturity,
the price is given by the following:

where i denotes the annual yield and n the number of years (periods) to maturity.7
Although the government cannot change the interest payments applicable to previously issued debt
when changing in ation rates affect the real value of the debt, the owner of this type of bond can be
protected against capital losses and realise capital gains if the bonds are sufficiently marketable. e
government, on the other hand, may, for example, be able to realise the bene ts by buying back unexpired
bonds during periods of rising in ation or when these buy-backs would reduce debt service cost.8 e
government itself (National Treasury), the central bank (the South African Reserve Bank) or private market-
makers assume the responsibility of marketing government bonds. Government securities are an important
instrument of monetary policy to the central bank, as interest rates (and hence the money supply control)
are in uenced by transactions in these bonds; these are so-called open market transactions.9 As nancial
markets develop (and in order for these markets to develop), the liquidity of bonds in the secondary capital
market becomes more important. At some stage, private nancial agencies are appointed as so-called
primary dealers or market makers; their functions are to quote (on behalf of the government) two-way
prices (selling and buying prices) for government bonds and to assume the responsibility of always buying
and selling securities in the secondary market.
When interest rates are volatile, the market value of bonds (and the net worth of government or net
indebtedness) changes accordingly. By valuing the government’s outstanding liabilities at current market
prices (so-called mark to market), a more accurate picture of the state of the scus is obtained. If the
nancial assets and liabilities of government were to be managed by a pro t-driven treasury, this valuation
would be done continuously in order to maximise pro ts and minimise losses as a going concern. However,
we will see later that when public debt management has macroeconomic stability as one of its objectives, it
is not solely driven by the pro t-and-loss bottom line.
How much does it cost the government to borrow? e immediate response is that the answer depends
on the interest rate. If the interest rate rises, so will the interest bill, and vice versa. Note that owing to the fact
(or to the extent) that government stock is usually issued at a xed rate of interest or coupon, a rising interest
rate affects only the interest cost pertaining to new debt or debt incurred to replace maturing debt, unless
existing bonds include variable interest bonds.
e interest cost is not the only cost. In modern capital markets, bonds are often issued at a discount,
which means that the government receives a smaller amount than that which will have to be repaid. e
reason for the discount on bonds is that the market rate of interest may be higher than the speci ed yield (or
coupon rate) on the bond when the bond is issued as a result of a number of market factors (such as changes
in economic conditions, monetary policy, demand, risk and so on). Instead of having to change the
announced coupon when a particular bond is issued, the amount of cash received is adjusted. e discount,
together with the coupon rate, therefore provides a better indication of the effective interest cost to the issuer
(or yield to the buyer). In fact, because it is assumed that all future coupon payments can be invested at the
current market rate during the remainder of the bond’s lifespan, this market interest rate is actually known
as the bond’s yield to maturity (YTM). e YTM changes on a daily basis as market conditions change and,
as a result, is beyond the control of the issuer.
Suppose that a bond of R100 (the nominal value) is issued at a discount of 5%. is means that the bond
will sell at a price of R100 – R(0,05 × 100) = R100 – R5 = R95. e investor thus pays R95 for an asset with a
nominal (book) value of R100. e amount received when a bond is issued at a discount (that is, the
discount price) differs from the nominal value. In this example, the government incurs an extra cost of R5
(in addition, that is, to the coupon or interest payment that is applicable to the bond). Suppose this bond has
to be repaid after ve years. In addition to the coupon payments paid over the ve-year period, the
government will have to repay R5 more in capital than the amount originally received, that is, the discount
price of R95 plus the R5 discount cost. e discount on a bond is thus de ned as the difference between the
nominal value and the discount price of the bond.
e gures in our example may appear insigni cant, but the amounts involved can be quite large. In
1998/99, for example, the discount on public debt in South Africa amounted to R6,4 billion, which
represented 12,8% of the total debt cost in that year.
e budget de cit, which is announced by the Minister of Finance in the annual budget speech, does not
account for discount on public debt on an accrual basis. If this were the case, it would result in a higher
budget de cit and state debt cost, and lower amounts of government debt in any particular nancial year.
A bond may, of course, also be issued at a premium. If the market interest rate were to be below the
coupon rate on the day of issue, the bond would be sold at a premium as the effective interest rate would be
lower than the coupon rate. e premium will close this gap. For example, during March 2005, the R15310
(with 2010 as year of maturity) traded at a price constituting a signi cant premium of about 24%. is was as
a result of the fact that the 13% per annum ( xed) coupon rate on that bond exceeded its market yield (YTM)
of about 7,5% by a large margin. is means that for every R1 million in nominal value of the R153 that the
government issued in March 2005, it received R1,24 million from the buyer. However, during scal year
2005/06, the government had to pay a coupon interest of R130 000 (13% × R1 000 000) to the holder of every
bond with a nominal value of R1 million. When an R153, bought in March 2005, matured in 2010, the
government only had to repay the holder R1 million.
ere are various other costs associated with debt, such as conversion costs, which are incurred when
bonds are redeemed before maturity and converted into other bonds, and the cost of raising loans. ese
costs are, however, relatively small.
17.6.3 Foreign borrowing
When foreign borrowing is undertaken, a foreign exchange cost may be incurred in the event of subsequent
exchange rate depreciation. Suppose that South Africa issues a one-year bond of €100 on 1 January. To
simplify the explanation, we assume an initial exchange rate of €1 = R10 on 1 January, a zero interest rate, no
discount on bonds and a 10% depreciation of the rand against the euro during the year. On 1 January, the
government receives R1 000 as the proceeds of the loan. On 31 December, the government has to repay €100.
In order to do this, rands have to be converted into euros. After the depreciation of the rand, the new
exchange rate is €1 = R11. e government will therefore require R1 100 to repay the loan. e depreciation
of the rand has caused an extra cost of R100. An exchange rate depreciation can therefore increase the cost of
foreign borrowing substantially. e opposite happens in the case of an appreciation of the rand against the
issue currency.
One must be careful, though, not to conclude that the cost associated with an exchange rate depreciation
rules out foreign loans altogether. Nominal interest rates are usually higher in a country with a depreciating
currency than in countries with stable or appreciating exchange rates. e latter type of country is normally
also the source of foreign nancing. If, in this example, the bond rate was 2% in Euroland and 10% in South
Africa, the total cost of the foreign loan after one year (assuming interest is paid at the end of the year) would
be the exchange rate depreciation cost of R100 plus the interest cost, which, when converted to rand,
amounts to R22 (calculated as 0,02 × €100 = 11). e total cost equals R100 + 22 = R122. e cost of a local
bond of equal value (ignoring discount cost) would be 0,10 × R1 000 = R100. Had the South African bond rate
been 12,2% (in other words, had the margin between the domestic and foreign interest rates been 10,2
percentage points), there would have been no difference between the cost of domestic borrowing (which
would be 0,122 × R1 000 = R122) and foreign borrowing. Owing to the changing value of the South African
rand against the currencies of countries in which foreign borrowing is undertaken, the scal authorities
recalculate (in rand terms) the value of maturing foreign loans in the year of repayment.
We now return to the objectives of public debt management, beginning with the minimisation of state
debt cost and macroeconomic stabilisation.

Public debt management objectives I and II: Minimisation of state debt

17.6.4
cost versus macroeconomic stabilisation
At the end of the scal year 2017/18, that is on 31 March 2018, total national government debt in South Africa
amounted to R2 243,9 billion. During that scal year the interest of government debt was R162,2 billion,
which was almost 3,5% of the GDP or 11,5% of the government expenditure budget. From the perspective of
cost effectiveness, no one will argue about the importance of debt cost minimisation. How is this effected?
To answer this question, one must keep in mind that a bond is an investment to a bond holder. e
return on this investment has two components: the regular coupon yield E (interest earnings) and the capital
gain (the difference between the price at which it is sold and bought), both of which the investor would want
to be as high as possible. From a cost minimisation perspective, the government as borrower would want the
opposite: the lowest interest possible and the smallest capital loss. Note that in a liquid market, the
government can buy back bonds issued previously, even at a capital gain (which would amount to a capital
loss for the bond holder).
Debt cost minimisation is the result of three factors:
• Firstly, the size of the budget de cit and, consequently, the total amount of debt. e lower the annual
budget de cit is, the smaller the total debt and the lower the debt service cost will be, all other things
being equal.
• Secondly, the interest rate level, which is determined in the money market (short-term rates) and capital
market (long-term rates). In South Africa, the monetary authorities play a key role in determining short-
term interest rates through the repo-rate mechanism. As we shall see when we discuss macroeconomic
stabilisation, the monetary and scal authorities may not always desire the same interest level.
• irdly, there exists a maturity structure for any debt level that minimises debt cost, in other words, an
optimal maturity structure. e government dominates the primary bond market. In 2016, for example,
general government was responsible for 89% of new marketable public sector bonds issued. e
government cannot, therefore, passively take the market interest rate as a given (as a small market
participant would). Government policies (of which scal policy is an important component) and the
open-market transactions in government bonds have a major impact on interest rates and thus on the
borrowing cost. An optimal debt strategy therefore has to calculate the cost of debt, taking account of the
impact on interest rates of the very same debt strategy.

e key factor in debt cost minimisation is the difference between short- and long-term interest rates. e
yield curve (Figure 17.2) shows time-to-maturity of all bonds on the horizontal axis and bond yields to
maturity on the vertical axis. A yield curve can be constructed for any of a number of bonds. We focus on
government bonds, for which the curve is drawn relatively easily as a result of the homogeneous features of
these bonds. Points on the curve represent the relationship between yield and time-to-maturity of
government securities, of which there would typically be a substantial number. Owing to market volatility,
the yield-to-maturity varies often so that the yield curve may change frequently, sometimes even daily.

Figure 17.2 Yield curves of government securities

Generally, there are three types of yield curves:

• A horizontal or at curve signi es no difference between short and long rates, so there would be no cost
advantage in changing the maturity structure of public debt.
• A negatively sloped or inverse yield curve is normally observed during periods of vigorous economic
expansion and close to the peak of the business cycle. Short-term yields (rates) are higher than long-term
yields (rates).
• When long-term yields (interest rates) are higher than short-term yields (rates), the yield curve has an
upward or a positive slope. Typically, this pattern is displayed during periods of economic recession
and moderate economic growth. In this case, higher long-term rates may re ect longer-term in ationary
expectations.

We will now discuss debt management with reference to the upward-sloping yield curve. (e reasoning in
respect of the inverse yield curve can easily be derived by inverting the arguments.)
 In the case of a positive yield curve11, the forward-looking cost-minimising treasurer will sell (issue)
short-term securities (which are relatively expensive12) and buy back long-term securities (which are
relatively cheap). A relatively bigger portfolio of securities with short-term expiry dates provides the
treasurer with much greater manoeuvrability to buy (lock into) long bonds once long rates begin to decline
relative to short rates. e maturity or term structure of public debt will therefore shorten and the debt
service cost will be reduced.
e increased supply of shorter-term bonds, on the other hand, will decrease their prices and put
upward pressure on short-term interest rates. is may well be in con ict with macroeconomic stabilisation
goals. e monetary authorities may have no need for higher interest rates at that point in time; on the
contrary, low(er) interest rates may be viewed as compatible with the pursuance of higher economic growth.
It is at this point that the cost-minimising objective of the scal authorities and the macroeconomic
stabilisation objective of the monetary authorities may well be in con ict.
e question is whether debt management should be pursued actively as a macroeconomic stabilisation
instrument. ere appears to be stronger support for debt management as an instrument of cost
minimisation than for pursuing macroeconomic stability. e argument is that the main contribution by the
scal authorities to macroeconomic stability concerns the size of the budget de cit, rather than the
nancing thereof. e potential con ict between cost minimisation and macroeconomic stabilisation can
give such confusing signals to the nancial markets that it would be more efficient if in debt management,
the scal authorities focused only on cost minimisation. e IMF and World Bank (2001) take this line. ey
recommend that where the level of nancial development allows, there should be a separation of debt
management and monetary policy objectives and accountabilities. In countries with well-developed
nancial markets, borrowing programmes are based on the economic and scal projections contained in
the government budget, and monetary policy is carried out independently from debt management. is
helps to ensure that debt management decisions are not perceived to be in uenced by inside information on
interest rate decisions and avoids perceptions of con icts of interest in market operations. In some countries
(Sweden, for example), a separate debt office was established with the sole brief of minimising public debt
cost, given the size of the budget de cit and public debt.
e counter argument is that the scal and monetary authorities share the responsibility for
macroeconomic stability and that proper policy coordination, rather than separate or independent policy
entities, is the answer. e Keynesian world of integrated policy-making is more in line with this view. e
South African government appears to have moved in the direction of greater distance between public debt
management and macroeconomic policy-making, but remains aware of its coordinating role in relation to
the monetary authorities. No separate debt office has (yet) been established.13
One aspect of public debt that has been acknowledged as an important factor in stabilisation policy is
foreign debt management. In Section 17.4, we explored the theoretical views regarding the case for foreign
debt. Suffice to say that during an economic boom, when I > S, foreign loans by government add to the
supply of savings, relieving domestic demand pressures. Note, though, that cost minimisation may be in
con ict with the goals of exchange rate policy. A situation may arise where foreign borrowing may be in the
interest of the accumulation of reserves, at the same time that (or precisely because) the domestic currency
is under pressure. Although the scal authorities may nd it too expensive to borrow abroad, the monetary
authorities may actually favour it. On the other hand, scal and exchange rate policy can also complement
each other. During 2010/11, for example, when there was concern about the strengthening of the South
African rand against major currencies, National Treasury contributed to a somewhat weaker currency by
borrowing in foreign rather than domestic capital markets.

Public debt management objective III: Development of domestic financial

17.6.5
markets
In South Africa, government bond issues have been a major factor in the development of the market for
loanable funds and the development of nancial markets in general. From the point of view of enhanced
efficiency, in which greater liquidity and better information play an important role, various capital market
developments may be traced to aspects of debt management in South Africa. Marketing of government debt
through primary dealers, which was introduced in 1998, not only improved liquidity but also reduced the
re nancing risk of government. To improve liquidity, a debt consolidation programme consisting of switches
and buy-backs was introduced in 2002. Illiquid bonds were switched into benchmark bonds and/or
repurchased. As the majority of the illiquid bonds had high coupons, the debt consolidation through which
they were replaced resulted in savings on debt service cost.
e National Treasury has also diversi ed the funding instruments available from xed income bonds
and treasury bills to in ation-linked bonds, variable-rate bonds and zero-coupon bonds. An innovation that
generated much interest is the xed-interest retail bond that the government launched on 24 May 2004,
which was designed to offer an inexpensive and attractive saving opportunity to small savers.

17.6.6 Public debt management objectives IV: Financial credibility

Debt management is not only concerned with minimising cost. It is also concerned with the ability to
borrow (see footnote 13 regarding the policy framework of the National Treasury). Do prospective investors
regard the security as a good investment? Does the issuer of the bond (the government) have nancial
credibility? South Africa’s re-entry into international nancial markets, the acquisition of international
sovereign credit ratings in 1994 and the subsequent systematic improvement in the country’s sovereign
credit ratings tell the story of the long road to nancial credibility and the hard work that this required.
Incidentally, a number of African countries have now acquired international credit ratings, albeit at
substantially different levels. Besides South Africa, countries as different as Botswana, Burkina Faso,
Cameroon, Egypt, Ghana, Madagascar, Mali, Morocco and Mozambique have subjected themselves to the
thorough investigation and scrutiny of international credit rating agencies. Credit rating agencies suffered
reputational damage when they did not detect the default risk underlying high-rated mortgage-backed
securities in the United States, a major cause of the international nancial crisis of 2007–2009. Nonetheless,
a number of countries, including South Africa, were downgraded after the crisis and ratings remain an
important yardstick of nancial creditworthiness.
ese agencies basically ask two questions when assessing creditworthiness:
• Firstly, can the country service its debt? is question revolves around a country’s economic ability as
well as the likelihood that it will pay the interest and repay the principal (debt).
• Secondly, will the country service its debt? is is a question about the political will of the country to
repay the debt.

ese questions are relevant irrespective of whether the government borrows domestically or
internationally. One of the asymmetries of economic life is that it is much easier for a government to impair
or destroy its nancial credibility than it is to build it up. e greater the credibility is, the lower the debt cost,
and the better the chances of raising loans per se. We highlight a number of factors that are decisive in
determining nancial credibility.
Market participants need to be convinced of scal sustainability. For example, during the rst half of the
1990s, scal sustainability in South Africa became a matter of concern (Calitz & Siebrits, 2003: 58–59). From
a macroeconomic perspective, the scal situation deteriorated from 1990 until 1994, when the long cyclical
downswing depressed tax revenues and government expenditures were raised by several extraordinary
transfer payments as well as the expansion of social services. Government revenue and expenditure trends
in the rst half of the 1990s resulted in the budget de cit peaking at 7,3% of GDP in 1992/93 and public debt
rising to 49,5% at the end of scal year 1995/96, which gave rise to a debate about whether or not a debt trap
was looming.14 A debt trap is said to be looming when the government has to borrow to pay interest on debt
and runs the risk of default. It is now well-known that the South African scal authorities, to their credit, not
only averted the debt trap, but also succeeded in systematically improving the government’s (and the
country’s) nancial credibility in the domestic and international capital markets. is was achieved by a
combination of measures: greatly improved tax administration and compliance, expenditure restraint and
the use of privatisation income to repay debt. Given the share of government expenditure in the economy,
the resultant reduction in the interest bill as a percentage of GDP released resources for reallocation to
higher priority expenditure areas, notably in the eld of social expenditure. An analysis by Calitz, Du Plessis
and Siebrits (2011) has shown that the sharp increase in the ratio of public debt to GDP during the middle of
the 1990s was less steep if the funding of the government employees’ pensions fund and the takeover of the
debt of the homelands of the apartheid state were counted as public debt when the obligations arose many
years earlier (instead of when the national government took over the debt during the rst half of the 1990s).
Retrospectively, the accusation of ‘weak aggregate scal discipline’ in the early 1990s (see Ajam & Aron,
2007: 746) was excessive, more so because the improved funding of the pension funds reduced, if not
removed, a major liability that otherwise would have required nancing in the future.
Market participants also need to be convinced that the market rules of the game are upheld and
reinforced by government. It is quite conceivable that, owing to distortional interventions in the nancial
markets, governments may encounter difficulties in raising loans in the domestic market or may only be able
to raise loans at high cost, even with relatively low budget de cits. We highlight three South African
examples that contributed to the nancial credibility of government. e rst is the abolition in 1989 of the
prescribed asset requirements in respect of insurance companies and private pension funds, referred to in
Section 17.3. In 2003, fourteen years later, the less intimidating and less distortional, albeit quite forceful,
technique of moral suasion was active in directing investment resources into targeted sectors of the
economy. e nancial sector charter (Financial Sector Charter Council, 2008) was developed voluntarily by
the nancial sector, naturally to avoid legislation of the kind aborted in 1989. Designed to underpin black
economic empowerment (BEE), the charter embodied targets (in total and for institutions) in respect of BEE
ownership, targeted investment in areas where gaps or backlogs in economic development and job creation
have not been adequately addressed by nancial institutions, addressed investment in education and so on.
e second example is the pre-1994 decision by the ANC to dispose of nationalisation as an economic
policy position. In February 1990, Nelson Mandela (who became President in 1994) stated that ‘the
nationalisation of the mines, banks and monopoly industry is the policy of the ANC and a change or
modi cation of our views in this regard is inconceivable’ (Nattrass, 1992: 624). Two years later, having been
confronted by the universal disapproval for nationalisation by world economic leaders at the World
Economic Forum in Davos, Switzerland, he told business people in Cape Town that he would try to persuade
the ANC to dispose of the policy, as it had become clear to him that South Africa would not be able to attract
foreign investment if investors felt that they had the ‘sword of Damocles’ of nationalisation hanging over
their heads. During 2010, a hefty debate about nationalisation of the mines was instigated by the ANC Youth
League. Despite government statements that nationalisation was not policy, support for this continues to
are up from time to time.
e last example of a government that understands and upholds the rules of the ( nancial markets)
game is the decision by the ANC government not to renege on public debt that accrued in the apartheid era.
ese examples illustrate actions that strengthened the development of a non-partisan style of
governance that values and pursues international best practices of good governance in a market-based
economy.

17.6.7 Contingent liabilities15

Finally, a word on contingent liabilities, which represent potential nancial claims against the government.
When triggered, a de nite nancial obligation or liability will arise. Contingent liabilities may be explicit,
such as government guarantees on foreign exchange borrowings by certain domestic borrowers,
government insurance schemes with respect to crop failures or natural disasters and instruments such as
put options on government securities.
Contingent liabilities may also be implicit, where the government does not have a contractual obligation
to provide assistance, but (ex post) decides to do so because it believes that the cost of not intervening is
unacceptable. Examples include possible interventions in respect of the nancial sector, state-owned
enterprises or sub-national governments.
According to the National Treasury, total contingent liabilities in South Africa at the end of the scal year
2018/19 amounted to R879,6 billion (or 17,7% of estimated GDP for 2018) (National Treasury, 2019a: 217).
Government guarantees have the effect of reducing the debt cost to the borrowing institution and are
therefore not free. In 2013/14, for example, Government received fees of R152 million on various guarantees
provided.
Unlike real government nancial obligations, however, contingent liabilities have a degree of
uncertainty: they may only be exercised if certain events occur and the size of the scal payout depends on
the structure of the undertaking. Experience indicates that these contingent liabilities can be very large,
particularly when they involve recapitalisation of the banking system by the government (as was the case
with huge bailouts of nancial institutions in various industrial countries during the international nancial
crisis of 2007–2009), or government obligations that arise from poorly designed programmes for
privatisation of government assets, or the bailout of state-owned enterprises such as was the case at various
times with Eskom and the South African Airways. If structured without appropriate incentives or controls,
contingent liabilities are often associated with moral hazard for the government, since making allowances
ahead of time can result in risky behaviour and thus increase the probability of these liabilities being
realised. As a result, governments need to balance the bene ts of disclosure with the moral hazard
consequences that may arise with respect to contingent liabilities. In Chapter 19 we discuss the issue of the
borrowing powers of sub-national governments in South Africa with reference, inter alia, to moral hazard.

Key concepts
• capital expenditure (page 375)
• consul (page 375)
• contingent liabilities (explicit, implicit) (page 391)
• crowding out (page 380)
• current expenditure (page 375)
• debt servicing (page 370)
• debt trap (page 372)
• discount on a bond (page 384)
• discount price (page 384)
• domestic or internal debt (page 377)
• foreign exchange cost (page 385)
• foreign or external debt (page 377)
• government bonds (page 370)
• inter-temporal burden (page 376)
• mark to market (page 384)
• maturity or term structure (page 387)
• net indebtedness (page 375)
• net tax or net fiscal burden (page 379)
• net worth (page 375)
• net worth of government (page 384)
• ownership distribution (of public debt) (page 374)
• primary dealers (market makers) in government bonds (page 384)
• principal of debt (page 370)
• public debt (page 370)
• public debt management (page 382)
• Ricardian equivalence (page 380)
• treasury bill (page 370)
• yield curve (horizontal or flat, inverse or negative, upward or positive) (page 387)
• zero-coupon bonds (page 370)

SUMMARY
• Public debt is the sum of all of the outstanding nancial liabilities of the public sector in respect of which
there is a primary legal responsibility to repay the original amount borrowed (the principal of debt) and
to pay interest (debt servicing). Most of the time, the reference is to national government debt.
• Debt is predominantly incurred by the sale of government bonds (long-term securities) and treasury bills
(short-term paper) to nance the government’s budget de cit.
• e average size of South African public debt as a percentage of GDP declined during the 1970s and
1980s, and then rose substantially during the early 1990s, until just after the political change in 1994.
During and after the international nancial crisis, South Africa incurred a substantial Keynesian-type
debt increase and concerns about scal sustainability were voiced in certain quarters. Although a
relatively low percentage of public debt has traditionally been sold in international nancial markets,
foreign holdings of domestic (rand-denominated) government bonds increased signi cantly through
secondary market bond transactions, especially between 2009 and 2013. is exposed the government to
the risk of a sudden out ow of foreign capital.
• In recent times, the balance-sheet view of public nances has gained much more prominence. e
government is not only the supplier of public goods and services. It is also the custodian of public assets
owned collectively by the citizens (taxpayers) of the country. e net worth of government has become
important as a measure of the quality of scal management and the impact on the next generation.
Because government debt represents an investment by scores of bond holders (investors) and successive
generations bene t from loan- nanced public infrastructure, timely repayment is crucial for scal
credibility even if major political changes occur.
• Various theories of public debt deal with the justi cation for public debt. A view that characterised
economic thinking in the 1940s and 1950s was that internal debt does not create a burden for the future
generation. On repayment of the debt by a future generation, income is transferred from one group of
citizens (those who do not hold bonds) to another group of citizens (the bond holders). When external
debt is used, however, the distribution of the burden depends on the way in which the funds are used,
the fact that interest payments constitute a net transfer of funds to the rest of the world and the fact that
bond holders are now external to the economy.
• Because generations overlap, even internal debt can cause an inter-temporal burden. It may well be that
older citizens who bene ted from a loan- nanced government programme are no longer alive. is
implies that they escape the burden of the tax, which has to be carried by members of younger
generations. e burden of the debt can thus be transferred to future generations.
• e inter-temporal scal burden reminds us that the distribution of scal bene ts and burdens between
generations cannot be measured in terms of what happens in a particular scal year. A lifetime
perspective is required. In generational accounting, the present value of lifetime taxes to be paid by a
representative person of each generation is compared to the present value of lifetime bene ts to be
enjoyed from government services. e difference is the ‘net tax’ (at present value), or the net scal
burden.
• Ricardian equivalence means that tax nance is equivalent to debt nance and activist scal policy (for
example, increasing the budget de cit to stimulate the economy) becomes ineffective. is happens
because rational citizens realise that a loan will have to be repaid from tax income at some future date
and will therefore increase their bequests by an amount equal to the increase in the tax burden of the
next generation. is constitutes a voluntary reduction in private spending, cancelling the impact on
domestic aggregate demand of the debt- nanced government expenditure. Various criticisms to this
approach have been raised and empirical evidence of its existence is ambiguous.
• Table 17.2 presents a summary of the different views on whether the government should tax or borrow,
with reference to allocation (efficiency), distribution (equity) and stabilisation considerations. e
choice between tax and debt on efficiency grounds depends on the type of tax used. A succession of low
tax rates occurs in the case of debt nancing and entails a lower total excess burden than a once-off up-
front tax payment. Debt nance arguably also serves inter- generational equity when used to nance
capital expenditure.
• Public debt management is de ned as decisions regarding the timing of borrowing, the term structure of
the existing debt, the desired future maturity structure, the nancial instruments, the cost of borrowing
 and the markets in which new debt is to be issued.
• e price of a bond is inversely related to its yield. e cost (for the Treasury) of a government bond
depends on the rate of interest, which may require the issuance of a bond at par, at a premium or at a
discount, management costs, conversion costs and foreign exchange cost (if issued in foreign currency).
• Public debt management has various objectives, which can be con icting. ey are the minimisation of
state debt cost, macroeconomic stabilisation, development of domestic nancial markets and nancial
credibility.
• Debt cost minimisation is the result of three factors: the size of the budget de cit, the interest rate level
and the extent to which an optimal maturity structure can be achieved. A forward-looking, cost-
minimising treasurer will use the yield curve for government bonds to choose between short- and long-
term bonds. e yield curve shows time-to-maturity of all bonds on the horizontal axis and bond yields
to maturity on the vertical axis. It can assume three shapes, depending on the state of the business cycle:
horizontal, negatively sloped (an inverse yield curve) and upward sloped.
• In the case of a positive yield curve, for example, the treasurer will sell (issue) short-term securities
(which are relatively expensive) and buy back long-term securities (which are relatively cheap). A
relatively bigger portfolio of securities with short-term expiry dates provides the treasurer with much
greater manoeuvrability to buy (lock into) long bonds once long rates begin to decline relative to short
rates. e maturity or term structure of public debt will therefore shorten and the debt service cost will
be reduced.
• Because cost-minimisation actions may be in con ict with macroeconomic stabilisation goals, this could
give such confusing signals to the nancial markets that it would be more efficient if the debt
management authorities focused only on cost minimisation. In some countries (Sweden, for example), a
separate debt office was established with the sole brief of minimising public debt cost, given the size of
the budget de cit and public debt.
• In South Africa, government bond issues have been a major factor in the development of the market for
loanable funds and the development of nancial markets in general. Marketing of government debt
through primary dealers, which was introduced in 1998, not only improved liquidity, but also reduced
the re nancing risk of government. e National Treasury has also diversi ed the funding instruments
available from xed income bonds and treasury bills to in ation-linked bonds, variable rate bonds and
zero-coupon bonds.
• South Africa’s re-entry into international nancial markets, the acquisition of international sovereign
credit ratings in 1994 and the subsequent systematic improvement in the country’s ratings tell the story
of the long road to, and hard work in, successfully building nancial credibility. International credit
rating agencies basically ask two questions when assessing creditworthiness: can the country service its
debt (a question about economic ability) and will the country repay its debt (a question about political
will)? ree examples are given of steps that enhanced the SA government’s credibility in the nancial
markets. e credit status of a number of countries, including South Africa, was downgraded after the
international nancial crisis. South Africa subsequently experienced further downgrading because of
rising public debt–GDP ratios, weak economic performance and political and economic uncertainties.
• Contingent liabilities represent potential nancial claims against the government and only become
public debt once the borrowing entity defaults. Contingent liabilities may be explicit or implicit.
Government guarantees reduce the debt cost to the borrowing institution and are therefore not free. If
structured without appropriate incentives or controls, contingent liabilities are often associated with
moral hazard for the government, since making allowances ahead of time can increase the probability of
these liabilities being realised.

MULTIPLE-CHOICE QUESTIONS
17.1 Which of the following statement(s) about public debt management is/are incorrect?
a. e price of a bond is proportional to the interest rate.
b. If the home currency is depreciating, it is always cheaper to issue bonds in foreign capital markets.
c. A positive yield curve suggests that long-term bonds should be sold.
 d. A horizontal yield curve suggests that zero-bonds should be issued.
17.2 Choose the alternative(s) that will correctly complete the following statement: e South African
balance of payments is adversely affected if …
a. the South African government borrows abroad.
b. non-residents buy bonds in the South African secondary bond market.
c. South African residents buy bonds in the Yankee bond market.
d. interest is paid to foreigners owning South African bonds.
17.3 Which of the following statements is/are correct?
a. A zero-coupon bond is always cheaper than a variable interest bond.
b. Public debt management has only the minimisation of debt servicing cost as goal.
c. A primary bond dealer is obliged to quote two-way prices on government bonds.
d. Contingent liabilities are not government debt.
17.4 Which of the following statements is/are correct?
a. In Keynesian thinking, debt nance is always justi ed.
b. Ricardian equivalence means that it is immaterial whether short- or long-term bonds are issued to
nance infrastructure projects.
c. Net indebtedness of the government is unaffected when the budget is entirely tax- nanced.
d. It is allocatively more efficient for the government to nance capital expenditure through loans
than a once-off tax increase.
17.5 Which of the following statements about equity is/are correct?
a. Public debt to nance the salaries of public servants promotes intra-generational equity.
b. Public debt to nance a new highway promotes inter-generational equity, ceteris paribus.
c. A bene t tax to nance a new highway promotes intra-generational equity.
d. It is impossible to know whether public debt to nance an infrastructural project will promote
inter-generational equity unless the inter-generational bene t is also known.

SHORT-ANSWER QUESTIONS
17.1 Suppose the South African government borrows $100 in the USA on 1 January. e exchange rate is R8
to $1.
a. How much will have to be repaid in rand after two years if the rand appreciates by 10% per annum
against the US dollar over the period?
b. What will be the net result of the loan transaction for the South African government at the end of
the period if the US interest rate is 5% per year? (Assume that interest is calculated and paid in
arrears on the last day of each year.)
17.2 In what way does Ricardian equivalence serve equity considerations?
17.3 ‘If the home currency is depreciating, the home government should under no circumstances issue
foreign bonds.’ Discuss this statement.
17.4 What should the debt management strategy of a cost-minimising treasury official be when the yield
curve has a negative slope?

ESSAY QUESTIONS
17.1 Compare the different theories of public debt in terms of efficiency, equity and macroeconomic
stabilisation considerations.
17.2 ‘Tax nance is always superior to debt nance on both efficiency and equity grounds.’ Discuss this
statement.
17.3 Explain how debt management pursues its different objectives and point out potential con icts with
monetary policy.
1 e authors thank Louis Fourie and Zichy Botha for their earlier critical comments and suggestions on sections of this chapter. e
usual disclaimer applies.
2 Sometimes ‘net indebtedness’ is only used in respect of the difference between the nancial liabilities and assets of government.
3 e discussion of these views relies substantially on Rosen and Gayer (2008: 467–472).
4 Note that the balance of payments will be affected when transactions in the secondary market result in a transfer of ownership to
non-residents, as was mentioned in Section 18.2.
5 We follow, with slight amendments, the de nition used by Abedian and Biggs (1998: 280–281). Substantial parts of the discussion
in this section rely on Abedian and Biggs (1998: 276–302).
6 e interested reader will notice this from National Treasury (2004a: Chapter 5).
7 is equation can be used to calculate the yield to maturity, which is the interest rate that makes the present value of the future
cash ow of a bond equal to the bond’s market price if the bond is held to maturity.
8 Buying back bonds is the same as repaying a loan before the due date.
9 For insight into these relationships, refer to the liquidity preference theory and the loanable funds theory; both are normally
studied in a monetary economics course.
10 e practice is to allocate a number such as this to a particular bond.
11 e slope of a normal yield curve is indeed upwards as the yield on longer-dated bonds must be higher than short-dated ones
owing to the so-called risk premium on the longer one.
12 is is as a result of the inverse relationship between the rate of interest and bond prices.
13 In 1996, the National Treasury (1996: 2) released a framework for debt, cash and risk management, stating that ‘… the Debt
Management Body has a direct management responsibility in ensuring the State’s continued orderly access to nancial markets,
both domestic and foreign: and contributing to the absorptive capacity for State debt within these markets, through on-going
market development, product innovation and proper co-ordination of its activities with the monetary management operations of
the South African Reserve Bank.’
14 See, for example, Van der Merwe (1993). is debate clearly indicated the extent to which the private sector also had lost faith in
the potential of anti-cyclical scal policy. Retrospectively, if not by design, the rising de cit portrayed a strong anti-cyclical policy
stance. In nancial circles, however, instead of being welcomed as an attempt to soften the impact of the cyclical downturn, this
caused alarm.
15 is section draws strongly on IMF and World Bank (2001: Section IV.5.2: 31–32).
Fiscal policy

Estian Calitz and Krige Siebrits

Fiscal policymaking and national budgeting are very complex tasks. Gerald Browne, Secretary of Finance in
South Africa from 1960 to 1977, aptly describes the experience of the scal policymaker as follows:

ose who have not personally taken part in such an exercise [of budgeting] may nd it difficult to
appreciate the tremendous pressures for higher expenditure to which the Treasury is exposed – pressures
applied, for the most part, with the best of motives and for expenditure on services of unquestioned merit.
e Minister of Finance and his aides are condemned to ght a lonely and thankless battle for a cause that
is seldom adequately understood.

Source: Browne, G.W.G. 1983. Fifty years of public nance. South African Journal of Economics, 51(1) March, 64.
Copyright © 1999–2015 John Wiley & Sons, Inc. All rights reserved.

e aim of this chapter is to outline the nature of scal policy with reference to theories and empirical
observations, some of which have been discussed in earlier chapters. is chapter also discusses various
aspects of scal policy in South Africa and comments on aspects of the ongoing process of scal reform in
sub-Saharan Africa.

Once you have studied this chapter, you should be able to:
define fiscal policy, and describe fiscal goals and instruments at the macroeconomic, sectoral and
microeconomic levels
discuss the evolution of views on the macroeconomic role of fiscal policy, focusing on the distinction
between the Keynesian and the structural approaches, and the choice between discretionary and rules-
based fiscal regimes
distinguish between the various definitions of budget balance and explain the economic significance of each
explain the importance of distinguishing between a cyclical and a structural budget deficit, and between
active and passive fiscal policy
explain the different types of fiscal frameworks
present the key features of the debate on fiscal rules versus fiscal discretion, also explaining which view you
support and why
describe salient features of fiscal policy in South Africa with reference to theory and fiscal frameworks, and
against the backdrop of international economic trends and aspects of the performance of the South African
economy
describe some of the features of fiscal reform in sub-Saharan Africa in recent years.
18.1 Introduction
e decisions of government concerning the allocation and distribution of resources are embodied in its
scal policies and re ected in its budgets. e term ‘ scal policy’ is normally used in relation to
macroeconomic policy and the contents of this chapter re ect that practice. Our discussion of the nature of
scal policy (Section 18.2) nonetheless recognises microeconomic goals and instruments of scal policy as
well. e reason why we do so is that scal policies aimed at achieving macroeconomic objectives are
seldom sustainable unless they consider or map out the implications for resource allocation at the sectoral
and micro levels as well. For example, if aggregate government expenditure has to be reduced to combat
in ation and all spending programmes are simply cut in equal measure, the efficiency and equity
consequences at the programme and project level of government can be profound. On the other hand, if the
government yields to pressures for more government expenditure at the programme and project level
without taking the consequences for the macro economy and the allocation of resources into account, it
could have serious implications for in ation, balance of payments stability and even long-term economic
growth (which might jeopardise the perceived sustainability of scal policy). Fiscal policymaking and
budgeting therefore constitute a juggling act of balancing ‘unlimited’ demands with limited resources. at,
after all, is what economics is all about.
In this chapter, we rst explore the nature of scal policy, identifying in the process the scal institutions
in South Africa and typically found, in varied form, in the rest of Africa. We then discuss the evolution of
views on the macroeconomic role of scal policy, with special attention to the worldwide shift during the
last quarter of the previous century from active Keynesian anti-cyclical scal policies to what may be
described as the structural approach to scal policymaking. In addition, we consider the apparent revival of
Keynesian economics following the international nancial crisis of 2007–2009 and the ensuing Great
Recession. is is followed by an overview of the nature of scal frameworks and a discussion of the choice
between rules-based and discretionary scal regimes. ereafter, salient aspects of scal policy in South
Africa are outlined. e chapter ends with comments on the ongoing process of scal reform in sub-Saharan
Africa.

18.2 The nature of fiscal policy

is section de nes scal policy and outlines its goals and most important policy instruments. It also
introduces the scal authorities in South Africa.

18.2.1 Definition
Fiscal policy may be de ned as decisions by national government regarding the nature, level and
composition of government expenditure, taxation and borrowing aimed at pursuing particular goals. Like all
forms of economic policy, scal policy has both an active element (when a deliberate step is implemented to
do something, for example, to increase the budget de cit) and a passive element (when there is a deliberate
decision to do nothing or to refrain from doing something, for example, when no tax increases are
announced in a particular budget).

18.2.2 Fiscal policy and the budget constraint

At the very beginning of this book (Section 1.1 of Chapter 1) we were reminded that the scarcity of resources
and the associated budget constraint constitute the core of the economic problem. e need to know and
understand the budget constraint and resulting need for prioritisation, also underpins scal policy choices.
ese choices dictate decisions at the macro level that affect the functioning of the entire economy, as well
as sectoral and microeconomic decisions that affect parts of the economy. Moreover, it also sets rules of
good nancial housekeeping in government institutions, which need to abide by these rules before they can
expect the same of the economy as a whole.
Our point of departure is therefore that the government has to set an example of good governance, in the
sense of conducting scal policy according to precepts such as those set out in South Africa’s Public Finance
Management Act (see Section 18.6.2). is requires a good understanding of two sets of causalities. e rst
is that the basis of the government’s scope for mobilising resources by means of taxation is a credible view
about the future generation of resources by the economy and how much of that can be used to nance the
functions that society entrusts to the government. e second is to know how the use of these resources
affects the welfare and behaviour of private economic agents – and in such a way that the generation of new
resources are optimised. ese bidirectional causalities are also visible from the ows illustrated in Figure
1.2 of Chapter 1.
As far as budgeting is concerned, this requires the setting of short- as well as medium-term revenue
constraints at the start of the budgeting process. is makes prioritisation essential from the outset and
ensures that budget requests do not become mere wish lists. It also ensures parliamentary assessment of the
executive’s proposed national budget as a comprehensive and integrated set of resource-constrained
expenditure programmes.
We will explain below that macroeconomic policy has developed over time to become an important
function of government. Fiscal policy is a vital element of this function. Macroeconomics developed largely
after the Great Depression of 1929–1933 and the government’s macroeconomic policy role expanded along
with the growing size of government on account of increases in the type and extent of public services
expected of government. e extent and pervasiveness of government’s associated resource mobilisation
enabled it to play a major role in the behaviour of economic agents.

18.2.3 Goals of fiscal policy

We distinguish the following macroeconomic goals of scal policy:
• Economic growth
• Job creation
• Price stability
• Balance of payments stability
• A socially acceptable distribution of income
• Poverty alleviation.

Note that this list contains none of the elements of the annual budget of the government, such as
government functions, programmes and taxes. e reason for this is that these elements are not goals; they
are the instruments that the government uses to pursue these goals. Note also that price stability, balance of
payments stability and cyclical economic growth are short-term goals; the others (including long-term
economic growth) are of a longer-term or structural nature.
e sectoral goals of scal policy include the following:
• e development of particular economic sectors, such as agriculture, tourism, mining, manufacturing or
the nancial markets
• e pursuance of social goals pertaining to sectors such as housing, education, health and welfare
(policies of this nature are often referred to as social policies; see Chapters 8, 9 and 10).

It is also possible to specify microeconomic goals of scal policy. Goals of this nature relate to scal action
aimed at a single product, activity, economic participant, or group of participants. Normally they can be
seen as sub-divisions of sectoral goals. e following are examples of microeconomic goals:
• Improving efficiency by addressing negative externalities in respect of a particular product (for example,
tobacco) or activity (for example, toxic waste disposal by a chemical plant); see the discussion of
externalities in Chapter 3 and the discussion of indirect taxes in Chapter 16
• Combating poverty (the equity consideration) by intervening in the market for a particular product (for
example, a bread subsidy)
• Pursuing goals with regard to a particular geographical area (suburban or rural), for example, where
government- nanced infrastructure and housing subsidies for low-income earners are incorporated in a
residential development project. Regional goals may even transcend national borders, as re ected in the
development of trans-border game parks or water supply facilities.

Fiscal policy is not the only tool for pursuing each of these sets of goals; in fact, it would be wrong to expect
of scal policy to be the panacea for all economic problems. Monetary policy, trade and industrial policy,
competition policy and labour policy are important allies in achieving these goals. It is quite often necessary
to prioritise the goals and also to recognise that they may be in con ict. For example, it may not always be
advisable to stimulate economic growth further, as doing so may fuel in ation. e government must then
decide whether economic growth or price stability should receive the highest priority. In these
circumstances, we say that there is a trade-off between economic growth and in ation (a related example is
the well-known Philips curve trade-off between in ation and unemployment). Fiscal policy differs
depending on whether the growth objective or price stability receives the highest priority.
Certain policies or policy instruments are more effective in pursuing some goals than others. An increase
in interest rates (a monetary policy measure) may, for example, achieve quicker results than a tax increase (a
scal policy measure) if private spending is to be reduced to combat in ation. e policy authorities must
therefore not only decide on the priority of policy goals, but also choose the most effective policy
instruments for the job at hand. We return to this point when we explain the different lags in scal
policymaking (Section 18.3.2).
What happened to efficiency and equity, you might ask? Are they not the ultimate goals of economic
policy? Have we not on numerous occasions emphasised these as the two pre-eminent considerations when
assessing any scal action? In earlier chapters, we have extensively studied various theories regarding public
goods, government expenditure and taxation. Efficiency and equity were recurrent themes in all of these
theories. We paid particular attention to the following:
• e conditions for Pareto efficiency in the allocation of resources between the supply of public and
private goods and the equity implications (Chapters 2, 3 and 5)
• e extent to which different voting rules produce efficient outcomes (Chapter 6)
• Identifying taxes that maximise efficiency (or minimise inefficiency) in the allocation of resources in the
private sector and assessing the equity implications of different taxes (Chapters 12 and 13)
• e trade-offs between efficiency and equity that have to be considered when dealing with the issues of
poverty and the distribution of income (Chapters 8, 9 and 10).

In Chapter 17, we paid attention to the choice between taxes and debt on the basis of efficiency criteria,
while considering the inter- and intra-generational distributional consequences of debt nancing.
e reason why we did not mention efficiency and equity before in this section is that they do not readily
lend themselves to the speci cation of quanti able targets for the economy as a whole. Instead of equity, we
therefore use more speci c goals for which we can formulate quantitative criteria, such as poverty
alleviation or a socially acceptable distribution of income. e government may decide, for instance, on a
programme to reduce the Gini coefficient. e Gini coefficient was explained in Section 8.1 of Chapter 8.
Under certain conditions, the promotion of economic growth and job creation will also serve the equity goal.
Efficiency goals can be developed at the level of government programmes and the tax system can be
designed with efficiency in mind, but it is not that easy at the macro level. In general terms, higher economic
growth may be seen as a re ection of greater efficiency, but this is by no means obvious. Low in ation may
also be a barometer of efficiency, just as high in ation might be indicative of a lack of it. e ‘ultimate’ goals
of efficiency and equity are thus included (or subsumed) in the list of macroeconomic goals of scal policy.

18.2.4 Instruments of fiscal policy

As in the case of goals, we also distinguish between macro and micro instruments of scal policy. e macro
instruments include total government expenditure, the economic categories of consumption and capital
expenditure (in other words, the composition of government expenditure), the total tax amount and the
budget de cit as well as the way in which the de cit is nanced. e sectoral or micro instruments include
the various expenditure votes and programmes (for example, education, health and defence) and the
concomitant criteria for the mobilisation and allocation of public and private resources, the different types
of taxes and their rates1, and the different dimensions of the public debt (such as maturity, ownership
structure and so on).

18.2.5 The fiscal authorities in South Africa

e key gure in scal policymaking is the Minister of Finance, who is given certain statutory powers by acts
of parliament. He or she has the authority to levy taxes, allocate state income (tax and non-tax revenue), and
borrow funds domestically and internationally. No state guarantees can be given for the borrowing of money
without the approval of the Minister of Finance. He or she is also responsible for the protection of the
country’s gold and foreign exchange reserves. e Minister has the authority to take and implement
decisions on some matters immediately; these include changing the rates of value-added tax2, excise duties
or the fuel levy during the course of the government’s nancial year, or providing guarantees (at a cost) for
foreign borrowing by parastatals (or state-owned enterprises or companies) such as Eskom and Transnet.
On other matters, such as changing income tax rates or implementing the appropriation of state monies in
the annual budget, parliamentary approval in the form of speci c acts of parliament is required before any
changes can be made. e Minister of Finance does not take important decisions without consulting and/or
obtaining the approval of Cabinet. He or she is accountable to parliament for all decisions made.
e Minister is responsible for the coordination of macroeconomic policy, and his or her statutory
powers cover all the scal policy instruments of government expenditure, taxation and borrowing. e
South African Constitution furthermore requires consultation between the Minister of Finance and the
South African Reserve Bank3 regarding the implementation of monetary policy, the Bank’s generic policy
function. In practice, therefore, the Finance Minister is responsible for macroeconomic policy formulation
and coordination, lays down the basic framework for monetary and exchange rate policy, and manages
scal policy.
e two key institutions that bear the responsibility for macroeconomic policymaking are the National
Treasury4 (macroeconomic and scal policy, expenditure allocation and control) and the South African
Reserve Bank (monetary and exchange rate policy). Another very important scal institution is the South
African Revenue Service (SARS). e responsibilities of SARS not only include tax collection and the
enforcement of tax law; SARS also plays an important supportive and advisory role in the determination of
tax policy. Close coordination between all of these institutions is essential for effective economic
policymaking.
ere is an old saying that monetary policy begins in the Treasury. is signi es much more than the fact
that public debt is nanced by the issuing of government bonds, which constitute the main instrument of
open-market policies by the central bank. It is a statement about the close links between scal and monetary
policy in the pursuit of macroeconomic goals. e economic impact of scal and monetary policies is such
that the scal and monetary authorities have to study and monitor the combined impact of these measures
on economic behaviour systematically and regularly. e scal policy menu that a particular country selects
has to be framed in the context of a coherent macroeconomic policy strategy that includes monetary policy
and a number of other policies (for example, trade and competition policy). e implementation of a
strategy of this nature requires regular consultation and active coordination among the Minister of Finance,
the Governor of the Reserve Bank and their respective staff. e issue of the coordination between the scal
and monetary authorities featured before in our discussion of public debt management in Chapter 17.
Another important form of coordination pertains to the formulation of tax policy, where close
cooperation between the National Treasury and SARS is essential.
e scal approval of the annual budget occurs by way of an act of parliament, the watchdog of the
public purse, thus giving effect to an important statement of scal policy at macro, sectoral and micro level.
Traditionally, the South African parliament only had the power to approve or reject the national budget in its
entirety. Two recent reforms aimed at strengthening the oversight role of parliament were added. e rst
was the Money Bills Amendment Procedure and Related Matters Act (Republic of South Africa, 2009), which
creates a procedure through which changes to appropriations may be considered in parliament. Secondly, a
Parliamentary Budget Office was established in 2013 in terms of Section 15 of the same Act to provide
independent, objective and professional advice and analyses to parliament on matters related to the budget
and other money bills.

18.3 The macroeconomic role of fiscal policy

Mainstream views on the macroeconomic role of scal policy have changed signi cantly over time,
re ecting the interaction between continuously evolving theoretical ideas and new policy challenges. is
section rst outlines the nature and shortcomings of the Keynesian approach to scal policy, which is also
known as scal activism or anti-cyclical scal policy. We also brie y contrast this with the Ricardian
equivalence view - an approach that pre-dated Keynesian thinking, but still has important proponents. is
is followed by a discussion of the structural view of scal policy that largely superseded the Keynesian
approach. e nal part of the section points out that the cyclical effects of scal policy have recently come
under renewed scrutiny, in that elements of the Keynesian or activist approach complement a revised
structural approach to the macroeconomic role of scal policy.

18.3.1 The Keynesian approach

e Keynesian approach (also known as anti-cyclical scal policymaking) came to dominate scal
policymaking after 1945. In the three decades that followed, most governments saw it as part of their task to
manage aggregate demand actively with the aim of achieving equality between aggregate demand and
aggregate supply at the full employment level of income. Such manipulation of aggregate demand
constitutes the active policy dimension of Keynesian scal policy. Keynesian economists and policymakers
also believed that the structure of income taxes and unemployment bene ts strengthens the demand-
stabilising effects of active scal policy. e argument was that income tax and unemployment bene ts act
as automatic or built-in stabilisers because changes in income would automatically (or passively) trigger
changes in tax revenue and transfer payments that would stabilise aggregate demand, income and output.5
Given any set of statutory taxes and social security commitments, automatic stabilisers function much like
rules in the sense of being non-discretionary. In this section we explain the active and passive elements of
Keynesian scal policy and then outline the shortcomings that caused it to fall out of favour from the 1970s
onwards.
Figure 18.1 introduces an analytical apparatus that will feature again later in this chapter.6 e gure
depicts government revenue and government expenditure as functions of national income (Y). e
horizontal line G0 represents government spending. We assume that the government sets its outlays
annually; hence, public expenditure is not systematically related to the level of economic activity. (is
assumption is realistic in the context of most developing countries, including South Africa, where
unemployment bene t schemes are rudimentary or even non-existent. e governments of most industrial
countries, however, maintain extensive unemployment bene t schemes. e government expenditure line
of industrial countries would slope downwards because lower levels of income are associated with higher
unemployment and, consequently, higher levels of government spending on unemployment bene ts.) e
government also sets tax rates, but the yields on the different taxes vary with the level of economic activity.
Hence, the tax revenue curve (T0) slopes upwards because individuals pay more tax when income increases
and vice versa. e positive relationship between income and tax revenues would be strengthened if some
taxes had progressive rate structures (a common example is the personal income tax system; see Section
13.3 in Chapter 13). Figure 18.1 also shows examples of the three possible budget outcomes. At income level
Y0, the budget is in balance at point A, where tax revenue (t0) equals government expenditure (g0). Income
level Y1 yields a budget de cit equal to the distance BC; here, government spending (g0) exceeds tax revenue
(t1). A budget surplus equal to distance DE occurs at income level Y2, where tax revenue (t2) exceeds
government spending (g0).
Figure 18.1 A framework for analysing Keynesian anti-cyclical fiscal policy

Figure 18.2 employs this framework to illustrate the distinction between active and passive scal policies.
Our point of departure is point A at income level Y0. e budget is in balance with tax revenue t0 and
government spending g0. Suppose the economy in question experiences an exogenous shock (say, a
decrease in export earnings) that reduces income to Y1. e fall in income reduces total tax revenue to t1
(point C on curve T0). e drop in tax revenue cushions the impact of the adverse shock because it
represents a reduction in the extent of leakage from the circular ow of income and expenditure in the
economy (see Section 1.5.3 in Chapter 1). e budget now exhibits a de cit (equal to g0 – t1 or the vertical
distance BC), which Keynesian economists traditionally regard as a stimulus to economic activity. Note that
the government took no active steps. e stabilising in uence and/or counter-cyclical stimulus to economic
activity are the entirely spontaneous results of structural aspects of the tax system, hence the name
‘automatic scal stabilisers’. e working of automatic scal stabilisers, or passive scal policies, is depicted
as movements along the tax revenue and government spending curves. To illustrate active scal policy, we
return to point A. e government now increases its outlays, shifting the government spending line upward
to G1. is step is a stimulus to economic activity that increases income and, with a lag, also tax revenue. e
extent of the stimulus and the time it takes to have its full effect on the economy depend on the strength of
the multiplier effect. One of the possible outcomes during this process of adjusting to the higher level of
public spending is point E, where the income level is Y2, government expenditure is g1 and tax revenue is t2.
A budget de cit equal to g1 – t2 (the vertical distance DE) has arisen. In contrast to the de cit depicted by the
distance BC, de cit DE resulted from active manipulation of a scal policy instrument. Active scal policies
are shown as shifts or rotations of the tax revenue and government spending curves.
Figure 18.2 The distinction between automatic stabilisers and active fiscal policy

As indicated, most Keynesian economists were optimistic about the ability of the automatic scal
stabilisers to moderate cyclical uctuations in economic activity. ey agreed, however, that the stabilisers
can only reduce this instability; they are not able to counteract it fully. erefore active measures were the
mainstay of the Keynesian approach to scal policy. In recessionary conditions, policymakers tried to
increase the budget de cit or change a budget surplus into a balanced budget or a budget de cit by
increasing government expenditure (shifting the G line upwards) or by reducing taxes (shifting the T curve to
the left or rotating the T curve in a northwesterly direction). ey tried to do the opposite in situations where
economies became over-heated during an economic boom and experienced (rising) in ation: reduce
government expenditure (shift the G line downwards) or increase taxes (shift the T curve to the right or tilt it
downwards).

Box 18.1 Different budget balances

Three concepts of the budget balance can be defined, each of which has a different use. In all three cases, a
negative balance signifies a deficit and a positive balance signifies a surplus. Firstly, the conventional balance
is equal to the difference between total revenue and total expenditure. Total revenue consists of tax and non-
tax current revenue (the latter includes entrepreneurial and property income as well as administrative fees and
charges), capital transfers (which includes the sale of fixed capital assets) and other receipts (such as
recoveries of loans and advances). This balance is a measure of the total loan finance and other financing that
the fiscal authorities require in a particular year to balance the budget. Budget balances are calculated at
current market prices and expressed in local currencies. Hence, the amounts cannot be compared readily
across countries or years. For this reason, it is customary to express the budget balance as a ratio or
percentage of nominal GDP (for example, 2% of GDP). One should keep the following in mind when interpreting
budget balances:
• In most countries (including South Africa), the budget balance is not a comprehensive reflection of the
borrowing requirement of the public sector as a whole because public enterprises and other extra-budgetary
institutions are responsible for a significant portion of public sector loan financing. Many countries therefore
also calculate and publish the net public sector borrowing requirement (PSBR) as a comprehensive
measure. The net PSBR consists of the net borrowing requirement of the general government, extra-
budgetary institutions, social security funds and non-financial public enterprises (in other words, the
borrowing requirement of these government entities after allowing for the refinancing of maturing debt). The
non-financial public enterprises exclude institutions such as the Development Bank of Southern Africa
(DBSA). Because the DBSA lends to public institutions, we would be counting some borrowing requirements
twice if the DBSA’s requirements were also included. Whilst the PSBR serves as an indicator of the impact
of the entire public sector on private savings and financial markets in general, it is rather difficult to manage
changes in the PSBR in policymaking actively because it is the outcome of a multitude of decentralised
decisions by parties within and outside national government departments.
• The state of the business cycle can significantly affect the size of budget balances. During an economic
downswing, government revenue (notably individual and company income tax) tends to be lower than the
longer-term trend, while government expenditure tends to rise above the trend (especially if the country has
a well-developed social security system). At the same time, nominal GDP (the denominator) is below the
trend line. Hence the budget deficit (in money terms and as a percentage of GDP) tends to be higher than
its trend value. The opposite happens during an economic upswing. Analysts sometimes take the impact of
the business cycle into account by calculating cyclically adjusted budget balances. We return to cyclically
adjusted budget balances in Section 18.3.4.
Secondly, the current balance is the difference between total current revenue (tax and non-tax revenue) and
total current expenditure (including interest payments). Calculations of the current balance therefore ignore
capital revenue (for example, privatisation income from the sale of non-financial assets) and capital expenditure
(for example, public outlays on the construction of dams, roads, schools, hospitals and so on). The current
balance is a measure of the extent of saving by government. A current surplus (a situation when the
government’s current revenue exceeds its current expenditure) implies that the government saves some of its
current revenue by using it to finance capital expenditure. Conversely, a current deficit implies that the
government dissaves: it has insufficient revenue to finance all of its current spending, and has to borrow or sell
assets to pay salaries and wages, interest on the public debt and so on. This is an example of what is
sometimes idiomatically described as ‘selling the family silver to buy groceries’. As a rule, borrowing can be
justified if it is used to finance capital expenditure, which should yield a return over a number of years. From an
intergenerational equity point of view, borrowing to finance current expenditure cannot be justified as it means
that future generations will have to pay for the consumption enjoyed by the current generation. Moreover,
government dissaving reduces the pool of savings that is available to finance capital formation in the economy.
In the national accounts, gross saving equals the amount of consumption of fixed capital (formerly known as
provision for depreciation), plus saving by households, corporations and government. If the government
dissaves, its contribution to gross saving is negative. The government sector then draws on the savings of the
household and corporate sectors to finance current outlays that usually do not contribute to economic growth.
Lastly, the primary balance is calculated as the difference between total revenue and total non-interest
expenditure. The primary balance should immediately be recognised as the conventional balance plus interest
expenditure. This is a measure of the government’s ability to service its debt (pay interest) through ordinary
revenue. It measures the impact of the budget on the government’s net indebtedness, that is, its liabilities net
of assets (see Section 17.3 of Chapter 17). If the primary balance is positive (in other words, there is a
primary surplus) and larger than the interest bill, government’s ordinary revenue (which is predominantly tax
income) is sufficient to pay all of the interest on public debt and redeem at least part of its debt or finance
some of the public investment. Net indebtedness is reduced. If the primary balance is positive (but smaller
than the interest bill), a portion of the interest bill is tax-financed. The rest is loan-financed (capitalised) and
added to the public debt, thus increasing net indebtedness. If the primary balance is negative (in other words,
there is a primary deficit), all of the interest and some of the current expenditure are loan-financed, thus
adding to the public debt. In a situation of this nature, government is borrowing to finance all its interest
payments as well as some non-interest current expenditure. This would also increase the net indebtedness of
the government. A negative primary balance is normally regarded as an unsound fiscal practice, particularly
when it increases the risk of runaway public debt. A general rule of thumb is that a negative primary balance
can be maintained for some time without an increase in the government’s debt–GDP ratio, provided the real
 rate of economic growth is higher than the real rate of interest in the economy. Another variation of this is that
if the primary balance is improving in the face of rising public debt–GDP ratios, the government is viewed as
being serious about fiscal sustainability.

e most widely used barometer of the extent of anti-cyclical scal policy is the budget de cit. Box 18.1
contains an overview of different budget balances.
Practical efforts to expand or contract an economy by means of scal policy are subject to various
limitations. ese limitations can render scal policy largely ineffectual and even cause it to produce
perverse results by making scal policy pro-cyclical rather than counter-cyclical. In the next section, we
discuss the implications of these and other in uences on the effectiveness of scal policy.

18.3.2 Shortcomings of anti-cyclical fiscal policy

As mentioned, the Keynesian approach to scal policy rose to prominence after World War II. Its heyday was
comparatively short, however: its support waned steadily from the mid-1970s onwards. Martin Eichenbaum
(1997: 236) summarised the opinions of mainstream macroeconomists at the end of the twentieth century as
follows:

In sharp contrast to the views that prevailed in the early 1960s, there is now widespread agreement that
counter-cyclical discretionary scal policy is neither desirable nor politically feasible. Practical debates
about stabilisation policy revolve almost exclusively around monetary policy.

Why did scal activism fall from favour so dramatically? e answer to this question lies in several practical
and theoretical developments in macroeconomics during the last third of the twentieth century and the rst
few years of the current century.
At least three sets of considerations made experts doubt whether it is possible to conduct anti-cyclical
scal policy effectively. e rst of these is that scal policymaking entails various stages, each with its own
delays or time lags. ere are four lags:
• e recognition lag is the delay between changes in economic activity and the recognition that the
changes have occurred. It takes time to prepare and release economic data such as the national accounts
(which provide information on economic growth and the state of the economy in general). In South
Africa, this data is published every three months in the Quarterly Bulletin of the South African Reserve
Bank. When the annual budget of the South African government is presented in February every year, the
latest available GDP gures are estimates based on gures from the end of the third quarter of the
previous year. If, for example, the Minister of Finance wants to present a budget to stimulate economic
growth, a signi cant margin of error is possible, since official information about the performance of the
economy is lagging by six months – and then the estimates are still subject to changes.
• e decision lag refers to the time that elapses between the recognition of the problem and the decision
on how to react. Various factors play a role in this regard, such as the analysis of different options, the
time required for discussion between officials and ministers, and, eventually, the speed with which
Cabinet takes a decision. e South African Constitution places a high premium on consultation and
various consultative forums exist for this purpose, such as the National Economic Development and
Labour Council (Nedlac) and the statutory Budget Council and Budget Forum (bodies of consultation
with provinces and local government respectively). e forums and various consultative processes, such
as the public hearings of the parliamentary committees, may all contribute to the decision lag. ese lags
are particularly evident when legislation is required. e national budget, for example, is presented in
February, but cannot be implemented before parliament has enacted it. In South Africa, this normally
happens more or less during June of each year (that is, about three months after the start of the scal
year). e government therefore has a standing legislative authority to spend funds provisionally until
the budget is approved by parliament. is includes provisions that enable the government to make
 limited changes during the course of the year when urgent matters arise. Matters of this nature would
otherwise have to stand over until the next year.
• e implementation lag refers to the period after the decision has been taken, but before it is
implemented. is lag mostly arises from procedures of orderly and accountable government. An
example is the time that it takes for government departments to implement approved capital expenditure
programmes, such as is required by tender procedures. e time lag between the approval of a capital
project such as a national road and its opening for traffic is quite long. Administrative procedures in the
private sector also in uence the implementation lag. Changing the VAT rate, for example, requires
adjustments to nancial documentation, cash registers and other automated business machines in the
private sector. e scal authorities have to allow time for these adjustments before they can implement
rate changes. Internet trade, on the other hand, reduces the implementation lag in the sense that
transaction conditions and documentation can be adjusted and communicated much quicker.
• e impact lag refers to the time it takes before an implemented policy measure begins to affect
economic behaviour. An increase in income tax, for instance, can take quite some time to realise its full
impact on private expenditure. Taxpayers may not immediately behave as though their after-tax
spending power has dropped. ey may reduce personal savings, for instance, in an attempt to maintain
their living st andard for some time. Another example of an impact lag arises when businesses delay
price increases because of higher taxes to reap some temporary competitive bene ts. On the expenditure
side of the budget, certain programmes (for example, welfare bene ts) have a much shorter impact lag
than others. e time lag may be quite long, for example, when transport or defence equipment with
long delivery lags is ordered.

ese lags make it particularly difficult for scal policymakers to react to uctuations in economic activity in
time.7 If the economy is overheating, for example, it could take so long to implement a tax increase that a
recession may already have superseded the upswing by the time that the intended dampening of demand
takes effect, by which time a stimulus would be more appropriate. Intended counter-cyclical scal policy
then becomes pro-cyclical in the sense that it deepens and prolongs (rather than smooths) cyclical
uctuations in economic activity. Pro-cyclical scal outcomes were common in industrial and developing
countries during the era of scal activism. In South Africa, Strydom (1987), for example, found that the
structure of the public nances had a destabilising rather than a stabilising effect on the economy from 1960
to 1986. More recently, Du Plessis, Smit and Sturzenegger (2007) also found evidence of scal pro-cyclicality
in South Africa for the period 1994–2006, as did ornton (2008) with regard to government consumption
expenditure in 37 low-income countries between 1960 and 2004. In general, the decision and
implementation lags are normally shorter in the case of monetary policy. is serves to increase monetary
policy’s ‘comparative advantage’ over scal policy as far as macroeconomic stabilisation is concerned. At the
time of the Great Recession, however, it was realised that monetary policy also has its limits when nominal
interest rates approach the lower bound of zero, a situation known as the liquidity trap.
Monetarists emphasise a second constraint to the effectiveness of anti-cyclical scal policy: the
possibility that an expansionary scal policy will push up interest rates and in this way crowd out private
expenditure. Suppose an initial equilibrium in the goods and money markets is disturbed by a budget de cit
incurred through an increase in government spending. What effect will it have on the economy? e
increase in government expenditure means that aggregate demand increases, which sets the multiplier
process in motion. e resultant increase in income leads to an increase in the demand for money. If the
supply of money remains constant in real terms, the excess demand for money causes interest rates to
increase. Higher interest rates dampen private investment and thus aggregate expenditure. is secondary
reduction in aggregate demand dampens the initial multiplier effect, resulting in a lower new equilibrium
level of income than would have applied if interest rates had remained unchanged. Given the negative
relationship between interest rates and investment, this type of nancial policy therefore dampens the rate
of private capital formation (in other words, private investment) in the economy. is phenomenon is called
crowding out and may be de ned as the dampening of private investment on account of increases in
interest rates associated with an increase in public expenditure, especially if the latter is debt- nanced.
Crowding out can occur when government, through its borrowing, competes with the private sector for
funds. is line of argument is associated with the neoclassical school of thought.
 is crowding out thus neutralises anti-cyclical scal policy measures by dampening the multiplier effect
of a scal policy stimulus. Monetarist analysis of scal policy also emphasises the effects of different ways of
nancing budget de cits. Monetarists are particularly concerned about the extent to which the government
nances its spending by means of money creation, which fuels in ation. is was also pointed out in Section
11.1 of Chapter 11.
Macroeconomists from the new classical school raise a third objection to Keynesian scal policies. ey
argue that private agents would anticipate systematic counter-cyclical policies and respond to them in ways
that neutralise their intended effects. e Ricardian equivalence theorem, which states that it is immaterial
whether governments use tax or debt nance, is a well-known example of this thinking. In brief, the
argument goes as follows. Government loans to nance budget de cits must be repaid in future from tax
revenue. Rational economic actors realise that government borrowing now means a higher tax burden for
them (or for their children) in future. ey would therefore respond to any increase in loan- nanced
government spending by reducing their consumption spending and increasing their saving by equivalent
amounts, thus ensuring that they can meet their future tax obligations. is type of behaviour would mean
that tax nancing has exactly the same effect on the level of aggregate demand as de cit nancing. On
balance, this implies that the scal authorities cannot manipulate aggregate demand by varying the level of
the budget de cit. e empirical validity of the Ricardian equivalence theorem is still being debated, but the
possibility that countervailing behaviour by private agents has at times contributed to the disappointing
outcomes of attempted anti-cyclical scal policies cannot be dismissed summarily.

18.3.2.1 Political constraints on anti-cyclical fiscal policy

In addition to the sets of considerations that limit the practical effectiveness of anti-cyclical scal policy that
are listed above, a second reason why scal activism fell from favour was growing evidence that
governments tend to be unwilling to implement these policies consistently (quite apart from their ability to
do so). Considerations of political popularity make most governments far more inclined to adopt
expansionary policies (increases in public spending and, at times, tax cuts) during recessions than to
impose the short-term hardships associated with contractionary policies (tax increases and spending cuts)
in economies facing the danger of overheating. Economists from the public choice school emphasise that
this asymmetric approach to scal stabilisation contributed to the sharp growth in government spending in
the second half of the twentieth century. One time-series study (Tanzi & Schuknecht, 1997) found that
general government expenditure in the industrial countries increased from levels of 10% of GDP or less in
1870 to between 45 and 50% in 1994. More recently, South Africa showed a similar though less dramatic
trend, with the ratio of general government consumption and investment increasing from an average of
18,3% (1960–1969) to 23,4% (2008–2017) (see Table 1.1, Chapter 1).
It is worth considering the public choice view in more detail, because it has signi cant implications for
the discussion of the rules-versus-discretion debate in scal policy in Section 18.5.1.

18.3.2.2 Public choice view

In Chapter 6, we studied the properties of the various social choice rules as well as the extent to and the
conditions under which an optimal allocation of resources would be possible. Chapter 6 also presented the
public choice argument that democratic institutions exhibit inherent biases towards an overexpansion of the
public sector. According to this view, higher than optimal levels of government expenditure are the results of
the responses of politicians, bureaucrats, voters and special interest groups to incentives in the democratic
process.
Public choice economists argue that there are two causes of this bias. Firstly, it is attributed to the alleged
co-existence of concentrated bene ts, and a diffused or widely spread distribution of the costs of
government expenditure programmes. Put differently, small groups tend to bene t from certain public
expenditure programmes, although the cost is spread over all (or a larger group) of taxpayers. A programme
or set of programmes with clearly de ned bene ts will have a better chance of being approved if the
advocating groups are well organised and the cost of nancing is obscure or hidden.
 A salary increase for highly unionised public officials is an example of this type of expenditure
programme. Debt nance, especially in ationary nancing through money creation, is an example of a very
diffuse nancing mechanism. In the case of loan nance, only a part of the burden of the cost is incurred in
the year in which the expenditure has to be voted. It may actually be a deliberate political strategy to sell
debt- nanced programmes to voters because it enables politicians to be vague about the cost, which is in
any case to be shared by absent future generations. is form of strategic behaviour may be particularly
prevalent prior to an election that the incumbent political party is in danger of losing or at the sub-national
level when the particular tier of government is highly dependent on transfer income from a higher tier of
government (see Sections 19.5 and 19.6 of Chapter 19). It is so much more tempting (and even easier) to sell
expenditure programmes to voters if another politician is accountable to a different constituency for the
nancing. Provincial politicians, for example, may use these tactics to force larger transfer payments from
the national government to the provinces, with the result that the tax or debt burden at the national level is
increased.8 ese practices rest on the assumed existence of scal illusion among the electorate, which may
be de ned as the belief that taxes are lower than they actually are, or will be.9
e second cause of the bias towards high levels of government expenditure is the separation of bene ts
and costs in the budgetary process, a practice implied at the beginning of this chapter (see Section 18.2.2).
is separation actually reinforces the rst cause. Expenditure and revenue budgets are designed and voted
on separately by parliaments. e one-year nature of budgeting is a further cause because annual budgeting
allows for the approval of expenditure programmes without information on the concomitant long-term
expenditure commitments and the long-term tax or debt implications of the total of all expenditure
programmes. e future gap between expenditure and revenue, which by de nition amounts to future debt
nancing, may increase without anyone being aware of it. is separation of revenue and expenditure
decisions is linked to the ability-to-pay approach to taxation, in other words, nancing based on equity
rather than efficiency considerations. Public choice economists therefore tend to argue strongly for the
increased application of the bene t principle of taxation (instead of the ability-to-pay principle). An
additional method of correcting for these problems is to apply medium-term scal planning (see Section
18.6.1). Nowadays, many countries have medium-term scal frameworks in which the multiyear
implications of budgetary choices are revealed, thus allaying some of the concerns of public choice
economists.
Public choice economists also argue that the alleged bias towards overexpansion of the public sector has
been fed by the Keynesian legacy of budget de cits explained earlier in this chapter. In many countries, the
size of the public debt and the budget de cit reached such proportions that their concomitant ne-tuning
properties were lost.
e public choice view on how to improve efficiency in the allocation of public resources has important
implications for debt nancing. Public choice economists argue that constitutional scal limits are required
to correct for the overexpansion bias, that is, they favour more rules and less discretion in scal policy. A
balanced budget – in other words, one without debt nancing – is an extreme version of this prognosis. In
the United States, the idea of a balanced government budget enjoyed substantial popularity well into the
rst decade of the twenty- rst century. is can be attributed to the strong correspondence in this regard
between public choice and neoclassical thinking, which gained the upper hand in macroeconomic policy
thinking in the United States at the turn of the century. We discuss the debate on scal rules in Section
18.5.1. Notwithstanding strong support for balanced budgets, the United States has recorded federal budget
de cits in every year from 2002 to 2017. Clearly, the in uence of this view waned a great deal at the time of
the international nancial crisis of 2007–2009.

18.3.2.3 Summary of views on the impact of debt

A summary of the different views about the impact of public debt on the economy appears in Chapter 17,
Section 17.5.4.

18.3.2.4 Anti-cyclical fiscal policy and structural economic problems

e joint appearance of unemployment and in ation (in other words, stag ation) after the oil price shocks
of 1973 also sharply dented the popularity of Keynesian anti-cyclical scal policy. Economists soon realised
that the economic problems of the time were structural rather than cyclical in nature, and that short-term
demand management measures are ineffective for solving structural problems such as supply shocks,
structural unemployment and cost-push in ation.

18.3.3 The structural approach to fiscal policy

Taken together, the aforementioned considerations gave rise to a strong perception that active anti-cyclical
scal policy is ineffective at stabilising the economy, perhaps to the point of causing more instead of less
volatility. Supply-side economists, however, advocated a particular brand of scal policy as a cure for
stag ation. Members of this school of thought believe that the public sectors of most countries are too large,
and that the concomitant increases in tax burdens and tax rates are major disincentives to work effort,
saving, investment and economic growth. ey therefore recommend a reduction in marginal tax rates (to
increase the incentive to work, save and invest and a concomitant reduction in government spending (to
create more scope for private sector activity). ‘Supply-side’ has now become a catch-all word for any policy
measure proposing to move the economy closer to its production possibility frontier or expanding the
frontier.
In a sense, the supply-siders re ected the mind shift that took place from the mid-1970s onwards. e
attention of scal policymakers shifted away from stabilisation policy – in the traditional Keynesian sense of
attempts to ‘ ne-tune’ the level of economic activity – to structural measures aimed at increasing the growth
and job creation capacity of the economy as well as the redistributive impact of the budget. is shift
received further impetus from the development of unsustainable scal imbalances (excessive budget de cits
and public debt burdens) in many industrial and developing countries during the 1980s and early 1990s, and
generally dominated scal policy thinking until the international nancial crisis of 2007–2009. For many
policymakers, reducing these imbalances became a more pressing objective than using scal policy for
cyclical stabilisation purposes, although some scholars have pointed out that signs of active scal policy
could already be observed early on in the rst decade of the twenty- rst century. e United States is an
example.
Key aspects of what could be described as the structural approach to scal policy include the following:
• Keeping the public debt (and the burden of servicing it) at a sustainable level by avoiding high budget
de cits or reducing it to acceptable levels (such as was required after the dramatic surge in de cits and
public debt levels as a result of the 2007–2009 international nancial crisis)
• Keeping the overall tax burden at a level that does not seriously prejudice incentives to work, save and
invest
• Keeping government spending in check to avoid crowding out of private activity, in ationary nancing,
and the cost-push and disincentive effects of an excessive tax burden.

One of the major outcomes of the widespread acceptance of the structural approach to scal policy has been
that more and more countries are adopting rules-based scal regimes in place of the discretionary regimes
that characterised the Keynesian era. We discuss this trend in more detail in Section 18.5. Another outcome
was that references to scal policy virtually disappeared from macroeconomic policy debates worldwide for
almost two decades (recall the statement of Martin Eichenbaum quoted earlier).
A good example of how the focus in scal policy analysis has changed is the contrasting reasons why
adherents of the Keynesian approach and the structural approach regard the extent of the budget balance as
important. During the heyday of Keynesian anti-cyclical scal policy, economists watched the budget
balance closely as an indication of whether the short-run impact of scal policy on aggregate demand is
expansionary or contractionary. e structural approach puts more emphasis on what the budget balance
suggests about the current and future sustainability of scal policy, and how the nancing of budget de cits
is likely to in uence the economy. is approach leaves room for passive anti-cyclical scal policy to the
extent that automatic stabilisers are effective.
 ere are several ways to nance a budget de cit. One option is to borrow from the central bank, which
amounts to the government using its overdraft facilities. is type of nancing increases the money supply
and is potentially in ationary. Governments should therefore avoid it as far as possible. Another undesirable
option is to run down the country’s foreign reserves. Unless a country has an exceptionally large stock of
reserves, nancing the budget de cit in this manner would soon deplete them and cause a foreign exchange
crisis. Most governments nance their de cits by borrowing from the domestic and international capital
markets. In countries with well-developed nancial systems (such as South Africa), governments borrow
from the capital markets by issuing bonds (government stock) on which they have to pay interest. is type
of borrowing therefore increases the public debt, the sustainability of which has also become an important
criterion for judging the soundness of scal policy and macroeconomic management in general. In practice,
it is very difficult to determine the sustainability of the public debt: doing so requires an assessment of
whether or not it is and will remain at levels that the particular country can repay and service (see Box 18.2).
Economists often use a non-increasing (in other words, a decreasing or constant) public debt–GDP ratio as a
benchmark to distinguish between sustainable and unsustainable scal policy (Chalk & Hemming, 2000: 3).
Apart from the possibility that it can become unsustainable (put differently, that the government would
have to default on its debt obligations), a large public debt also has other undesirable effects. For one thing,
government borrowing places a burden on future generations, who have to pay the interest and repay the
debt. As indicated in the discussion of the current budget balance, this type of inter-generational
redistribution is especially problematic when the government uses the borrowed funds to nance current
expenditure that only bene ts the current generation. Another disadvantage of a large public debt is the
associated interest burden that, for a given level of public expenditure, leaves the government with less
money to use for productive purposes.

18.3.4 Renewed interest in active fiscal policy

Even before the international nancial crisis of 2007–2009 and the ensuing Great Recession, there were clear
signs that policymakers were showing greater interest in the cyclical effects of scal policy than had been the
case only a few years earlier. is development should not be interpreted as a fully edged return to scal
activism of the traditional Keynesian variety. At the time of writing, leading economists were still grappling
with a full understanding of the causes and consequences of the international nancial crisis, and the
appropriate macroeconomic and regulatory response.
Gradually it became clear, however, that a number of important lessons had been learnt from the crisis.
Blanchard, Dell’Ariccia and Mauro (2010) mention three of these: low in ation limits the scope for using
monetary policy in de ationary recessions, countercyclical scal policy remains an important tool, and the
regulation of nancial markets is not neutral in a macroeconomic sense. e latter acknowledges that the
regulation of nancial institutions can be pro- or anticyclical. For example, if reserve requirements cause
banks to increase their deposits with the central bank during an economic recession, it reduces the
availability of credit at a time when an economic recovery would be facilitated by an increase in the supply
of credit.
Nonetheless, many economists still maintains that, except maybe in most extreme situations, monetary
policy is a better tool for stabilisation purposes than scal policy and that activist scal policies should
generally be avoided. Alan Blinder (2006: 54) puts it as follows:

Under normal circumstances, monetary policy is a far better candidate for the stabilisation job than scal
policy. It should therefore take rst chair. at said, however, there will be occasional abnormal
circumstances in which monetary policy can use a little help, or maybe a lot, in stimulating the economy –
such as when recessions are extremely long and (or) extremely deep, when nominal interest rates approach
zero or when signi cant weakness in aggregate demand arises abruptly. To be prepared for such
contingencies, it makes sense to keep one or more scal policy vehicles tuned up and parked in the garage,
and perhaps even to adopt institutional structures that make it easier to pull them out and take them for a
spin when needed.
ere are two reasons for the tentative comeback of scal policy in discussions of macroeconomic
stabilisation. First, automatic scal stabilisers often become destabilising when in ation is high. In many
countries, lower in ation rates have now rejuvenated the working of the automatic scal stabilisers, thus
restoring to scal policymakers a tool for in uencing economic activity that is not subject to some of the
shortcomings of activist policies. Second, the decisions to bail out certain nancial institutions when
nancial markets in various countries were at perceived risk of systemic failure in 2007 and 2008, together
with the ineffectiveness of monetary policy to counter the severe recessionary effects when interest rates
approached the zero bound, seemingly left many governments with little choice but to fall back on scal
measures. During these times, monetary activism took on a new dimension, such as when the US Federal
Reserve Bank adopted a balance-sheet measure of quantitative easing (so-called QEs), whereby the supply
of money was increased.10 Because of the condition of countries’ public nances prior to the crisis (see
Tables 18.1 and 18.2), the scope for active scal policies varied and a measure such as quantitative easing
thus added to the arsenal of policy measures. On the whole, however, the approach to the macroeconomic
role of scal policy clearly shifted – at least temporarily – towards Keynesian anti-cyclical scal policy.
Appropriate indicators are needed to determine whether the scal policy stance complements or
undermines the actions of the monetary authorities. We introduced the cyclically adjusted budget de cit in
Box 18.1, and now discuss it in more detail with the aid of Figure 18.3 to show how countries have begun to
use cyclically adjusted (or structural) budget balances in assessing and designing scal policy.
e set-up of Figure 18.3 is similar to that of Figures 18.1 and 18.2: it also depicts government tax revenue
and government spending as functions of income (Y). As before, the total tax revenue curve (T) slopes
upwards and the government spending curve (G) has been drawn as a horizontal line. e economy
experiences full employment when income is YF. At this level of income, tax revenue and government
spending are t0 and g0 respectively. e associated budget de cit (distance AB) is not distorted by the effects
that any particular state of the business cycle may have on the levels of government revenue or expenditure.
As such, it represents the ‘underlying’ budget balance that results from the tax structure and level of
government spending at the full-employment, potential or trend-line level of income. e benchmark level
of the budget balance is variously known as the structural budget balance, the full-employment income
budget balance, the potential-income budget balance, the trend-line budget balance or the cyclically
adjusted budget balance.
e calculation of structural budget balances is easier said than done. Whereas the conventional budget
balance is usually expressed as a percentage of the estimated or actual gross domestic product, the
structural budget balance is expressed as a percentage of the potential output of the economy or the full-
employment level of income. Potential output and the full-employment level of income cannot be measured
accurately. ey can only be estimated using techniques that remain controversial. is explains why Vito
Tanzi (2005: 9) once referred to cyclically adjusted budget balances as ‘counter-factual variables’ and ‘virtual
variables’. Structural budget balances are nonetheless very useful indicators because they enable
policymakers to assess the extent and impact of observed budget balances in a dynamic manner and,
therefore, more accurately. Consider, for example, the situation at income level Y1 in Figure 18.3. e
economy is well below the full-employment level of income, and the levels of tax revenue t1 and g0 result in a
budget de cit depicted by the distance CE. Provided that they have a reasonably accurate estimate of
potential output, policymakers would be able to distinguish between the structural component of this de cit
(distance CD, which equals distance AB) and its cyclical component (distance DE). Given the assumption in
our example that government spending is not sensitive to the state of the business cycle, the cyclical
component of the de cit can be ascribed fully to a revenue effect. Tax revenue is lower than what it is likely
to be at the full-employment income level by the distance DE (= t1t0) because income is below the full-
employment level. is difference between potential (or full-employment) output and actual output is
referred to as the output gap.
Figure 18.3 Structural and cyclical budget balances

Hence, structural budget balances enable policymakers to distinguish between the permanent and
transitory components of observed budget balances. is distinction can help policymakers to avoid errors,
a typical example of which is taking long-lasting decisions on taxes and expenditures based on transitory
movements in revenues. e situation at income level Y2 illustrates this possibility. Here tax revenue is t2 and
government spending is g0, yielding a budget surplus equal to the distance FH. Many governments
(supported by interest groups) would interpret the emergence of a budget surplus as an indication that they
can and should increase public spending, for example, to help the poor or to expand the country’s physical
infrastructure. Increases in public spending, however, are seldom easily reversible: once appointed, public
servants cannot be retrenched at will, and newly created assets (such as schools and hospitals) have running
expenses and require maintenance. Policymakers should therefore ensure that sufficient nancing will be
available on a sustained basis to cover recurring costs associated with public outlays before expanding them.
Here the observed budget surplus appears to be a temporary (cyclical) phenomenon: income is well above
the full-employment level (by the distance Y2YF) and this raised tax revenue from t0 to t2. What Figure 18.3
tells us, is that income should eventually revert to the full-employment level YF, with a concomitant fall in tax
revenue to t0 and the re-emergence of the structural budget de cit. In fact, the budget surplus may be the
very de ating factor that moves the economy back to YF . If, however, the government uses the windfall
revenue to nance spending programmes of a long-term nature, the horizontal line G0 would shift upwards
to G1 to cut the T-function at F so that the structural de cit increases from AB to (AB + FH). e distinction
between the cyclical and the structural budget balances could therefore help policymakers to avoid the
ratchet effects that many economists blame for persistent budget de cits and secular increases in
government spending levels. ese effects were an important reason why Keynesian scal activism lost its
appeal in the 1970s.
 A brief reflection on the fiscal consequences of the Great
18.4
Recession
Even though governments resorted to some form of Keynesian scal activism during and after the
international nancial crisis, much concern remains about countries’ scal scope to stimulate the economy.
e levels of debt and budget de cits signal warning lights at a time when governments are under pressure
to stimulate economic recovery and to restore scal sustainability. Tables 18.1 and 18.2 show the scal
predicament of some industrial as well as developing countries eight years after the crisis.
In retrospect the words by Spilimbergo, Symansky, Blanchard and Cottarelli (2008: 2) sound prophetic.
ey concluded that the measures to be contained in an optimal scal package should be timely (because
the need for action was immediate), large (because the current and expected decrease in private demand
was exceptionally large), lasting (because the downturn would last for some time), diversi ed (because of
the unusual degree of uncertainty associated with any single measure), contingent (because the need to
reduce the perceived probability of another ‘Great Depression’ required a commitment to do more, if
needed), collective (since each country with scal space had to contribute) and sustainable (so as not to
have led to a debt explosion and adverse reactions of nancial markets). ey then argued that spending
increases as well as targeted tax cuts and transfers were likely to have the highest multipliers. General tax
cuts or subsidies, either for consumers or for rms, were likely to have lower multipliers.
In reality, different countries applied different scal measures. Countries such as India, Indonesia and
Mexico mainly used tax reductions. e United States, Australia, Korea, France and Russia made use of a mix
of tax and expenditure measures. e United Kingdom, Canada, Germany and Turkey mostly resorted to
expenditure measures. Countries such as Greece, Portugal and Ireland had no room for stimulatory
measures, and had to rely on bailout support from other countries and the International Monetary Fund, in
addition to harsh measures of scal restraint.

Table 18.1 The evolution of general government fiscal positions in South Africa and selected advanced economies
(% of GDP, selected years from 2006 to 2016)

Source: IMF World Economic Outlook, April 2018.

Table 18.2 The evolution of general government fiscal positions in South Africa and selected emerging markets and
other developing economies (% of GDP, selected years from 2006 to 2016)

Source: IMF World Economic Outlook, April 2018.

When we compare scal trends across groups of countries from 2006 (the year before the international
nancial crisis) to 2016, IMF (2018) data show signi cant differences. e biggest increases in budget de cits
and public debt occurred in industrial countries. Yet, even after downgrading of their sovereign debt, they
retained much higher credit ratings than many emerging market economies. A country’s creditworthiness
obviously depends on other factors too, such as growth performance and potential, political stability, the
structure of nancial markets, property rights, security of contracts and various other risks to foreign
investment.
After the nancial crisis, South Africa’s scal position was much less disconcerting than that of some of
the advanced economies, but worse than some emerging market economies. One aspect of the scal debate
was about the speed with which economic recovery should be pursued with scal activism, in other words,
what the nature of scal consolidation should have been. Fiscal consolidation simply means the
combination of policies used by a government to reduce its de cits and accumulated stock of debt to
acceptable levels in a particular timeframe. At face value, this would suggest a more conservative scal
approach in high-debt countries, which would not have generated as speedy a recovery, but would have put
scal sustainability less at risk. As it turned out, the general government budget de cits were reduced
sharply in countries such as the UK and the USA, but debt ratios continued to increase, largely because of
low economic growth.11 Box 18.3 gives an overview of literature ndings regarding scal consolidation
outcomes, including the rather controversial interpretation of such ndings known as the ‘expansionary
scal contraction hypothesis’.
What added to misgivings about the effectiveness of active scal policies as growth stimuli, was that
consumers appeared to respond to some of the measures (for instance, in the United States) by replenishing
lost savings (or wealth) rather than increasing consumption. Such behaviour, which resembled Ricardian
equivalence, undermined the Keynesian increases in aggregate demand envisaged by policymakers.
Another complication was that a number of smaller countries (for example, Greece and Ireland) exhibited
scal unsustainability, and had to be bailed out by the International Monetary Fund and the European
Central Bank. e conditions accompanying the bailouts entailed major scal austerity measures, which
halted the period of active scal policies and dampened the strength of the economic recovery. One of the
major steps to reduce the structural de cits in European countries was the raising of the retirement age, that
is, the age at which people become entitled to pension bene ts. Although the retirement bene ts of the
welfare states of Europe varied, in general they had for a long time been regarded as increasingly
unaffordable and unsustainable, although very hard to change. e grim realities of the Great Recession as
well as the cost to taxpayers of the bailouts and the active scal measures seemed to have made it politically
feasible to increase the retirement age. is had the effect of reducing future scal commitments and social
security costs, which were socio-politically easier to digest than a future cut in bene ts or an increase in
taxes.
Fiscal balances also matter for the assessment of scal sustainability. Some countries had budget
surpluses before the crisis, like Germany, Chile and South Africa. ese countries, as well as low-de cit
countries not only encountered a less severe recession, but were also able to generate quicker economic
recoveries. is was also experienced by many developing countries that had been applying prudent scal
policies. Barring a few exceptions such as China and India, however, these countries apparently did not
achieve exceptional growth recoveries. It may well be that the structural approach to scal policy is needed
for most countries to increase their potential output, rather than to achieve mere cyclical upturns. Keynesian
policy then becomes a tool that is put to use only when a major shock leaves policymakers with no real
alternative – but then also just to address short-term demand de ciencies. Even so, there are limits to what
scal measures can do to achieve scal sustainability. It requires the harnessing of the full range of structural
economic policies to enhance the countries’ potential output, amongst which measures are needed that
increase countries’ international competitiveness.

Box 18.2 Fiscal sustainability

Fiscal sustainability refers to a government’s continued ability to service its debt (in other words, pay the
interest on its domestic and foreign debt timeously) and repay the loan amount (principal) when it becomes
due. In more technical terms, we distinguish between fiscal solvency and fiscal sustainability. A government is
fiscally solvent if the present value of the flow of all future revenues exceeds the value of outstanding debt plus
the discounted value of future government expenditure. This is a necessary though not a sufficient condition for
fiscal sustainability. The concept of fiscal sustainability adds the requirement of a track record regarding the
generation of revenue and the management of expenditure, together with a credible expectation that the track
record will be repeated in future.
It is difficult to judge fiscal solvency and sustainability as described here in practice. The concept of fiscal
sustainability can be made operational by arguing that states maintain fiscal sustainability when the
government is regarded as solvent without the need for any changes to the present stance of fiscal policy.
Attempts to judge the sustainability of fiscal policy in individual countries often employ primary balance trends
as a predictor of the intent to restore a sustainable fiscal stance when rising fiscal conventional deficits
threaten macroeconomic stability or when the debt burden is rising to levels that generate concerns. A widely
used rule of thumb states that primary deficits are consistent with stable government debt–GDP ratios only if
and for so long as the real rate of economic growth exceeds the real rate of interest. Applications of this rule of
thumb usually rely on the following equation for calculating changes in the ratio of government debt to GDP:

Change in the ratio of debt to GDP = Bt/Yt – Bt–1/Yt–1 = (r – g) Bt–1/Tt–1 + (Gt – Tt)/Yt + (Mt – Mt–1)/Yt

where (r – g) = difference between real rates of interest and economic growth

Bt–1/Yt–1 = previous year’s debt–GDP ratio
(Gt – Tt)/Yt = primary balance as ratio of GDP
(Mt – Mt)/Yt = seigniorage (in other words, government revenue on account of an expansion of the money
supply) as ratio of GDP.
This equation, sometimes also referred to as reflective of the inter-temporal budget constraint, gives rise to
another diagnostic rule of thumb about fiscal prudence: if the primary balance is improving in the face of rising
public debt–GDP ratios, the government is viewed as being serious about maintaining or restoring fiscal
sustainability.
During the past twenty to thirty years, much attention has been given to the nature and measurement of
scal sustainability, and to scal consolidation strategies to restore scal sustainability. Fiscal sustainability
remains a somewhat elusive concept because it depends on assumptions about future tax compliance and
government expenditure discipline, and the projected output gap, that is, the difference between the actual
and the potential GDP. Several scholars have investigated scal policy in South Africa over the years. Most of
them concluded that scal policy in South Africa has been sustainable (see, for example, Burger, Stuart,
Jooste & Cuevas, 2012). However, there has been growing concern about the sustainability of scal policy
because of the increase in South Africa’s public debt–GDP ratio in the aftermath of the international
nancial crisis and the ensuing Great Recession (see Table 18.1). Ultimately, South Africa’s international
sovereign debt was downgraded to speculative (‘junk-bond’) status by two of the three major credit ratings
agencies during 2017. To address the growing scal dilemma, Burger, Calitz and Siebrits (2016) mentioned
an increase in public investment relative to government consumption expenditure as a means of improving
the asset side of the government’s balance sheet (or, as the accounting profession would say, the statement
of nancial position). After President Ramaphosa replaced President Zuma at the end of 2017, serious steps
were taken with a view to rebuilding the economy and restoring scal sustainability.

Box 18.3 Fiscal consolidation approaches and the ‘expansionary fiscal contraction hypothesis’12

Fiscal deficits (and, by implication, public debt burdens) can be reduced in three ways: by reducing government
expenditure, by increasing government revenue, and by a combination of the two. Several empirical studies
compared the effectiveness of these types of deficit-reduction strategies. Most studies of this nature began by
formulating exact, albeit arbitrary, criteria for judging the success of fiscal consolidation efforts.13 Next, the
criteria were used to distinguish those efforts that brought significant reductions in fiscal deficits from those
that did not. A further distinction was made between durable successful efforts and episodes in which
reductions in budget deficits were soon reversed. The researchers then identified features associated with the
success and durability of consolidation efforts. Some studies simply compared changes in the average values
of fiscal aggregates in the groups of episodes, while others used econometric techniques to identify the
features of successful and durable fiscal consolidations.
Most analyses of attempted fiscal consolidations in OECD countries found that efforts based mainly on cuts
in transfer payment and the government wage bill were more likely to have achieved significant reductions in
fiscal deficits than those based mainly on tax increases. Another key result of the majority of these studies was
that episodes of deficit reduction based mainly on reductions in the same categories of government spending
were more durable than ones that relied mainly on revenue increases, in the sense of having been less likely to
have been reversed within a few years after satisfying the chosen criteria for a successful consolidation.
Analyses of emerging market and other developing economies also linked the success and persistence of fiscal
consolidations to the extent of cutbacks in current government spending. In such countries, however, revenue
increases were seemingly also important elements of durable reductions in fiscal deficits, especially in cases
where revenue collection was weak. Another important finding was that protecting or increasing the share of
capital spending in total government expenditure during consolidation episodes increased the probabilities of
success and persistence.
On balance, these results suggested that a ‘one-size-fits-all’ approach to fiscal consolidation should be
avoided. A plausible interpretation of the differences in the findings for OECD and developing countries is that
individual countries should design consolidation plans consisting of mixtures of revenue increases and
spending reductions capable of correcting the specific causes of their fiscal imbalances. Excessive growth in
spending on the remuneration of government employees and social transfer payments were major causes of
fiscal problems in most OECD countries; hence, such items should have been targeted for reductions in
attempts to reduce budget deficits. In many developing countries, by contrast, weak revenue collection efforts
often contributed markedly to fiscal imbalances. Such countries usually had considerable scope for increasing
tax revenues without raising tax rates, for example, by improving the administration of the tax system and by
curbing tax evasion. The finding that the success and durability of fiscal consolidations in developing countries
were linked to the share of capital outlays in total government expenditure confirmed that deficits can be
reduced in more or less appropriate ways. Some governments reduced deficits by reducing asset formation (i.e.
public investment) or by accumulating hidden liabilities (for example, borrowing via off-budget agencies). Both
strategies might put fiscal sustainability at risk: underinvestment in infrastructure might harm the future growth
 performance of the economy, while off-budget borrowing ultimately remains liabilities of the public sector. The
implication of this interpretation of existing research is that the literature on fiscal consolidation does not
suggest a simple set of prescriptions for reducing the fiscal deficit and public debt. Recommendations should
be derived from a thorough analysis of the factors influencing the size of and trends in these fiscal aggregates.
GDP growth facilitates attempts to reduce deficit–GDP and public debt–GDP ratios by boosting the
denominator of such ratios. Hence, the success and persistence of fiscal consolidation efforts are also linked
to the effects of such efforts on output growth. Simple Keynesian models imply that fiscal tightening dampens
economic activity in the short to medium run: Spending cutbacks and revenue increases reduce aggregate
demand and income directly and via multiplier effects. Recent studies of the output implications of deficit
reduction have focused on the possibility and likelihood of non-Keynesian effects. Drawing on case studies and
cross-country empirical analyses, adherents of the ‘expansionary fiscal contraction hypothesis’ have claimed
that fiscal consolidation can have positive effects on output that sometimes outweigh the negative ones
highlighted by Keynesian models. According to some authors fiscal adjustment can stimulate growth via the
demand side and the supply side of the economy. Credible consolidations that change expectations about
future fiscal policy can appease fears of more dramatic and disruptive future policy changes. The resulting
boost to the confidence of private agents and the reduced need for precautionary saving might stimulate private
consumption and investment. This boost to aggregate demand would be reinforced if the consolidation
contributes to lower interest rates by decreasing pressure on capital markets and risk premia for default on the
public debt. In addition, fiscal consolidations based mainly on cuts in the government’s wage bill can boost the
supply side of the economy if spillover effects result in downward pressure on unit labour costs. Reductions in
unit labour costs increase the competitiveness of private firms.
This possible supply-side effect suggests that the influence of fiscal consolidation on economic growth is
also linked to its composition (that is, the relative weights of revenue increases and expenditure reductions).
The distinction between spending-focused and revenue-focused consolidation efforts possibly matters for the
demand-side effects on economic growth as well. It has been pointed out that the positive credibility effects of
decisive steps to rein in spending on large and politically sensitive items, such as reductions in public
employment and tightening of the eligibility criteria for transfer payments in OECD countries, should eclipse
those of tax hikes. This argument suggests that expenditure-based consolidations should stimulate private
investment and consumption more than revenue-based consolidations would.
The notion that deficit reduction can be a painless or, for that matter, a growth-stimulating process appeals
greatly to politicians, many of whom fear backlashes against fiscal austerity. The ‘expansionary fiscal
contraction hypothesis’ remains controversial, though. Some empirical studies found no evidence of non-
Keynesian effects in the wake of attempts to reduce fiscal imbalances. There has also been criticism of the
practice of using fiscal outcomes as the basis for identifying episodes of fiscal contraction. Arguing that this
practice often leads to unwarranted inferences about the intentions of policymakers and inadequate attention
to the effects of exogenous factors on fiscal outcomes, some critics argued for a narrative approach that bases
identification of fiscal consolidation episodes on careful study of policymakers’ pronouncements and actions.
The results of the application of narrative approaches amounted to a rejection of the ‘expansionary fiscal
contraction hypothesis’. Given the dearth of empirical research on the validity of this hypothesis in emerging
markets and other developing countries and the inconclusiveness of results for advanced countries, it would be
unwise to expect that fiscal consolidation would always or, for that matter, usually stimulate economic growth.14
The possibility that it might not dampen economic growth significantly, if at all, is intriguing, however, and
should be kept in mind when reflecting on the design of fiscal consolidation programmes.

18.5 Fiscal policymaking frameworks 15

Much attention is given nowadays to connections between scal policymaking frameworks and scal
outcomes. A country’s scal policymaking framework consists of all the institutions that structure scal
policymaking processes. e following are the best-known institutions of this nature.
• Numerical scal rules are quantitative restrictions on the absolute or relative levels of scal aggregates.
e most common categories of numerical scal rules are: limits on the extent of the public debt
(expressed as amounts or as ratios of the gross domestic product or GDP); limits on various de nitions of
scal balances (expressed as ratios of the GDP); limits on the absolute levels, growth rates or GDP shares
 of public spending aggregates; and upper or lower limits on government revenue (expressed as ratios of
GDP).
• Procedural scal rules are the details of budget processes, that is, the arrangements governing the
formulation of budget proposals by executive branches of governments, the approval of budget
proposals by legislatures and the implementation of budget laws. Such rules determine the distribution
of scal policymaking powers, for example, that between the Minister of Finance and the other cabinet
ministers and that between the executive and the legislative branches of government. Procedural scal
rules are also known as budget-process rules.
• Medium-term expenditure frameworks are rolling revenue and expenditure projections presented
against the backdrop of economic and scal goals and the prospects of the economy. e purposes of
such frameworks are to enhance the transparency of budget processes, strengthen links between policy
priorities and longer-term spending plans, and improve expenditure control. e medium-term
expenditure frameworks of countries are closely linked to their budget-process rules.
• Fiscal responsibility laws specify the medium-term paths of key scal aggregates, outline annual and
medium-term strategies for achieving policy objectives, and establish frameworks for regular reporting
on scal trends and auditing of scal information. e main aim of scal responsibility laws is to make
policymakers more accountable by increasing the transparency of scal processes.

e scal policymaking frameworks of a growing number of countries also contain scal councils, which
are non-partisan agencies consisting of independent scal policy experts. Such agencies have monitoring
and advisory tasks that range from costing of budgetary initiatives to advising policymakers on scal policy
options; monitoring compliance with numerical rules; analysing scal trends; and generating independent
economic and scal forecasts for policymaking purposes. Fiscal councils differ from the institutions listed
above in being organisations rather than rules. ey are usually regarded as elements of scal frameworks,
however, because they in uence aspects of scal policymaking processes in the same ways that those
institutions do.
e reason for the current interest in scal policymaking frameworks is the belief that appropriate
policymaking institutions can contribute to better scal outcomes. More speci cally, a growing body of
theory and empirical evidence suggests that scal policy is more sustainable and more supportive of
macroeconomic stability when underpinned by well-designed scal policymaking frameworks. e
remainder of this section discusses this theory and evidence. It rst comments on the long-running
discretion-versus-rules debate in scal policy, then discusses the in uence of scal policymaking
frameworks on scal outcomes, and ends with a concluding comment.

18.5.1 Rules versus discretion in fiscal policy

Dornbusch, Fischer and Startz (2004: 198) formulated the choice between the rules-based and discretionary
scal policymaking regimes as follows: ‘Should … the scal authority conduct policy in accordance with pre-
announced rules that describe precisely how their policy variables will be determined in all future
situations, or should they be allowed to use their discretion in determining the values of the policy variables
at different times?’ Debates about the pros and cons of these two types of regimes have been features of
macroeconomics for a very long time. Such debates traditionally focused on the choice between discretion
and numerical scal rules, and this section maintains that perspective. However, the arguments also apply
to the other institutions or rules that make up scal policymaking frameworks (e.g. procedural scal rules
and scal responsibility laws) - as will become apparent in Section 18.5.2.
Policymakers operating in discretionary regimes have more exibility (that is, more scope to adjust
policy instruments) than their counterparts in rules-based regimes do. ere are two possible reasons why it
might make sense to limit scal policymakers’ exibility by means of numerical rules.
• Rules may function as binding constraints that improve the effectiveness of scal policymaking by
making it very difficult (if not impossible) for the authorities to stray from sound policies. e argument
that rules are necessary to prevent policymakers from lapsing into unsustainable or procyclical scal
policies rests on the premise that policymakers cannot be trusted to manage the public nances well all
 the time. Policymakers might not be competent enough to do so, for example, or might exhibit utility-
maximising behaviour that distorts policy choices (recall the discussion of government failure in Chapter
6, Section 6.7).
• e adoption of and adherence to rules may be a mechanism for scal policymakers to make credible
commitments to sound policies. Such commitments may help to make policies more effective by
in uencing the expectations and responses of private economic agents. Consider an example: To
encourage investment, governments would like to reassure prospective investors that they would not
resort to unsound scal practices that lead to high in ation and nancial crises (for example, running
excessive budget de cits and nancing them by borrowing from the central bank).16e argument is that
the adoption of rules that limit the size of budget de cits would be ideal mechanisms to signal
commitment to scal discipline, thus fostering investment and economic growth. Note that the premise
of this justi cation for rules differs from that of the binding constraints argument. e credible
commitment argument does not rest on the belief that governments should be constrained by rules
because they are inclined to stray from prudent scal policies. Instead, it holds that governments with
strong commitments to restoring or maintaining scal discipline should choose to constrain themselves
by means of rules to convince private actors of their resolve.

e popularity of discretionary and rules-based scal policymaking regimes has changed over time.
Keynesian macroeconomics emphasised the sluggishness of market adjustment to shocks and the
stabilising properties of active anti-cyclical scal policies. e dominant view during the heyday of
Keynesian macroeconomics was that rules-constrained policymakers cannot pursue activist policies
effectively. Fiscal discretion was therefore the norm in scal policymaking from the end of World War II
until the mid-1970s. Recall from Section 18.3.3, however, that numerical scal rules have grown in
popularity since the emergence of the structural approach to scal policy. Both sets of arguments outlined
above contributed to this development. For one thing, the shortcomings of activist scal policies identi ed
in Section 18.3.2 were interpreted widely as clear proof of the inadequacy of discretionary policy regimes. A
number of countries adopted rules in the belief that doing so would prevent the problems associated with
activist policies by making it very difficult (if not impossible) for the scal authorities to stray from sound
policies. e binding constraints justi cation for scal rules underpinned such reforms of scal
policymaking frameworks. e case for scal rules was bolstered further by the widespread adoption in
macroeconomics of the rational expectations hypothesis, which states that all agents make optimal use of
the information at their disposal by taking into consideration past, current and the expected future states of
their environment – including anticipated economic policies – when they make decisions. One of the
implications of this hypothesis is that announced policies only affect the behaviour of the agents in the
manner desired by policymakers if the policies are deemed credible, that is, if agents believe that
policymakers will implement them fully. As was suggested earlier, proponents of scal rules argue that the
adoption of numerical rules represents credible commitments to sound policies that should make scal
policies more effective at in uencing the expectations and behaviour of private agents.
e empirical record of numerical scal rules is mixed. To be sure, a growing number of studies suggest
that scal outcomes tend to become more prudent and less destabilising after the adoption or strengthening
of such rules. Still, many governments nd it easy to ignore, abandon, suspend or circumvent quantitative
limits on scal aggregates. Two well-known tactics to circumvent rules are to base budgets on unrealistic
GDP or government revenue forecasts and using ‘creative accounting’ to hide breaches of numerical limits.
e vulnerability of numerical rules was evident during the Great Recession, for example, when the Stability
and Growth Pact rules failed to prevent large increases in the budget de cits of all European Union
countries, a severe public debt crisis in Greece and near-crises in Portugal and Spain.17 is shows that the
binding constraints argument for scal rules has limited validity.
ere are two main reasons why many rules-based scal regimes fail, however. e rst reason is that no
individuals or agencies can force sovereign government to adhere to rules. e second is that policymakers
have limited control over scal outcomes. As indicated above, numerical scal rules typically establish
upper limits for the ratios of key scal aggregates to GDP (for example, budget balances and the level of
public debt). e values of these ratios are strongly affected by factors beyond the immediate control of
governments, including GDP shocks (which impact upon the denominator and the tax revenue component
of the numerator) and restrictions on the scope for changing the level of public spending in the short run
(for example, the contractual nature of major expenditure items such as compensation of employees,
interest on public debt, procurement programmes, and certain subsidies and transfer payments).
On paper, the credible commitment justi cation is a powerful argument for scal rules. For example,
adopting numerical rules seems to be one of the few options available to a new government without a record
of scal prudence that wishes to signal its commitment to sound scal policies. Introducing rules may yield
credibility bene ts in such circumstances. Ultimately, however, the durability of such bene ts will depend
on the government’s ability to honour its commitment by adhering to those rules. To avoid undermining
their credibility by breaching rules, policymakers may have little choice but to respond actively to small or
transitory increases in budget de cits or public indebtedness caused by negative GDP shocks, among other
things. Yet such responses may not be effective given that policymakers cannot control the values of key
scal aggregates with precision. Moreover, the same policy lags that caused problems for active scal
policies aimed at stabilising the business cycle will also cause problems for active scal policies aimed at
complying with scal rules. Hence, rules may force policymakers to adopt ultra-conservative forecasting and
budgeting techniques that result in excessively contractionary scal outcomes. Other common undesirable
effects of attempts to comply with rules are inappropriate forms of tax increases and expenditure cuts.
Experience has shown that policymakers often respond to shocks that threaten compliance with rules by
using measures that are likely to provoke little political resistance, instead of those that are most appropriate
from an economic point of view. For example, they often reduce growth-promoting capital spending
programmes while maintaining subsidies to loss-making public enterprises.

18.5.2 Fiscal policymaking frameworks and fiscal outcomes

Section 18.5.1 pointed out that there is a strong case for scal rules based on theoretical considerations and
evidence of the weaknesses of discretionary scal policymaking. Yet it also argued that rules-based scal
policymaking has a mixed record, in part because it faces formidable practical hurdles (such as the limited
controllability of important scal aggregates). ese considerations raise two questions about the
relationship between scal policymaking frameworks and scal outcomes. First, are the other elements of
scal policymaking frameworks (procedural scal rules, medium-term expenditure frameworks, scal
responsibility laws and scal councils) more likely to contribute to good scal outcomes than numerical
scal rules have been? Second, can the effectiveness of numerical scal rules be enhanced by combining
them with these other elements of scal policymaking frameworks?
With regard to the rst questions, theory and empirical evidence from advanced and developing
countries suggest that well-designed budget-process rules can contribute to good scal outcomes. e best-
known argument is that scal discipline is enhanced if the budget process curtails the in uence of the
spending ministers and departments as well as the legislature, and assigns strong powers to the nance
minister and the treasury. e premise of this argument is that national budgeting is a common-pool
problem: most public spending programmes are nanced from general tax revenues, but bene t speci c
groups of voters (such as the inhabitants of particular geographical areas). is implies that the bene ciaries
of public spending programmes often pay for fractions of the bene ts they receive. Hence, the argument
goes, politicians and voters overestimate the net marginal social bene ts of these programmes and demand
excessive levels of public government spending. A plausible solution for the common-pool problem is to
concentrate decision-making authority in the hands of nance ministers and treasuries - the parties who are
most likely to acknowledge the overall budget constraint and to enforce it against the demands of spending
ministers, legislatures and interest groups. A key aspect of the argument is that the budget process should be
‘top-down’, which means that nance ministers should determine aggregate expenditure envelopes and
budget balances before decisions are taken about allocations to spending programmes.
e other three (medium-term expenditure frameworks, scal responsibility laws and scal councils) are
fairly new elements of scal policymaking frameworks, and it might be too soon to draw de nitive
conclusions about their effectiveness. Yet the early indications are that these three elements can also
contribute to better scal outcomes, most notably by enhancing transparency and accountability in scal
policymaking. Fiscal transparency means being open to the public and to nancial markets about the
structure and functions of government as well as scal projections, policy intentions and outcomes.
Advocates of transparency argue that it improves scal policymaking by making the scal authorities more
accountable. Some scal policy experts believe that transparency-enhancing reforms are more effective
mechanisms to make scal policies credible than adopting scal rules.18
e extent to which transparency-based frameworks constrain policymakers depends on their details.
On balance, however, transparency-based frameworks are more exible than rules, but less exible than
discretionary ones. e greater exibility of transparency-based regimes gives them a major advantage over
rules-based regimes (especially to policymakers who are strongly committed to prudent scal policies and
who therefore do not need binding by rules). Another advantage of transparency-based regimes over rules-
based ones is their superior ability to strengthen the effectiveness of market discipline over scal
policymakers. ‘Market discipline’ refers to the phenomenon that the nancial markets restrain the de facto
independence of scal policymakers: the markets quickly ‘punish’ scal laxness by suspending lending,
withdrawing capital and increasing risk premiums on borrowed funds. On the whole, the in uence of
market discipline on scal policymaking is benign, but there is the danger that governments could become
excessively concerned with gaining or maintaining the con dence of markets and allow the short-term
oriented preferences of market participants to bias their macroeconomic policy decisions. Rules-based
regimes tend to emphasise a small number of summary indicators of scal policy and in this way foster the
tendency of market participants to judge scal policy on a too narrow basis. Transparency-based regimes,
on the other hand, can strengthen the effectiveness of market discipline by ensuring that the bigger picture
receives due emphasis.
Turning to the second question, it has become increasingly clear that complementary transparency-
enhancing elements can make rules-based scal regimes more effective. Recall that governments
sometimes attempt to circumvent rules intended as binding constraints by basing budgets on unrealistic
forecasts or by applying ‘creative accounting’ tricks. e detailed reporting requirements of medium-term
expenditure frameworks and scal responsibility laws as well as the monitoring and whistle-blowing
activities of scal councils can thwart such attempts to mislead voters and nancial markets. In addition,
transparency is particularly helpful for policymakers using rules to signal commitments to sound scal
policies. We pointed out before the need to protect their own credibility and that of the rules-based regime
might force policymakers to respond whenever compliance is threatened, even if such reactions are costly in
economic terms. Transparency-focused policymaking frameworks include tools and structures for clear
communication with outside parties. ese aspects of scal policymaking frameworks and objective
assessments of policy responses by scal councils can help policymakers to avoid knee-jerk reactions and
preserve their credibility even if scal shocks cause occasional breaches of numerical rules.
e ability of scal councils to strengthen the effectiveness of numerical rules depends on their degree of
independence from governments, the scope of their mandates and the quality of their outputs. e
importance of independence stems from the high likelihood of tension between governments and scal
councils. In discharging their functions, scal councils are likely to criticise governments from time to time.
Hence, legal and other safeguards are needed to protect such agencies against retaliation by offended
governments, whether in the form of public criticism, budget and staff cuts, or disbandment. Empirical
evidence con rms that governments’ scal forecasts and scal outcomes improve when they are monitored
by scal councils with high levels of operational or legal independence and strong media pro les.

18.5.3 Concluding comment

Appropriate scal policymaking frameworks can contribute to better scal outcomes, but their potential
should not be exaggerated. e ultimate determinant of a country’s scal outcomes is not the nature of is
policymaking framework, but the strength of its government’s commitment to scal discipline. A
policymaking framework cannot substitute for government commitment in the sense that the mere
existence of institutions cannot prevent irresponsible policies.

18.6 Fiscal policy in South Africa

is section highlights and brie y discusses key features of scal policy in South Africa. Section 18.6.1
discusses the evolution of the macroeconomic (stabilisation) role of scal policy in South Africa. We focus
mainly on the period since 1990, but refer to longer-term trends and developments where appropriate.
Section 18.6.2 comments on aspects of the scal policymaking framework in South Africa.

18.6.1 Macroeconomic aspects of fiscal policy in South Africa19

As was the case in most other countries, the South African scal authorities adopted Keynesian policies after
World War II. Fiscal policy in South Africa continued to re ect aspects of Keynesian thinking well into the
1980s and even continued to achieve anti-cyclical effects from time to time thereafter. Towards the end of
the 1970s, however, there were indications that the authorities were beginning to abandon active anti-
cyclical stabilisation policy in favour of longer-term scal planning, a changing mindset that prepared the
way for formal medium-term expenditure planning some 20 years later. Taxation and government
expenditure were no longer used actively or deliberately to stimulate economic growth in times of recession,
nor to dampen demand during boom times. South Africa’s progressive income tax was no longer capable of
exerting strong automatic stabilizing effects owing to bracket creep, which was not compensated for by
indexing income tax brackets to in ation. Moreover, the Treasury was under so much pressure to reduce
government expenditure that it had lost the option of stimulating or cooling-off the economy by changing
the level of government expenditure.
e longer-term focus that gradually replaced the Keynesian approach emphasised scal discipline,
which the authorities regarded as a prerequisite for improving the longer-term growth and job creation
potential of the South African economy. e strong emphasis on the pursuit of price stability, based on the
view that lower in ation is a necessary condition for sustained economic growth, further con rmed the
longer-term and structural focus of scal policy. Statements in the annual budget speeches of successive
Ministers of Finance suggest that price stability was the most important macroeconomic objective of scal
policy from 1994 until 1997, and the second most important in 1990, 1993 and 1998. During the 1990s,
government budgets contained almost no measures aimed at deliberately stimulating economic growth in
the short term. When growth regained pride of place as an objective of scal policy in 1999, the scal
authorities emphasised their intention to promote growth by microeconomic reforms to boost the supply
side of the economy, instead of demand stimulation.20 e government’s decision to adopt medium-term
expenditure planning in 1998 further cemented the movement away from expenditure ne-tuning.
During the 1990s, the focus on scal discipline and other structural aspects of scal policy was closely
associated with the stabilisation or even the reduction of the public sector’s claim on resources, including
the total pool of savings. is development should be interpreted against the backdrop of longer-term trends
in the major scal aggregates.
Sections 1.5.2 of Chapter 1 and 7.2.1 of Chapter 7 showed that the total tax burden and government
expenditure increased, on balance, as percentages of GDP during the 1960s, 1970s and 1980s. e budget of
the national government was in de cit throughout the 1960s and the 1970s, and the loan debt of the national
government increased more than sevenfold in nominal terms from 1961 to 1981. For most of this period, the
economy grew even more rapidly, however, and the debt ratio (i.e. the national government debt expressed
as a percentage of GDP) therefore decreased signi cantly. On average, the national budget de cit–GDP ratio
was no higher during the 1980s than during the preceding two decades. Yet because of sluggish economic
growth, these de cits resulted in a faster growth of the public debt than the gross domestic product. Hence
the government debt ratio increased by about six percentage points of GDP during the 1980s. is
experience reminds us of the simple arithmetic reality that a strongly-growing economy can accommodate
higher budget de cits without or with manageable increases in the public debt–GDP ratio.
e period from 1989 to 1995 saw a marked deterioration in the overall scal situation: the long recession
from March 1989 to May 1993 depressed tax revenues, while several extraordinary transfer payments and the
expansion of social services sharply increased government expenditure. e budget de cit ballooned from
1,4% of GDP in 1989/90 to 7,3% in 1992/93. e resulting sharp increase in national government debt (from
36,4% on 31 March 1989 to 49,5% on 31 March 1996) caused widespread concern about the sustainability of
scal policy.21 Some South African economists even predicted that the government was heading for a debt
trap in which it would not be able to service its debt and might nd it increasingly difficult to raise loans in
the capital market. ese fears con rmed that the private sector had lost faith in the potential of Keynesian
anti-cyclical scal policy. e government did not deliberately increase the budget de cit to stimulate the
economy, but the rising de cit during a deep recession nevertheless represented a strong anti-cyclical policy
stance.22 In the heyday of Keynesian anti-cyclical scal policy, the private sector would have welcomed this
as an appropriate attempt to soften the cyclical downturn. In the early 1990s, however, the relatively high
de cits and rising public debt caused alarm in nancial circles.
It is important to keep in mind that the debt ratio never reached a particularly high level in international
terms. It remained, for example, well below the sixty-per-cent-of-GDP threshold that countries had to
achieve to qualify for membership of the European Economic and Monetary Union (EMU). Analysts of scal
policy were more worried about the growth of the public debt than about its level. e situation was also the
rst test of the democratic government’s commitment to scal discipline. e rst two Ministers of Finance
after the political transition in South Africa, Derek Keys and Chris Liebenberg, were businesspersons whose
perceived political neutrality gave them much credibility in local and international nancial circles. e
appointment of ANC politician Trevor Manuel as Minister of Finance in 1996 raised a crucial question:
would the democratic government maintain scal discipline or would it pursue unsustainable populist scal
policies to achieve its well-publicised goals of ghting poverty and redistributing income?
It was against this background that the South African government adopted the growth, employment and
redistribution (GEAR) strategy in June 1996. From a scal policy point of view, the two most important goals
of GEAR were to turn the deteriorating scal situation around and to quell fears about the democratic
government’s perceived lack of commitment to scal prudence. Hence the scal goals of the GEAR strategy
for the period 1996 to 2000 included a step-wise reduction in the budget de cit to 3% of GDP, maintenance
of the total tax burden at 25% of GDP, the reduction of general government consumption expenditure as a
percentage of GDP and the gradual elimination of general government dissaving. Apart from making major
progress towards effecting a socially more acceptable distribution of income, the application of the strategy
succeeded on both counts. From 1996 to 2001, a combination of buoyant revenue growth, strict control over
public spending and wise use of privatisation receipts reduced the conventional budget de cit from 5,1% of
GDP to 1,9%. Concern about a possible debt trap also evaporated as the debt ratio dropped from a peak of
49,9% of GDP in 1999 to 43,9% in 2001.
e achievements of the rst ten years after the political transition brought the authorities an enviable
reputation for sound scal policymaking23, as was evident from the consistent improvement in South Africa’s
international credit ratings since the rst official ratings were issued in 1994. It has also drawn warm praise
from experts on scal policy. Tanzi (2004: 539), for example, recognised this in an article about episodes of
successful scal reform:

Outside of the Americas, South Africa merits a mention because … it has followed, in recent years, a steady
path towards scal adjustment, trying to use its public resources sparingly and efficiently and creating an
efficient tax system while resisting the temptation of magic solutions. e country has managed to reduce
its scal de cit by about ve to six per cent of GDP over the past decade through a careful reallocation of
spending and tax reform …

e stated scal policy stance became more expansionary after 2001. A major aim of this stance has been to
raise the growth rate of the South African economy through investment in social services and the country’s
physical infrastructure. e data do not re ect the more expansionary scal stance fully, however, because
government revenue regularly exceeded budget forecasts and kept the budget de cit below 2,5% of GDP.
Assisted by relatively brisk economic growth, these modest budget de cits reduced the public debt ratio
further to 29,0% of GDP on 31 March 2007. e scal authorities also succeeded in eliminating government
dissaving: whereas general government current expenditure exceeded general government current income
by 7,3% of GDP in 1992 (the extent of dissaving, in other words), general government saving amounted to
1,2% of GDP in 2006 and 2,4% of GDP in 2007. e longer-term perspective provided by Table 1.1 (Chapter 1)
further ampli es the signi cance of the scal consolidation between the middle of the 1990s and the rst
half of the 2000s, during which time the share of public sector resource use and public sector resource
mobilisation in the economy decreased signi cantly from the average levels recorded for the period 1990–
1994. No wonder that, in an analysis of scal consolidation between 1990 and 2014, it was found that
government expenditure and not revenue was the scal factor that adjusted to restore scal sustainability
during periods of rising debt (see Burger & Calitz, 2014). is curtailment of the longer-term growth of
government is remarkable in that it occurred after the country’s constitutional change and its adoption of
constitutional social rights in recognition of all human rights. is expenditure pattern actually contradicts
some theories of public sector growth, which predict that the share of public spending in national income is
likely to increase in countries with unequal income distributions if lower-income groups receive voting
rights.24 Section 18.3.4 discussed the recent world-wide revival of interest in the cyclical effects of scal
policy and Section 18.4 looked at the momentum gained by scal activism in response to the international
nancial crisis of 2007–2009. South Africa has been no exception in this regard: of late, several researchers
have attempted to estimate the size of the automatic scal stabilisers and to establish how scal policy has
affected the stability of the macroeconomy since 1994. e automatic stabilisers in South Africa appear to be
of the order of 0,5% of GDP (Du Plessis & Boshoff, 2007: 10; Swanepoel & Schoeman, 2003: 813). is
suggests that these stabilisers exert a modest, but by no means negligible, in uence on uctuations in
economic activity. e ndings of efforts to determine whether scal policy has been pro-cyclical or anti-
cyclical since 1994 have been sensitive to the empirical methods employed, but a careful analysis of the
evidence (Du Plessis, Smit & Sturzenegger, 2007) concluded that scal policy had not signi cantly affected
the stability of economic activity in South Africa since 1994.25
Up to the time of the international nancial crisis of 2007–2009, the approach followed by the scal
authorities in South Africa was largely reminiscent of the view that regards monetary policy as a more potent
instrument for stabilisation policy than scal policy. Attempts to ‘ ne tune’ the level of economic activity by
means of scal policy (in other words, active scal policy) was therefore generally avoided. is did not
mean, however, that the scal authorities were ignoring the cyclical state of the economy or the need for
mutually supportive scal and monetary policy stances when formulating the budget. A clear example of
this approach was the inclusion of estimates of the structural budget balance in the 2007 Medium-Term
Budget Policy Statement (National Treasury, 2007d) and the 2008 Budget Review (National Treasury, 2008a).
e 2007 Medium-Term Budget Policy Statement (National Treasury, 2007d: 3) commented as follows on
these matters:

Fiscal policy over the past few years has increasingly taken account of the economic cycle. Government
revenue performance uctuates in response to economic fortunes, in uenced in turn by the economic
performance of our major trading partners, commodity prices, interest rate cycles, in ation trends and
business pro tability. ese are complex factors that affect the economy and tax revenue in ways that
cannot be fully modelled or predicted. In the face of potentially destabilising cyclical factors, it is important
to adopt a policy stance oriented towards stable long-term growth.

In this MTBPS, the National Treasury introduces the concept of a structural budget balance as a
contribution to more systematic and consistent adaptation of the scal stance to cyclical factors. Simply
put, when economic conditions are good, as they are now, we must invest and save in a manner that allows
us to maintain public spending and societal welfare when economic conditions turn less favourable, as
they inevitably will.

e 2008 Budget Review (National Treasury, 2008a: 42) contains a clear example of how the scal authorities
use estimates of the structural (or cyclically adjusted) budget balance to inform policy decisions:

e 2008 Budget takes account of a more unsettled global economic environment. While total revenue
growth is expected to moderate in line with economic activity, strong commodity prices are generating
robust tax revenues from the mining sector. As a result, the cyclical element of tax revenue remains
signi cant. Taking these factors into account, government is budgeting for a scal surplus, which keeps the
cyclically adjusted budget balance from deteriorating. By saving a share of the cyclical revenue, the scus is
 protected from cyclical and external volatility, and ensures that the state does not contribute to pressure on
in ation and the current account de cit.

Some commentators criticised the authorities for budgeting for a surplus in the 2008/09 nancial year,
arguing that it would have been better to increase government spending further to address various social
needs. e quoted statement, however, con rms that contemporary views of the macroeconomic role of
scal policy in South Africa and elsewhere attempt to balance short-term considerations and the ever-
present longer-term issue of the sustainability of the public nances.
e scal surplus turned out to be a blessing in disguise. As indicated earlier, at the start of the
international nancial crisis South Africa was one of the developing countries that was able to utilise its
scal space to counter the ensuing recession with active scal policy measures, without running the same
scal sustainability risk experienced by some industrial countries. In 2010/11, however, the scal surplus did
turn into a de cit equivalent to 5,3% of GDP. is was the outcome of reduced revenue and the maintenance
of medium-term expenditure patterns on economic and social services, but was also re ective of a structural
budget de cit. At the same time, the government committed itself to a process of scal consolidation. e
2011/12 budget envisaged a reduction of the de cit–GDP ratio to 3,8% in 2013/14. e scal consolidation
turned out to have been less successful than in comparable emerging market economies: in fact, the reality
that the debt–GDP ratio in South Africa had increased faster than in peer emerging market countries had
become a matter of concern.
A more comprehensive assessment of scal sustainability requires that trends in the government budget
de cit and public debt be considered together with public assets, that is, to take account of trends in the
balance sheet of government. In this regard Burger, Calitz & Siebrits (2016) concluded that one sensible way
to restore scal sustainability might be to stabilise the debt–GDP ratio at its post-crisis level but by way of the
substitution of much-needed infrastructure capital expenditure for current expenditure. is will entail a
reversal of the falling xed capital–GDP ratio, which would be a positive impetus for economic growth.

18.6.2 The fiscal policymaking framework in South Africa

South Africa has adopted two institutional innovations in monetary policy after 1994: constitutional
protection of the independence of the central bank (1996) and in ation targeting (2000). e South African
scal authorities have implemented multiyear expenditure planning, but have not committed themselves to
numerical scal rules enshrined in legislation (yet). As indicated in Section 18.5, however, rules are growing
in popularity in other countries and it is therefore likely that they will remain under consideration in South
Africa.
South Africa has not used formal numerical scal rules since 1976, when the dual-budget system that
had been in use since 1910 was abolished. Under this system, the authorities had to nance current
spending from tax and other current revenues via the Revenue Account and capital spending from loans via
the Loan Account. Its purpose was to maintain current balance, that is, equality between the current
revenues and current expenditures of government (the principle of current balance is sometimes called the
‘golden rule of scal policy’). We pointed out earlier that the South African authorities have adopted annual
or medium-term scal targets from time to time in recent years to give effect to the more structural approach
to scal policy. ese targets, however, never achieved the status of formal numerical rules. From 1985 until
the mid-1990s, scal policy in South Africa was guided by a 3% de cit guideline. is guideline had no
theoretical basis and proved to be ineffective when the budget de cit rose sharply during the early 1990s. In
1995, the post-apartheid government embarked on the very successful scal adjustment effort that was
discussed in Section 18.6.1. is effort was guided by the goals of the GEAR strategy. e GEAR goals were
pursued in a exible manner, as became apparent when the timeframe for reaching the 3% de cit goal was
extended by one year when the Asian crisis of 1997/98 precipitated an economic downswing. With the
exception of a central bank borrowing rule and constitutional restrictions in the inter-governmental scal
system, there were no permanent restrictions on scal policymaking during the adjustment.26 More recently,
the scal authorities introduced nominal expenditure ceilings in the 2014/15 budget. e ceilings are not
fully edged numerical rules, however, because they lack a legal basis and enforcement mechanisms.
 From 1998 onwards, the South African government adopted various measures to make scal
policymaking more transparent. e overarching framework for this endeavour is the Public Finance
Management Act (PFMA) of 1999. e Act does not put limits on the absolute or relative values of scal
aggregates, that is, it does not prescribe numerical scal rules. Instead, it aims to address the accountability
dimension of scal transparency by emphasising regular nancial reporting, sound internal expenditure
controls, independent audit and supervision of control systems, improved accounting standards and
training of nancial managers, and greater emphasis on outputs and performance monitoring. e Act also
compels the South African scal authorities to disclose their longer-term objectives and views about future
trends in scal policy annually. e introduction of the Medium-Term expenditure Framework (MTEF) in
1998/99 gave effect to Article 28 of the PFMA. Hence, South Africa’s current scal policymaking framework is
very similar to the transparency-based ones discussed in Section 18.5.2.
Both regimes that have been used since 1994 (target-guided discretion from 1994 to 1998 and
transparency-based discretion since then) have worked well. e successful completion of the scal
adjustment effort in 2001 and the maintenance of scal discipline thereafter were made possible primarily
by the scal authorities’ strong commitment to scal prudence. However, the exibility of these regimes
proved useful in a turbulent milieu marked by, inter alia, political democratisation, strong popular pressure
for more expansionary policies and three currency crises (1996, 1997/98 and 2001). eir discretionary
nature was not a drawback from a credibility point of view. e authorities have earned considerable
credibility for themselves and the regime, having established an enviable reputation for maintaining scal
discipline. In fact, Frankel, Smit and Sturzenegger (2006: 71) suggested that the degree of credibility enjoyed
by the South African scal authorities at the time made scal rules unnecessary:

South African authorities have worked hard to earn credibility which has the bene t of allowing some
margin of discretion. What would then be the logic of constraining themselves with scal rules which are
just bound to be violated when shocks require the use of discretion?

e excellent reputation of the scal authorities until the international nancial crisis was also evident from
South Africa’s relatively good international credit ratings, which improved consistently since the rst official
ratings were conducted in 1994. ese considerations suggest that there was no compelling reason why
South Africa should have adopted numerical scal rules. Rules were unlikely to have added credibility
bene ts over and above those already enjoyed by policymakers and the transparency-based regime,
especially if it is taken into account that rules as such could not have constrained future generations of
policymakers. Moreover, the adoption of numerical rules would have denied policymakers the valuable
exibility that they have used wisely – until recently at least. However, the high scal cost of social security
bene ts such as the proposed national health insurance system will, if implemented, be a far greater
challenge to scal discretion and might well strengthen the case for scal rules in future. is is one of the
reasons why scal sustainability has become a matter of concern in the aftermath of the international
nancial crisis and the ensuing Great Recession. Counter-cyclical scal policy resulted in the ratio of public
debt to GDP rising rapidly from 26,5% in 2007/08 to 43,9% in 2013/14. e ratio subsequently increased
further to 55,6% in 2018/19 (National Treasury, 2019a: 215). South Africa’s international sovereign credit was
downgraded in 2012 by both Moody’s and Standard and Poor’s, the most prominent rating agencies, but the
country did retain its investment-grade rating (National Treasury, 2014a: 66).27 Subsequently matters
deteriorated further, however, as was reported earlier with reference to the country’s downgrading to below
investment grade.
As mentioned earlier, the scal authorities’ increasing concerns caused National Treasury to already
introduce rolling cash ceilings for nominal main budget expenditure in 2012. e ceilings are ‘soft rules’ that
lack a legal basis and enforcement mechanisms. e rules have worked well from an accounting point of
view: Main budget spending has remained below the ceiling in every year from 2012/13 onwards (National
Treasury, 2015a: 117; National Treasury, 2016: 32). Yet this achievement did not prevent the upward trend in
the government expenditure–GDP ratio, relatively large budget de cits and persistent growth in the public
debt–GDP ratio. With the bene t of hindsight, it is clear that the ceilings were too high to compensate for the
effects of the anaemic rate of GDP growth on government revenue as well as the government spending–GDP
ratio.

18.7 Fiscal reforms in sub-Saharan Africa 28

Many of the scal issues with which South Africa has been grappling also feature on the scal agendas of
other countries in sub-Saharan Africa. e public nances of most sub-Saharan African countries are
inherently fragile as they have narrow revenue bases and government spending is under great pressure.
During the 1970s and 1980s, policy mistakes and global economic developments rapidly destabilised the
vulnerable scal systems of many sub-Saharan African countries. Since then, efforts to correct these
imbalances and to prevent them from re-emerging have dominated scal policymaking in the region. When
assessing scal trends in sub-Saharan Africa, one should keep in mind that this part of Africa consists of 48
countries whose scal situations and structures differ in many respects. Generalisation is always risky.
Several broad shifts are nonetheless taking place. We highlight a few positive developments.
• Budget balances have improved in many countries, especially from the mid-1990s onwards. It appears as
if the trend toward lower budget de cits in sub-Saharan Africa mainly re ects expenditure cutbacks
rather than revenue increases. Even in 2009, during the Great Recession, national government budget
balances in 27 (about half ) of the African countries were better than –3% of GDP. By 2013, this ratio had
worsened slightly, with the average budget balance for all African countries amounting to a
comparatively modest –3,3% of GDP (African Development Bank, 2016). A considerable number of sub-
Saharan African countries have reduced their dependence on taxes on international trade, partly
because these taxes distort prices and partly to comply with World Trade Organization agreements. A
number of countries recouped the lost revenue by introducing value-added tax. Countries in all parts of
sub-Saharan Africa now use this high-yielding and relatively non-distorting tax. Reductions in income
tax rates have been common and many countries have successfully broadened their tax bases and/or
improved tax collection. Of the thirteen countries for which data was available for the period 1987 to
2002, eleven countries reduced their highest marginal income tax rates on individuals and twelve
countries reduced their highest rate on corporations. More recently (in 2006), Egypt and Mauritius
reduced their highest marginal income tax rates, from 34% to 20% and from 30% to 15% respectively
(Global Finance, 2019). Reductions in corporate tax rates continued to occur: in 2008/09, rates were
reduced in Benin (from 38% to 30%), Cape Verde (from 30% to 25%), Sudan (from 30% to 15%) and Togo
(from 37% to 30%) (World Bank & PriceWaterhouseCoopers, 2010). e impact of the Great Recession
subsequently put a damper on corporate tax reductions. e only country registering a major tax
reduction was Zimbabwe, whose highest marginal corporate tax rate was reduced from about 31% to
about 26% between 2010 and 2012 (Global Finance, 2014).
• Factors such as faster economic growth, accelerated debt relief and smaller scal de cits have reduced
the debt burdens of a number of sub-Saharan African countries. e resulting reduction in the interest
burdens of these countries has released budgetary resources for more productive ends. In contrast to the
1980s, the norm during the 1990s was no longer to reduce public investment sharply when scal
problems occur. is, together with indications that government spending on education and healthcare
generally did not fall nearly as much during reform periods as is often claimed by critics of structural
adjustment, suggests that government-spending priorities have improved in many sub-Saharan African
countries. Before the international nancial crisis, public nances in a number of countries (for example,
Mozambique, South Africa, Tanzania and Uganda) were on a much sounder footing than a decade
earlier, and they were in a better position to promote growth, efficient resource allocation, and the
reduction of poverty and inequality. Empirical evidence indicates, however, that scal policy instruments
are often pro-cyclical (and practically never counter-cyclical [Carmignani, 2010]). e average scal
balance for African countries during 1986–1990 amounted to –7,5%, which improved to 2,6% during
2004–2008. e subsequent weakening of the economies together with stimulus packages caused scal
balances in Africa to deteriorate on average by around 6,5 percentage points of GDP, that is, from a
surplus of 2,2% of GDP in 2008 to a de cit of 4,4% of GDP in 2009, thus mitigating the downturn of
aggregate demand.29
e scal challenges nonetheless remain formidable. Narrow tax bases and underdeveloped nancial
markets still make many African countries highly dependent on grants and foreign borrowing as sources of
nancing for government spending. Spending pressures and susceptibility to external shocks remain high,
and the gains of the recent past are therefore fragile. Moreover, governance and service delivery challenges
remain acute, as was affirmed by the ndings of a survey commissioned by the United Nations Economic
Commission for Africa (2004). e international nancial crisis of 2007 and beyond has revealed that
policymakers’ room to manoeuvre in fragile countries in sub-Saharan Africa is limited in periods of crisis
because of low scal space and limited institutional capacity (Allen & Giovannetti, 2011). In African
countries, successful scal consolidations were achieved using discretionary scal regimes (for example, in
South Africa) or regimes bound by conditionalities of the Bretton Woods institutions.

Key concepts
• active or passive fiscal policies (page 405)
• automatic or built-in stabiliser (page 403)
• budget-process rules (page 426)
• common-pool problem (page 430)
• contractionary (fiscal) policy (page 411)
• conventional balance (page 406)
• crowding out (page 410)
• current balance (page 437)
• cyclically adjusted budget balance (page 416)
• decision lag (page 408)
• discretionary policy (page 428)
• dissaving (page 407)
• expansionary (fiscal) policy (page 411)
• fiscal activism or anti-cyclical fiscal policy (page 403)
• fiscal consolidation (page 421)
• fiscal councils (page 426)
• fiscal discretion (page 428)
• fiscal illusion (page 412)
• fiscal policy (page 398)
• fiscal policymaking framework (page 426)
• fiscal responsibility laws (page 426)
• fiscal solvency (page 422)
• fiscal sustainability (page 422)
• fiscal transparency (page 430)
• full-employment income budget balance (page 416)
• impact lag (page 409)
• implementation lag (page 409)
• Keynesian approach (page 403)
• macroeconomic goals of fiscal policy (page 399)
• macro instruments (page 401)
• market discipline (with reference to fiscal policymaking) (page 430)
• medium-term expenditure frameworks (page 426)
• microeconomic goals of fiscal policy (page 400)
• multiplier (page 405)
• National Treasury (page 402)
• net borrowing requirement (page 406)
• net indebtedness (page 407)
• numerical fiscal rules (page 426)
• ordinary revenue (page 407)
• output gap (page 417)
• potential-income budget balance (page 416)
• primary balance (page 407)
• primary deficit (page 407)
• primary surplus (page 407)
• procedural fiscal rules (page 426)
• Public Finance Management Act (page 399)
• public sector borrowing requirement (page 406)
• rational expectations hypothesis (page 428)
• recognition lag (page 408)
• Ricardian equivalence (page 403)
• sectoral goals of fiscal policy (page 399)
• sectoral or micro instruments (page 401)
• stagflation (page 413)
• structural approach to fiscal policy (page 414)
• structural budget balance (page 416)
• supply-side economists (page 413)
• total revenue (of government) (page 406)
• transparency-enhancing reforms (page 430)
• trend-line budget balance (page 416)

SUMMARY
• Fiscal policy are decisions by national government regarding the nature, level and composition of
government expenditure, taxation and borrowing, aimed at pursuing particular goals. It has an active
and a passive element.
• Fiscal policy has macro, sectoral and micro goals and instruments. Fiscal policy is not a panacea for all
economic problems. Monetary policy, trade and industrial policy, competition policy and labour policy
are important allies, and some are more effective than others in pursuing particular goals.
• e Minister of Finance is the key gure in scal policy and is given certain statutory powers by acts of
parliament. Various supporting institutions exist, notably the National Treasury, whose policymaking
role is undertaken in close coordination and cooperation with the South African Reserve Bank.
• e nal approval of the annual budget occurs by way of a law of parliament, the watchdog of the public
purse. Two recent reforms have strengthened the oversight role of parliament.
• Mainstream views on the macroeconomic role of scal policy have changed signi cantly over time. e
Keynesian approach dominated thinking for three decades after World War II, following which new
classical and structural views gained supremacy. Keynesian thinking regained some credibility during
and after the international nancial crisis of 2007/08. e passive (automatic stabilisers) and active
dimensions are explained in Figures 18.1 and 18.2.
• Different budget balances are used in the planning, implementation and assessment of scal policy,
namely the conventional, current and primary balance. ese and other related concepts are explained
in Box 18.1. Sometimes cyclically adjusted balances are calculated to analyse the underlying or structural
balance, which is important for long-term scal sustainability.
• ere are three reasons why anti-cyclical scal policy is deemed ineffective. Firstly, it entails various
stages, each with its own delays or time lags, namely the recognition, decision, implementation and
impact lag. Secondly, there is the possibility that an expansionary scal policy will push up interest rates
and in this way crowd out private investment. is dampens the multiplier effect of a scal policy
stimulus. irdly, new classical macroeconomists argue that private agents would anticipate systematic
counter-cyclical policies and respond to them in ways that neutralise their intended effects. e
tendency of governments to apply popular expansionary measures eagerly during economic recessions,
but not to reverse these measures symmetrically during upswings, revealed a further operational
 shortcoming, as pointed out by public choice economists. is tendency contributes to increases in the
share of government in the economy and to rising public debt ratios.
• Supply-side economists and proponents of the structural approach to scal policy have voiced
opposition to scal policy activism. e latter puts more emphasis on the current and future
sustainability of scal policy, and how the nancing of budget de cits is likely to in uence the economy.
is approach leaves room for passive anti-cyclical scal policy to the extent that automatic stabilisers
are effective.
• e international nancial crisis kindled a renewed interest in Keynesian activism, but many economists
still maintain that, except perhaps in most extreme situations, monetary policy is a better tool for
stabilisation purposes. ere are two reasons for the tentative comeback of scal policy: owing to lower
in ation, automatic scal stabilisers became more effective and monetary policy lost its effectiveness
once interest rates approached the zero bound. is, in turn, was countered by quantitative easing.
• e importance of distinguishing between cyclical and structural budget balances is explained with
reference to Figure 18.3. e distinction can help policymakers to avoid errors, a typical example of
which is taking long-lasting decisions on taxes and expenditures based on transitory movements in
revenues.
• e scal consequences of the international nancial crisis are explained with reference to Tables 18.1
and 18.2. It illustrates the difference between countries with structural de cits and surpluses, and the
countries’ big differences in the ratio to GDP of primary balances and public debt. What made the choice
of scal measures so problematic was the uncertainty about the nature and depth of the crisis, the likely
duration of the economic downswing, the reaction time to and strength of different stimulatory
measures, and the difficult trade-off between short-term scal stimulation (with the associated higher
de cit and debt levels) and longer-term scal sustainability (associated with lower de cits and debt
levels). One set of authors concluded that an optimal scal policy package should be timely, lasting,
diversi ed, contingent, collective and sustainable. ey then argued that spending increases as well as
targeted tax cuts and transfers were likely to have the largest multipliers.
• Fiscal sustainability, as explained in Box 18.2, refers to a government’s continued ability to service its
debt (in other words, pay the interest on its domestic and foreign debt timeously) and repay the loan
amount (principal) when it becomes due. It should be distinguished from scal solvency. Fiscal
sustainability remains a somewhat elusive concept because it depends on assumptions about future tax
compliance and government expenditure discipline, and the projected output gap, that is, the difference
between the actual and potential GDP.
• e mixed track record of discretionary scal policy has resulted in the adoption by industrial and
developing countries of scal rules, which are permanent restraints on scal policy, typically de ned in
terms of indicators of overall scal performance such as expenditure, current balances, overall balances
or public debt. Fiscal rules are discussed as one of four elements of scal frameworks. Fiscal rules
essentially have a negative purpose, namely to prevent policymakers (who allegedly cannot be trusted to
do the right thing) from lapsing into scal pro igacy.
• Proponents of scal rules argue that the adoption of these rules represents a credible commitment to
sound policies that should lessen or solve the credibility problem in a rational expectations world. is
needs to be distinguished from binding constraints on scal policy. ere are, however, two main
reasons why governments nd it relatively easy to circumvent, ignore, suspend or abandon scal rules:
all rules are subject to sovereign political authority and policymakers have only limited control over scal
outcomes. It is therefore argued that the rational economic agents postulated by modern
macroeconomic theory would recognise the fragility of rules and not regard rules-based commitments
as credible.
• Two other criticisms of rules have been raised. Firstly, rules may force governments to adopt ultra-
conservative forecasting and budgeting techniques that result in excessively contractionary scal
outcomes or to undertake undesirable forms of tax increases and expenditure cuts. Secondly, neither
economic theory nor experience provides a basis for adopting rules that transcend business cycles and
structural economic changes for long periods.
• Some scal policy experts believe that transparency-enhancing reforms are more effective than scal
rules to make scal policies credible. eir greater exibility gives them a major advantage over rules-
based regimes (especially to policymakers who are strongly committed to prudent scal policies and
who therefore do not need binding by rules). Another advantage is their superior ability to strengthen the
effectiveness of market discipline over scal policymakers. is refers to the speed with which the
markets punish scal laxness by suspending lending, withdrawing capital and increasing risk premiums
on borrowed funds.
• Adopting rules may well be useful when a new government without a record of accomplishment wishes
to signal its commitment to sound scal policies.
• ere is a sense in which rules and discretion meet each other. Discretion with increased transparency
and accountability (and the underlying institutional reforms to correct for destabilising incentives on the
side of policymakers) amounts to constrained exibility. Rules with escape clauses or rules specifying
medium-term boundaries or constraints to be met over the course of the business cycle amount to
exible constraints.
• A fairly recent institutional development complementing rules and adding to transparency has been the
establishment of scal councils. A scal council typically consists of a group of independent experts with
designated responsibilities to analyse and comment publicly on scal policy. ree types are
distinguished. eir ability to strengthen the effectiveness of numerical rules depends on their degree of
independence, their remits and the status of their forecasts.
• Medium-term scal or expenditure planning is essential in industrial and developing countries from a
macro, sectoral and microeconomic point of view. South Africa’s medium-term budget policy statement
(MTBPS) (published annually since November 1997) is an example of this type of planning.
• Fiscal policy in South Africa continued to re ect aspects of Keynesian thinking well into the 1980s,
although there were earlier signs of abandoning it in favour of longer-term scal planning. e longer-
term focus emphasised scal discipline, which the authorities regarded as a prerequisite for improving
the longer-term growth and job creation potential of the South African economy. During the 1990s, the
focus on scal discipline and other structural aspects of scal policy was closely associated with the
stabilisation or even the reduction of the public sector’s claim on resources, including the total pool of
savings.
• e period from 1989 to 1995 saw a marked deterioration in the overall scal situation: a sharply rising
budget de cit and public debt, with fears of a debt trap. Analysts were more worried about the growth in
the public debt than about its level. e situation was also the rst test of the democratic government’s
commitment to scal discipline. From a scal policy point of view, the two most important goals of
GEAR were to turn the deteriorating scal situation around and to quell fears about the democratic
government’s perceived lack of commitment to scal prudence. e post-apartheid government
succeeded with both and was credited for achieving scal sustainability in a transparency-based
discretionary manner (in other words, without numerical scal rules). e Public Finance Management
Act of 1999 played an important role in this regard.
• However, scal sustainability in South Africa has become a matter of concern for various reasons in the
aftermath of the international nancial crisis and the ensuing Great Recession.
• e public nances of most sub-Saharan African countries are inherently fragile as they have narrow
revenue bases and government spending is under great pressure. During the 1970s and 1980s, policy
mistakes and global economic developments rapidly destabilised the vulnerable scal systems of many
sub-Saharan African countries. Since then, efforts to correct these imbalances and to prevent them from
re-emerging have dominated scal policymaking in the region. A few positive developments are
highlighted.
• e international nancial crisis has revealed that policymakers’ room for manoeuvring in fragile
countries in sub-Saharan Africa is limited in periods of crisis because of low scal space and limited
institutional capacity. In African countries, successful scal consolidations were achieved using
discretionary scal regimes (for example, in South Africa) or regimes bound by conditionalities of the
Bretton Woods institutions.
MULTIPLE-CHOICE QUESTIONS
18.1 Which of the following is a macroeconomic instrument of scal policy?
a. e bank rate
b. e excise tax rate on tobacco products
c. A government subsidy on white bread
d. e primary budget balance
e. All of the above options are correct.
18.2 e primary budget balance is equal to …
a. total government revenue minus total government expenditure.
b. the conventional budget balance plus interest expenditure.
c. the conventional budget balance minus interest expenditure
d. the conventional budget balance plus capital expenditure.
18.3 e following statement best describes Keynesian thinking on scal policy:
a. In total, the conventional budget balances should be equal to zero over the course of the business
cycle.
b. e structural de cit should equal government dissaving.
c. e cyclically adjusted conventional budget balance should always be zero.
d. Automatic stabilisers are an important instrument of active scal policy.
18.4 Empirical research indicates that the best way to restore scal sustainability is by …
a. increasing the personal income tax rates.
b. reducing public investment.
c. issuing new government bonds.
d. reducing the government wage bill.

SHORT-ANSWER QUESTIONS
18.1 ‘e conventional budget de cit is superior to other de nitions of budget de cit as a measure of the
stance of scal policy.’ Do you agree? Explain your answer.
18.2 ‘Owing to lags, scal policy is irrelevant.’ Do you agree with this statement? Substantiate your answer.
18.3 Use a diagram to explain the distinction between automatic scal stabilisers and active scal policies.
18.4 How would you know that a government was serious about ensuring scal sustainability?
18.5 Discuss the structural approach to scal policymaking.
18.6 Discuss the advantages and disadvantages of numerical scal rules. Is scal policy in South Africa
driven by rules or discretion? Substantiate your answer.

ESSAY QUESTIONS
18.1 Discuss the principles of Keynesian anti-cyclical scal policy and the reasons for its fall from favour
since the mid-1970s.
18.2 Use a diagram to explain the distinction between automatic scal stabilisers and active scal policies.
18.3 Why did the Keynesian approach to scal policy lose its appeal during the 1970s, and then regain some
attractiveness during and after the Great Recession?
18.4 Use a diagram to explain the distinction between the structural and the cyclical components of budget
balances, and discuss the practical signi cance of this distinction.
18.5 Discuss the difference between binding constraints and commitment devices in scal policy. Which of
them will best serve credible scal policymaking and why?

1 e economic impact of some of these taxes is largely limited to a particular sector or a limited number of sectors in the economy
(for example, the excise tax on tobacco, the levy on plastic bags or the mining tax). Others such as value added tax, the fuel tax and
 income tax on individuals and companies, exert their in uence throughout the economy. Changes in these taxes may therefore
affect the macroeconomic performance of the country.
2 Section 77 of the Value Added Tax Act (Act 89 of 1991) authorises the Minister to make such adjustments, with the proviso that
Parliament has to legislate this within six months after noti cation.
3 For more information, visit the following website: http://www.resbank.co.za.
4 For more information, visit the following website: http://www.treasury.gov.za.
5 Section 12.4 in Chapter 12 introduced the notion of automatic stabilisers.
6 is framework was adapted from El-Khouri (2002: 213–215).
7 An analysis of the difference between actual and budgeted gures in South Africa for the period 2000/01–2010/11 found that the
margin of forecast error in respect of each of revenue, expenditure and GDP had been quite big at times (Calitz, Siebrits & Stuart,
2013). e authors pointed out that scal credibility would be severely tested if such errors were to coincide in any particular year.
8 Incidentally, taxation through in ation as a result of bracket creep, which was discussed in Chapter 13 (Section 13.4.4), falls in the
same category.
9 In 1903, the Italian scal economist, Amilcare Puviani, recognised that the existence of scal illusion or the potential for its
establishment would enable or ensure the continued tax nancing of growing demands for public goods.
10 Quantitative easing (QE) is an unconventional monetary policy whereby central banks stimulate the economy when standard
monetary policies (notably interest rate measures) have become ineffective. It entails buying speci ed amounts of nancial assets
from commercial banks and other private institutions, thus increasing the monetary base and lowering the yield on those nancial
assets. is is distinguished from the more usual open-market policy of buying or selling government bonds in order to keep inter-
bank interest rates at a speci ed target value. ree rounds of QE took place in the US: in 2008, 2009 and 2011.
11 e United States and the United Kingdom implemented large scal stimulus packages. Blinder and Zandi (2010) reported that the
United States was to spend close to $1 trillion (roughly 7% of GDP) on scal stimulus. Note from Table 18.1 that both the UK and
the US had substantially higher debt–GDP ratios in 2009 than in 2006 and that both countries experienced further increases since
then. By comparison, the South African ratio was still falling by 2007 and had increased to only 30,1% in 2009.
12 e contents of this box comes from Calitz and Siebrits (2018: 1–4).
13 Even though earlier in this chapter a de nition of successful scal consolidation was offered, de nitions differ from author to
author. A consensus view has not emerged.
14 According to Sundaram and Chowdhury (2016), the historical experiences of Eurozone and other economies suggested that the
probability of successful scal consolidation is low. Factors that affected the probability of success included global business cycles,
monetary policy, exchange rate policy and structural reforms.
15 is section draws on parts of Siebrits and Calitz (2004b) and Siebrits (2018). ese writings contain references to many important
publications on scal policymaking frameworks.
16 Various spillover and other effects provide a rationale for the adoption of scal rules in a regional bloc (such as the European
Union). It could also be argued, however, that rules are credibility-enhancing devices that re ect the commitment of the member
states to economic integration.
17 Fiscal rules did however seem to retain their value as points of scal responsibility to which countries should return by
implementing medium-term scal consolidation measures, albeit within time frames varying greatly across countries.
18 Australia, New Zealand and the United Kingdom pioneered the adoption of transparency-enhancing measures aimed at making
scal policy-makers more accountable. e scal frameworks of these countries do not prescribe speci ed numerical targets, but
require the scal authorities to disclose their scal objectives regularly. e objectives may or may not take the form of quantitative
targets. Section 18.6.2 shows that South Africa’s current scal policymaking framework also focuses on the notions of transparency
and accountability.
19 is section draws on Calitz and Siebrits (2003), Siebrits and Calitz (2004b) and Burger, Calitz and Siebrits (2016).
20 e shift to supply-side factors is in line with the thrust of new growth theory, which argues that growth-promoting capital
formation is not limited to investments in privately owned physical capital (for example, factories and machinery). New growth
theory indicates that additions to any of the following components of the capital stock may yield increasing returns by creating
externalities that bene t a range of sectors and industries: the existing physical infrastructure (for example, roads and electricity
networks), accumulated human capital acquired through education, training and healthcare, and technical expertise acquired
through learning-by-doing and research and development (see Section 7.6 of Chapter 7). ese kinds of measures obviously are
important ingredients of any attempt to improve a country’s international competitiveness.
21 Some of this high debt re ected the consolidation of subnational regional debt, which, together with the new Provincial
Government Borrowing Act, laid the basis for much more prudent subsequent borrowing. Also the proper (full actuarial) funding
of the government employees’ pension funds (GEPF) removed a potential future contingent liability. (During 2018, however, it
became apparent from the actuary’s report on the GEPF that the fully funded status has come under threat.) e funding status
nevertheless needs to be taken into account when comparing South Africa’s debt–GDP ratio to that of countries with underfunded,
pay-as-you-go pension funds. e debt–GDP ratios of such countries would be higher if the actual pension liabilities are taken into
account as well. For more on this, see Calitz, Du Plessis and Siebrits (2011).
22 is impression of counter-cyclical scal policy is not borne out by econometric analysis of the pattern over longer periods. Various
studies have found scal policy to be pro- rather than counter-cyclical. Swanepoel and Schoeman (2003) found that scal policy
was strongly pro-cyclical between 1986 and 2000. Burger and Jimmy (2006) also found that scal policy was pro-cyclical between
1975 and 2004.
23 For an assessment of scal policies post South Africa’s political transition, see Ajam and Aron (2007).
24 is is the main line of argument developed by Meltzer and Richard (1981); see Section 7.4.2 of Chapter 7.
25 Nonetheless, Du Plessis, Smit and Sturzenegger (2007) found that scal policy itself was pro-cyclical between 2002 and 2006.
26 Section 13(f ) of the South African Reserve Bank Act of 1989 (as amended) prohibits lending to the South African government.
27 One of the reasons for the downgrades was the ‘(s)hrinking headroom for counter-cyclical policy actions, given the deterioration in
the government’s debt metrics since 2008, the uncertain revenue prospects and the already low level of interest rates’ (Moody’s
Investors Service, 2012).
28 is section draws on Calitz and Siebrits (2007).
29 For this and other information on the past trends and economic outlook of African economies, see African Economic Outlook
(African Development Bank, 2014).
Fiscal federalism

Tania Ajam

e aim of this chapter is to study the rationale for scal federalism and the nature of inter-governmental
scal relations from an efficiency and equity perective, with particular reference to the South African
context.
We start by examining the economic rationale for scal decentralisation as opposed to the arguments for
scal centralisation. is is followed by an explanation of the considerations on which the assignment of tax
powers and expenditure functions to sub-national governments are based. Next we discuss tax competition
and tax harmonisation as well as borrowing powers and debt management at sub-national level, before
proceeding with an analysis of different kinds of inter-governmental grants. e chapter concludes with an
overview of inter-governmental scal issues within the spheres of provincial and local government in South
Africa.

Once you have studied this chapter, you should be able to:
explain why sub-national governments exist at all, and compare the merits of fiscal decentralisation and
centralisation
describe the Tiebout model
describe the assignment of expenditure functions and revenue sources (tax powers) to the national,
provincial and local spheres of government in South Africa
distinguish between tax competition and tax harmonisation
explain the reasons for, and the nature of, inter-governmental grants
list the types of inter-governmental grants
explain the issues that surround borrowing powers and debt management at sub-national level
review the role of the South African Financial and Fiscal Commission (FFC) in sharing revenue across the
three spheres of government
explain how economies of agglomeration contribute to the existence and growth of cities
discuss trends and issues in provincial and local government financing in South Africa.

19.1 The economic rationale for fiscal decentralisation

In Chapter 1, we introduced the concept of general government (Figure 1.1). is concept signi es that
governments typically consist of more than one level. at is, there may be provincial or local tiers of
government in addition to national government.
e term ‘sub-national government’ (in other words, provincial and local government) refers to those
levels of government that have smaller geographic jurisdictions compared to the national government.
National government’s jurisdiction would, of course, extend to the whole country, whereas the jurisdiction
of a municipality, for instance, would only be within the borders of that particular municipality. In some
countries, provincial governments are called state or regional governments, and the national government is
called the central or federal government. e local tier of government typically consists of several
municipalities.
Budgetary decisions are generally made at different levels of government. e greater the discretion that
sub-national governments are authorised to exercise (for example, decisions about spending, taxation and
borrowing), the more decentralised the scal system is. In a state with more than one level of government,
inter-governmental scal relations, also called ‘ scal federalism’, is concerned with the structure of public
nances: how taxing, spending, borrowing and regulatory functions are allocated among the different tiers
of government as well as the nature of transfers between national, provincial and local governments. In
contrast to approaches followed in political science and constitutional law, the generic meaning of the term
‘federalism’ in economics implies decentralisation, and scal federalism deals with the scal implications of
a decentralised system of multilevel government.
e justi cation for a decentralised system embodying sub-national decision-making powers stems
partly from our earlier analyses presented in Part 1 of the book. Firstly (and as discussed in Chapter 2), it
may improve allocative efficiency or the ability of the public sector to produce the level and mix of public
services that citizens demand and that correspond with their preferences. However, this depends on the
nature of the speci c public (and merit) goods produced and delivered by government, which differs from
country to country. Some public goods are national in scope (for example, defence) or indeed global (as
discussed in Section 3.9 of Chapter 3). Local public goods, however, confer bene ts that are con ned to a
limited geographical area. For instance, the transmission of a regional or local radio programme would
bene t only those people within the broadcasting range of the transmitter, and public playgrounds would
generally serve the recreational needs of children living close to them. Local public goods are therefore
speci c to a particular location. By electing to locate in a particular geographic jurisdiction, consumers can
therefore choose the quantity and type of local public goods they receive. is theme is explored in greater
detail in Section 19.1.1, where we discuss the Tiebout model and the allocative role of government.
Local and provincial governments therefore exist as a result of the fact that the spatial incidence of public
goods differs. In practice, however, the boundaries of sub-national governments are often historically or
politically determined. erefore these may not coincide exactly with the bene t areas of the public goods
that sub-national governments produce. As a result, spatial externalities may exist, that is, spillovers of
costs and bene ts at the boundaries between sub-national government jurisdictions.
A second rationale for decentralisation is provided by public choice theory generally, and the limitations
of a centralised majority-based democratic system in particular, as discussed in Chapter 6 and brie y
elaborated on in Section 19.1.2 below.

19.1.1 The Tiebout model

Tiebout (1956) asserted that if there were a large enough number of local government jurisdictions and each
of these local governments offered a different mix of local public goods and taxes, individuals would reveal
their true preferences for local public goods by choosing a particular local government jurisdiction in which
to live. In this model, citizens (who have different tastes) are mobile and choose to settle in the local
government jurisdictions that produce the mix of tax and public good outputs that correspond most closely
to their preferences. eir choice of location thus reveals their preferences for public goods in the same way
that their choice of private goods purchased in the market reveals their preferences for private goods.

Just as the consumer may be visualised as walking to a private market place to buy his or her goods, we
place him or her in the position of walking to a community where the prices (taxes) of community services
 are set. Both trips take the consumer to market. ere is no way in which the consumer can avoid revealing
his or her preferences in a spatial economy.

Tiebout, 1956: 422

e greater the number of communities and the greater the variation in taxes and public services offered,
the closer consumers will be to satisfying their preferences. Under these conditions, local public goods can
be decentralised in a way that is immune to the free-rider problem. Tiebout’s notion of ‘voting with one’s
feet’ permits the revelation of preferences by allowing people to sort themselves into groups with similar
tastes. Furthermore, the equilibrium that will be achieved by voting with one’s feet will be Pareto efficient.
e Tiebout model thus describes a theoretical solution for the problem of preference revelation, a
phenomenon that inhibits the achievement of allocative efficiency (see Chapters 2, 3 and 6).
It must be noted that the Tiebout model is based on the following restrictive assumptions:
• All citizens are fully mobile.
• Individuals have full information about the local public goods offered by each jurisdiction.
• ere is a large number of jurisdictions to choose from, spanning the full range of public good
combinations desired by citizens.
• ere are no geographic employment restrictions: people receive income from capital only and are not
tied to a particular location through job or family ties.
• ere are no spillovers across jurisdictions.
• ere are no economies of scale in the production of public goods.

If there are economies of scale in the production of public goods and hence declining average cost (for
example, the cost of an additional listener to a local radio programme or of an additional road user may be
zero), then a local public goods equilibrium may not exist at all. Preference revelation once again becomes a
problem.
If there is only a limited number of communities, they may compete with each other to attract outsiders.
While this behaviour (analogous to a monopolistically competitive rm) may provide an incentive towards
an efficient production of public services, the mix and level of public services provided may not be Pareto
efficient. If there are fewer communities than types of individuals, a person might not be able to nd a
jurisdiction where people’s tastes match his or her own.
Finally, there are issues concerning redistribution. As there is an element of redistribution involved in
the provision of local public goods (for example, health and education services), the wealthy may attempt to
avoid this redistribution by segregating themselves from the poor (Atkinson & Stiglitz, 1980).
Although the Tiebout model is based on a number of stringent assumptions, it does clearly demonstrate
that a decentralised scal system – which can accommodate a diversity of preferences for public goods – can
be welfare-increasing in relation to a centralised system that imposes a standardised public good-tax mix on
people, no matter what their tastes (see Box 19.1). Fiscal decentralisation can in principle contribute to a
more efficient provision of local public goods by aligning expenditures more closely with local priorities.

Box 19.1 The Tiebout model extended

Since its debut in 1956, the Tiebout model has become the cornerstone of fiscal federalism and new local
government finance literature, inspiring literally hundreds of journal articles and books, which later extended the
very simple, highly abstract initial model. The original model and its subsequent refinements have captured the
attention of economists as well as political scientists, geographers and other social scientists (Fischel, 2006).
One type of extension to the Tiebout Model under which its efficiency conditions continued to hold was the
addition of a political collective choice mechanism, most often through imposing a median voter mechanism.
This rendered the model more descriptively realistic, since in practice, residents do not only have the option of
relocating to other communities in order to change their public good consumption patterns (that is, ‘voting with
their feet’, referred to above), but they can also influence the mix of public goods offered at local level by voting
in elections. Other researchers have added a focus on the role of municipal managers in service provision
(Fischel, 2006).
Other significant extensions to the model include expanding the role of property taxation (in practice, this is
a major revenue source for many municipalities), land use regulation and the impact of zoning restrictions, the
relationship with housing markets and residential sorting (in other words, whether communities consist of
homogenous communities of a single income class or heterogeneous communities with a mix of income
classes). In the original Tiebout model, taxes are regarded as merely the ‘price’ of a basket of local public
goods, and the model made no attempt to explain the form and nature of these taxes or local government
budget processes. Extensions to the Tiebout model that incorporated property taxes, however, evinced an
incentive for citizens to free-ride. Citizens could purchase a house with a value substantially below the value of
the neighbouring houses, thereby paying less property tax than other homeowners in the community, but still
consuming the same basket of local public goods (Oates, 2006). Hamilton (1975) resolved this dilemma of
partial free-riding by adding a zoning rule that enforced a minimum threshold in housing consumption (through
zoning by-laws, for instance). Zoning is the practice of regulating land use (for example, specifying the minimum
size of a house or the density of construction) to limit the growth in a particular community, to achieve optimal
size for public service provision and to avoid ‘free riders’ who would consume more in public services than they
would pay in property taxes. Since all the households in the community would have more or less the same
property value (as a result of zoning), homeowners in a particular community would have similar property tax
burdens for consuming the same bundle of local public goods, which would be equal to the costs of service
provision.
The Tiebout model focused exclusively on the efficiency of public service provision and ignored the
distributional impacts completely. However, the distributional consequences of homogenous Tiebout
communities caused concern since there would be an incentive for rich or white communities to segregate
themselves in wealthy enclaves, excluding poor individuals or possibly even those of a different race or religion:
‘Arguably, the reason that the debate about the micro-foundations of the Tiebout model has enjoyed such a
long life and generated such strong feelings is the possibility that sorting has less to do with preferences about
service-and-tax bundles and more to do with the preferences of well-healed citizens, especially White people, to
sort themselves into racially and economically homogenous enclaves’ (Bickers et al., 2006: 59).
Zoning requirements in many countries explicitly may not exclude potential residents on the basis of race,
religion and so on, but political scientists have expressed concerns that the application of zoning by-laws or
resident income thresholds could indirectly have an exclusionary impact in practice.
Furthermore, the stratification of communities on the basis of income would limit local cross-subsidisation
and redistribution, so that efficiency in local public provision could be achieved, but at the cost of conflicting
with other social values, such as inclusivity and diversity (Oates, 2006).
The Tiebout model was never intended solely as a theoretical mechanism for the revelation of local public
good preferences. It also aimed to explain the actual behaviours observed in local government (particular urban
metropolitan municipalities). The Tiebout model experienced an academic resurgence when Oates first
attempted to test an important implication of the model empirically in 1969. If citizens were willing and able to
‘shop around’ across local government jurisdictions for the mix of local public goods and associated taxes that
most closely matched their preferences, as the Tiebout model would postulate, then both the quality of local
public goods and services and local property tax liabilities should be reflected in local property values. In other
words, fiscal differentials should be ‘capitalized’ into the prices of houses across different municipal
jurisdictions. Since Oates’ seminal application in 1969, a myriad of empirical studies (primarily in developed
countries) have exhibited significant capitalisation among homes in the same housing market. Some debate
remains, however, about the degree of capitalisation that occurred (Oates, 2006).
After Oates, a vast empirical literature emerged, testing either whether the assumptions on which the
Tiebout model is based prevail in practice or the implications of the model, which are empirically testable.
Empirical testing has focused mainly on the developed world, particularly the United States, the United
Kingdom and Canada. The evidence has been mixed over the last 50 years, with certain patterns and trends
consistent with the model and others conflicting (Baiker et al., 2010). Dowding et al. (1994) surveyed over 200
journal articles and books on the Tiebout model. Their findings in relation to some of the testable implications
of the extended Tiebout model include the following:
• The larger the number of local jurisdictions is, the greater the satisfaction levels of citizens. Mixed but
marginal evidence was found for this proposition.
• The larger the number of jurisdictions in the same metropolitan area is, the greater the competition among
them to attract residents. This was difficult to test rigorously and hardly any credible evidence in support of
 this hypothesis was found.
• The larger the number of competing local jurisdictions is, the more homogenous each will be. Only weak
evidence was available to support this proposition.
• The rich may band together to avoid paying higher taxes to cross-subsidise local public service delivery to the
poor. There is empirical evidence in the developed world to suggest that rich households do relocate to avoid
redistributive local taxation.
• The higher the quality of local services offered, the higher the property values in a local jurisdiction (as
indicated by house prices, for example), and the higher the property tax level , the lower the value of
properties. There is evidence that the locational decisions of households are influenced by both taxes and
services, and to a lesser extent that there is a relationship between local taxes and services and property
values.
• The decisions of households to relocate are affected by the different taxes levied, and baskets of services
offered, by different local jurisdictions. Migration patterns can be explained by differences in tax and service
packages and welfare benefits (in other words, fiscally induced migration). Differences in tax and service
packages do seem to affect migration decisions, but the evidence is also open to other interpretations.

While the empirical jury is in many respects still out on the practical application of the Tiebout model, its
theoretical contribution remains enormously in uential. Instead of viewing public good provision in
municipalities and provinces as a collective choice exercised by static communities, the Tiebout model
highlighted the rational decisions and dynamic responses of mobile individuals who move to match their
preferences. is not only introduced a more de ned spatial focus, but also focused attention on the role of
migration and mobility. e Tiebout model remains a fertile area for further research, especially within the
developing world as urbanisation proceeds rapidly.

19.1.2 Public choice perspective on fiscal federalism

From a public good perspective, we have already discussed (in Chapter 6) Arrow’s impossibility theorem
and the maximising behaviour of politicians, bureaucrats and special interest groups within a centralised
democratic system. Both explicit and implicit logrolling1 on the part of political parties and individual
politicians as well as the empire-building motives of bureaucrats, coupled with rational ignorance among
the broad electorate, can give rise to an oversupply of public goods in the economy. To the extent that they
do, the net effect is clearly sub-optimal, with goods and services not being allocated in accordance with the
relative preferences of the community as a whole.
is Leviathan hypothesis, rst proposed by Brennan and Buchanan (1980), views government as a
revenue-maximising monopolist that seeks to exploit its citizens systematically by maximising the tax
revenue that it extracts from the economy. According to this perspective, scal decentralisation would place
a powerful restraint on the government’s Leviathan tendencies. Devolution of taxing and spending powers
to sub-national governments would act as a disciplinary force on the size of government by forging a closer
link between raising funds and spending funds. For instance, any additional expenditure by a sub-national
jurisdiction may have to be funded by increased sub-national taxation. Centralised scal systems break this
link, encouraging the growth of government. In centralised scal systems, local residents have more
opportunities to lobby for spending programmes that are nanced out of nationally collected revenues or
national loans.

19.1.3 Other reasons for fiscal decentralisation

Competition among sub-national jurisdictions may enhance innovation. Successful provincial and local
government experiments may be replicated elsewhere and the failures discarded. is argument will be
examined in greater detail later (see the discussion on dynamic efficiency in Section 19.3.3).
Furthermore, there may be a high cost associated with decision-making if it is completely centralised.
Owing to the smaller groups involved, the devolution of spending and taxing powers may reduce the cost of
decision-making. Decision-making by smaller groups would most probably result in different levels of
optimal majority at local level than if the same issue were to be decided nationally (see Section 6.6, Figure
6.2). Fiscal decentralisation could also encourage public participation in decision-making since local and
provincial governments may be closer to the communities they serve and may foster scal accountability.

19.2 Reasons for fiscal centralisation

Although there are advantages associated with scal federalism, there are also factors that favour
centralisation.
Firstly, spatial externalities may arise when the bene t or costs of a public service spill over to residents
of another jurisdiction. Goods with external bene ts are likely to be underproduced, since each sub-national
government is concerned primarily with the welfare of its own residents. Similarly, public goods with
signi cant external costs may be overproduced since the residents of a jurisdiction do not bear the full social
cost. Under these circumstances, it may be preferable to have a centralised provision to ‘internalise’ these
costs and bene ts.
Secondly, centralised provision of public services may be justi ed by economies of scale; that is, certain
services (such as transport systems and water) may require areas larger than a single sub-national
jurisdiction for cost-effective provision.
irdly, centralised provision may lead to lower administration and compliance costs in the nancing of
public services. For example, using one computer system for the whole country or one revenue collection
system serving national and provincial governments may prevent the cost of duplication.
In practice, no country has a completely decentralised or completely centralised system. While the
provision of certain goods is preferable at national level, others are best provided at sub-national level. e
crucial question is therefore the following: what is the optimal degree of scal decentralisation? is brings
us to the assignment problem.

19.3Taxing and spending at sub-national level: The assignment

issue
e assignment problem is concerned with how spending and taxation responsibilities should be
distributed among national and sub-national governments when taking considerations of macroeconomic
stabilisation, distribution (equity) and allocative efficiency into account. Fiscal federalism literature provides
broad guidelines on this fundamental issue.

19.3.1 Stabilisation function

ere is general consensus that macroeconomic policy should be assigned to central government. Sub-
national governments cannot and should not conduct monetary policy. If the power to create money were
decentralised to regional entities, there would be strong incentives for sub-national governments to print
money to nance public service provision, rather than raising sub-national taxes or imposing user charges.
is type of behaviour would clearly lead to in ationary pressures, thus adversely affecting the national
economy and compromising national government policies for which the particular sub-national
government bears no responsibility.
e conventional wisdom in the scal federalism literature is that a scal stabilisation policy would be
ineffective at the sub-national level. Provincial and local economies tend to be ‘open’ (in other words, they
‘import’ and ‘export’ from other provinces or local jurisdictions large shares of what they produce or
consume). If a single sub-national government were to pursue an expansionary scal policy, for example,
much of the increase in demand would be lost to outside jurisdictions owing to the openness of such
economies. If, for example, a provincial government were to cut taxes substantially in order to stimulate the
provincial economy, most of the newly generated spending would ow out of the provincial economy in
payment for goods and services produced elsewhere. e ultimate impact on employment levels in the
province would be very small. Fiscal policy by sub-national governments is thus likely to prove impotent
since the extent of import leakages would substantially reduce any multiplier effects. Taxes suitable for
macroeconomic stabilisation, such as personal and corporate income tax, should therefore be centralised
(Musgrave, 1983).

19.3.2 Distribution function

In the scal federalism literature, it is generally argued that only a centralised redistribution policy by central
government is likely to be effective. e argument is that any effort to redistribute income by a single sub-
national government (for example, by increasing taxes on high-income earners and rms, and spending the
proceeds on the poor) would ultimately be self-defeating. ere would be an in ux of poor migrants into the
jurisdiction, attracted by the scal bene ts (transfer payments such as social grants, or increased access or
quality of public services). is would be accompanied by an exodus of high-income earners and businesses
from that jurisdiction. It then becomes more difficult for the jurisdiction to attain its distributional goals,
given the dwindling tax base. Sub-national governments may therefore end up in a worse distributional
position, which may in fact clash with the redistributive objectives of the national government.

19.3.3 Allocation function

Probably the most compelling economic case for scal decentralisation is its potential to secure efficiency
gains. e static arguments linking scal decentralisation with improved efficiency include the following:
• Uniform centralised policy forces every region to consume the same mix of taxes and public spending,
even though tastes and attitudes may vary widely across regions in a large country with many cultural
and ethnic groups. Each decentralised jurisdiction could tailor its service and tax package more closely
to the preferences of its citizenry. For instance, the residents of particular provinces might want
education in particular languages. Education provision, and its associated expenditure, may therefore be
allocated to provincial governments rather than the national government. Politically, decentralisation,
which accommodates diversity, may be necessary to induce various regions to remain part of the
federation (for example, Quebec in Canada).
• Different public goods have different spatial characteristics. Some bene t the entire country (for
example, defence) whereas others bene t only a province (for example, forestry services) or a locality
(for example, street lighting). e defence expenditure responsibility could therefore be assigned to
national government, forestry services to provincial governments and street lighting to municipalities.
Public services are provided most efficiently by a jurisdiction that has control over the minimum
geographic area that would internalise the bene ts and costs of this provision.
• Lower-tier governments may have more information about the needs and priorities of their citizens as
well as region-speci c conditions and prices than national governments, which could improve
programme design and service delivery.
• Diseconomies of scale and increasing bureaucratic inefficiency arise when spending programmes
become too large, that is, when they serve too large a geographical area.

e dynamic efficiency arguments point out that scal decentralisation can stimulate innovation.
Contestability in the public sector arena may have similar bene cial effects to competition in private
markets. Centralisation of functions may mean that national governments may be prone to inertia. With little
experimentation, practices within government may become rigid and perpetuate themselves, even when the
underlying logic for their introduction no longer holds true. Variety in policy design and application at sub-
national level is seen as desirable as it diversi es the country’s exposure to disastrous policy experiments.
Successful policy experiments at sub-national level can be replicated by other tiers of government as best
practice and the failures can be discarded.
 Improved allocative efficiency in a decentralised system depends heavily on the political and
institutional mechanisms through which sub-national governments can be made aware of their electorates’
preferences and are held accountable for their actions. However, in many developing countries, these
democratic structures are not in place or if they are nominally in place, de facto do not function.
Furthermore, sub-national governments may lack capacity and may be prone to corruption. us, while
efficiency gains owing to scal decentralisation are attainable in principle, the way in which scal
decentralisation is implemented determines whether these are in fact realised. In a sense, while scal
decentralisation can attenuate one form of government failure, it may introduce other forms. e above
discussion also shows clearly that while a decentralised system may promote efficiency, it may prove
detrimental to the equity goal (in other words, redistributive goals) and may even compromise stabilisation
objectives.

19.3.4 Tax assignment

Expenditure assignment refers to the allocation of functions and associated spending responsibilities to the
various levels of government. In the same vein, tax assignment refers to the assignment of tax sources to
different tiers of government.
It is intuitive that tax assignment should complement expenditure assignment. In principle, the greater
the spending responsibilities assigned to a particular level of government are, the greater the tax revenue
sources that should be assigned to it. As the difference between expenditure and tax assignments increases,
so sub-national governments become more dependent on grants from national government in order to meet
their spending obligations. As a result, sub-national governments may become more responsive to the
preferences of national government rather than their citizenry, and the ability of the electorate to enforce
scal accountability decreases.
In determining the most appropriate tax assignment, important factors for consideration are equity
(ensuring vertical and horizontal equity among individual taxpayers as well as across regions) and
administrative and allocative efficiency (minimising the cost of collection and compliance as well as
minimising any market distortions). In the light of the issues of equity and efficiency, Musgrave (1983)
proposes the following assignment guidelines:
• Progressive redistributive taxes should be assigned to the national government (for example, personal
and corporate income taxes).
• Taxes appropriate for macroeconomic stabilisation should likewise be centralised (for example, value-
added tax and personal income tax). As its corollary, taxes assigned to the sub-national governments
should be less sensitive to economic and business uctuations, that is, they should be cyclically stable
(for example, motor vehicle taxes).
• Unequal tax bases among jurisdictions should be assigned to the national government (for example,
mining tax).
• Taxes on mobile factors of production should be centralised (for example, corporate income tax or value-
added tax where companies are able to shift the accounting base of the tax to lower-tax jurisdictions).
• Residence-based taxes such as excise taxes should be assigned to the provinces.
• e local authorities should levy taxes on immobile factors of production, for example, property taxes.
• All levels of government may charge user charges and bene t taxes.

In practice, assignment of expenditure responsibilities and taxation and borrowing powers are often the
result of constitution negotiating processes or other political and legislative processes with several
stakeholders with multiple (often con icting) objectives. In the pursuant compromises, the negotiated
consensus may produce expenditure, taxation and borrowing powers which are not congruent. In this case,
intergovernmental grants may be introduced to mitigate the resultant scal imbalances (discuss below in
Section 19.6) or formal or informal intergovernmental scal institutions may evolve to mitigate some of the
resultant tensions.
19.4 Tax competition versus tax harmonisation
When different sub-national governments impose different tax rates, citizens and businesses may react by
moving to jurisdictions with lower tax rates. Tax competition occurs when sub-national governments adjust
(lower) their tax rates to attract mobile factors of production (notably capital) from other jurisdictions. Tax
harmonisation occurs when sub-national governments coordinate their tax policies (for instance, by
limiting the degree of variation in tax rates levied or by de ning the tax bases in a uniform way). Note that
this distinction corresponds to our earlier discussion of the same concepts introduced in Chapter 17, Section
17.5.4.
Tax competition was initially regarded as distortionary, non-neutral and leading to sub-optimal
outcomes, and it was thought that it could be recti ed by means of tax harmonisation. e rationale was that
if one province decides to pursue a competitive tax strategy, the other provinces would respond likewise.
is ‘beggar-thy-neighbour’ downward spiral caused by provinces attempting to undercut each other could
eventually lead to identical but sub-optimally low tax rates on mobile production factors. In addition, the
distribution of mobile factors (particularly capital) would be distorted. Uncoordinated tax policies could
therefore lead to market distortions with regards to mobile factors of production as well as tradable goods
and services.
More recent thinking sees tax competition as a positive in uence and efficiency-enhancing.
Decentralised tax powers could promote innovation, as sub-national governments would be able to
experiment with various scal packages. In the public sector analogue of private market competition and
discipline, policy successes could be emulated elsewhere and failures abandoned. It could also permit sub-
national governments to tailor tax mixes to their citizens’ preferences and encourage accountability. If
governments are providing services that individuals and rms want and are willing to pay for, the adverse
effects of tax competition may be limited. If government overspends and tries to place the tax burden on
those who do not derive any bene t from the service, tax competition may be construed as a positive spur to
increased government responsiveness. One reason for the about-face is the intensifying global competition.
Owing to international mobility of capital, investment that is merely displaced to another region at least
remains within the country instead of migrating across national borders.

19.5 Borrowing powers and debt management at sub-national level

Sub-national governments generally have more limited capacity than central government in issuing debt
obligations. In the interest of coordinated macroeconomic policy and the achievement of overall
macroeconomic objectives, there must be a central government supervision of the debt operations of sub-
national governments. Because the national economy is larger and more diverse, central government can
absorb the shocks that a single region would nd too great to deal with.
Another related aspect of sub-national debt is scal exposure, which refers to the total amount of the
direct liabilities of government (for example, bonds) and the contingent liabilities (for example, government
guarantees). Sub-national borrowing requires an active market in government securities assisted by bond-
rating agencies. Fiscally irresponsible behaviour of sub-national governments is then penalised with
increased interest rates. e discipline of the market may, however, be undermined by contagion effects and
negative pecuniary externalities. Financial contagion refers to a situation where a nancial crisis in one
government or in respect of one particular type of nancial instrument triggers a loss of con dence in
investors, which precipitates similar crises in other similar governments or similar classes of nancial
instruments. In this context, one province’s inability to service debt could cause a loss of investor con dence
in other provinces. Provincial tax bases are narrower and more elastic than national tax bases as a result of
factor mobility. An adverse shock may render a province unable to service its debt, precipitating a nancial
panic. e perception that the nancial panic is contagious might impose a negative externality on the other
provinces. Alternatively, the higher interest rates in response to increased perceived risk will affect all of the
provinces.
Effective market discipline on the borrowing activities of provinces assumes, however, that private agents
have sufficient information to assess provincial risk pro les accurately. For instance, if sub-national
governments do not have their own substantial tax revenue sources, potential lenders and bond-rating
agencies will set their interest rates based on their perceptions of the terms under which revenue sharing or
inter-governmental grants are likely to be made. In some developing countries, payments to sub-national
governments are often suspended when central governments run into nancial problems. In practice,
information asymmetries are rife, creating the conditions for ‘moral hazard’ behaviour.
Moral hazard occurs in a situation where the actions of one party with respect to a contract cannot be
monitored by the other party or parties to the contract. is permits opportunistic behaviour on the part of
the party whose actions are ‘hidden’, to the detriment of the other less informed parties. A classic example of
moral hazard can be found in the insurance industry, where drivers (whose driving style and ability cannot
be observed by the insurance company) start to drive more recklessly once fully insured. Moral hazard
behaviour and sub-national debt are described below.
In the instance of sub-national debt, moral hazard entails that one party (the lending institution) may
behave in scally imprudent or risky ways, such as by extending risky loans to provincial governments of
dubious creditworthiness, knowing that the debt will be either explicitly or implicitly guaranteed by the
other party (the national government). To complicate the problem, provincial governments know that the
national government is explicitly or implicitly underwriting their debt and therefore also have an incentive
to act scally imprudently. ere is thus moral hazard between the lender and the provincial government as
well as between the provincial governments and the national government.
Even if a government explicitly refuses to bail out a sub-national government, this may have very little
credibility with markets. Although it would be best for a government to say in advance that it will not bail out
sub-national governments in such an eventuality, it has every incentive to renege on its undertaking. In
other words, no matter how emphatically national government refuses to aid bankrupt provinces, there will
always be a strong incentive for the government to assist sub-national governments should they nd
themselves in nancial trouble. e long-term costs of impassively standing by while a sub-national
government fails may be so great that national governments invariably act as the government of last resort.
Anticipating this behaviour, markets would react as if the national government had implicitly
underwritten sub-national government lending. e implicit guarantee by central government (which
generally has a better credit rating than sub-national governments) will mean that sub-national
governments will be able to borrow on more favourable terms than would have applied otherwise, thereby
encouraging increased debt.
In addition, under these circumstances, moral hazard behaviour by creditors could exacerbate the
situation. e closer a sub-national government is to nancial crisis and the more liabilities it accumulates,
the more likely it is that the national government will have to step in and bail it out. Banks and nancial
institutions, anticipating this, may lend more rather than less to the sub-national government. is may
result in even more unsustainable sub-national debt, which could have a destabilising effect on the macro
economy.
As political circumstances render it untenable in most cases for a central government to allow a sub-
national government to go bankrupt2, national government may ultimately nd itself liable for any default
on public debt, no matter which tier of government does the borrowing. is would apply even if the actual
letter of the debt contract exempts the national government from any liability. ere is thus a need for
national government regulation in order to allow sub-national borrowing, but under conditions that
minimise national government risk. In an international setting, irresponsible behaviour by or on behalf of a
sub-national government could harm the country’s credibility and credit rating, and thus impose a negative
externality on other spheres of government.

19.6 Inter-governmental grants

Inter-governmental grants are transfer payments from one sphere of government (for example, national
government) to another sphere of government (for example, a provincial or local government). As discussed
in Section 19.6.5, they may be introduced, for example, to correct a vertical scal imbalance where the own
revenue sources of sub-national governments fall short of the expenditure responsibilities assigned to them.
Inter-governmental grants may be unconditional or conditional. Unconditional grants may be spent by
recipient governments as they see t. Conditional grants must be spent on the speci c service stipulated by
the grantor (in other words, the sphere of government that is making the grant).
Grants may also be matching or non-matching. In a matching grant, the grantor government will match
a certain percentage of each rand of spending by the sub-national government on the same activity. A non-
matching grant is just a lump-sum allocation that does not depend on the level of sub-national expenditure.

19.6.1 Unconditional non-matching grants

An unconditional non-matching grant is a lump-sum transfer to sub-national government on which no
constraints are placed as to how it is to be spent. e national government could recommend that the grant
be spent on certain public goods – referred to below as ‘grant-aided public goods’ – but the choice ultimately
lies with the recipient government. e recipient government may spend it on any public good or service, or
provide tax relief to its citizens. is grant acts to increase the income of the recipient government, but does
not alter the relative price of any particular public good. A non-matching grant is in effect an income
supplement. A block grant in the South African context refers to a type of unconditional non-matching
grant where a global lump sum is transferred to a sub-national government to be spent at its discretion. is
is also referred to as revenue sharing.
e effect of introducing a grant of this nature is illustrated in Figure 19.1. We measure spending on the
grant-aided public good in rand, on the horizontal axis. Spending on all other public goods is measured on
the vertical axis. e line AB shows the government’s budget constraint before receiving a grant. e line CD
shows the government’s new budget constraint after receiving the grant. I0 and I1 are the indifference curves
of the median voter, indicating society’s relative preferences.
Point E0 shows the initial equilibrium, which could be thought of as re ecting the median voter’s
preferences (in other words, preferred combination of grant-aided and other public goods). An
unconditional grant (of BD rand) shifts the recipient government’s budget constraint from AB to CD. e
new equilibrium is at E1, signifying a higher level of social welfare. ere is an increase in grant-aided public
good expenditure by the sub-national government (GH). is increase is, however, less than the amount of
the grant (BD). Because unconditional grants may be spent on any public good or to nance tax breaks, they
have the least stimulatory impact on the recipient government’s consumption of the grant-aided public
good.
Figure 19.1 Unconditional non-matching grant

19.6.2 Conditional non-matching grants

Conditional non-matching grants provide recipient governments with a given amount of funds (without
sub-national matching), with the condition that these funds be used for a particular purpose. For example, a
conditional grant might be for spending on healthcare only. As shown in Figure 19.2, the sub-national
government’s budget line will therefore shift outwards by the amount of the grant (AF) from the original
budget line AB to the post-grant budget line AFD.
From the sub-national government’s perspective, OJ (equal to AF) of the grant-aided good is ‘free’.
erefore at the new equilibrium E1, at least OJ of the grant-aided public good will be produced and
consumed. Note that this particular community can still reduce its own spending on the grant-aided good as
long as the full grant (in other words, AF = 0J) is spent as prescribed. At E1, therefore, the extra spending on
the subsidised good (GM) is less than the grant (OJ), as part of the initial spending on the subsidised good by
the sub-national government was diverted to the other goods (LK). Grants of this type are most appropriate
to subsidise activities that are considered low priority by sub-national governments, but considered to be
high-priority activities by national government.

19.6.3 Conditional matching grants (open-ended)

Matching conditional grants, which are cost-sharing arrangements, may be open-ended or closed-ended.
In the case of open-ended matching grants, the national government pays some proportion of the cost of
providing a particular public good or service. e sub-national government provides the rest of the funds
needed. It therefore, in effect, reduces the price of that particular public service (say healthcare) for the
recipient government. Since the grant is open-ended, the sub-national government can use as much of the
grant at the new price as it wishes, as long as it matches the national government’s contribution by the stated
percentage.

Figure 19.2 Conditional non-matching grant

As shown in Figure 19.3, a 33⅓% subsidy on healthcare provision or expenditure (in other words, R2 of
sub-national government funds for each R1 of grant) would rotate the budget line outwards from AB to AD.
(If the slope of the original budget line were 1, then the slope of the budget constraint after the grant would
be atter at ⅔, re ecting the change in the relative price of the two goods. e public good subsidised by the
grant becomes relatively cheaper.) Owing to this cost-sharing arrangement, the sub-national government
can afford 50% more healthcare services at any level of other goods and services. As in the case of a selective
tax on income (see Section 13.4.1 of Chapter 13), a grant that changes the relative price of public goods has
an income and a substitution effect. In this case, the income effect means that the public (as represented by
the sub-national government) is better off, and can thus consume more of both the grant-aided goods and
the other public goods. e substitution effect involves the substitution of the grant-aided good for other
public goods. e net effect determines the position of E1, the new equilibrium. As long as E1 lies to the right
of E0, more of the subsidised public good is purchased. Both the income and substitution effects would
prompt the sub-national government to increase expenditure on the public good.
Figure 19.3 Conditional matching grant (open-ended)

If relative preferences were such that E1 lay to the left of E0, the income effect would dominate the
substitution effect to such an extent that less of the subsidised good will be purchased than before the grant
(in other words, the subsidised good or service is an inferior or Giffen good or service).
In general, open-ended matching grants are regarded as most appropriate for correcting inefficiencies in
public good production that result from positive externalities (Shah, 1995). Positive externalities, or bene t
spillovers, occur when the provision of goods and services by one sub-national government bene ts other
sub-national governments, which do not, however, bear the cost of provision. In this case, there would be an
incentive for the sub-national government to underprovide that public good or service unless it was
subsidised. Note that open-ended matching grants may bene t richer sub-national governments more than
poorer ones, who might not be able to match national government expenditure. It can be shown
geometrically that if E1 were to lie directly above E0, the cost to the sub-national government of the new
bundle of public goods would be the same as the pre-grant combination. e response of a poor community
to a conditional grant may well be to seek a combination of goods that does not increase or even decrease
the total cost in respect of all public goods; that is, E1 will be directly above or even to the left of E0.

19.6.4 Conditional matching grants (closed-ended)

ere are also closed-ended matching grants, where the national government pays some proportion of the
cost of providing a particular public good or service up to a certain limit. e effect of a closed-ended
matching grant is illustrated in Figure 19.4.
Figure 19.4 Conditional matching grant (closed-ended)

When there is a 33⅓% subsidy on, for instance, healthcare up to a limit, the budget line will move from AB to
ACD. Costs of healthcare provision will be shared along AC until the subsidy limit (at spending level OJ) is
reached. Beyond the subsidy limit, healthcare is unsubsidised and the sub-national government faces the
full price of provision; hence the steeper slope of the section CD of the new budget line (the slope of CD is
the same as that of AB). At the new equilibrium, E1, more healthcare will be provided than would have been
the case without the grant.
Grantor governments generally prefer closed-ended matching transfers as these allow them to retain
control over their budgets.

19.6.5 The rationale for inter-governmental grants

e main arguments for inter-governmental transfers are summarised below. e design of the grant should
be appropriate to the objective it seeks to attain.
• Fiscal imbalances between expenditure needs and revenue generation capacities of sub-national
government can be addressed. Under circumstances where it is not feasible to devolve increased tax
powers to sub-national government, unconditional non-matching grants (in other words, block grants)
should be considered. Sometimes revenue is collected at national level, and then transferred to sub-
national governments as block grants to address scal imbalances. is is known as revenue sharing.
Where national government devolves an additional expenditure responsibility to sub-national
governments without the assignment of sufficient additional own revenue sources or a concomitant
increase in inter-governmental grants, this may lead to an unfunded mandate for sub-national
governments. Unfunded mandates undermine the effective and equitable delivery of services by sub-
national governments.
 Conditional non-matching grants are appropriate to ensure minimum standards in the provision of
• public goods and services across the nation.
• Conditional matching transfers (open-ended) are suitable to compensate for bene t spillovers. e rate
of subsidisation should re ect the degree of bene t spillover.

A conditional matching grant (closed-ended) may be considered to assist sub-national governments

nancially while promoting expenditures on an activity considered by the national government to be of a
high priority, but at the same time affording the national government better control over its own budget (see
Box 19.2).

Box 19.2 Conditional grants in 2019/20

Conditional grants are made from national government departments to provincial governments and to
municipalities. These grants are appropriated (that is, budgeted for) in the annual Division of Revenue Act
(DoRA) passed by Parliament at the same time as the Budget. To a much lesser extent, provincial governments
may also occasionally give conditional grants to municipalities. Conditional grants are earmarked allocations
that are ring-fenced for a specific purpose by national government, and provinces and municipalities can only
use these grants for the purpose for which they were appropriated. But, as we have seen in Section 19.6.2,
nothing prevents a sub-national government from spending less of its own funds on the grant-aided good or
service than before the grant was announced.
The amount budgeted for each conditional grant as well as indicative allocations for the next two fiscal years
are listed in the DoRA. Each conditional grant has its own grant framework, which spells out in detail the
conditions attached to that particular grant, the service delivery outputs or outcomes expected from that grant,
the criteria used to divide each grant among provinces or municipalities, a summary of the audited actual
spending on that grant in the previous year, a grant payment schedule, and how and by whom the grant’s
performance will be monitored. Grant recipients report quarterly on their spending of the grant and on their
delivery performance. Should provincial departments or municipalities receiving a grant not adhere to its
conditions, the DoRA empowers the National Treasury to stop or withhold payments and to relocate grant
payments to other recipients.
In 2019/20, there were 26 provincial conditional grants, collectively amounting to R106,7 billion. Examples
of provincial conditional grants include the following:
• A Comprehensive Agricultural Support Grant for emerging farmers from the national Department of
Agriculture, Forestry and Fishing to the nine provincial Departments of Agriculture
• The National School Nutrition Programme Grant from the national Department of Basic Education to
provincial Departments of Education
• A Comprehensive HIV/AIDS and TB Grant, a health facility revitalisation grant and a National Tertiary Services
grant for specialised health services from the national Department of Health to provincial counterparts
• The Human Settlements Development Grant for housing and related infrastructure from the national
Department of Human Settlements to the provincial Departments of Housing.

Most of these grants were direct conditional grants disbursed directly to, and to be spent by, recipient provincial
governments. There were, however, indirect grants, in other words, grants that are spent by a national
government department on behalf of a provincial government that does not have the capacity to spend the
grant effectively. One example is the school infrastructure backlogs grant, where the national Department of
Basic Education builds schools in provinces lacking the requisite capacity.
As can be seen in Table 19.3, R44,8 billion of direct conditional grants was budgeted to be paid directly to
municipalities in 2018/19, of which R15,3 billion was for municipal infrastructure, R11,3 billion was for urban
settlements development and R6,3 billion was for public transport. In addition, R7,9 billion was budgeted for as
indirect conditional grants in 2018/19, mainly for national integrated electrification programme (R3,3 billion),
but also regional bulk water and sanitation infrastructure (R2,9 billion). In the case of indirect grants, which are
grants-in-kind, national government spends the allocation on behalf of municipalities that lack the capacity to
roll out infrastructure themselves.
Examples of local government conditional grants include the following:
 • The Municipality Infrastructure Grant is paid by the Department of Cooperative Government and Traditional
Affairs to individual municipalities for the provision of basic services such as water and sanitation, roads and
social infrastructure to poor households in non-metropolitan municipalities.
• The Public Transport Network Grant, administered by the national Department of Transport, funds the public
infrastructure and operations of integrated public transport networks in 13 cities.
• The National Electrification Programme funds municipalities and Eskom to sustain progress in connecting
poor households to electricity.
• The Municipal Systems Improvement Grant assists municipalities in building capacity for management,
planning and technical skills.

19.7 Inter-governmental issues in South Africa

e section below uses the scal federalism theoretical framework to analyse the structure and performance
of the South African inter-governmental system. It covers the constitutional assignment of revenue and
expenditure responsibilities to provincial and local government in South Africa as well as revenue-sharing
processes, formulae and institutions. e section concludes by focusing on speci c provincial and
municipal nancing challenges.

19.7.1 Constitutional issues

Constitutional issues rst focus on expenditure assignment, and then on revenue and borrowing matters.

19.7.1.1 Expenditure assignment

e South African constitution establishes a state with three spheres of government: national, provincial and
local. It assigns to each of these three spheres of government certain powers or functions. ese
competencies may be concurrent (shared responsibility of national and provincial governments) or
exclusive (sole responsibility and discretion of the province or sole responsibility of national government).
Functional areas of concurrent legislative competencies are listed in Schedule 4 of the Constitution and
exclusive provincial responsibilities are detailed in Schedule 5.
Schedule 5 of the Constitution speci es that certain expenditure responsibilities be devolved completely
to the provincial sphere (for example, provincial roads and abattoirs). Others (as described in Schedule 4)
are administered jointly as concurrent competencies (for example, primary and secondary education and
health). Some functions remain at national level (for example, foreign affairs and defence).
Provincial legislation in respect of these Schedule 5 functions takes precedence over national legislation,
except when national legislation is necessary to establish national norms and standards, to maintain
economic unity, to protect the common market in respect of the mobility of goods, services, capital and
labour or to promote economic activities across provincial borders. Provinces therefore have a limited
degree of scal (and political) autonomy, although this is weighed against the national interest.
Local government competencies are detailed in Part B of Schedule 4 and Schedule 5. Examples of
concurrent functions of local government include water, sanitation, air pollution, electricity and gas
reticulation, municipal health services and so on. Exclusive local government competencies include
beaches and amusement facilities, cleansing, dog licensing, local amenities, sport facilities and municipal
roads.

19.7.1.2 Revenue assignment and borrowing powers

e Constitution permits a province to impose taxes, levies and duties other than income tax, value-added
tax, general sales tax and rates on property or customs duties. Most productive taxes are reserved for
national government.
 A province may also levy at-rate surcharges on the tax bases of any tax (including individual income
tax), levy or duty imposed by national legislation except corporate income tax, value-added tax, rates on
property and customs duties.
Provinces may levy these taxes provided that they do not prejudice national economic policies,
economic activities across provincial boundaries, and national goods and services or factor mobility.
Additional own revenue3 raised by provinces or municipalities may not be deducted from their share of
revenue raised nationally or from other allocations made out of national government revenue. is is to
provide an incentive for provinces to increase their tax effort.
To supplement own revenues and scal transfers from national government through the revenue-
sharing formula (see Section 19.7.2), provinces are also empowered to raise loans. e Borrowing Powers of
Provincial Governments Act of 1996 and the Public Finance Management Act of 1999 set out the conditions
under which provinces may borrow. Loans may only be raised to nance capital expenditure (for example,
bridges and other infrastructure) and not for current expenditure (for example, wages). e only exception
to the ban on borrowing in order to nance current expenditure is for bridging nance, in which case the
loans must be redeemed within the same scal year. No national government guarantee is available in
respect of provincial borrowing.
Local governments are entitled by the Constitution to impose rates on property and surcharges on fees
for services provided by or on behalf of the municipality (for example, for electricity or sewerage). In the
past, municipalities were able to raise Regional Services Council (RSC) levies as a source of revenue. ese
were, however, abolished since they were unconstitutional. Until a new alternative revenue source is made
available to municipalities, they will receive a temporary grant to replace lost RSC levy income. With the
proposed restructuring of the electricity distribution industry, municipalities could also lose their electricity
user charge income, which could further compromise local scal capacity. Municipalities are also allowed to
borrow, subject to the same restrictions as those described above for the provinces. e Municipal Finance
Management Act (2003) describes the framework for municipal borrowing.

Inter-governmental transfers and the Financial and Fiscal Commission

19.7.2
(FFC)
In South Africa, most taxes are raised at national level because collection is easier to administer and it avoids
the administrative duplication and higher compliance costs for taxpayers associated with a more
decentralised system. However, the Constitution assigns provinces with certain responsibilities regarding
the delivery of goods and services, either individually or jointly with national government. ere is an
imbalance between the expenditure mandate of sub-national levels of government and the nancial
resources that they can raise on their own account. is mismatch is referred to as vertical scal imbalance
and arises as a result of the limited capacity of the provinces to raise revenue for themselves independently
of national government. For most provinces, income raised within the province as ‘own revenue’ (mainly
from car licences and hospital fees) amounts to less than 5% of the provincial budget (see FFC, 1996), a
percentage that has subsequently declined even further in some provinces. In terms of tax assignment
criteria and the constitutional mandate of provinces, the scope to expand the provincial revenue base is thus
limited.
e Constitution states that provinces are entitled to an equitable share of the revenue collected
nationally, in line with their new expenditure responsibilities and functions. e process by which
government incomes are pooled and subsequently divided among national and sub-national governments
is referred to as revenue sharing.
e Financial and Fiscal Commission (FFC) is an independent body established in terms of the
Constitution to make impartial recommendations to the national parliament and the nine provincial
legislatures on nancial and scal matters such as the following: equitable allocations to the three tiers of
government from the national revenue pool, intentions of provincial governments to levy taxes and
surcharges, raising of loans by lower tier governments and the criteria to be used for these purposes. We now
focus on the process of making equitable allocations to national, provincial and local government from
nationally collected revenues.
19.7.2.1 The vertical division of revenue
e allocation of funds from nationally collected revenue entails a vertical division of such revenue between
the national, provincial and local spheres of government. e vertical division of revenue for the scal years
2015/16–2021/22 is shown in Table 19.1. First a ‘top slice’, which consists mainly of funds to service the debt
and provision for a contingency reserve, is subtracted from the nationally collected revenue pool. e
remainder is then split vertically among the national, provincial and local spheres. Note that the national
share includes the budgets for national departments, but excludes conditional grants to provinces and local
government. e contingency reserve is an unallocated pool of funds set aside for new priorities in future
years or to deal with unforeseeable events that are difficult to budget for precisely (such as natural disasters,
or foreign and domestic economic shocks).
From Table 19.1, it is evident that in 2018/19, after the ‘top slice’ (for state debt service costs and the
contingency reserve) was subtracted, R1 327,6 billion remained for division among the three spheres.
National government’s share of this amounted to 48,1% in 2019/20, and was expected to decrease marginally
to 47,8% over the next two years. In 2019/20, provincial governments received 43,0% of total revenue after
the top slice was subtracted (consisting of their ‘equitable share’, which is an unconditional grant, as well as
various conditional grants from national government). It was envisaged that the provincial share would
increase marginally to 43,1% between 2019/20 and 2021/22. Local government received only 8,9% after the
top slice was subtracted in 2019/20. e relatively small local government share has raised concerns about
adequacy in the light of provision of free basic services, service backlogs in poorer municipalities, migration
from neighbouring states and the developmental mandate of municipalities.
Table 19.1 Vertical division of revenue in billions of rands in South Africa, 2015/16–2021/22

Source: National Treasury. 2019a. Budget Review 2019.

While the FFC formulae and recommendations are regarded as important inputs in the calculation of
these vertical divisions, the Commission’s recommendations are apparently not decisive. e government
regards the vertical division of funds between the different spheres of government as a political policy
judgement that re ects the relative priority of functions assigned to each sphere of government and not
something that can be captured in a formula (Ministry of Finance, 1999: 59).

19.7.2.2 The horizontal division of revenue of the provincial pool

e vertical division of revenue is followed by a horizontal division of the provincial and local pools of
resources available among the nine provinces and the 278 municipalities respectively. To ensure that each
provincial government receives an equitable allocation in order to meet its constitutional expenditure
obligations, the FFC (1996) proposed a revenue-sharing formula over a three- to ve-year period for the
horizontal division of resources among the nine provinces. e FFC asserted that formula funding is more
objective and less prone to manipulation by politicians and civil servants. In addition, it would enable
provinces to predict the revenues that would accrue to them over the period in which the formula was in
force with greater certainty. e original FFC formula was mainly population-driven, with the population in
a province being an indicator of the scal need of the province, given explicit minimum standards of service
such as learner–educator ratios or average number of primary health care clinic visits. It was also weighted
in favour of rural people as a proxy for backlogs and for poverty.4 e formula aimed to equalise rural
weighted spending per capita across the provinces. As such, it addressed the horizontal scal imbalances
among provinces that arise from differences in revenue (tax) capacity in relation to their expenditure
responsibilities.
However, the FFC is only an advisory body and its recommendations are not binding on the Budget
Council, which actually does the division of revenues. e Budget Council consists of the Minister of
Finance and the nine provincial MECs for Finance, with the FFC as observers. e actual formula used by
the Budget Council to determine each province’s equitable share is based on the provinces’ demographic
and economic pro les (which are an indication of the relative need for public services across provinces),
given implicit service standards. e formula used in the 2019/20 division of revenue consisted of the
following:
• An education share (48%), based on the size of the school-age population (ages ve to seventeen) and the
number of learners (Grade R–12) enrolled in public ordinary schools
• A health share (27%) based on a combination of the health risk pro le of each province and its health
system case load (for example, women of child-bearing age and older persons consume more health
services than the population average; therefore a province with a greater share of women of child-
bearing age and of older persons would have a greater health risk pro le, which is taken into account in
the formula).
• A basic component (16%) derived from each province’s share of the national population
• An institutional grant (5%) divided equally among the provinces, which recognizes that some costs of
running a provincial government are not related to the size of a province’s population or factors included
in other formula components.
• A poverty component (3%) based on the province’s share of poor households (in other words, people
falling in the lowest 40% of household incomes in the 2010/11 Income and Expenditure Survey [StatsSA
2012]), reinforcing redistribution in the formula.
• An economic output share (1%) based on the Gross Domestic Product by Region (GDP-R) data (GDP-R,
published by Statistics South Africa, measures the gross domestic product produced in each province).

e formula determines the equitable share that is given to the provinces as an unconditional block grant. In
order to determine the total allocation for each province, the conditional grants that each province receives
from national government must be added to the equitable share. Table 19.2 shows the budgeted allocations
to the provinces for 2019/20.
Table 19.2 Budgeted provincial allocations in billions of rands for 2019/20

Source: National Treasury. 2019a. Budget Review 2019.

Because the provincial formula is largely population driven, provinces with large populations such as
Gauteng and KwaZulu Natal tend to receive larger allocations than provinces with smaller populations such
as the Northern Cape.
Since 1996, the formula has been revised several times at ve-year intervals to respond to new census
data and new policy objectives. New census results typically reveal that as a result of migration between
provinces, the populations of certain provinces have increased and others have declined, with an equivalent
impact on the demand for public services. As the demographic data in the formula are updated to re ect
these changes, the share received by each province is adjusted upwards or downwards accordingly. e new
shares are not effective immediately, but are phased in over three years to enable provinces receiving lower
shares of nationally collected revenue to scale back their operations and provinces that are the recipients of
additional resources to put in place the capacity to spend their increased allocations effectively.

19.7.2.3 The horizontal division of revenue of the local government pool

Table 19.1 shows that in 2018/19, local government received R117,3 billion or 8,8% of nationally collected
revenues available for sharing, after top-slicing debt servicing costs and the contingency reserve. is
amount comprises both unconditional transfers such as the local government equitable share and
conditional grants to municipalities, as illustrated in Table 19.3 below.
In 2018/19, municipalities received 75,2 billion through the local government equitable share and other
related unconditional grants. e eight metropolitan municipalities (in large cities) also received an
additional R12,5 billion in unconditional transfers in 2018/19 as a share of the General Fuel Levy.
A further R44,8 billion was paid to municipalities in 2018/19 as direct conditional grants to be spent
mainly on municipal infrastructure. Finally, the national government spent R7,9 billion as indirect
conditional grants (in other words, grants in kind) on behalf of low capacity municipalities that could not
spend effectively themselves.
Table 19.3 Transfers to local government, 2018/19–2021/22

Source: National Treasury. 2019a. Budget Review 2019.

As will be illustrated later, local governments are (in aggregate) not as reliant as provinces on transfers
from other spheres of government. ere is, however, substantial variation among municipalities. Some
poorer municipalities rely on grants for up to 92% of their income (for example, the Bohlabela municipality),
while some urban municipalities raise up to 97% of their own income (National Treasury, 2005a: 30).
Transfers to the local sphere include unconditional equitable share grants, direct conditional grants and
indirect conditional grants in kind (for example, the Water Services Infrastructure grant). Conditional grants
are generally for municipal infrastructure and other capital expenditure, capacity building or in support of
restructuring.
 e individual municipality’s claim to nationally collected revenue depends not only on the total size of
the vertical division, but also on the nature of the horizontal division among municipalities. e FFC
formula for the division of the local government resource pool, which is the basis of the method used by the
Department of Cooperative Governance and Traditional Affairs and the National Treasury to distribute the
equitable share, bases the claim for a share on the relative needs of jurisdictions, after taking the tax capacity
into account. e purpose of the formula is to provide nancial assistance to those municipalities that
cannot provide basic services to the poor from their own tax base.
Essentially, the latest version of the local government equitable share formula consists of ve
components:
• A basic services component for water, refuse removal, sanitation, environmental healthcare services and
electricity reticulation to poor households earning less than R2 300 per month; for each of the subsidised
services, there are two levels of support: a full subsidy for those households that actually receive services
and a partial subsidy for unserviced households set at one-third of the subsidy to serviced households
• An institutional support component to support the basic administrative and governance capacity in local
governments; it consists of a base allocation that will go to every municipality regardless of size, and a
variable allocation depending on the population in the municipality and number of councillors
• A community services component that funds services that bene t communities as a whole rather than
individual households (which are provided for in the basic services component) such as municipal
health services, re services, municipal roads, cemeteries, planning, storm water management, street
lighting and parks
• A revenue-raising capacity correction component to take into account the estimated revenue capacity of
each municipality; it is applied to the institutional and community services components of the formula
to ensure that these funds are targeted at poorer municipalities that are least likely to be able to fund
these functions from their own revenues
• A correction and stabilising factor to ensure that municipalities are given at least 90% of what they had
been promised in the previous MTEF round of allocations in the current formula, and to ensure that
allocations would not be negative owing to the revenue raising correction (National Treasury, 2018a).

Like the provincial equitable share formula, the structure of the local government equitable share formula
itself is reviewed every ve years to factor in new policy priorities. In addition, the demographic information
and other data used by the formula are updated regularly, for instance, when a new census is conducted that
shows that as a result of urbanisation, populations of rural and small towns may have declined and that
populations of larger cities have increased. ese changes mean that some municipalities may experience
big drops in their equitable share allocations and other municipalities may receive substantial gains. ese
allocation changes are phased in gradually over ve years in order to smooth out the impact of the changes
and to allow municipalities’ spending patterns to adjust to their changed allocation.
e introduction of free basic services at municipal level in 2003/04 placed increasing pressure on the
local governments’ equitable share as well as on their capital budgets. It became necessary for them to roll
out services to previously underserved communities characterised by high levels of poverty and
unemployment, and with little ability to pay cost-re ective municipal tariffs for basic services. While
backlogs in the provision of water, sanitation and other municipal services persist, requiring new
infrastructure, municipalities also battle to fund the maintenance and upgrade of existing municipal
infrastructure. It is a struggle to nd people with the necessary technical engineering skills to do so. ese
pressures have been exacerbated by cuts to, or slow growth in, key conditional grants to local government in
2019/20 due to scal consolidation and de cit reduction imperatives implemented by National Treasury. As
re ected in Table 19.3, for instance, the Municipal Infrastructure Grant declined to R14,8 billion in 2019/20
from R15,3 billion the previous year; though the Urban Settlements Development grant increased to R12,0
billion in 2019/20 from R11,3 billion the previous year; and the National Integrated Electricity Grant was
reduced to R1,86 billion in 2019/20 from R1,9 billion the previous year. is decline in conditional grant
funding may further undermine municipal nancial sustainability and basic service backlog reduction, as
the number of indigent consumers qualifying for free basic services has increased in the aftermath of the
Great Recession of 2008/09. Given the need to conserve electricity and water, and to nance signi cant
infrastructure maintenance backlogs, the prices of bulk water and electricity have increased markedly,
putting upward pressure on municipal rates and on the tariffs that municipalities charge for these services.

19.7.3 Provincial financing issues

Until 1996, provinces were only spending agencies for the national government, disbursing funds according
to the policies and priorities determined at national level. In the past, provinces were treated in much the
same way as national departments in the budget process, rather than distinct spheres of government. ey
were mainly concerned with the implementation of national policy. ey could not set their own priorities,
nor did they have much accountability. Problems could always be blamed on national government.
Estimating and evaluating expenditures across functions and provinces were done by government officials
in Pretoria, with little regard for provincial priorities. Under the provisions of the Constitution of 1996, the
provincial governments have much more latitude to determine their own spending patterns, given their own
revenue resources and the equitable share of nationally collected revenue which they receive as
unconditional transfers.
Provincial budgets should therefore increasingly embody the provincial governments’ responses to
regional challenges and opportunities for development within the nine provinces. Under these
circumstances, coordination between the spheres of government in setting expenditure priorities becomes
crucial, so that differing needs can be provided for without jeopardising national goals.
However, the nine provinces have vastly differing capacities for nancial management and expenditure
control. Accountability and efficiency thus depend on the strengthening of managerial and administrative
capacity in all provinces, especially the weaker ones. Given South Africa’s history of government
overspending, sound nancial management is particularly important so that provinces can perform the
spending functions devolved to them without the risk of provincial overspending. Provincial overspending
would, of course, jeopardise national de cit targets and other stabilisation objectives.
It is also important that the share of national revenue received by the province be adequate to ful l
provincial spending obligations and that unfunded mandates are avoided. If nancing is not commensurate
with the new distribution of responsibilities across spheres of government, this could also lead to persistent
pressures that encourage provincial overspending.
Probably one of the most pressing challenges facing provincial governments is the need to diversify their
tax bases and reduce dependence on national government. Provinces in South Africa have very little own
revenue capacity compared to sub-national governments in other countries and thus are scally highly
dependent on central government. Provincial own revenues currently represent less than 5% of provincial
expenditure on average. ese revenues are mainly derived from motor vehicle licences, hospital fees and
gambling proceeds.
ere are concerns that the overreliance on funding from central government could undermine
provincial scal autonomy. Furthermore, there is a weak link between the revenue-raising responsibility,
which is mainly at the national level, and the responsibility for spending decisions, which is provincial. is
could dilute scal accountability in the sense that provincial executives are not called upon to justify
expenditure patterns to provincial electorates as taxpayers. is could also induce perverse incentives such
as inefficient increases in expenditure, since costs are de facto being shifted onto national government, as
well as deviation from provincial electorate preferences. Fiscal federalism’s supposed bene t of allocative
efficiency is thus weaker. Tax legislation has recently been passed that should allow provinces limited leeway
to extend their tax bases.
e FFC has proposed that provinces be allowed to ‘piggy-back’ a provincial surcharge on the national
personal income tax, in line with constitutional provisions. For a number of technical reasons, it is unlikely
that provincial governments will be able to implement a surcharge on the personal income tax in the near
future, although they may elect to levy smaller taxes in terms of the Provincial Taxation Regulation Process
Act of 2001. If provinces were empowered to tax, it would probably only bene t the better-off provinces,
which have viable tax bases (for example, Gauteng, the Western Cape and possibly KwaZulu-Natal). In 2005,
the Western Cape Provincial Government proposed a provincial fuel levy surcharge of between 10 and 50
cents as a new own revenue source, which was approved by the then Minister of Finance, Trevor Manual,
but was never implemented.
e other major challenge for provincial governments is to ensure allocative efficiency in employing
public funds in order to achieve desired basic education, health and other developmental outcomes. While
national government sets overall policy as well as national norms and standards for concurrent functions
such as health and education, actual delivery of services by schools and hospitals is implemented and
budgeted for by provincial governments. While South Africa spends much more per capita than other
African and developing countries, our basic education and health outcomes are much poorer, and continue
to be characterised by stark inequalities (Ajam & Aron, 2007).

19.7.4 Urban economics

Cities around the world, with their high concentrations of economic activities and dense populations,
provide signi cant opportunities for stimulating economic growth and development, but also introduce
considerable challenges in relation to pollution, traffic congestion, public transport and crime as well as the
efficient and equitable provision and nancing of services such as water, sanitation and electricity. South
African cities also have to contend with the racially skewed spatial distribution of economic activity and land
use as a result of apartheid segregation, where high concentrations of poverty and backlogs in housing and
infrastructure co-exist with wealth and rst-world living standards.
Urban economics is a eld of study that uses the analytical tools of economics to explain the spatial and
economic organisation of cities and metropolitan areas to analyse their special economic problems and
explore solutions. Urban economics considers land use, human settlement patterns and densities, transport
costs and geographic location factors in production and consumption decisions. Unlike other economic
disciplines, which virtually ignore geography, urban economics focuses on these spatial relationships to
explain the economic motivations underlying why and how cities are formed, function and develop over
time. Policies designed to change the distribution of populations and economic activity within cities,
between cities, and between urban and rural areas also fall within the domain of urban economics
(Heilbrun & McGuire, 1987; Henderson, 1988).
A vast array of context-speci c geographic, historical, political, cultural and sociological factors has given
rise to the growth of cities. Economic analysis of why cities exist goes back to Alfred Marshall’s Principles of
Economics in 1890 and has largely centred on the concept of agglomeration economies. Agglomeration
economies are economies of scale external to the rm, which could lead spatially concentrated economies
to have higher productivity and growth rates.
Internal economies of scale (also called increasing returns to scale) relate to a production technology
within a rm where the cost of producing an additional unit of output (that is, the marginal cost of a
product) reduces as the volume of output (in other words, the scale of production) increases. External
economies of scale (or agglomeration economies, as they are also called) refer to cost reductions that rms
realise when locating near each other, which increases productivity of individual rms and/or clusters of
rms in general. Unlike internal economies of scale, agglomeration economies result from the collective
presence and actions of a group of other rms in the vicinity and falls outside the control of any individual
rm. Agglomeration economies refer to cost savings to the individual rm that depend on the existence of
the scale of the industry to which the rm belongs or other complementary industries (Parr, 2002).
Agglomeration economies may arise as a result of a number of causal factors such as the following
(Ciccone & Hall, 1996; Glaeser & Gottlieb, 2009):
• Transport cost savings: e more concentrated labour, capital and other factors of production are (in
other words, the greater their spatial density), the lower the transport costs incurred. If, for example,
production of the various intermediate components that comprise a manufactured good exhibits
constant returns to scale, but transport increases with distance, then the production of all goods within
the geographical area would have increasing returns to scale. In other words, the ratio of outputs to
inputs will increase with greater density. Customers located near production facilities would also reduce
costs.
• Labour market efficiencies: Concentrations of labour may make it easier for workers to match up with a
rm’s skill requirements and vice versa. Concentration (or pooling) of labour markets may reduce
employment risk for workers and allow them to move from less productive to more productive rms
(which may be in a position to pay higher wages). In event of retrenchment, their likelihood of nding
another job in industrial concentration is greater. Furthermore, the depth of the labour market in terms
of specialist skills is likely to be greater in spatially dense urban areas.
• Sharing of inputs and infrastructure: Agglomeration economies may result from sharing, for example,
rental and specialised equipment as well as electricity generation and reticulation infrastructure.
• Information sharing, learning and innovation: Physical proximity may stimulate low-cost sharing of
information, the spread of ideas and hence increased productivity through innovation. e greater
competition borne of physical proximity may also encourage innovation. On the other hand, the age of
information and communication technology (ITC) has reduced the importance of physical proximity in
cost saving in many sectors.
• Greater consumption possibilities: Cities may offer an array of services not available elsewhere, for
example, opera houses or universities. eir density may make certain public services more sustainable,
for example, light rail public transport, large sport stadia and specialised schools.

A number of studies have attempted to establish the existence of agglomeration economies in different
countries and time periods, and to quantify them using different speci cations of agglomeration economies.
While there has been widespread empirical support for the existence of agglomeration economies, precise
measurement of this phenomenon and isolation of its underlying causes proved to be far more elusive
(Glaeser & Gottlieb, 2009; Melo, Graham & Noland, 2009). is lack of understanding of the extent, nature
and causes of agglomeration economies is problematic from a public policy perspective, since it is not clear
whether existing fast growing and productive urban concentrations could, or indeed should, be replicated.
Many urban and regional development strategies based on clustering economic activity are based on the
potential to achieve agglomeration economies (Parr, 2002).
With increase in urban density and hence economics of agglomeration come increasing diseconomies of
agglomeration, including congestion, overcrowding and pollution. Using a cross-country data set of 105
countries between 1960 and 2000, Brülhart and Sbergami (2009) found evidence that agglomeration
promotes GDP growth at the initial phases of an economy’s development (when transport and
communication infrastructure and skills are scarce, and the access to capital markets is limited). Once
countries attain a certain level of GDP per capita, however, agglomeration may have no impact on growth or
even a negative impact on growth.
A policy implication of this is that there may be a trade-off between national growth and inter-regional
equity. Realising increased agglomeration economies through greater urbanisation might stimulate
aggregate growth of the national economy. It could also, however, exacerbate inter-regional or urban rural
inequality (Brülhart & Sbergami, 2009). is tension has also surfaced in South Africa, where the increased
focus on cities as dynamic centres of growth and employment may raise concerns that rural areas are being
neglected (see Box 19.3). e challenge is to ensure that linkages between urban and rural areas are
mutually bene cial and growth-enhancing.

Box 19.3 South African urban development opportunities and challenges

In African and Asian nations, there has been a trend towards ever larger proportions of the population being
resident in urban areas, ranging from small towns of 15 000–20 000 people to large cities with millions of
inhabitants. It is estimated that 60% of their populations will be urbanised by 2050, creating huge
opportunities for economic growth and combatting poverty, but also many developmental challenges (growth of
informal settlements, access to basic services such as water, sanitation and electricity, congestion, pollution,
increased carbon footprint and so on [Department of Cooperative Government and Traditional Affairs, 2013]).
In South Africa, the percentage of the population living in urban areas was 63% in 2013. This is expected to
increase to 80% in 2050 (Department of Cooperative Government and Traditional Affairs, 2013). Cities and
large towns are the economic powerhouses of South Africa, comprising 80% of the country’s gross value added
 (GVA). It is therefore not surprising that they attract rural migrants in search of job opportunities and access to
better public services. The built footprint of South African cities is estimated to grow at between 3 and 5% per
annum. Not only has South Africa to deal with the challenges of urbanisation, it also has to contend with
persisting spatial, economic and social segregation on racial lines despite the demise of apartheid two
decades ago. South African cities tend to be sprawling and have fragmented spatial forms with some of the
lowest densities in the world, which makes provision of certain public goods (such as public transport and other
urban infrastructure networks) less financially viable (Department of Cooperative Government and Traditional
Affairs, 2013).

19.7.5 Local government financing issues

Local government or municipal nance (or urban public nance, as it is sometimes called) is an important
component of urban economics that considers how cities raise revenues (through user charges, property
rates and inter-governmental grants) in order to deliver services and promote economic growth and
development.
Unlike the provinces, local government in South Africa has a more substantial tax base in aggregate and
the 278 municipalities that make up the local government sphere generate more than 90% of their aggregate
budget as own revenues. Government transfers are therefore a much smaller percentage of local
government revenue than at the provincial level (see Box 20.4).

Box 19.4 Local government finances in 2018/19

In the 2018/19 local government financial year, aggregate municipal operating expenditure amounted to
R336,3 billion in total. In addition, municipalities collectively spent R55,4 billion on their capital budgets for
that year. The combined operating and capital spend of all municipalities amounted to a total of R391,7 billion
in 2018/19.
The eight metropolitan municipalities5 alone made up 57,9% of the combined operating and capital budget
of all municipalities in 2017/18 (R235,5 billion) which reflects the concentration of economic activity and
revenue bases in urban areas (National Treasury, 2019c).
Local government capital expenditures on the construction and rehabilitation of municipal infrastructure of
R55,4 billion in 2018/19 were financed through the following means:
• Transfers and subsidies (including conditional grants from national government) of R33,3 billion (60,1%)
• External loans of R8,0 billion (14,4%)
• Historical operating surpluses and other internal contributions of R13,1 billion (23,6%)
• Public contributions and donations, totalling R1,0 billion (1,8%).

This indicates that the local government sphere as a whole is very dependent on national government for
infrastructure funding on capital budgets. This is particularly true for rural and poor municipalities that have
limited revenue bases.
Operating revenue for all municipalities amounted in aggregate to R342,5 billion in 2017/18 and was
derived from the following sources:
• Service charges for water, electricity and sanitation of R172,1 billion (49,3%)
• Operational transfers R78,2 billion (22,4%)
• Property rates of 67,4 billion (19,3%)
• Revenue from investments R4,7 billion (1,3%)
• Other own revenues of R27,0 billion (National Treasury, 2019c).

The aggregate municipal operating expenditure of R336,3 billion in 21018/19 consisted of the following:
• Personnel expenditure (30,1%)
• Bulk purchases such as electricity bought from Eskom and retailed to residents, and bulk purchases in
connection with water and sewage service provision (33,2%)
• Finance charges such as interest and redemption on loans (2,5%)
 • Depreciation and asset impairment (7,3%)
• Remuneration of councillors (1,2%)
• Other expenditure such as general overheads and administration of the council (24,7%).

In the Western Cape and Gauteng, municipalities in aggregate are less reliant on inter-governmental grants and
have greater access to own revenue sources. In these provinces, municipal own revenues amounted to 83,6%
and 85,9% of total operating revenue respectively in 2017/18. Direct infrastructure conditional grants
constituted 73,8% and 49,7% of capital funding in the Western Cape and Gauteng respectively. This suggests
that the municipalities in provinces with greater levels of economic activity and residents with higher income
levels tend to be more self-financing, more sustainable and less dependent on inter-governmental grants.
This is in contrast to municipalities in the Eastern Cape, Mpumalanga and Limpopo, where municipal fiscal
bases are constrained by high levels of unemployment and poverty, and own revenues (raised by the
municipalities themselves) constitute a significantly lower proportion of total operating revenue. Municipal own
revenues in the Eastern Cape, Mpumalanga and Limpopo comprised 71,9%, 71,1% and 59,2% of operating
revenue in 2017/18 respectively. Direct infrastructure grants from national government funded a higher
proportion of aggregate municipal capital budgets in these provinces (In 2017/18 78,4%, 85,6% and 83,9%
respectively). Municipalities in poorer provinces tend to be more dependent on grants from national and
provincial governments, and less self-financing.
Outstanding consumer accounts for municipal services amounted to R165,5 billion as at 30 June 2019, of
which R4,2 billion was deemed irrecoverable and written off as bad debt (up from R143,2 the previous financial
year). These increases in debts owed by consumers to municipalities indicate their weak ability to recover the
revenue that is due to them. Revenue management remains one of the weaknesses of local government.
In 2018/19, 125 of the 257 municipalities found themselves in financial distress. The symptoms of
financial distress included insufficient cash reserves to fund their operations, overspending on operating
budgets but underspending on capital budgets, increases in consumer debt owed to municipalities, high levels
of water and electricity losses, inadequate repairs and maintenance spending and deteriorating audit
outcomes.

e Constitution distinguishes three main categories of municipalities: metropolitan councils (category A),
local councils (category B) and district municipalities (category C). ere is considerable variation both
within and across these categories in terms of revenue and expenditure patterns. Metropolitan councils
occur in the big cities, for example, Johannesburg, Tshwane, Cape Town and Ethekwini. Outside of the
metropolitan councils, groups of adjacent local councils comprise a municipal district, which has its own
district council. South Africa therefore has a two-tiered local government system outside of the urban
metropolitan areas.
e key issue confronting local governments is sustainability. Muni-cipalities are under extreme pressure
(as evidenced by service delivery protests) to improve access to services as well as the quality of the services
they deliver (for example, eradication of the bucket system and provision of effective sanitation, safe
drinking water and electricity). Rural municipalities, in particular, are also faced with infrastructure backlogs
that must be eliminated as well as existing infrastructure that must be maintained. Urban municipalities are
under pressure to invest in economic infrastructure such as public transport, which is required to underpin
economic growth. All of these create spending pressures on municipal budgets, but their revenue sources
are also severely constrained. A culture of non-payment for municipal services has led to the accumulation
of arrears and pressured the revenue side of municipal budgets. Many municipalities have experienced great
difficulties in recovering the user charges owed to them by households for services rendered.
One of the challenges of local government is to improve nancial management to ensure that budgets
are adhered to. is would include the introduction of uniform accounting standards and compliance with
GAMAP (Generally Accepted Municipal Accounting Practices). Financial reporting systems also tended to
be weak at municipal level, precluding early warning systems and effective monitoring as well as evaluation
of nancial and service-delivery performance. While improved credit control, debt collection and other
forms of nancial management can certainly contribute to the sustainability of poor municipalities, their
nancial condition will remain vulnerable unless the underlying structural conditions of unemployment,
poverty and the skewed spatial distribution of economic activity are also addressed.
 Since 1994, local governments have been undergoing a fundamental transformation that culminated in
the redemarcation of municipalities in December 2000. e number of municipalities was reduced from 843
to 278 in 2014. Other important changes include the reassignment of functions between district and local
municipalities, the impact of the restructuring of the electricity industry, the devolution of health and
certain public transport functions, the funding of free basic services and the introduction of a new property
rates system. It is important to assess the impact of all these factors cumulatively on local government
nance. At present, however, there are too many transformational processes that are still either under way or
newly completed for the true nancial position of individual municipalities to be determined. e Municipal
Finance Management Act of 2003 and the Property Rates Act of 2004 introduced a uniform valuation system
that should go some way to providing a legal framework for enhancing the nancial viability of
municipalities. e challenge which remains is to fully implement these acts, and to identify sustainable,
new revenue streams for municipalities.
As service delivery protests at municipal level intensify, there is an urgent need for poorer and rural
municipalities to develop the managerial and technical capacity to deliver services effectively to their
residents. In addition, they must eliminate the waste, mismanagement and outright corruption that are
brought to light so often by the Auditor General. Finally, the poor performance of many public entities such
as Eskom, PRASA and some of the water boards seriously undermine the ability of municipalities to delivery
basic services, create inclusive cities and towns conducive to job creation, harness the emerging
technologies of the Fourth Industrial Revolution, deal with water scarcity and food security, and develop
resilience to climate change. Better national government oversight and effective, independent regulation of
state owned enterprises to combat corruption and inefficiency will be crucial to support municipalities in
confronting the many challenges facing them.

Key concepts
• assignment problem (page 456)
• agglomeration economies (page 478)
• block grant (page 462)
• closed-ended matching grants (page 465)
• conditional grants (page 461)
• expenditure assignment (page 458)
• financial contagion (page 460)
• fiscal federalism (page 450)
• horizontal fiscal imbalances (page 472)
• inter-governmental fiscal relations (page 450)
• inter-governmental grants (page 461)
• matching grant (page 461)
• moral hazard (page 460)
• non-matching grant (page 461)
• open-ended matching grants (page 463)
• revenue sharing (page 466)
• spatial externalities (page 450)
• sub-national government (page 450)
• tax assignment (page 458)
• tax competition (page 459)
• tax harmonisation (page 459)
• Tiebout model (page 483)
• unconditional grants (page 461)
• unfunded mandate (page 466)
• urban economics (page 478)
• vertical fiscal imbalance (page 461)
SUMMARY
• Most governments around the world have different sub-national levels or tiers. ese include
provincial/state or regional governments and/or local governments or municipalities.
• Fiscal federalism (also known as inter-governmental scal relations) is a body of economic theory that
tries to explain why these different levels or tiers of government exist and which functions are best
centralised (that is, performed at national level) or decentralised (that is, performed by provincial and/or
local governments).
• e Tiebout model demonstrates that a decentralised scal system – which can accommodate a diversity
of preferences for public goods – can be welfare-increasing in relation to a centralised system that
imposes a standardised public good-tax mix on people, no matter what their tastes. Tiebout’s notion of
‘voting with one’s feet’ permits the revelation of preferences by allowing people to sort themselves into
groups with similar tastes
• Fiscal federalism theory also explains which revenue instruments (such as taxes and user charges),
which expenditure responsibilities and which borrowing powers are best assigned to each sphere of
government.
• Fiscal federalism also looks at the design of inter-governmental grants that are allocated by national
government to sub-national governments, either unconditionally (discretionary) or conditionally (in
other words, earmarked for a speci c purpose only).
• Urban economics focuses on the economic rationale for the existence of cities based on agglomeration
economies and their growth and development (for example, as a result of increasing urbanisation and
rural-urban migration).
• Local government nance is a specialised sub-discipline of economics looking at how cities raise funds
(through taxes such as property rates as well as user charges for water and electricity) and how they
spend these funds in order to deliver services (for example, the rollout of municipal water and electricity
infrastructure).
• South Africa has three spheres of government: national government, nine provincial governments and
278 municipalities, each with their own expenditure functions and revenue sources, as determined by
the Constitution.
• Most of the buoyant revenue sources are collected at national government level by the South African
Revenue Services. Provincial governments as well as certain poor and rural municipalities have limited
own revenue sources, but signi cant expenditure responsibilities.
• As a result of this vertical scal gap, the South African Constitution outlined a process whereby nationally
collected revenues are divided equitably across the three spheres of government (vertical division of
revenue) and within each sphere to each provincial government and municipality (horizontal division of
revenue).

MULTIPLE-CHOICE QUESTIONS
19.1 Which of the following statements are correct? e delivery of public goods with the following
characteristics are better decentralised to sub-national governments:
i. ere are limited economies of scale involved in their production.
ii. ey generate spatial externalities.
iii. Citizen preferences and tastes for these public goods vary substantially across jurisdictions within
a country.
a. All of the above
b. i and ii only
c. i and iii only
d. ii and iii only
19.2 Which of the following statements are correct? e following types of taxes are best centralised in
national government:
 i. Property taxes
ii. Value-added tax (VAT)
iii. Corporate income tax
a. All of the above
b. i and ii only
c. i and iii only
d. ii and iii only
19.3 Which of the following is not an assumption of the Tiebout model?
a. Citizens have full information on the public goods offered by each jurisdiction.
b. ere are a large number of jurisdictions for citizens to choose from.
c. All factors of production are fully mobile.
d. ere are no spillovers across jurisdictions.
19.4 Certain provinces, such as the Western Cape and Gauteng, have academic hospitals that offer
specialised health services such as heart transplants, while other provinces have no academic hospitals,
but refer patients to those provinces that do offer specialised services. If provinces had to nance
specialised services in academic hospitals from their own revenue sources, they would underprovide.
What type of inter-governmental grant from national government does scal federalism theory suggest
is most suitable for ensuring minimum standards of access to these services?
a. Unconditional non-matching grants
b. Conditional non-matching grants
c. Conditional matching grants
d. None of the above

SHORT-ANSWER QUESTIONS
19.1 Explain what a local public good is and provide an example.
19.2 In a decentralised scal system with different municipalities or provinces, negative and positive spatial
externalities may exist. Explain what a spatial externality is and give an example of each.

ESSAY QUESTIONS
19.1 e scal federalism literature contends that stabilisation and distribution functions are best performed
at national level, whereas allocative functions are best performed at sub-national level. Can you explain
why?
19.2 Explain the assumptions and main arguments of the Tiebout model.
19.3 ‘A scally decentralised system is always more efficient and more equitable than a scally centralised
system.’ Discuss.
19.4 Which type of inter-governmental grant would be most suitable to ensure minimum educational
standards across all provinces? Why? Illustrate your answer by means of a diagram.
19.5 ‘Tax competition by sub-national governments always has negative effects.’ Do you agree?
19.6 Should provincial debt be formally guaranteed by the national government? Why (not)?
19.7 How does South African expenditure and revenue assignment compare with the guidelines of scal
federalism theory?
19.8 What are the key issues and trends in provincial and local government nances in South Africa?

1 Logrolliing refers to the reciprocal exchange of support or favours among political parties or politicians whereby one will vote for
the other’s proposed project or legislation in exchange for similar support.
2 e United States is an exception. Cities have been allowed to declare bankruptcy, most famous of which is New York City (1975)
and Detroit (2013).
3 Own revenue is taken to refer to revenue raised within the province or on behalf of the province. ese include provincial taxes,
user charges, licence fees and so on.
4 Although the original 1996 FFC formula did suggest costed minimum service norms and standards, their formula was never used.
e provincial formula actually used by National Treasury and approved by the Budget Council only has implicit service standards.
5 Johannesburg, Cape Town, eekwini, Ekurhuleni, Tshwane, Mangaung, Buffalo City and Nelson Mandela.
References
Abedian, I. & Biggs, M. (eds). 1998. Economic globalization and scal policy. Cape Town: Oxford University Press.
Ade, M., Rossouw, J. & Gwatidzo, T. 2017. Analysis of tax harmonisation in the SADC. ERSA working paper 684.
Economic Research Southern Africa (ERSA). [Online]. Available:
https://econrsa.org/system/ les/publications/working_papers/working_paper_684.pdf [Accessed 15 July 2018].
African Development Bank. 2001. African Development Report. Oxford: Oxford University Press.
African Development Bank. 2004. African Development Report. Oxford: Oxford University Press.
African Development Bank. 2014. African Economic Outlook 2014. Available:
https://www.un.org/en/africa/osaa/pdf/pubs/2014afrecooutlook-afdb.pdf [Accessed 17 October 2019].
African Development Bank. 2016. African Economic Outlook 2016. Available:
https://www.afdb.org/ leadmin/uploads/afdb/Documents/Publications/AEO_2016_Report_Full_English.pdf
[Accessed 11 August 2019].
African Economic Outlook. 2010. Past scal prudence and disin ation have created space for expansionary macro
policies. [Online]. Available: http://www.africaneconomicoutlook.org/en/outlook/macroeconomic-situation-and-
prospects/facing-the-crisis-with-new-policy-responses/ [Accessed 14 February 2011].
Ajam, T. & Aron, J. 2007. Fiscal renaissance in a democratic South Africa. Journal of African Economies, 16(5): 745–781.
Alence, R. 1998. e economic policy-making process in South Africa: Report on a survey of an informed panel, late
1997. Pretoria: Human Science Research Council.
Allen, F. & Giovannetti, G. 2011. e effects of the nancial crisis on Sub-Saharan Africa. Review of Development Finance,
1(1): 1–27.
Alm, J. 2012. Measuring, explaining, and controlling tax evasion. International Tax and Public Finance, 19(1): 54–77.
Alvaredo, F & Atkinson, A.B. 2010. Colonial rule, Apartheid and natural resources: Top incomes in South Africa, 1903–
2007. Discussion Paper Series no. 8155. Centre for Economic Policy Research. [Online]. Available:
http://www.parisschoolofeconomics.eu/IMG/pdf/DP8155_SouthAfrica.pdf [Accessed 30 January 2015].
Amra, R. 2013. Back to the drawing board? A critical evaluation of South Africa’s tariff setting methodology. Paper
presented at the Biennial Conference of the Economic Society of South Africa. 25–27 September 2013,
Bloemfontein. [Online]. Available: http://www.essa2013.org.za/fullpaper/essa2013_2671.pdf [Accessed 11 June
2018].
Anderson, P. & Baumberg, B. 2006. Alcohol in Europe: A public health perspective, A report for the European
Commission, Institute of Alcohol Studies, UK, June 2006.
Ardington, E. & Lund, F. 1995. Pensions and development: How the social security system can complement programmes of
reconstruction and development. Development Paper 61. Midrand: Development Bank of Southern Africa.
Arrow, K.J. 1951. Social choice and individual values. New York: John Wiley.
Arrow, K.J. 1962. e economic implications of learning by doing. Review of Economic Studies, 29.
Aschauer, D.A. 1989. Is public expenditure productive? Journal of Monetary Economics, 23(2).
Atkinson, A.B. & Stiglitz, J.E. 1980. Lectures on Public Economics. Maidenhead UK: McGraw-Hill.
Auerbach, A.J. 2009. Implementing the new scal policy activism. American Economic Review: Papers & Proceedings,
99(2): 543–549.
Bahl, R. 1998. Land versus property taxes in developing and transition countries. Lincoln Institute of Land Policy
conference on land value taxation in contemporary societies. Phoenix. 11–13 January 1998.
Baiker, K., Clemens, J. & Singhal, M. 2010. Fiscalfederalism in the United States. [Online]. Available:
http://www.hks.harvard.edu/fs/msingha/BaickerClemensSinghal_FiscFed_Jun10.pdf [Accessed 1 April 2015].
Bank of Namibia. 2006. Integrated Paper on recent economic developments in SADC. 1 December 2006. [Online].
Available:
http://www.sadcbankers.org/SADC/SADC.nsf/LADV/3CD719E2A8CEBD7442257257002EEFBC/$File/RED+Oct20
06+(Final).pdf [Accessed 27 March 2008].
Bannock, G., Baxter, R.E. & Rees, R. 1971. e Penguin Dictionary of Economics. London/New York: Allen Lane/e
Viking Press.
Barr, N. 1989. Social insurance as an efficiency device. Journal of Public Policy, 9(1): 59–82.
Barreix, A. & Roca, J. 2007. Strengthening a scal pillar: e Uruguayan dual income tax. Cepal Review, 92: 121–140.
Bastagli, F., Hagen-Zanker, J., Harman, L., Sturge, G., Barca, V., Schmidt, T. & Pellerano, L. 2016. Cash transfers: what does
the evidence say? A rigorous review of impacts and the role of design and implementation features. Research report.
London: Overseas Development Institute.
Baumol, W.J. 1967. Macro-economics of unbalanced growth: e anatomy of urban crisis. American Economic Review,
57: 415–426.
Baumol, W.J. 1982. Contestable markets and the theory of industry structure. San Diego: Harcourt.
Becker, G.S. 1985. Public policies, pressure groups, and dead weight costs. Journal of Public Economics, 28(3): 329–347.
Bernardi, L. Barreix, A. Marenzi, A. & Profeta, P. 2008. Tax systems and tax reforms in Latin America. London: Routledge.
Bhorat, H. 2004. Labour market challenges in post-apartheid South Africa. South African Journal of Economics, 72(5):
940–977.
Bickers, B.N., Salucci, L. & Stein, R.M. 2006. Assessing the Micro-Foundations of the Tiebout Model. Urban Affairs
Review, 42(1): 57–80.
Bird, R.M. 1971. Wagner’s ‘law’ of expanding state activity. Public Finance, 26(1): 1–26.
Bird, R.M. 1992. Tax policy and economic development. Baltimore: Johns Hopkins.
Bird, R.M. 2008. e BBLR approach to tax reform in emerging countries. International Studies Program Working Paper
08-04. Andrew Young School of Policy Studies. Georgia State University. December 2008: 1–30.
Bird, R.M. & Zolt, E.M. 2010. Dual income taxation and developing countries. Columbia Journal of Tax Law, 1: 174–217.
Black, P.A. 1993. Affirmative action: Rational response to a changing environment. South African Journal of Economics,
61(4): 317–323.
Black, P.A. 1996. Affirmative action in South Africa: Rational discrimination according to Akerlof? South African Journal
of Economics, 64(1): 74–82.
Black, P.A. 2004. Economic impact analysis: A methodological note. South African Journal of Economics, 72(5): 1068–
1074.
Black, P.A. 2008. Alcohol taxes versus preventative measures: A theoretical note. South African Journal of Economics,
76(4): 607–611.
Black, P.A. & Dollery, B.E. 1992. Leading issues in South African Microeconomics: Selected readings. Halfway House:
Southern Book Publishers.
Black, P.A. & Mohamed, A.I. 2006. ‘Sin’ taxes and poor households: Unanticipated effects. South African Journal of
Economics, 74(1): 131–136.
Black, P.A. & Saxby, G. 1996. Differential investment multipliers: An application of Weiss and Gooding. South African
Journal of Economic and Management Sciences, 9(4).
Blanchard, O., Dell’Ariccia, G. & Mauro, P. 2010. Rethinking macroeconomic policy. IMF Staff Position Note. Washington,
DC. Available: http://www.imf.org/external/pubs/ft/spn/2010/spn1003.pdf.
Blecher, M., Davén, J., Kollipara, A., Maharaji, Y., Mansvelder, A. & Gaarekwe, O. 2017. Health spending at a time of low
economic growth and scal constraint. In: Padarath A. & Barron, P. (eds). South African health review 2017. Durban:
Health Systems Trust: 25–44.
Blecher, M., Kollipara, A., De Jager, P. & Zulu, N. 2011. Health nancing. In: Padarath, A. & English, R. (eds). South
African health review 2011. Durban: Health Systems Trust: 29–47.
Blinder, A.S. 1987. e rules-versus-discretion debate in the light of recent experience. Welwirtscha iches Archiv, 123:
399–413.
Blinder, A.S. 2006. e case against the case against discretionary scal policy. In: Kopcke, R.W., Tootell, G.M.B. & Triest,
R.K. (eds). e macroeconomics of scal policy. Cambridge, Mass: MIT Press.
Blinder, A.S. & Zandi, M. 2010. How the Great Recession was brought to an end. 27 July. [Online]. Available:
http://scholar.google.co.za/scholar?cluster=16640258663503337300&hl=en&as_sdt=0,5. [[PDF] from dismal.com].
[Accessed 17 February 2011].
Boadway, R. 2005. Income tax reform for a globalized world: e case for a dual income tax. Journal of Asian Economics,
16: 910–927.
Boadway, R. & Shah, A. 1995. Perspectives on the role of investment incentives in developing countries. In: Shah, A. (ed).
Fiscal incentives for investment and innovation. Washington, DC: World Bank.
Boadway, R., Chamberlain, E. & Emmerson, C. 2010. Taxation of wealth and wealth transfers. In: Dimensions of tax
design: e Mirrlees review chaired by Sir James Mirrlees. Oxford: Oxford University Press: 737–814.
Boadway, R.W. & Wildasin, D.E. 1984. Public sector economics. 2nd ed. Boston: Little, Brown and Co.
Bohm, P. 1978. Social efficiency. London: Macmillan.
Borcherding, T.E. (ed). 1977. Budgets and bureaucrats. Durham: Duke University Press.
Boshoff, W.H. 2008. Cigarette demand in South Africa over 1996-2006: e role of price, income and health awareness.
South African Journal of Economics, (76)1: 118–131.
Boskin, M.J. (ed). 1996. Frontiers of tax reform. Stanford: Hoover Institution Press.
Bowles, S. 2004. Microeconomics. Princeton: Princeton University Press.
Brennan, G. & Buchanan, J.M. 1980. e power to tax: Analytical foundations of a scal federalism constitution.
Cambridge, New York: Cambridge University Press.
Brewer, M., Browne, J. & Johnson, P. 2012. e 50p income tax rate: what is known and what will be known? In
Emmerson, C., Johnson, P. & Miller, H. (eds). e IFS Green Budget. London: Institute for Fiscal Studies, 180–196.
Bromberger, N. 1982. Government policies affecting the distribution of income, 1940–1980. In: Schrire, R. (ed). 1982.
South Africa: Public policy perspectives. Cape Town: Juta: 165–203.
Brown, C.V. & Jackson, P.M. 1990. Public sector economics. Oxford: Basil Blackwell.
Browne, G.W.G. 1983. Fifty years of public nance. South African Journal of Economics, 51(1): 134–173.
Browning, E.K. & Browning, J.M. 1994. Public nance and the price system. 4th ed. New York: Macmillan.
Brülhart, M. & Sbergami, F. 2009. Agglomeration and growth: Cross-country evidence. Journal of Urban Economics, 65:
48–63.
Buchanan, J.M., Tollison, R.D. & Tullock, G. (eds). 1980. Towards a theory of the rentseeking society. College Station: Texas
A & M University Press.
Buchanan, J.M. & Tullock, G. 1962. e calculus of consent. Ann Arbor Mich.: University of Michigan Press.
Bundy, C. 1988. e rise and fall of the South African peasantry. 2nd ed. Cape Town: David Philip.
Bureau of Market Research. 2017. Household Income and Expenditure Trends and Patterns, 2013–2017. Research Report
No 488. Pretoria: Unisa.
Burger, P. 2014. Facing the conundrum: how useful is the ‘developmental state’ concept in South Africa? South African
Journal of Economics, 82(2): 159–180.
Burger, R., Bredenkamp, C., Grobler, C. & Van der Berg, S. 2012. Have public health spending and access in South Africa
become more equitable since the end of apartheid? Development Southern Africa, 29(5): 681–703.
Burger, P. & Calitz, E. 2014. Twenty-year review of South African scal policy. Unpublished commissioned report written
for the National Treasury. Bloemfontein: University of the Free State, and Stellenbosch: University of Stellenbosch.
Burger, P., Calitz, E. & Siebrits, F.K. 2016. Fiscal consolidation and the public sector balance sheet in South Africa. South
African Journal of Economics, 64(4): 501–519.
Burger, P. & Jimmy, C. 2006. Should South Africa have a scal rule? South African Journal of Economics, 74(4): 642–669.
Burger, P., Stuart, I. & Jooste, C. & Cuevas, A. 2012. Fiscal sustainability and the scal reaction function for South Africa:
assessment of the past and future policy applications. South African Journal of Economics, 80(2): 209–227.
Business Day. 2017. Alarm as Eskom tops list of tender benders. 11 October 2017. [Online]. Available:
https://www.businesslive.co.za/bd/companies/2017-10-11-eskom-responsible-for-lions-share-of-r37bn-in-
procurement-deviations/ [Accessed 14 June 2018].
Calitz, E. 1986. Aspekte van die vraagstuk van staatsbestedingsprioriteite met spesiale verwysing na die Republiek van
Suid-Afrika: ’n funksione elekonomiese ondersoek. Stellenbosch: University of Stellenbosch. (DCom thesis)
Calitz, E. 1992. e limits to public expenditure. In: Howe, G. & Le Roux, P. (eds). 1992. Transforming the economy: Policy
options for South Africa. Indicator Project SA, University of Natal and Institute for Social Development, University of
the Western Cape.
Calitz, E. 2002. Comparative international perspectives on structural economic reform in South Africa. BMR Research
Report 2002/01, Pretoria: University of South Africa, Bureau for Market Research, Pretoria.
Calitz, E., Du Plessis. S.A. & Siebrits, F.K. 2011. An alternative perspective on South Africa’s public debt, 1962-1994. South
African Journal of Economics, 79(2): 161–172.
Calitz, E. & Siebrits, F.K. 2002. Changes in the role of government in the SA economy. Absa Economic Perspective, Second
Quarter: 15–23.
Calitz, E. & Siebrits, F.K. 2003. Fiscal policy in the 1990s. South African Journal of Economic History, 18(1&2): 50–75.
Calitz, E. & Siebrits, F.K. 2007. e legacy and challenge of scal policy in sub-Saharan Africa. South African Journal of
Economics, 75(2): 221–235.
Calitz, E. & Siebrits, F.K. 2015. Fiscal policy. In: Black, P.A., Calitz, E. & Steenekamp, T.J. (eds). Public Economics. 6th ed.
Cape Town: Oxford University Press: 343–388.
Calitz, E. & Siebrits, F.K. 2018. Review of and proposals for reducing Ghana’s scal de cit and public debt. Commissioned
and unpublished report for Institute for Economic Affairs, Ghana.
Calitz, E., Siebrits, F.K. & Stuart, I. 2013. e accuracy of scal projections in South Africa. Stellenbosch Working Papers
24/13. Stellenbosch: Stellenbosch University (Department of Economics & Bureau of Economic Research).
[Online]. Available: http://www.ekon.sun.ac.za/wpapers [Accessed 27 January 2015].
Carmignani, F. 2010. Cyclical scal policy in Africa. Journal of Policy Modeling, 32(2): 254–267.
Case, A. & Deaton, A. 1998. Large cash transfers to the elderly in South Africa. Economic Journal, 108(450): 1330–61.
Central Statistical Service. 1996. Living in South Africa: Selected ndings of the 1995 October household survey. Pretoria:
Government Printer.
Chalk, N. & Hemming, R. 2000. Assessing scal sustainability in theory and practice. IMF Working Paper, WP/00/81.
Washington, DC: International Monetary Fund.
Chernew, M.E. & Newhouse, J.P. 2011. Health care spending growth. In: Pauly, M.V., McGuire, T.G. & Barros, P.P. (eds).
Handbook of health economics. Volume 2. Amsterdam: Elsevier: 1–43.
Chia, N.C. & Whalley, J. 1995. Patterns in investment: Tax incentives among developing countries. In: Shah, A. (ed).
Fiscal incentives for investment and innovation. Washington: World Bank.
Chopra, M., Daviaud, E., Pattinson R., Fonn S. & Lawn J.E. 2009. Saving the lives of South Africa’s mothers, babies, and
children: can the health system deliver? e Lancet, 374(9692): 835–46.
Christian, C.S. & Crisp, N. 2012. Management in the South African public health sector: An X-inefficiency perspective.
Development Southern Africa, 29(5): 725–737.
Ciccone, A. & Hall, R.E. 1996. Productivity and the density of economic activity. American Economic Review, 86(1): 54–
70.
Cirillo, C. & Tebaldi, R. 2016. Social protection in Africa: Inventory of non-contributory programmes. Brasilia:
International Policy Centre for Inclusive Growth & New York: United Nations Children’s Fund.
Cnossen, S. 1990. e case for selective taxes on goods and services in developing countries. In: Bird, R. & Oldman, O.
(eds). Taxation in developing countries. 4th ed. Baltimore: Johns Hopkins.
Cnossen, S. 2003. How much tax coordination in the European Union? International Tax and Public Finance, 10: 625–
649.
Cnossen, S. 2015. Mobilising VAT revenues in African countries. International Tax and Public Finance, 22(6): 1077–1108.
Coady, D., Grosh, M. & Hoddinott, J. 2004. Targeting outcomes redux. World Bank Research Observer, 19(1): 61–85.
Coetzee, M. 2013. Finding the bene ts: estimating the impact of the South African child support grant. South African
Journal of Economics, 81(3): 427–450.
Cohen, J.M. & Cohen, M.J. 1960. e Penguin Dictionary of Quotations. New York: Penguin.
Committee of Urban Transport Officials (CUTA). 2002. Guidelines for conducting the economic evaluation of urban
transport projects. 3rd ed. Cape Town: Government Printer.
Compensation Fund. 2018. Annual report 2017/18. Pretoria: Department of Labour.
Connolly, S. & Munro, A. 1999. Economics of the public sector. Harlow: Prentice Hall.
Cook, P. & Kirkpatrick, C. 1997. Globalisation, regionalisation and ird World development. Regional Studies, 31(1): 55–
66.
Corlett, W.J. & Hague, D.C. 1953. Complementarity and the excess burden of taxation. Review of Economic Studies, 21:
21–30.
Cowen, T. & Crampton, E. 2011. Market failure or success: the new debate. Fairfax: George Mason University Press.
Crouch, L. & Mabogoane, T. 1998. When the residuals matter more than the coefficients: An educational perspective.
Studies in Economics and Econometrics, 22(2): 1–14.
Das, J., Do, Q-T. & Özler, B. 2005. Reassessing conditional cash transfer programs. World Bank Research Observer, 20(1):
57–80.
Dasgupta, A.K. & Pearce, D.W. 1974. Cost bene t analysis: eory and practice. London: Macmillan Press.
Davis Tax Committee (DTC). 2014. First interim report on Small and Medium Enterprises for the Minister of Finance,
January 2014. [Online]. Available:
https://www.taxcom.org.za/docs/20150806%20DTC%20First%20Interim%20Report%20on%20SMEs.pdf [Accessed
15 June 2018].
Davis Tax Committee (DTC). 2015. First interim report on Estate Duty for the Minister of Finance, January 2015. [Online].
Available:
https://www.taxcom.org.za/docs/20150723%20DTC%20First%20Interim%20Report%20on%20Estate%20Duty%20-
%20website.pdf [Accessed 30 June 2018].
Davis Tax Committee (DTC). 2016a. Second and nal report on Hard-rock Mining for the Minister of Finance, December
2016. [Online]. Available: https://www.taxcom.org.za/docs/20171113%20Second%20and%20 nal%20hard-
rock%20mining%20report%20on%20website.pdf [Accessed 15 June 2018].
Davis Tax Committee (DTC). 2016b. Report on Oil and Gas for the Minister of Finance, September 2016. [Online].
Available: https://www.taxcom.org.za/docs/20171113%20Oil%20and%20Gas%20Report%20on%20website.pdf
[Accessed 15 June 2018].
Davis Tax Committee (DTC). 2016c. Small and Medium Enterprises: taxation considerations, second and nal report,
April 2016. [Online]. Available: http://www.taxcom.org.za/docs/20160414%20DTC%20Final%20SME%20Report.pdf
[Accessed 15 June 2018].
Davis Tax Committee (DTC). 2016d. Second and nal report on Estate Duty for the Minister of Finance, April 2016.
[Online]. Available:
https://www.taxcom.org.za/docs/20160428%20DTC%20Final%20Report%20on%20Estate%20Duty%20-
%20website.pdf [Accessed 30 June 2018].
Davis Tax Committee (DTC). 2018a. Report on the efficiency of South Africa’s corporate income tax system for the
Minister of Finance, March 2018. [Online]. Available:
https://www.taxcom.org.za/docs/20180411%20Final%20DTC%20CIT%20Report%20-%20to%20Minister.pdf
[Accessed 15 June 2018].
Davis Tax Committee (DTC). 2018b. Report on Feasibility of a Wealth Tax in South Africa for the Minister of Finance,
March 2018. [Online]. Available:
https://www.taxcom.org.za/docs/20180329%20Final%20DTC%20Wealth%20Tax%20Report%20-
%20To%20Minister.pdf [Accessed 30 June 2018].
Davis Tax Committee (DTC). 2018c. Final report on VAT for the Minister of Finance, March 2018. [Online]. Available:
https://www.taxcom.org.za/docs/20180329%20Final%20DTC%20VAT%20Report%20to%20the%20Minister.pdf
[Accessed 15 July 2018].
Davis Tax Committee (DTC). 2018d. Closing report on the work done by the Davis Tax Committee, March 2018. [Online].
Available: https://www.taxcom.org.za/docs/20180329%20DTC%20Closing%20Report(2).pdf [Accessed 15 July
2018].
De Wulf, L. 1975. Fiscal incidence studies in developing countries: Survey and critique. IMF Staff Working Papers. 22(1)
March: 61–131.
Debrun, X. & Kapoor, R. 2010. Fiscal Policy and Macroeconomic Stability: Automatic Stabilizers Work, Always and
Everywhere. IMF Working Paper, WP/10/111. Washington, DC: International Monetary Fund.
Decker, J.M. & Crompton, J.L. 1993. Attracting Footloose Companies: An Investigation of the Business Location Decision
Process. Journal of Professional Services Marketing, 9(1): 69–94.
Den Hertog, J. 2010. Review of economic theories of regulation. Discussion Paper Series No 10-18. Utrecht: Utrecht School
of Economics (Tjalling C. Koopmans Institute).
Department of Agriculture, Forestry and Fisheries. 2012. Policy for the small scale sheries sector in South Africa.
Government Gazette. Republic of South Africa. Vol. 564, No. 35455, 20 June 2012. Pretoria. Department of Basic
Education. 2018. Annual performance plan 2018/19. Pretoria: Department of Basic Education.
Department of Cooperative Government and Traditional Affairs. 2013. Towards an integrated urban development
framework: A discussion document. [Online]. Available:
http://www.cogta.gov.za/index.php/documents/doc_view/1036-discussiondocument-for-integrated-
urbandevelopment-framework.raw?tmpl=component [Accessed 25 February 2014].
Department of Finance. 1996. Macroeconomic strategy on growth, employment and redistribution. Pretoria: Government
Printer.
Department of Finance. 1997. Medium term budget policy statements. Pretoria: Government Printer.
Department of Finance. 1998. Budget review 1998. Pretoria: Government Printer.
Department of Finance. 1999. National expenditure survey 1999. Pretoria: Government Printer.
Department of Finance. 2000. Budget review 2000. Pretoria: Government Printer.
Department of Housing. 1995. Annual Report. Pretoria: Government Printer.
Department of Housing. 1996. Annual Report. Pretoria: Government Printer.
Department of Minerals and Energy. 2008. Electricity pricing policy (EPP) of the South African electricity supply
industry. Government Gazette No 31741 (19 December 2008). Pretoria.
Department of Welfare. 1997. White Paper for Social Welfare. Pretoria: Government Printer.
Dornbusch, R., Fischer, S., Mohr, P. & Rogers, C. 1994. Macroeconomics. 3rd ed. Johannesburg: Lexicon.
Dowding, K., John, P. & Biggs, S. 1994. Tiebout: A Survey of the Empirical Literature. Urban Studies, 31(4): 767–797.
Downs, A. 1957. An economic theory of democracy. New York: Harper & Row.
Du Plessis, S.A. & Boshoff, W. 2007. A scal rule to produce counter-cyclical scal policy in South Africa. Stellenbosch
Economic Working Papers 13/07. Stellenbosch: University of Stellenbosch (Department of Economics) and Bureau
for Economic Research.
Du Plessis, S.A., Smit, B.W. & Sturzenegger, F. 2007. e cyclicality of monetary and scal policy in South Africa since
1994. South African Journal of Economics, 75(3): 391–411.
Easson, A.J. 1992. Tax incentives for foreign direct investment in developing countries. Australian Tax Forum, 9: 387–439.
Econex. 2010. e appropriateness of using excise taxes to address the question of external costs related to alcohol
consumption in the South African context and alternative interventions. Report prepared by Econex Pty (Ltd) for the
South African Breweries Limited. Team leader: Professor Philip Black. Econex: Trade, Competition & Applied
Economics: Cape Town.
Economist 2012a. Taxing the rich in America: e politics of plutocracy – America’s rich should pay more, but there is no
need to raise their income-tax rates, 21 January, 21. [Online]. Available:
http://www.economist.com/node/21543165 [Accessed 30 January 2015].
Economist 2012b. Squeezing the rich: An already highly taxed country may be burdened even more. 3, March, 3.
[Online]. Available: http://www.economist.com/node/21548980 [Accessed 30 January 2015].
Eichenbaum, M. 1997. Some thoughts on practical stabilization policy. American Economic Review Papers and
Proceedings, 87(2): 236–239.
El-Khouri, S. 2002. Fiscal policy and macroeconomic management. In: Khan, M.S., Nsouli, S.M. & Wong, C-H. (eds).
Macroeconomic management: Programs and policies. Washington, DC: International Monetary Fund.
Engelbrecht, B. & Crisp, N. 2010. Improving the performance of the South African health system. In: Padarath, A. & Fonn,
S. (eds). South African health review 2010. Durban: Health Systems Trust: 195–203.
Eskom, 2018. Moody’s downgrades Eskom’s credit rating. [Online]. Available:
http://www.eskom.co.za/news/Pages/2018Jan26.aspx [Accessed 14 June 2018].
Estrin, S. & Pelletier, A. 2018. Privatisation in developing countries: What are the lessons of recent experience? World
Bank Research Observer, 33: 65–102.
Evans D.B., Marten, R. & Etienne, C. 2012. Universal health coverage is a development issue. e Lancet, 380(9845): 864–
865.
Evans, D.K., Hausladen, S., Kosec, K. & Reese, N. 2014. Community-based conditional cash transfers in Tanzania: results
from a randomized trial. Washington, DC: World Bank.
Evans, D.K. & Popova, A. 2017. Cash transfers and temptation goods. Economic Development and Cultural Change,
65(2): 189–221.
Fan, S. & Rao, S. 2003. Public spending in developing countries: Trends, determination, and impact. EPTD Discussion
Paper No. 99. Washington, DC: International Food Policy Research Institute.
Faria, A.G.A. 1995. Source versus residence principle; Relief from double taxation; Aspects of tax treaties; International
capital ows. In: Shome, P. (ed). 1995. Tax policy handbook. Washington, DC: International Monetary Fund.
Feldstein, M. 1997. How big should government be? National Tax Journal, 50(2): 197–213.
Filmer, D., Hammer, S. & Pritchett, L.H. 2000. Weak links in the chain: A diagnosis of health policy in poor countries.
World Bank Research Observer, 15(2): 199–224.
Fin24. 2018. More than 200 senior Eskom managers give Ramaphosa deadline to act. [Online]. Available:
https://www. n24.com/Economy/Eskom/more-than-200-senior-eskom-managers-give-ramaphosa-deadline-to-
act-20180120 [Accessed 14 June 2018].
Financial and Fiscal Commission (FFC). 1996. e Financial and Fiscal Commission’s recommendations for the allocation
of nancial resources to the national Government and the provincial Governments for the 1997/98 nancial year.
Midrand: Financial and Fiscal Commission.
Financial sector Charter Council. 2008. Annual Review: Report on the Transformation of the Financial Sector in South
Africa. [Online]. Available: https://fstc.org.za/pdf/reports-and-reviews/2008-Annual-Review-on-
Transformation.pdf [Accessed 16 October 2019].
Financial Services Board. 2018. Registrar of Pension Funds annual report 2017. Pretoria: Financial Services Board.
Fischel, W.A. Footloose at fty: An introduction to the Tiebout Anniversary Essays. In: Fischel, W.A. (ed). 2006. e
Tiebout Model at Fifty: Essays in Public Economics in Honor of Wallace Oates. Cambridge, Mass: Lincoln Institute of
Land Policy: 1–20.
Fisman, R. & Wei, S. 2004. Tax rates and tax evasion: Evidence from ‘missing imports’ in China. Journal of Political
Economy, 112(2): 471–500.
Fjeldstad, O. 2004. What’s trust got to do with it? Non-payment of service charges in local authorities in South Africa.
Journal of Modern African Studies, 42(4): 539–562.
Fourie, F.C.v.N. 1997. How to think and reason in macroeconomics. Kenwyn: Juta.
Frankel, J., Smit, B.W. & Sturzenegger, F. 2006. South Africa: Macroeconomic challenges after a decade of success. CID
Working Paper 133. Cambridge, Mass: Harvard University (Center for International Development).
Franzsen Commission. 1968. First Report of the Commission of Inquiry into scal and monetary policy in South Africa
(Chairman: D.G. Franzsen). RP 24/1969. November 1968. Pretoria: Government Printer.
Freeman, A.M. 1983. Intermediate microeconomic analysis. New York: Harper.
Frey, B.S. 1978. Modern political economy. Oxford: Martin Robertson.
Fullerton, D. 1999. Marginal effective tax rate. Research Report, originally published in e Encyclopedia of Taxation and
Tax Policy, edited by Cordes, J., Ebel, R. and Gravelle, J. Washington, DC: Urban Institute Press. [Online]. Available:
https://www.urban.org/sites/default/ les/alfresco/publication-pdfs/1000538-Marginal-Effective-Tax-Rate.PDF
[Accessed 15 June 2018].
Gale, W.G., & Scholz, J.K. 1994. IRAs and household savings. American Economic Review, 84: 1233–1260.
Gcabo, R. & Robinson, Z. 2007. Tax compliance and behavioural response in South Africa: An alternative investigation.
South African Journal of Economic and Management Sciences, 10(3): 357–370.
Giertz, S.H. 2009. e elasticity of taxable income: In uences on economic efficiency and tax revenues, and implications
for tax policy. In Viard, A.D. (ed). In Tax Policy Lessons from the 2000s. Washington DC: AEI Press: 101–136.
Gillis, M., Shoup, C.S. & Sicat, G.P. (eds). 1990. Value added taxation in developing countries. Washington, DC: World
Bank.
Glaeser, E.L. & Gottlieb, J.D. 2009. e wealth of cities: Agglomeration economies and spatial equilibrium in the United
States. Journal of Economic Literature, 47(4): 983–1028.
Global Finance. 2019. Income Tax Rates. [Online]. Available: https://www.gfmag.com/global-data/economic-
data/personal-income-tax-rates?page=2 [Accessed 17 October 2019].
Go, D.S., Kearney, M., Robinson, S. & ierfelder, K. 2005. An analysis of South Africa’s value added tax. World Bank
Policy Research Working Paper 3671: 1–21.
Goudge, J., Gilson, L., Russel S., Gumede, T. & Mills, A. 2009. Affordability, availability and acceptability barriers to health
care for the chronically ill: Longitudinal case studies from South Africa. BMC Health Services Research, 9(75): 1–18.
Grabowski, R. 1994. e successful developmental state: Where does it come from? World Development, 22(3): 413–433.
Grobler, C. & Stuart, I. 2007. Health care provider choice. South African Journal of Economics, 75(2): 327–350.
Grosh, M. 1994. Administering targeted social programs in Latin America. World Bank Regional and Sectoral Studies No.
12808. Washington, DC: World Bank.
Gupta, S., Schiller, C. & Ma, H. 1999. Privatisation, social impact and social safety nets. IMF Working Paper, WP/99/68.
Washington, DC: International Monetary Fund.
Gupta, S., Verhoeven, M. & Tiongson, E. 1999. Does higher government expenditure buy better results in education and
health care? IMF Working Paper 99/21. Washington, DC: International Monetary Fund.
Gustafsson, M. 2011. e when and how of leaving school: e policy implications of new evidence on secondary
schooling in South Africa. Stellenbosch Economic Working Papers 09/11. Stellenbosch: Stellenbosch University
(Department of Economics & Bureau of Economic Research).
Gustafsson, M. & Patel, F. 2006. Undoing the apartheid legacy: Pro-poor spending shifts in the South African public
school system. Perspectives in Education, 24(2): 65–77.
Hall, R.E. & Rabushka, A. 1983. Low tax, simple tax, at tax. New York: McGraw-Hill.
Hall, R.E. & Rabushka, A. 1995. e at tax. Stanford: Hoover Institution Press.
Hamilton, B.W. 1975. Zoning and property taxation in a system of local governments. Urban Studies, 12: 205–211.
Harberger, A.C. 1962. e incidence of the corporation income tax. Journal of Political Economy, 70(3): 215–240.
Harrod, R. 1952. Economic essays. New York: Harcourt, Brace & Co.
Hau er, A. 2001. Taxation in a global economy. Cambridge: Cambridge University Press.
Hayek, F.A. 1960. e constitution of liberty. London: Routledge & Kegan Paul.
Heilbrun, J & McGuire, P.A. 1987. Urban Economics and Public Policy. New York: St. Martin’s Press.
Hemming, R. & Kell, M. 2001. Promoting scal responsibility: Transparency, rules and independent scal authorities.
Bank of Italy Workshop on Fiscal Rules. Peruglia, Italy: 1–3 February 2001.
Henderson, J. V. 1988. Urban Development: eory, Fact and Illusion. New York: Oxford University Press.
Heracleous, L. 1999. Privatisation: Global trends and the Singapore experience. International Journal of Public
Economics, 12(5).
Hochman, H.M. & Rodgers, J.D. 1969. Pareto optimal redistribution. American Economic Review, 59(4): 531–541.
Hoddinott, J. 2007. Social protection: To target or not to target. IDS Bulletin, 38(3): 90–94.
Hofmeyr, A., Burns, J. & Visser, M. 2007. Income inequality, reciprocity and public good provision: An experimental
analysis. South African Journal of Economics, 75(3): 508–520.
http://www.theguardian.com/business/2011/aug/15/warren-buffett-higher-taxes-super-rich [Accessed 30 January
2015].
Hyman, D.N. 1999. Public Finance. 6th ed. Fort Worth, Texas: e Dryden Press.
Hyman, D.N. 2010. Public nance. 10th ed. Mason: South-Western Cengage Learning.
Inchauste, G., Lustig, N., Maboshe, M., Pur eld, C. & Woolard, I. 2015. e distributional impact of scal policy in South
Africa. Policy Research Working Paper 7194. Washington, DC: World Bank. Available:
http://documents.worldbank.org/curated/en/502441468299632287/pdf/WPS7194.pdf [Accessed 14 July 2018].
International Monetary Fund (IMF). 2000. Republic of Poland: Article IV Consultation – Staff Report. Washington, DC:
IMF. [Online]. Available: http://www.imf.org/external/pubs/ft/scr/2001/cr0156.pdf. [Accessed 12 June 2008].
International Monetary Fund (IMF). 2001. Government nance statistics yearbook. Washington, DC: IMF. Available:
http://www.imf.org/external/np/mae/pdebt/2000/eng/guide.pdf [Accessed 12 June 2008].
International Monetary Fund (IMF). 2002. Assessing Sustainability. Policy Development and Review Department.
Washington, DC: IMF. [Online]. Available: https://www.imf.org/external/np/pdr/sus/2002/eng/052802.pdf
[Accessed 27 August 2018].
International Monetary Fund (IMF). 2006. Government nance statistics yearbook, 2006: Database and Browser, March
2006. Washington, DC: IMF.
International Monetary Fund (IMF). 2008. Government nance statistics yearbook. Washington, DC: IMF.
International Monetary Fund (IMF). 2010. World Economic Outlook. October. [Online]. Available: http://www.imf.org./
[Accessed 13 February 2011].
International Monetary Fund (IMF). 2011. Fiscal Monitor Update. January. Washington, DC: IMF. [Online]. Available:
http://www.imf.org/external/pubs/ft/fm/2011/01/update/fmindex.htm [Accessed 30 March 2011].
International Monetary Fund (IMF). 2013. United Kingdom: Technical Assistance Report – Assessment of HMRC’s Tax Gap
Analysis. IMF Country Report No. 13/314. [Online]. Available:
https://www.imf.org/external/pubs/ft/scr/2013/cr13314.pdf [Accessed 23 August 2018].
International Monetary Fund (IMF). 2015. South Africa Technical Assistance Report – Revenue Administration Gap
Analysis Program – the Value-Added Tax Gap. [Online]. Available:
https://www.imf.org/external/pubs/ft/scr/2015/cr15180.pdf [Accessed 29 May 2018].
International Monetary Fund (IMF). 2018. World Economic Outlook April 2018. Available:
https://www.imf.org/external/pubs/ft/weo/2018/01/weodata/index.aspx.
International Monetary Fund and World Bank. 2001. Guidelines for public debt management. Prepared by the Staff of the
International Monetary Fund and the World Bank.
International Social Security Association. 2018. Understanding social security. [Online]. Available:
https://ww1.issa.int/Understanding%20social%20security [Accessed 3 December 2018].
Ipsos. 2018. 2018 National Tobacco Market Study Executive Summary Report. [Online]. Available:
http://www.tobaccosa.co.za/wp-content/uploads/IPSOS-2018-National-Tobacco-Market-Study-Executive-
Summary-1.pdf [Accessed 24 October 2018].
James, S. 2016. Tax incentives around the world. In: Tavares-Lehman, A.T. et al. (ed). Rethinking Investment Incentives:
Trends and Policy Options. New York: Columbia University Press, 153–175.
Janisch, C.A. 1996. An analysis of the burdens and bene ts of taxes and government expenditure in the South African
economy for the year 1993/94. Pietermaritzburg: University of Natal. (MCom thesis)
Jappelli, T. & Pistaferri, L. 2002. Tax incentives for household saving and borrowing. Working Paper 83, Centre for Studies
in Economics and Finance. [Online]. Available: http://ideas.repec.org/p/sef/csefwp/83.html [Accessed 12 June
2008].
Jorgenson, D.W. and Yun, K. 2001. Investment, Volume 3, Lifting the burden: Tax reform, the cost of capital, and US
economic growth. Cambridge, MA: MIT Press.
Kagel, J.H. & Roth, A.E. 1995. e handbook of experimental economics. Princeton, NJ: Princeton University Press.
Kahn, A.E. 1988. e economics of regulation: principles and institutions. Cambridge, MA: MIT Press.
Kahn, M.H. 2006. Determinants of corruption in developing countries: e limits of conventional economics. In: Rose-
Ackerman, S. (ed). International handbook on the economics of corruption. Cheltenham: Edward Elgar Publishing.
Katz Commission. 1994. Interim Report of the Commission of Inquiry into certain aspects of the tax structure of South
Africa (Chairman: M.M. Katz). Pretoria: Government Printer.
Katz Commission. 1995. ird Interim Report of the Commission of Inquiry into certain aspects of the tax structure of
South Africa (Chairman: M.M. Katz). Pretoria: Government Printer.
Katz Commission. 1997a. Fourth Interim Report of the Commission of Inquiry into certain aspects of the tax structure of
South Africa (Chairman: M.M. Katz). Pretoria: Government Printer.
Katz Commission. 1997b. Fifth Interim Report of the Commission of Inquiry into certain aspects of the tax structure of
South Africa (Chairman: M.M. Katz). Pretoria: Government Printer.
Katz Commission. 1998. Eighth Interim Report of the Commission of Inquiry into certain aspects of the tax structure of
South Africa (Chairman: M.M. Katz): e Implications of Introducing a Land Tax in South Africa. Pretoria:
Government Printer.
Keen, M. 1989. Pareto-improving indirect tax harmonisation. European Economic Review, 33(1): 1–12.
Keen, M. 2012. e anatomy of the VAT. National Tax Journal, 66(2): 423–466.
Keen, M., Kim, Y. & Varsano, R. 2008. e ‘ at tax(es)’: Principles and experience. International Tax Public Finance, 15:
712–751.
Keen, M. & Mansour, M. 2009. Revenue mobilization in Sub-Saharan Africa: Challenges from globalization. IMF Working
Paper, WP/09/157. Washington, DC: International Monetary Fund: 1–47.
Kemp, J.H. 2017. e elasticity of taxable Income: the case of South Africa. Economic Research Southern Africa (ERSA).
Working Paper 702: 1–37. ). [Online]. Available:
 https://econrsa.org/system/ les/publications/working_papers/working_paper_702.pdf [Accessed 4 June 2018].
Kessides, I.N. 2005. Infrastructure privatisation: Promises and perils. World Bank Research Observer, 20(1): 81–108.
Khalilzadeh-Shirazi, J. & Shah, A. 1991. Tax policy in developing countries. Washington: World Bank.
Klasen, S. 1996. Poverty and inequality in South Africa. Mimeograph. (Accepted for publication in Social Indicator
Research.) Cambridge: Centre for History and Economics: Kings College.
Klemm, A. & Van Parys, S. 2010. Empirical evidence on the effects of tax incentives. Working Paper, 2010/673.
Universiteit Gent, Faculteit Economie en Bedrijfskunde, September 2010: 1–31.
Kopits, G. & Craig, J. 1998. Transparency in government operations. IMF Occasional Paper 158. Washington, DC:
International Monetary Fund.
Kopits, G. & Symansky, S. 1998. Fiscal policy rules. IMF Occasional Paper 162. Washington DC: International Monetary
Fund.
Kopits, G. 2001. Fiscal rules: Useful policy framework or unnecessary ornament? IMF Working Paper, WP/01/145.
Washington DC: International Monetary Fund.
Kruger, J.J. 1992. State provision of social security: Some theoretical, comparative and historical perspectives with reference
to South Africa. Stellenbosch: University of Stellenbosch. (MCom thesis)
Lachman, D. & Bercuson, K. 1992. Economic policies for a new South Africa. IMF Occasional Paper 91. Washington, DC:
International Monetary Fund.
Laffer, A.2004. e Laffer Curve: Past, Present, and Future. Backgrounder #1765 on Taxes. e Heritage Foundation.
[Online]. Available: http://www.heritage.org/research/reports/2004/06/the-laffer-curve-past-present-and-furure
[Accessed 2 February 2015].
Lambsdorff, J.G. 2002. Corruption and rent-seeking. Public Choice, 113: 97–125.
Lancaster, K. & Lipsey, R.G. 1957. e general theory of the second best. Review of Economics and Statistics, 24: 11–32.
Lane, T. 1993. Market discipline. IMF Staff Papers, 40(1): 53–88.
Leach, D.F. 1997. Concentration–pro ts monopoly versus efficiency debate: South African evidence. Contemporary
Economic Policy, 15(2): 12–23.
Leftwich, A. 1995. Bringing politics back in: Towards a model of the developmental state. e Journal of Development
Studies, 31(3): 400–427.
Leibbrandt, M., Lilenstein, K., Shenker, C. & Woolard, I. 2013. e in uence of social transfers on labour supply: a South
African and international review. SALDRU Working Paper 112. Cape Town: University of Cape Town (Southern
African Labour and Development Research Unit).
Leibbrandt, M., Poswell, L., Naidoo, P., Welch, M. & Woolard, I. 2004. South African poverty and inequality: Measuring
the changes. Paper for Transformation Audit of Institute of Justice and Reconciliation. Cape Town. [Online].
Available: http://www.transformationaudit.co.za/articles/Poverty.pdf [Accessed November 2004].
Leibbrandt, M., Woolard, I., Finn, A. & Argent, J. 2010. Trends in South African income distribution and poverty since the
fall of apartheid. OECD Social, Employment and Migration Working Papers 101. Paris: Organisation for Economic
Co-operation and Development.
Leibenstein, H. 1966. Allocative efficiency versus X-efficiency. American Economic Review, 56(3): 392–415.
Leibenstein, H. 1978. General X-efficiency theory and economic development. New York: Oxford University Press.
Leicester, A., Levell, P. & Rasul, I. 2012. Tax and bene t policy: insights from behavioural economics. IFS Commentary
C125, Institute for Fiscal Studies, UK. [Online]. Available:
http://www.ucl.ac.uk/~uctpimr/research/IFScomm125.pdf [Accessed 29 May 2018].
Leistner, G.M.E. 1968. Table insert. Africa Institute Bulletin, VI(6): 175–7.
Lemboe, C.J. 2010. Cigarette taxes and smuggling in South Africa: Causes and consequences. Stellenbosch: University of
Stellenbosch. (MCom thesis)
Lim, D. 1993. Recent trends in the size and growth of government in developing countries. In: Gemmell, N. (ed). 1993.
e growth of the public sector: eories and international evidence. Aldershot: Edward Elgar.
Lindahl, E. 1958. Just taxation – a positive solution. In: Musgrave, R.A. & Peacock, A.T. (eds). Classics in the theory of
public nance. New York: St Martin’s Press.
Lindauer, D.L. & Velenchik, A.D. 1992. Government spending in developing countries: Trends, causes, and
consequences. World Bank Research Observer, 7(1): 59–78.
Lipsey, R.G. & Chrystal, K.A. 1995. An introduction to positive economics. 8th ed. New York: Oxford University Press.
Lund Committee. 1996. Report of the Lund Committee on child and family support. Pretoria: Government Printer.
Lusinyan, L. & ornton, I. 2007. e revenue-expenditure nexus: Historical evidence for South Africa. South African
Journal of Economics, 75(3): 496–507.
Lustig, N. & Higgins, S. 2017. e CEQ assessment: Measuring the impact of scal policy on inequality and poverty. In:
Lustig, N. (ed). Commitment to equity handbook: Estimating the impact of scal policy on inequality and poverty.
 Washington, DC: Brookings Institution Press & New Orleans, LA: Tulane University (CEQ Institute): 1–50.
Manuel, T.A. 2003. Finding the right path. Finance and Development, 40(3): 18–20.
Manyema, M., Veerman, L.J., Chola, L., Tugendhaft, A., Sartorius, B., et al. 2014. e Potential Impact of a 20% Tax on
Sugar-Sweetened Beverages on Obesity in South African Adults: A Mathematical Model. PLoS ONE, 9(8):
e105287.doi:10.1371/journal.pone.0105287. [Online]. Available: http://journals.plos.org/plosone/article?
id=10.1371/journal.pone.0105287 [Accessed 17 May 2018].
Margo Commission. 1987. Report of the Commission of Inquiry into the tax structure of the Republic of South Africa
(Chairman: C.S. Margo). RP34/1987. Pretoria: Government Printer.
Marshall, A., 1890. e Principles of Economics. London: Macmillan.
Mather, D., Britz, P.J., Hecht, T. & Sauer, W.H.H. (eds). 2003. An Economic and Sectoral Study of the South African Fishing
Industry: Volume 1. Economic and regulatory principles, survey results, transformation and socio-economic impact.
Department of Ichthyology and Fisheries Science, and Department of Economics and Economic History. Rhodes
University: Grahamstown.
Mbewe, S. & Woolard, I. 2016. Cross-Sectional Features of Wealth Inequality in South Africa: Evidence from e National
Income Dynamics Study. Cape Town: SALDRU, University of Cape Town. SALDRU Working Paper Number 185/
NIDS Discussion Paper 2016/12. Available: http://www.nids.uct.ac.za/images/papers/2016_12_NIDSW4.pdf
[Accessed 29 June 2018].
McConnell, C.R. & Brue, S.L. 2005. Economics, principles, problems, and policies. 16th ed. McGrawHill: Boston.
McGrath, M.D. 1983. e distribution of personal income in South Africa in selected years over the period from 1945 to
1980. Durban: University of Natal. (PhD thesis)
McGrath, M.D., Janish, C. & Horner, C. 1997. Redistribution through the scal system in the South African economy.
Conference of the Economic Society of South Africa, Potchefstroom, 8 September 1997.
McManus, J. & Warren, N. 2006. e case for measuring tax gap. eJournal of Tax Research, 4(1): 61–79. [Online]. Available:
http://www.atax.unsw.edu.au/ejtr/ [Accessed 13 December 2010].
Meghir, C. & Phillips, D. 2010. Labour supply and taxes. In: Dimensions of tax design: e Mirrlees review chaired by Sir
James Mirrlees, Oxford: Oxford University Press.
Meier, N.G., Breitenbach, M. & Kekana, R.D. 2008. An economic appraisal of the impact of traffic diversion – the N1 toll
road and its alternative. South African Journal of Economics, 76: 4.
Meintjes, C.J. 1992. Impediments on the labour absorption capacity of the South African economy. EBM Research
Conference, Port Elizabeth, 30 November 1992.
Melo, P.C, Graham, D.J. & Noland, R. 2009. A meta-analysis of estimates of urban agglomeration economies. Regional
Science and Urban Economics, 39: 332–243.
Meltzer, A.H. & Richard, S.F. 1981. A rational theory of the size of government. Journal of Political Economy, 89(5): 914–
927.
Merriman, D. 2003. Understanding, measuring and combating tobacco smuggling. In: World Bank Economics of tobacco
Toolkit. Washington, DC: World Bank, Chapter 7.
Meth, C., Naidoo, R. & Shipman, B. 1996. Report of the Task Team on unemployment insurance and related coverage
issues. Pretoria: Department of Labour.
Minister of Finance. 2019. Budget Speech. Pretoria: National Treasury.
Ministry of Finance. 1999. Medium Term Budget Policy Statement 1998. Pretoria.
Mintz, J.M. & Seade, J. 1991. Cash ow or income: e choice of base for company taxation. e World Bank Observer,
6(2): 177–190.
Mohapatra, S., Patra, J., Popova, S., Duhig, A. & Rehm, J. 2010. Social cost of heavy drinking and alcohol dependence in
high-income countries. International Journal of Public Health, 55(3): 149–57.
Mohr, P.J. 1994. Can South Africa avoid macroeconomic populism? Development Southern Africa, 11(1): 1–32.
Mohr, P. & associates. 2015. Economics for South African students. 5th ed. Pretoria: Van Schaik Publishers.
Mohr, P.J., Fourie, L.J. & associates. 2008. Economics for South African students. 4th ed. Pretoria: Van Schaik Publishers.
Moody’s Investors Service. 2012. Rating Action: Moody’s downgrades South Africa’s government bond rating to Baa1;
outlook remains negative. [Online]. Available: https://www.moodys.com/research/Moodysdowngrades-South-
Africas-government-bond-rating-to-Baa1-outlook-- PR_256159 [Accessed 28 January 2015].
Moreno-Serra, R. & Smith, P. 2012. Does progress towards universal health coverage improve population health? e
Lancet, 380(9845): 917–923.
Mosley, L. 2000. Room to move: International nancial markets and national welfare states. International Organization,
54(4): 737–773.
Motala, S. 2006. Education resourcing in post-apartheid South Africa: e impact of nance equity reforms in public
schooling. Perspectives in Education, 24(2): 79–93.
Mouton Committee. 1992. Report of the Committee of Investigation into a retirement provision system for South Africa.
Volumes 1 & 2. Pretoria: Department of Finance.
Muchapondwa, E., Carlsson, F. & Kohlin, G. (2008). Wildlife management in Zimbabwe: Evidence from a contingent
valuation study. South African Journal of Economics, 76(2).
Mukand, S. 1999. Globalization and the ‘con dence game’. Discussion Paper 99–24. Medford, MA: Tufts University,
Department of Economics.
Mullis, I.V.S., Martin, M.O., Foy, P. & Hooper, M. 2017. International results in reading — PIRLS 2016. Boston: Boston
College (Lynch School of Education TIMSS & PIRLS International Study Centre).
Musgrave, R.A. 1959. e theory of public nance: A study in public economy. Tokyo: McGraw-Hill.
Musgrave, R.A. 1969. Fiscal systems. New Haven: Yale University Press.
Musgrave, R.A. 1983. Who should tax, where and what? In: McClure, C.E. (ed). Tax assignment in federal countries.
Canberra: Australian National University Press.
Musgrave, R.A. 1987. Public nance. In: Eatwell, J., Milgate, M. & Newman, P. (eds). 1987. e new Palgrave. New York:
McMillan Press: 1055–1061.
Musgrave, R.A. & Musgrave, P.B. 1989. Public nance in theory and practice. 5th ed. New York: McGraw-Hill.
Mutasa, G. 2010. e disincentive effects of disability grant on labour supply in South Africa: a cohort analysis. Cape Town:
University of Cape Town (Development Policy Research Unit).
Nahman, A. & Antrobus, G. 2005. Trade and the environmental Kuznets curve: A literature review. South Africa Journal
of Economics, 73(1): 105–120.
National Treasury. 1996. Proposed Framework of Philosophies and Principles of Debt, Cash and Risk Management.
Pretoria.
National Treasury. 2001. Intergovernmental scal review 2001. Pretoria: National Treasury.
National Treasury. 2002. Budget review 2002. Pretoria: Government Printer.
National Treasury. 2004a. Budget review 2004. Pretoria: Government Printer. [Online]. Available:
http://www.treasury.gov.za [Accessed 31 March 2008].
National Treasury. 2004b. Intergovernmental scal review 2004. Pretoria: National Treasury.
National Treasury. 2004c. Trends in intergovernmental nances: 2000/01–2006/7. Pretoria: Government Printer.
National Treasury. 2005a. Budget Review 2005. Pretoria: Government Printer. [Online]. Available;
http://www.treasury.gov.za [Accessed 17 June 2008].
National Treasury. 2005b. Estimates of national expenditure 2005. Pretoria: Government Printer.
National Treasury. 2005c. Intergovernmental scal review 2005. Pretoria: National Treasury.
National Treasury. 2007a. Budget review 2007. Pretoria: Government Printer. [Online]. Available:
http://www.treasury.gov.za [Accessed 17 June 2008].
National Treasury. 2007b. Estimates of national expenditure 2007. Pretoria: National Treasury.
National Treasury. 2007c. Local government budgets and expenditure review 2001/02 – 2007/08. Pretoria: Government
Printer. [Online]. Available: http://www.treasury.gov.za [Accessed 17 June 2008].
National Treasury. 2007d. Medium-term budget policy statement 2007. Pretoria: Government Printer.
National Treasury. 2008a. Budget review 2008. Pretoria: Government Printer. [Online]. Available:
http://www.treasury.gov.za [Accessed 17 June 2008].
National Treasury. 2008b. Estimates of national expenditure 2008. Pretoria: Government Printer. [Online]. Available:
http://www.treasury.gov.za [Accessed 17 June 2008].
National Treasury. 2009a. Budget review 2009. Pretoria: Government Printer.
National Treasury. 2009b. Estimates of national expenditure 2009. Pretoria: National Treasury.
National Treasury. 2010a. Budget review 2010. Cape Town: FormeSet Printers.
National Treasury. 2010b. Estimates of national expenditure 2010. Pretoria: National Treasury.
National Treasury. 2010c. Local government adopted capital and operating expenditure budgets for the 2010/11 Medium
Term Revenue and Expenditure Framework (MTREF), 2 December 2010, [Online]. Available:
http://www.treasury.gov.za [Accessed 1 March 2011].
National Treasury. 2011a. Budget review 2011. Pretoria: Government Printer. [Online]. Available:
http://www.treasury.gov.za [Accessed 1 March 2011].
National Treasury. 2011b. Estimates of national expenditure 2011. Pretoria: National Treasury.
National Treasury. 2012. Estimates of national expenditure 2012. Pretoria: National Treasury.
National Treasury. 2013. Estimates of national expenditure 2013. Pretoria: National Treasury.
National Treasury. 2014a. Budget Review 2014. Pretoria: National Treasury.
National Treasury. 2014b. Strengthening retirement savings: an overview of proposals announced in the 2012 Budget.
Pretoria: National Treasury.
National Treasury. 2015a. Budget Review 2015. Pretoria: National Treasury.
National Treasury. 2015b. Provincial budgets and expenditure review, 2010/11–2016/17. Pretoria: National Treasury.
National Treasury. 2015c. Budget Review 2015. Pretoria. Government Printer. [Online]. Available:
http://www.treasury.gov.za/documents/national%20budget/2015/review/FullReview.pdf [Accessed 16 June 2018].
National Treasury. 2016. Budget Review 2016. Pretoria: National Treasury.
National Treasury. 2017a. Budget Review 2017. Pretoria. [Online]. Available: http://www.treasury.gov.za [Accessed 4 June
2018].
National Treasury. 2017b. Estimates of national expenditure 2017. Pretoria: National Treasury.
National Treasury. 2017c. Local government adopted capital and operating expenditure budgets for the 2017/18 Medium
Term Revenue and Expenditure Framework (MTREF), 7 November 2017, [Online]. Available:
http://www.treasury.gov.za [Accessed 1 March 2018].
National Treasury. 2018a. Budget review 2018. Pretoria: Government Printer. [Online]. Available:
http://www.treasury.gov.za [Accessed 1 March 2018].
National Treasury. 2018b. Estimates of national expenditure 2018. Pretoria: National Treasury.
National Treasury. 2018c. Final Report: Forensic Investigation into Various Allegations at Transnet And Eskom: Tender
Number NT 022-2016 RFQ 026-2017. Available:
http://www.treasury.gov.za/comm_media/press/2018/Final%20Report%20-%20Fundudzi%20-
%20Eskom%2015112018.pdf [Accessed 5 March 2019].
National Treasury. 2019a. Budget Review 2019. Government Printer. [Online]. Available:
http://www.treasury.gov.za/documents/national%20budget/2019/review/FullBR.pdf [Accessed 1 March 2019].
National Treasury. 2019b. Estimates of national expenditure 2019. Pretoria: National Treasury.
National Treasury. 2019c. Media Statement Local Government Revenue and Expenditure: Fourth Quarter Local
Government Section 71 Report for the period 1 July 2018 – 30 June 2019. Available:
http://mfma.treasury.gov.za/Media_Releases/s71/1819/4th_1819/Documents/00.%20Media%20statement%20-
%204th%20Q%20S71%20Publication%20-%2003%20Sept%202019%20%28 nal%29.pdf [Accessed 8 November
2019].
National Treasury and SARS (South African Revenue Service (SARS). 2013. Tax Statistics 2013. National Treasury:
Pretoria.
National Treasury and South African Revenue Services (SARS). 2017. 2017 Tax Statistics. Joint publication between
National Treasury and SARS. [Online]. Available:
http://www.sars.gov.za/AllDocs/Documents/Tax%20Stats/Tax%20Stats%202017/Tax%20Stats%202017%20Publicati
on.pdf [Accessed 29 May 2018].
National Treasury and South African Revenue Services (SARS). 2018. 2018 Tax Statistics. Joint publication between
National Treasury and SARS. [Online]. Available:
http://www.sars.gov.za/AllDocs/Documents/Tax%20Stats/Tax%20Stats%202018/Tax%20Stats%202018.pdf
[Accessed 1 March 2019].
Native Economics Commission. 1932. Report of the Native Economic Commission 1930–32. Pretoria: Government
Printer.
Nattrass, J. 1988. e South African economy: Its growth and change. 2nd ed. Cape Town: Oxford University Press.
Nattrass, N. 1992. e ANC’s economic policy: A critical perspective. In: Schirre, R. 1992. Wealth or poverty: Critical
choices for South Africa. Cape Town: Oxford University Press.
Nelslon, R.R. & Winter, S.E. An evolutionary theory of economic change. Cambridge, MA: Harvard University Press.
Newbery, D. & Stern, N. (eds). 1987. e theory of taxation for developing countries. Washington, DC: World Bank.
Nicholson, W. & Snyder, C. 2010. eory and application of intermediate microeconomics. 11th ed. South-Western,
Cengage Learning.
Niskanen, W.A. 1971. Bureaucracy and representative government. Chicago: Aldine-Atherton.
Norregaard, J. & Khan, T.S. 2007. Tax policy: Recent trends and coming challenges. IMF Working Paper, WP/07/274.
Washington, DC: International Monetary Fund: 1–59.
North, C.C. 1991. Institutions. Journal of Economic Perspectives, 5: 97–112.
Nozick, R. 1974. Anarchy, state and utopia. New York: Basic Books.
Nyamongo, M.E. & Schoeman, N.J. 2007. Tax reform and the progressivity of personal income tax in South Africa. South
African Journal of Economics, 75(3): 478–495.
Oates, W.E. 1972. Fiscal federalism. New York: Harcourt Brace Jonvanovich.
Oates, W.E. 2006. e many faces of the Tiebout model. In: Fischel, W.A. (ed). 2006. e Tiebout Model at Fifty: Essays in
Public Economics in Honor of Wallace Oates. Cambridge, MA: Lincoln Institute of Land Policy: 21–45.
Oberholzer, R. 2008. Attitudes of South African taxpayers towards taxation: A pilot study. Accountancy Business and the
Public Interest, 7(1): 44–69. [Online]. Available:
 http://www.adobe.com/supportservice/custsupport/LIBRARY/acwin.htm [Accessed 9 December 2010].
Organisation for Economic Co-operation and Development (OECD). 1988. Taxation of net wealth, capital transfers and
capital gains on individuals. Paris: OECD.
Organisation for Economic Co-operation and Development (OECD). 2017. Tax Policy Reforms 2017: OECD and Selected
Partner Economies, Paris: OECD Publishing. [Online]. Available: http://dx.doi.org/10.1787/9789264279919-en
[Accessed 15 August 2018].
Organisation for Economic Co-operation and Development (OECD). Stat. 2019. Available: https://stats.oecd.org/#
[Accessed 7 March 2019].
Orthofer, A. 2016. Wealth inequality in South Africa: Evidence from survey and tax data. Working Paper 15. Research
Project on Employment, Income Distribution and Inclusive Growth (REDI3x3). Available:
http://www.redi3x3.org/sites/default/ les/Orthofer%202016%20REDI3x3%20Working%20Paper%2015%20-
%20Wealth%20inequality.pdf [Accessed 29 June 2018].
Ouattara, A.D. 1997. Globalization’s challenges for Africa. IMF Survey, 26(11): 177.
Palmer, N., Mills, A., Wadee, H., Gilson, L. & Schneider, H. 2003. A new face for private providers in developing countries:
What implications for public health? Bulletin of the World Health Organization, 81(4): 292–297.
Parker, D. 2002. Economic regulation: A review of issues. Annals of Public and Cooperative Economics, 73(4).
Parker, D. & Kirkpatrick, C. 2005a. Privatisation in developing countries: A review of the evidence and the policy lessons.
Journal of Development Studies, 41(4): 513–541.
Parker, D. & Kirkpatrick, C. 2005b. Regulating prices and pro ts in utility industries in low-income countries: Rate of
return, price cap or sliding-scale regulation? International Journal of Public Sector Management, 18(3): 241–255.
Parliamentary Budget Office. 2017. Analysis of Eskom’s nancial position. Cape Town: Parliament of the Republic of
South Africa. [Online]. Available:
https://www.parliament.gov.za/storage/app/media/PBO/Analysis_of_Eskom_ nances_Report_to_SCOA_presente
d_8_March_2017.pdf [Accessed 13 June 2018].
Parr, J.B. 2002. Agglomeration economies: Ambiguities and confusions. Environment and Planning, 34: 717–731.
Parry, C. 2009. Supply versus demand reduction: Best practices for addressing alcohol abuse in the South African
Context. 1st Biennial Regional Liquor Regulation Conference, Johannesburg. 25–26 March, 2009.
Peacock, A.T. & Wiseman, J. 1967. e growth of public expenditure in the United Kingdom. London: George Allen &
Unwin.
Pearce D.W. (ed). 1986. e MIT Dictionary of Modern Economics. Cambridge, Mass.: MIT Press.
Pienaar, W.J. 2008. Economic valuation of the proposed road between Gobabis and Grootfontein, Namibia. South
African Journal of Economics, 76: 4.
Portfolio Committee on Basic Education. 2016. Overview and analysis of SAQMEC IV study results. Cape Town:
Parliament of the Republic of South Africa. [Online]. Available: http://pmg-assets.s3-website-eu-west-
1.amazonaws.com/160913overview.pdf [Accessed 9 July 2019].
Prest, A.R. & Turvey, R. 1978. Cost-bene t analysis: A survey. In: AEA and RES. Surveys of economic theory. London:
MacMillan.
Quartz Africa. 2018. How two South African women stopped Zuma and Putin’s $76 billion Russian nuclear deal.
[Online]. Available: https://qz.com/1260877/how-two-south-african-women-stopped-zuma-and-putins-76-
billion-nuclear-deal/ [Accessed 14 June 2018].
Ramsey, F.P. 1927. A contribution to the theory of taxation. Economic Journal, 37: 47–61.
Ranchhod, V. 2006. e effect of the South African old age pension on labour supply of the elderly. South African Journal
of Economics, 74(4): 725–744.
Rankin, N. 2006. e Regulatory Environment in SME’s: Evidence from a South African Firm Level Data: Development
Policy Research Unit: University of Cape Town: Working Paper 06/113. [Online]. Available:
http://www.health.uct.ac.za/sites/default/ les/image_tool/images/36/DPRU%20WP06-113.pdf [Accessed 15 June
2018].
Rawls, J. 1971. A theory of justice. Cambridge, Mass.: Harvard University Press.
Reddy, V., Prinsloo, C., Netshingani, T., Moletsane, R., Juan, A. & Janse van Rensburg, D. 2010. An investigation into
educator leave in the South African ordinary public schooling system. Pretoria: HSRC Press.
Reddy, V., Visser, M., Winnaar, L., Arends, F., Juan, A., Prinsloo, C.H. & Isdale, K. 2016. TIMSS 2015: Highlights of
mathematics and science achievement of Grade 9 South African learners. Pretoria: Human Sciences Research
Council.
Registrar of Pension Funds. 2005. Forty seventh annual report. Pretoria: Financial Services Board. [Online]. Available:
ftp://ftp.fsb.co.za/public/pension/Reports/PFAR2005.pdf [Accessed 8 April 2008].
Republic of Namibia. 2018. Estimates of Revenue, Income and Expenditure, 1 April 2018 to 31 March 2021. [Online].
Available: https://mof.gov.na/documents/35641/36556/Estimates+2018-19+%282%29.pdf/c3ac5462-03ad-27b0-
 0e4a-7b391cab122b [Accessed 7 March 2019].
Republic of South Africa. 1969. Bantu Taxation Act, No 92. Pretoria: Government Printer. [Laws.]
Republic of South Africa. 2003. Local Government: Municipal Finance Management Act, No 56 of 2003, Government
Gazette, 464, 13 February 2004. [Laws.]
Republic of South Africa. 2004. Local Government: Municipal Property Rates Act, No 6. Pretoria: Government Printer.
[Laws.]
Republic of South Africa. 2009. Money Bills Amendment Procedure and Related Matters Act, No 9 of 2010. Pretoria:
Government Printer. [Laws.]
Richardson, J. 1986. e local historian’s encyclopedia. 2nd ed. London: Historical Publications.
Romer, P.M. 1986. Increasing returns and long-run growth. Journal of Political Economy, 94.
Rosen, H.S. 2002. Public nance. 6th ed. New York: McGraw-Hill.
Rosen, H.S. & Gayer, T. 2008. Public nance. 8th ed. Boston: McGraw-Hill/Irvin.
Rosen, H.S. & Gayer, T. 2010. Public Finance. Global Edition. 10th ed. Berkshire: McGraw-Hill Education.
Rosen, H.S. & Gayer, T. 2014. Public Finance. 10th edition. New York: McGraw-Hill Irwin.
Rosenthal, L. 1983. Subsidies to the personal sector. In: Millward, R., Parker, D., Rosenthal, L., Sumner, M.T. & Topham,
N. (eds). 1983. Public sector economics. London and New York: Longman.
Rostow, W.W. 1971. Politics and the stages of growth. Cambridge: Cambridge University Press.
Sadka, E. & Tanzi, V. 1993. A tax on gross assets of enterprises as a form of presumptive taxation. IBFD Bulletin, 66–73
(February).
Saez, E. 2004. Reported Incomes and Marginal Tax Rates, 1960–2000: Evidence and Policy Implications. NBER Working
Paper No. 10273. Cambridge: National Bureau of Economic Research, 1–57.
Saez, E., Slemrod, J & Giertz, S.H. 2012. e elasticity of taxable income with respect to marginal tax rates: A critical
review. Journal of Economic Literature, 50 (1): 3–50.
Sahn, D.E., Dorosh, P.A. & Younger, S.D. 1997. Structural adjustment reconsidered: Economic policy and poverty in Africa.
Cambridge: Cambridge University Press.
Salvatore, D. 1998. International economics. 6th ed. New Jersey: Prentice-Hall.
Samuelson, P.A. 1954. e pure theory of public expenditure. Review of Economics and Statistics, 36(4): 387–389.
Samuelson, P.A. 1955. A diagrammatic exposition of a theory of public expenditure. Review of Economics and Statistics,
37(4): 350–356.
Sandford, C. & Hasseldine, J. 1992. e compliance costs of business taxes in New Zealand. Wellington: Institute of Policy
Studies, Victoria University.
Santos, G. 2005. Urban congestion charging: A comparison between London and Singapore. Transport Review, 25(5):
511–534.
Saunders, P. & Klau, F. 1985. e role of the public sector: Causes and consequences of the growth of government. OECD
Economic Studies, 4 (Special Edition): 11–239.
Schelling, T. 1985. Choice and consequence. Cambridge: Harvard University Press.
Schmalensee, R & Stavins, R.N. 2017. Lessons learned from three decades of experience with cap- and-trade. Review of
Environmental Economics and Policy, 11(1): 59–79.
Schumpeter, J. 1954. Capitalism, socialism and democracy. 4th ed. London: Allen & Unwin.
Schumpeter, J.A. 1987. Capitalism, socialism and democracy ( rst UK edition 1943). London: Unwin Paperbacks.
Shah, A. 1995. Introduction to scal incentives for investment and innovation. In: Shah, A. (ed). Fiscal incentives for
investment and innovation. Washington, DC: World Bank.
Shome, P. & Schutte, C. 1993. Cash ow tax. IMF Working Paper WP/93/2, Fiscal Affairs Department. Washington, DC:
International Monetary Fund.
Shoven, J.B. 1976. e incidence and efficiency effect of taxes on income from capital. Journal of Political Economy, 84:
1261–1283.
Siebrits, F.K. 1998. Government spending in an international perspective. In: Abedian, I. & Biggs, M. (eds). Economic
globalisation and scal policy. Cape Town: Oxford University Press.
Siebrits, F.K. 2018. How to use policymaking institutions to restrain leviathan. Draft paper commissioned by e Institute
of Economic Affairs, Ghana (IEA). Stellenbosch: Stellenbosch University (Department of Economics).
Siebrits, F.K. & Calitz, E. 2004a. Observations about scal policy in sub-Saharan Africa. Absa Economic Perspective,
Fourth Quarter: 2–9. [Online]. Available:
http://www.absa.co.za/ABSA/PDF les/EP_2004_Q1_English_Focus_Web_4.pdf
Siebrits, F.K. & Calitz, E. 2004b. Should South Africa adopt numerical scal rules? South African Journal of Economics,
72(4): 759–783.
Simons, H. 1936. Rules versus authorities in monetary policy. Journal of Political Economy, 44(1): 1–30.
Skinner, J. 1991. If agricultural land taxation is so efficient, why is it so rarely used? e World Bank Economic Review,
5(1): 113–133.
Slemrod, J. & Sorum, N. 1984. e compliance cost of the U.S. individual income tax system. National Tax Journal, 38(4).
Smith Committee. 1995. Report of the Committee on strategy and policy review of retirement provision in South Africa.
Pretoria: Department of Finance.
Sørensen, P.B. & Johnson, S.M. 2010. Taxing capital income: Options for reform in Australia. In: Melbourne Institute –
Australia’s future tax and transfer policy conference. Melbourne: Melbourne Institute of Applied Economic and
Social Research: 179–235.
Sørensen, P.B. 2004. Measuring taxes on capital and labor: an overview of methods and issues. In: Measuring the Tax
Burden on Capital and Labor. MA: MIT Press, 1–33.
Sørensen, P.B. 2005. Neutral taxation of shareholder income. International Tax and Public Finance, 12: 777–801.
Sørensen, P.B. 2007. e Nordic dual income tax: Principles, practices, and relevance for Canada. Canadian Tax Journal,
55(3): 557–602.
South Africa. 1998. Developing a culture of good governance. Report of the Presidential Commission on the reform and
transformation of the Public Service in South Africa. Pretoria: Government Printer.
South African Financial Sector. 2003. Financial Sector Charter. [Online]. Available: http://fsc.cmcnetworks,net
South Africa Foundation. 1998. Big and small business in South Africa: Where the twine meets. SAF Viewpoint,
December.
South African Reserve Bank. 2018a. Quarterly Bulletin, No 290, December 2018. [Online]. Available:
https://www.resbank.co.za/Publications/Detail-Item-View/Pages/Publications.aspx?sarbweb=3b6aa07d-92ab-
441f-b7bf-bb7dfb1bedb4&sarblist=21b5222e-7125-4e55-bb65-56fd3333371e&sarbitem=8985 [Accessed 25
February 2019].
South African Reserve Bank. 2018b. Quarterly Bulletin December 2018. Statistical tables. Pretoria: SARB [Online].
Available:
https://www.resbank.co.za/Lists/News%20and%20Publications/Attachments/8985/10Statistical%20tables%20%E2
%80%93%20National%20accounts.pdf [Accessed 6 March 2019].
South African Reserve Bank. Various years. Electronic data. [Online]. Available:
https://www.resbank.co.za/Research/Statistics/Pages/OnlineDownloadFacility.aspx [Accessed 20 August 2018].
South African Reserve Bank. Various years. Quarterly Bulletin (various issues). Pretoria: South African Reserve Bank.
[Online]. Available: https://www.resbank.co.za/Publications/QuarterlyBulletins/Pages/QuarterlyBulletins-
Home.aspx [Accessed 20 August 2018].
South African Revenue Services (SARS). 2008. Budget Tax Guide 2008/09. Pretoria: South African Revenue Services.
[Online]. Available:
http://www. nance.gov.za/documents/national%20budget/2008/guides/Budget%20Pocketguide%202008.pdf
[Accessed 12 June 2008].
South African Revenue Services (SARS). 2010a. Annual Report 2009–2010. Pretoria: Government Printer. [Online].
Available: http://www.sars.gov.za [Accessed 13 December 2010].
South African Revenue Service (SARS). 2010b. Reference guide – Environmental levy on carbon dioxide emissions of new
motor vehicles manufactured in the Republic, SE-EL-GU-03. Pretoria: SARS. [Online]. Available:
http://www.sars.gov.za [Accessed 16 March 2011].
South African Revenue Services (SARS). 2011. Pocket Tax Guide, Budget 2011. Pretoria: South African Revenue Services.
[Online]. Available: http://www.sars.gov.za [Accessed 28 February 2011].
South African Revenue Service (SARS). 2018. Annual Report 2017/18 South African Revenue Service. RP245/2018.
Pretoria: SARS [Online]. Available: http://www.sars.gov.za/AllDocs/SARSEntDoclib/AnnualReports/SARS-AR-
23%20-%20Annual%20Report%202017-2018.pdf [Accessed 1 March 2019].
South African Revenue Services (SARS). 2019. Budget 2019 Tax Guide. [Online]. Available:
http://www.treasury.gov.za/documents/national%20budget/2019/sars/Budget%202019%20Tax%20Guide.pdf
[Accessed 4 March 2019].
South African Social Security Agency. 2009. Annual report 2008/09. Pretoria: South African Social Security Agency.
South African Social Security Agency. 2017. You and your grants 2017/18. Pretoria: South African Social Security Agency.
South African Social Security Agency. 2018. Annual report 2017/18. Pretoria: South African Social Security Agency.
South African Wind Energy Association. 2019. Stats and facts. [Online]. Available: https://sawea.org.za/stats-and-facts-
sawea/ [Accessed 4 April 2019].
Spaull, N. 2019. Priorities for education reform in South Africa. Input document for an Economic Colloquium organised
by National Treasury. Pretoria (19 January 2019).
Spilimbergo, A., Symansky, S., Blanchard, O. & Cottarelli, C. 2008. Fiscal Policy for the Crisis. IMF Staff Position Note
SPN/08/01, 29 December. Washington, DC: International Monetary Fund.
Statistics South Africa (StatsSA). 2002a. Earning and spending in South Africa: Selected ndings and comparisons from
the income and expenditure surveys of October 1995 and October 2000. Pretoria: Statistics South Africa.
Statistics South Africa (StatsSA). 2002b. General household survey 2002. Statistical Release P0318. Pretoria: Statistics
South Africa.
Statistics South Africa (StatsSA). 2002c. Income and expenditure of households, 2000 South Africa. Statistical release
P0111. Pretoria: Statistics South Africa.
Statistics South Africa (StatsSA). 2005. Labour Force Survey, September 2004. Pretoria: Statistics South Africa.
Statistics South Africa (StatsSA). 2008. Income and expenditure of households, 2005/06. Statistical release P0100. Pretoria:
Statistics South Africa.
Statistics South Africa (StatsSA). 2017a. Financial statistics of consolidated general government 2015/2016. Statistical
Release P9119.4. Pretoria: Statistics South Africa.
Statistics South Africa (StatsSA). 2017b. General household survey 2017. Statistical Release P0318. Pretoria: Statistics
South Africa.
Statistics South Africa (StatsSA). 2017c. Living Conditions of Households in South Africa: An analysis of household
expenditure and income data using the LCS 2014/2015. Statistical Release, P0310. [Online]. Available:
http://www.statssa.gov.za/?page_id=1854&PPN=P0310&SCH=6811 [Accessed 17 May 2018].
Statistics South Africa (StatsSA). 2018a. Financial statistics of consolidated general government 2016/2017. Statistical
Release P9119.4. Pretoria: Statistics South Africa.
Statistics South Africa (StatsSA). 2018b. Quarterly Labour Force Survey Quarter 1: 2018. Statistical release P0211.
[Online]. Available: http://www.statssa.gov.za/publications/P0211/P02111stQuarter2018.pdf [Accessed 2 March
2019].
Statistics South Africa (StatsSA). 2019a. General household survey 2018. Statistical Release P0318. Pretoria: Statistics
South Africa.
Statistics South Africa (StatsSA). 2019b. Quarterly labour force survey trends 2008–2018 Quarter 4. [Online]. Available:
http://www.statssa.gov.za/?page_id=1854&PPN=P0211 11 [Accessed 24 February 2019].
Statistics South Africa (StatsSA). Various issues. Mid-year population estimates. Statistical Release P0302. Pretoria:
Statistics South Africa.
Steenekamp, T.J. 1994. Moet armes belasting betaal? South African Journal of Economics, 62(4): 371–392.
Steenekamp, T.J. 1996. Some aspects of corporate taxation in South Africa: the Katz Commission. South African Journal
of Economics, 64(1): 1–19.
Steenekamp, T.J. 2007. Tax performance in South Africa: a comparative study. Southern African Business Review, 11(3):
1–16.
Steenekamp, T.J. 2012. Taxing the rich at higher rates in South Africa? Southern African Business Review,16(3): 1–29.
Steenekamp, T.J. & Döckel, J.A. 1993. Taxation and tax reform in LDCs: Lessons for South Africa. Development Southern
Africa, 10(3): 319–333.
Steyn, G. 2006. Investment and uncertainty: Historical experience with power sector investment in South Africa and its
implications for current challenges. Working paper prepared for the Management Programme in Infrastructure
Reform and Regulation at the Graduate School of Business at the University of Cape Town, 15 March 2006. [Online].
Available: http://www.gsb.uct.ac.za/ les/Eskom-InvestmentUncertainty.pdf [Accessed 13 June 2018].
Stiglitz, J.E. 1984. eories of wage rigidity. NBER Working Paper 1442. Washington, DC.
Stotsky, J. & WoldeMariam, A. 2002. Central American tax reform: Trends and possibilities. IMF Working Paper,
WP/02/227: 1–41.
Stowell, D. 2005. Climate trading: Development of greenhouse gas markets. New York: Palgrave MacMillan.
Strydom, P.D.F. 1987. Structural imbalances in the South African economy. e Economic Society of South Africa
Conference. Pretoria, South Africa.
Sundaram, J.K. & Chowdhury, A. 2016. Expansionary scal consolidation myth. Inter Press Service (11-08-2016).
Available: http://www.globalissues.org/news/2016/08/11/22398 [Accessed 8 February 2017].
Sunday Times. 2018. New Eskom Board announced – Jabu Mabuza named as new chairman. [Online]. Available:
https://www.timeslive.co.za/sunday-times/business/2018-01-20-new-eskom-board-announced-jabu-ma”buza-
named-as-new-chairman/ [Accessed 14 June 2018].
Sunley, E.M. 1989. e treatment of companies under cash ow taxes: Some administrative, transitional, and
international issues. Working paper 189 of Country Economics Department. Washington, DC: World Bank.
Swanepoel, J.A. & Schoeman, N.J. 2003. Countercyclical scal policy in South Africa: Role and impact of automatic scal
stabilisers. South African Journal of Economic and Management Sciences, 6(4): 802–822.
Tanzi, V. 1995. Taxation in an integrating world. Washington, DC: Brookings.
Tanzi, V. 1996. Globalization, tax competition and the future of tax systems. IMF Working Paper, WP/96/141.
Washington, DC: International Monetary Fund.
Tanzi, V. 2004. Globalization and the need for scal reform in developing countries. Journal of Policy Modeling, 26: 525–
542.
Tanzi, V. 2005. e economic role of the state in the 21st century. Cato Journal, 25(3): 617–638.
Tanzi, V. & Schuknecht, L. 1995. e growth of government and the reform of the state in industrial countries. IMF
Working Paper 95/130. Washington, DC: International Monetary Fund.
Tanzi, V. & Zee, H.H. 2000. Tax policy for emerging markets: Developing countries. National Tax Journal, LIII(2): 299–
322.
Taylor, N. & Reddi, B. 2013. Writing and learning mathematics. In: Taylor, N., Van der Berg, S. & Mabogoane, T. (eds).
Creating effective schools. Cape Town: Pearson: 177–199.
aler, R.H. 2016. Behavioral Economics: Past, Present, and Future. American Economic Review. 106 (7): 1577–1600.
e Guardian. 2011. Warren Buffett calls for higher taxes for US super-rich. [Online].
ornton, J. 2008. Explaining Procyclical Fiscal Policy in African Countries. Journal of African Economies, 17(3): 451–464.
Tiebout, C.M. 1956. A pure theory of local expenditures. Journal of Political Economy, 64: 416–424.
Todaro, M.P. & Smith, S.C. 2015. Economic development. 15th ed. Harlow: Pearson.
Toye, J. 2000. Fiscal crisis and scal reform in developing countries. Cambridge Journal of Economics, 24: 21–44.
Trong, K. 2010. e relative risk-percentage equity index: Measuring equity in reading achievement across PIRLS 2006
countries. Hamburg: International Association for the Evaluation of Education.
Unemployment Insurance Fund. 2018. Annual report 2018. Pretoria: Department of Labour.
UNESCO. 2010. Education for All Global Monitoring Report 2010. Paris: UNESCO Publishing.
UNESCO. 2014. Education for All Global Monitoring Report 2014. Paris: UNESCO Publishing.
UNESCO. 2017. Education for All Global Monitoring Report 2017. Paris: UNESCO Publishing.
UNESCO. 2019. Education for All Global Monitoring Report 2019. Paris: UNESCO Publishing.
United Nations Development Programme. 2004. Human development report 2004. New York: Oxford University Press.
United Nations Economic Commission for Africa. 2004. Striving for good governance in Africa. Addis Ababa.
Van Aarle, B. & Garretsen, H. 2003. Keynesian, non-Keynesian or no effects of scal policy changes? e EMU case.
Journal of Macroeconomics, 25(2): 213–240.
Van den Heever, A. 2012. e role of insurance in the achievement of universal coverage within a developing country
context: South Africa as a case study. BMC Public Health, 12(Supplement 1): 1–13.
Van der Berg, S. 1991. Redirecting government expenditure. In: Moll, P., Nattrass, N. & Loots, L. (eds). 1991.
Redistribution: How can it work in South Africa? Cape Town: David Philip: 74–85.
Van der Berg, S. 1992. Social reform and the reallocation of social expenditures. In: Schrire, R. (ed). 1992. Wealth or
poverty? Critical choices for South Africa. Cape Town: Oxford University Press: 121–142.
Van der Berg, S. 1997. South African social security under apartheid and beyond. Development Southern Africa, 14(4):
481–503.
Van der Berg, S. 2001a. e means test for social assistance grants and its recent evolution. Social Work, 37(2): 125–142.
Van der Berg, S. 2001b. Trends in racial scal incidence in South Africa. South African Journal of Economics, 69(2): 243–
268.
Van der Berg, S. 2005. Fiscal expenditure incidence in South Africa, 1995 and 2000. Final report to National Treasury on
aspects of expenditure incidence. February.
Van der Berg, S. 2006. e targeting of public spending on school education, 1995 and 2000. Perspectives in Education,
24(2): 49–63.
Van der Berg, S. 2009. Fiscal incidence of social spending in South Africa, 2006. A report to National Treasury, 28
February 2009. [Online]. Available:
http://www.treasury.gov.za/publications/other/Fiscal%20Incidence%20Study/Report%20on%20 scal%20incidenc
e%20in%202006.pdf [Accessed 30 January 2015].
Van der Berg, S. 2014. e transition from apartheid: Social spending shifts preceded political reform. Economic History
of Developing Regions, 29(2): 234–244.
Van der Berg, S. & Burger, R. 2002. e stories behind the numbers: An investigation of efforts to deliver services to the
South African poor. Study for World Bank as background paper to World Development Report 2004. Stellenbosch.
November. [Online]. Available: http://econ.worldbandk.org/ les/28003_van_der_berg.pdf
Van der Berg, S. & Burger, R. 2003. Education and socio-economic differentials: A study of school performance in the
Western Cape. South African Journal of Economics, 71(3): 496–522.
Van der Berg, S., Burger, C., Burger, R., De Vos, M., Du Rand, G., Gustafsson, M., Moses, E., Shepherd, D., Spaull, N.,
Taylor, S., Van Broekhuizen, H. & Von Fintel, D. 2011. Low quality education as a poverty trap. Stellenbosch
Economic Working Papers 25/11. Stellenbosch: Stellenbosch University (Department of Economics & Bureau of
Economic Research).
Van der Berg, S. & Hofmeyr, H. 2018. Education in South Africa. Background note for the Republic of South Africa
Systematic Country Diagnostic. Washington, DC: World Bank.
Van der Berg, S. & Louw, M. 2004. Changing patterns of South African income distribution: Towards time series
estimates of distribution and poverty. South African Journal of Economics, 72(3): 546–572.
Van der Berg, S., Louw. M. & Yu, D. 2008. Post-transition poverty trends based on an alternative data source. South
African Journal of Economics, 76(1): 58–76.
Van der Berg, S. & Moses, E. 2012. How better targeting of social spending affects social delivery in South Africa.
Development Southern Africa, 29(1): 127–139.
Van der Berg, S. & Siebrits, F.K. 2010. Social assistance reform during a period of scal stress. Stellenbosch Economic
Working Papers 17/10. Stellenbosch: University of Stellenbosch (Department of Economics & Bureau for Economic
Research).
Van der Merwe, E.J. 1993. Is South Africa in a debt trap? South African Reserve Bank Occasional Paper 6, Pretoria:
Government Printer.
Van Heerden, J.H. 1996. e distribution of personal wealth in South Africa. South African Journal of Economics, 64(4):
278–292.
Van Heerden, Y. 2013. Personal income tax reform to secure the South African revenue base using a micro-simulation tax
model, Unpublished PhD in Economics in the Faculty of Economic and Management Sciences at the University of
Pretoria, November 2013.
Van Heerden, Y. & Schoeman, N.J. 2010. An Empirical Dissemination of the Personal Income Tax Regime in South Africa
Using a Microsimulation Tax Model. University of Pretoria, Department of Economics Working Paper, No.2010–25:
1–30.
Van Oordt, M. 2018. Zero-rating versus Cash Transfers under the VAT. Fiscal Studies, 39(3): 489–515.
Van Zyl, C., Botha, Z., Goodspeed, I. & Skerritt, P. 2009. Understanding South African nancial markets. Pretoria: Van
Schaik Publishers.
Venkatakrishnan, H. & Spaull, N. 2014. What do we know about primary teachers’ mathematical content knowledge in
South Africa? An analysis of SACMEQ 2007. Stellenbosch Economic Working Papers 13/14. Stellenbosch:
Stellenbosch University (Department of Economics & Bureau of Economic Research).
Wagner, A. 1883. ree extracts on public nance. In: Musgrave, R.A. & Peacock, A.T. (eds). 1958. Classics in the theory of
public nance. London: Macmillan.
Whiteford, A. & McGrath, M. 1994. Inequality in the size distribution of income in South Africa. Occasional Papers 10.
Stellenbosch: Stellenbosch Economic Project.
Williams, M.J. 2007. e social and economic impacts of South Africa’s child support grant. EPRI Working Paper 39. Cape
Town: Economic Policy Research Institute.
Wilson, J.D. 1999. eories of tax competition. National Tax Journal, 52(2): 269–304.
Woolard, I. & Leibbrandt, M. 2001. Measuring poverty in South Africa. In: Bhorat, H., Leibbrandt, M., Maziya, M., Van der
Berg, S. & Woolard, I. (eds). 2001. Fighting poverty: Labour markets and inequality in South Africa. Cape Town: UCT
Press.
Woolard, I. & Leibbrandt, M. 2010. e evolution and impact of unconditional cash transfers in South Africa. SALDRU
Working Paper No. 51. Cape Town: University of Cape Town (Southern African Labour and Development Research
Unit).
World Bank. 1991. Lessons of tax reform. Washington, DC: World Bank.
World Bank. 1994. World development report 1994. Washington, DC: World Bank.
World Bank. 1997. Congo – Poverty Assessment. Report No. 16043-COB. Washington, DC: World Bank.
World Bank. 2000. Swaziland – Reducing poverty through shared growth. Report No 19658-SW. Washington, DC: World
Bank.
World Bank. 2004a. African Development Indicators 2004. Washington, DC: World Bank.
World Bank. 2004b. World Development Indicators 2004. Washington, DC: World Bank.
World Bank. 2004c. World development report 2004. Washington, DC: World Bank.
World Bank. 2006. South Africa – Sector study of the effective tax burden (English). Washington, DC: World Bank.
[Online]. Available:
http://documents.worldbank.org/curated/en/426631468165266422/pdf/588250WP0South10BOX353820B01PUBLI
C1.pdf [Accessed 14 June 2018].
World Bank. 2010. World Development Indicators 2010. CD Rom. Washington, D.C.: World Bank.
World Bank. 2015. South Africa Country-level scal policy notes Sector Study of Effective Tax Burden & Effectiveness of
Investment Incentives in South Africa-Part I. Washington, DC: World Bank. [Online]. Available:
http://www.taxcom.org.za/docs/Sector%20Study%20of%20Effective%20Tax%20Burden%20in%20South%20Africa%
20-%20Part%201%20-%20June%202015.pdf [Accessed 15 June 2018].
World Bank. 2016. Southern Africa Country-level scal policy notes. South Africa: Sector Study of Effective Tax Burden and
Effectiveness of Investment Incentives in South Africa – Firm Level Analysis. Washington, DC: World Bank. [Online].
Available:
http://www.taxcom.org.za/docs/Sector%20Study%20of%20Effective%20Tax%20Burden%20in%20South%20Africa%
20-%20Part%202%20-%20September%202016%20(updated).pdf [Accessed 14 June 2018].
World Bank. 2018a. Overcoming poverty and inequality in South Africa: an assessment of drivers, constraints and
opportunities. Washington, DC: World Bank.
World Bank. 2018b. World development indicators 2018. [Online]. Available: https://data.worldbank.org/products/wdi
[Accessed 14 July 2018].
World Bank and Namibia Statistics Agency. 2017. Does scal policy bene t the poor and reduce inequality in Namibia?
Washington, DC: World Bank.
World Bank & PriceWaterhouseCoopers. 2010. Paying Taxes. [Online]. Available:
http://www.scribd.com/doc/29675385/Paying-Taxes-2010-Global-Report-via-World-Bank [Accessed 14 February
2011].
World Health Organization. 2001. Macroeconomics and health: Investing in health for economic development. Report of
the Commission on Macroeconomics and Health. Geneva: World Health Organization.
World Health Organization. 2018. Global tuberculosis report 2018. Geneva: World Health Organization.
World Health Organization. 2019. Number of new HIV infections — Data by country. [Online]. Available:
http://apps.who.int/gho/data/node.main.HIVINCIDENCE?lang=en [Accessed 6 July 2019].
Zee, H.H., Stotsky, J.G. & Ley, E. 2002. Tax incentives for business investment: A primer for policy makers in developing
countries. World Development, 30(9): 1497–1516.
Zizzamia, R., Schotte, S. & Leibbrandt, M. 2019. Snakes and ladders and loaded dice: poverty dynamics and inequality in
South Africa between 2008-2017. SALDRU Working Paper 235. Cape Town: University of Cape Town (Southern
African Labour and Development Research Unit).
Zodrow, G.R. 2003. Tax competition and tax coordination in the European Union. International Tax and Public Finance,
10: 651–671.
Answers to multiple-choice questions
e solutions to all of the end-of-chapter multiple-choice questions are provided here.

Chapter 1
1.1 a, b, c and d
1.2 a, b, c and d
1.3 d
1.4 b
1.5 c

Chapter 2
2.1 d
2.2 a and b
2.3 c

Chapter 3
3.1 a and b
3.2 a and c
3.3 a and c
3.4 d

Chapter 4
4.1 a, b and c
4.2 b
4.3 a and c

Chapter 5
5.1 b
5.2 a, b and c
5.3 d
5.4 a, c and d

Chapter 6
6.1 c and d
6.2 b and c
6.3 b and c
6.4 d

Chapter 7
7.1 d
7.2 b
7.3 a
7.4 d
7.5 d
Chapter 8
8.1 a, b and d
8.2 a and b
8.3 b and c
8.4 a, b and c

Chapter 9
9.1 a and d
9.2 b, c and d
9.3 a, b, c and d
9.4 b and d

Chapter 10
10.1 b and d
10.2 b, c and d
10.3 a and b
10.4 b and c

Chapter 11
11.1 d
11.2 b
11.3 c
11.4 b, c, d and e

Chapter 12
12.1 c
12.2 b
12.3 a
12.4 b, c and d

Chapter 13
13.1 c
13.2 e
13.3 b
13.4 e

Chapter 14
14.1 b
14.2 a and c
14.3 c

Chapter 15
15.1 d
15.2 b and c
15.3 c
15.4 a

Chapter 16
16.1 d
16.2 c and d
16.3 e
16.4 b and c

Chapter 17
17.1 a
17.2 c and d
17.3 c and d
17.4 b, c and d
17.5 b and d

Chapter 18
18.1 d
18.2 b
18.3 a
18.4 d

Chapter 19
19.1 c
19.2 d
19.3 c
19.4 b