SCHOOL OF GRADUATE STUDIES

INSTITUTE OF AGRICULTURE & DEVELOPMENT STUDIES

TRENDES AND DETERMINANTES OF INCOME INEQUALITYIN ETHIOPIA

By

ASSEFA BELAY

DECEMBER 2020

ADDIS ABABA, ETHIOPIA

 TRENDES AND DETERMINANTES OF INCOME INEQUALITYIN ETHIOPIA

A THESIS SUBMITTED TO THE ST. MARY’S UNIVERSITY SCHOOL OF

GRADUATE STUDIES PARTIAL FULFILLMENTS OF THE REQUIREMENT FOR
THE DEGREE OF MASTER OF ART IN DEVELOPMENTAL ECONOMICS

By

ASSEFA BELAY

DECEMBER 2020

ADDIS ABABA, ETHIOPIA

i
 DECLARATION
I, the under signed, declare that this thesis is my original work, prepared under
the guidance of KurabachewMenber (PhD).All sources of material used
while working on this thesis have been duly acknowledged. I further confirm that the
thesis has not been submitted either in part or in full to any other higher learning institution for
the purpose of earning any type of degree.

Assefa Belay Signature:_________________________

ii
 ENDORSEMENT
This thesis has been submitted to St. Mary‟s University, school of Graduate Studies
for examination with my approval as a university advisor.

KurabachewMenber (PhD) ------------------------

Advisor Signature

St. Mary‟s University, Addis Ababa December, 2020

iii
 APPROVED BY BOARD OF EXAMINERS
As members of board of examining of the final MA thesis open defense, we certify
that we have read and evaluated the thesis prepared by Assefa Belay under
the title “Trends and Determinants of Income Inequality the Case of Ethiopia” we
recommend that this thesis to be accepted as fulfilling the thesis requirement for the Degree of
Master of Art in Development Economics.

______________________ ____________________________

Chairperson Signature

______________________ ______________________________

Advisor Signature

________________________ _______________________________

Internal Examiner Signature

_________________________ _____________________________

External Examiner Signature

iv
 ACKNOWLEDGEMENT
First and for most I would like to thanks almighty God and Tsion Mariam for giving me healthy
and strength to reach this stage and become successful. It is not exaggeration to say that without
insisting the Almighty God, Jesus Christ and Tsion Mariam, would not have been in a place to
complete successfully this thesis. Thus, glory to them. My gratitude and appreciation goes to my
advisor Dr. Kurabachew Menber for his constructive comments, technical support, welcoming
approach in every step of my work and helped me in shaped this study. I would like to extend my
gratitude to my families, especially my wife „(Tseheynesh F.) without whose moral and financial
support, my achievement was not possible. And also I want to say thank you all to my relatives
and colleagues who have helped me while I was writing this research. Last, but not least, I want
to thank St .Mary‟s university for sponsoring my stay in the university.

v
 TABLE OF CONTENTS
DECLARATION ............................................................................................................................ ii

ENDORSEMENT .......................................................................................................................... iii

APPROVALSHEET ...................................................................................................................... iv

ACKNOWLEDGEMENT .............................................................................................................. v

LIST OF FIGURES ....................................................................................................................... ix

LIST OF TABLES .......................................................................................................................... x

ACRONYMS ................................................................................................................................. xi

ABSTRACT ................................................................................................................................. xiii

CHAPTER ONE ............................................................................................................................. 1

INTRODUCTION .......................................................................................................................... 1

1.1 Background of the Study .................................................................................................. 1

1.2 Statement of the Problem ................................................................................................. 2

1.3 Objective of the Study ...................................................................................................... 4

1.3.1 General Objective ........................................................................................................ 4

1.3.2 Specific objective: .......................................................................................................... 4

1.4 Significance of the Study ................................................................................................. 4

1.5 Scope and Limitation of Study ........................................................................................ 5

1.6 Organization of the Study/Thesis ......................................................................................... 5

CHAPTER TWO ............................................................................................................................ 6

LITERATURES REVIEW ............................................................................................................ 6

2.1 Theoretical Literature ...................................................................................................... 6

2.1.1 Income Inequality and its Measurments ................................................................... 6

2.1.2 Income Inequality and its Determinants ................................................................... 9

vi
 2.1.3 Effects of income Inequality on Economic Growth .............................................. 13

2.1.4 Effects of Income Inequality on Human Capital .................................................... 13

2.1.5 Effects of Income Inequality on Policies of the Country ........................................ 14

2.1.6 Effects of Income Inequality on the Market Structure............................................ 15

2.1.7 Effects of Income Inequality on Environment ........................................................ 15

2.1.8 Effects of Income Inequality on National Stability and Social Cohesion .............. 16

2.1.9 Effects of Income Inequality on Finance ................................................................ 16

2.2 Empirical Literatures Review......................................................................................... 16

2.3 Conceptual frameworkof Study .................................................................................... 18

CHAPTER THREE ...................................................................................................................... 19

RESEARCH METHODOLOGY.................................................................................................. 19

3.1 Research Approach and Design ..................................................................................... 19

3.2 Data Source and Types of Data ...................................................................................... 20

3.3 Econometrics Model Specification ................................................................................ 20

3.4 Estimation Procedure ..................................................................................................... 21

3.5 Model Specification ....................................................................................................... 21

3.6 Unit Root Test ................................................................................................................ 24

3.7 Description of variables ................................................................................................. 25

3.8 Methods of Data Analysis .............................................................................................. 27

CHAPTER FOUR ......................................................................................................................... 28

DATA ANALYSIS AND INTERPRETATION .......................................................................... 28

4.1 Descriptive analysis........................................................................................................ 28

4.1.1 Trend of Income Inequality .................................................................................... 28

4.1.2 Trends of GDP per capital / Economic growth/...................................................... 29

4.1.3 Trend of net primary school enrollment rate or Education ..................................... 30

vii
 4.1.4 Trend of Trade openness ......................................................................................... 31

4.1.5 Trend of Unemployment rate .................................................................................. 32

4.1.6 Trend of Inflation in Ethiopia ................................................................................. 33

4.2 Econometric Analysis .................................................................................................... 34

4.2.1 The Unit Root Test Analysis................................................................................... 34

4.3 Model Stability and Dignostic Test ............................................................................... 37

4.3.1 Test of Multicolinearity .......................................................................................... 37

4.3.2 Functional form (Ramsey RESET test) .................................................................. 38

4.3.3 Test of Hetroscedasticity......................................................................................... 39

4.3.4 Tests for Autocorrelation ........................................................................................ 40

4.3.5 Test for normality ................................................................................................... 40

4.4 Long Run ARDL Bounds Tests for Co-integration ....................................................... 43

4.5 ARDL Model Estimation ............................................................................................... 44

4.5.1 Interpretation of the ARDL model estimation coefficients .................................... 46

4.6 Short-Run Error Correction Model (ECM) ....................................................................... 47

4.7 Summa of Result………………………………………………………………………......49

CHAPTER FIVE .......................................................................................................................... 50

CONCLUSION AND RECOMMENDATION ............................................................................ 50

5.1 Conclusion...................................................................................................................... 50

5.2 Recommendation ............................................................................................................ 51

REFERENCES ............................................................................................................................. 53

APPENDIX I ................................................................................................................................ 56

APPENDIX: II .............................................................................................................................. 57

APPENDIX: III............................................................................................................................. 57

Appendix: IV ................................................................................................................................ 58

viii
 LIST OF FIGURES

Figure 2-1: Gini coefficient ............................................................................................................ 8

Figure 2-2: Kuznets curve ............................................................................................................. 11
Figure 2-3: Conceptual frame work of the study .......................................................................... 19
Figure 4-1:Trends of income inequality in Ethiopia from 1988 to 2018 ...................................... 29
Figure 4-2: Trends of GDP per capital in Ethiopia from 1988 to 2018 ........................................ 30
Figure 4-3: Trends of Net Primary School Enrollments rate in Ethiopia from 1988 to 2018 ...... 31
Figure 4-4: Trends of Trade openness in Ethiopia from 1988 to 2018 ....................................... 32
Figure 4-5: Trends of unemployment rate in Ethiopia from 1988 to 2018 ................................. 33
Figure 4-6: Trends of Inflation rate in Ethiopia from 1988 to 2018 ........................................... 34
Figure 4-7: Normality Test ........................................................................................................... 41
Figure 4-8: Plot of Cumulative Sum of Recursive Residuals ...................................................... 42
Figure 4-9: Plot of Cumulative Sum of Squares of Recursive Residuals ..................................... 42

ix
 LIST OF TABLES

Table 3-1: Description of variables............................................................................................... 27

Table 4-1: Unit root test (Augmented Dickey-Fuller test)............................................................ 35
Table 4-2: Unit root test (Phillips-Perron test statistic test) .......................................................... 36
Table 4-3: VIF test for Multicolinearity ....................................................................................... 38
Table 4-4: Functional form (Ramsey RESET Test) ..................................................................... 39
Table 4-5: Test of Hetroscedasticity ............................................................................................. 39
Table 4-6: F-Bounds test ............................................................................................................... 44
Table 4-7: Estimated long run Coefficients .................................................................................. 45
Table 4-8: Error Correction Representation for the Selected ARDL ........................................... 48

x
 ACRONYMS
ADF Augmented Dickey Fuller

ARDLM Auto Regressive Distributed Lag Model

CSA Central Statistics Agency

CUSMSQ Cumulative Sum of Squares

CUSUM Cumulative Sum

DW Durbin Watson

EEA Ethiopia Economic Association

EKC Environmental Kuznets Curve

EPL Employment Protection Legislation

FRED Federal Reserve Economic Data

GDP Gross Domestic Product

IMF International Monetary Found

KPSS Kwiatowski Phillips Schmidt Shin

MOFED Ministry of Finance and Economics

NBE National Bank of Ethiopia

OLS Ordinal Least Square

PP Philip-Perron

xi
UNCTAD United Nation Conference on Trade and Development

UNSESCO United Nation Educational Scientific and Cultural Organization

VARM Vector Error Correction Model

VECM Vector Error Correction Model

VIF Variance Inflation Factor

xii
 ABSTRACT
Income inequality has been regarded as one of the most serious problem that most countries
(both developed and developing countries) face today. Inequality can be signal of lack of income
mobility and opportunity reflections of persistent disadvantage for particular segments of the
society. The main objective of this study is to analyze the trends and examine the determinants of
income inequality in Ethiopia from (1988 to2018). The data collected from World Development
Indicator (WDI), the Global Economy, World Data Bank and National Bank of Ethiopia
websites. The method of analysis was both descriptive and econometrics analysis. In descriptive
analysis, the trend of income inequality, real GDP per capita, unemployment rate, net primary
school enrollment rate trade openness and inflation rate have been analyzed. To check the
verifiability of the estimated long run model, some diagnostic test is undertaken. This paper used
Autoregressive Distributed Lag (ARDL) and Error Correction Model (ECM) in order to
investigate the long-run and short run relationship between the dependent variable (income
inequality) and its determinants. To test stationary Augmented Dickey –Fuller (ADF) test and
Phillpes Perron (PP) test were used. All the determinants have got with a sign as expected by the
paper. The finding of the Bounds test shows that there is a stable long run relationship between
income inequality and real GDP per capital, School of enrollment rate, trade openness,
unemployment rate and general inflation rate. The study results real GDP per capita,
unemployment rate and inflation rate have a positive impact on income inequality. The
remaining has negative impact. In the long run analysis, real GDP per capita, net primary
school enrollment rate, unemployment rate and constant are statistically significant .The error
correction coefficient, estimated at -0.84277 is highly significant, has the correct negative sign,
and imply a very high speed of adjustment to equilibrium. According to the econometrics
analysis, real GDP per capita and unemployment rate are the main determinants of income
inequality for Ethiopia based on ARDL model estimation result. According to the thesis, the
paper gives some policy recommendations. Like the government or other responsible body
should focus on the countries growth and development, decreasing unemployment rate, Inflation
rate the expansion of education access and the development of international trade in order to
reduce the income gap of the people. And the policies should consider the poor to participate
them from the countries benefit.

Keywords: Economic Growth, Income Inequality, ARDL, ECM, Ethiopia.

xiii
 CHAPTER ONE

INTRODUCTION
1.1 Background of the Study

Economic inequality is most obviously shown by people‟s different positions within the
economic distribution - income, pay, and wealth (The Equality Trust 2012-2016). There are three
main types of economic inequality. Such as income inequality, pay inequality, wealth inequality.
The first one is income inequality which is the extent to which income is distributed unevenly in
a group of people. Income is not just the money received through pay, but all the money received
from employment (wages, salaries, bonuses etc.), investments, such as interest on savings
accounts and dividends from shares of stock, savings, state benefits, pensions (state, personal,
company) and rent. The second one is pay inequality. Pay refers to payment from employment
only. This can be on an hourly, monthly or annual basis, is typically paid weekly or monthly and
may also include bonuses. Pay inequality therefore describes the difference between people‟s
pay and this may be within one company or across all pay received. The last one is wealth
inequality. Wealth refers to the total amount of assets of an individual or household. This may
include financial assets, such as bonds and stocks, property and private pension rights. Wealth
inequality therefore refers to the unequal distribution of assets in a group of people, (The
Equality Trust 2012-2015).The word income means “the money that a person, a region, a
country, etc… earns from investing money, from business etc… ”.And inequality defined as
“something that is unfair; the state of being unfair, unjust”. Together, income inequality means
“unfair or unjust distribution of money, earns from investing, from business etc…” (Oxford,
2017). There are a number of factors that drive income inequality, Such as technological change,
change in labor market institutions, redistributive policy, education and other social, economic,
political and demographical factors. Technology has led to improvements in productivity and
well-being by leaps and bounds, but has also played a central role in driving up the skill premium
resulting in increased labor inequality. This is because technological changes can
disproportionately raise the demand for capital and skilled labor over law – skilled and unskilled
Labor by eliminating many jobs through automation or upgrading the skill level required to

1
attaining or keeping those jobs (Card and Dinardo 2012; Acemoglu 1998). More flexible labor
market institution can foster economic dynamism by reallocating resources to more productive
firms and enabling firm restructuring. If it‟s not flexible income inequality will be increased.
Education can play an important role in reducing income inequality, or it determines
occupational choice, access to jobs, and the level of pay, and plays a pivotal role or ability and
productivity in the job market. The distribution of income will be unfair when education is not
well address to the people. And the above advantages will be lost.

Income inequality influences the macroeconomic and social activities indifferent ways.
According to previous IMF studies, income inequality (which is measured by GINI coefficient)
negatively affect growth and its sustainability (Beerg and Ostry, 2014). By depriving the ability
of lower – income households to stay healthy and accumulate physical and human capital higher
inequality lowers growth of the country (Galor and Moav, 2014). For example, it leads to under-
investment in education as poor children ends up in lower quality schools and are less able to go
to college. Because of this, labor productivity could be lower than it would have been in more
equitable world (Stiglitz, 2012). Many empirical and theoretical studies indicates that the rising
influence of the rich and stagnant incomes of the poor and the middle class have a causal effect
on crises, and thus directly hurts short and long term growth. Similarly, higher inequality in
advanced economies is associated with the global financial crisis. This global imbalance can be
challenged for macroeconomic and financial stability and thus growth (Kumholf, 2013). Extreme
inequality can be associated with conflict by damaging trust and social cohesion. Conflicts may
arise from the management of common resources. In other words, inequality affects the
economies of conflict by intensifying the power of a certain group and then reducing the
opportunity costs of initiating and joining a violent conflict (Lichbach, 1989).

1.2 Statement of the Problem

Income inequality has been regarded as one of the most serious problem that most countries
(both developed and developing countries) face today. Inequality with in most advanced,
emerging markets and developing countries is a phenomenon that has received considerable
attention. President Obama called widening income inequality has the “defining challenge of this
time „(Obama,2014) .Inequality can be signal of lack of income mobility and opportunity

2
reflections of persistent disadvantage for particular segments of the society. Widening inequality
also has significant implications for growth and Marco economic making power in the hands of a
few, lead to a suboptimal use of human resources, cause investment reducing political and
economic instability, and raise crisis risk. The economic and social fallout from the global
financial crisis and the resultant head winds to global growth and employment have heightened
the attention to rising income inequality. (Era dabble Norris, KalpanaKocchar 2015).According
to many evidences, the rich becomes richer and the poor becomes poorer. This shows the
presence of high wealth concentration. This means the newly created wealth is concentrated in
already wealthy individuals because people who already hold wealth have the resource to invest,
which creates new wealth. This wealth concentration process makes income inequality a vicious
cycle. Its effect may be transform to future generations. The children with rich family have an
economic advantage of getting quality education good health care. As a result, they may have a
higher chance of earning a higher income than their poor peers. This creates a vicious cycle of
inequality. Now a day, the issue becomes a headache for politicians. Higher inequality lower
growth by depressing the ability of lower income house to stay healthy and accumulate physical
and human capital (Galor and Moav 2004) for instance, it can lead to under-investment in
schools and are less able to go on to college. As result, labor productivity could be lower than it
would have been in a more equitable world (Stieglitz 2017).

Extreme inequality may damage trust and social cohesion and thus is also associated with
conflicts, which discourage investment. Conflicts are particular prevalent in the management of
common resources, for example, inequality makes resolving disputes more difficult, see, for
example bard hare (2008).Although in Ethiopia, there is a high level of unequal distribution of
resources exists between people. The countries level of national income inequality measured by
GINI coefficient in2017/18 is 0.33, rural income inequality is 0.28 and urban income inequality
is 0.38. Which is low, but it grows rapidly. It is especially so in the Ethiopia where poverty is
widespread and where, give low likely to be significant. This research was used only 31 years
data, because of time constraint but it is better to use 60 and above year‟s data to analyze tends
and its determinants by using macro variables, this is interesting since it is important for policy
makers, academics and others to understand the forces behind distribution of income in order to
trickle the problem in most efficient way. Even though there are many studies on the issues at

3
international level, few attempts were made to examine the determinants of income inequality a
case of Ethiopia by using micro variables. This research is tries to fill the gaps that are not
address by other researcher. Much of the income inequality literature focuses on the relationship
between income inequality and economic growth, but a different strand of literature takes a step
back and instead looks at the causes of income inequality and they were not address‟ trends and
determinants of income inequality by using macro variables, so this research was fill the gaps
that was not address by other researchers by analyzing trends determinants of income inequality
by using macro variables.

1.3 Objective of the Study

1.3.1 General Objective

The ultimate objective of this research is to analyze the trends and examine the determinants of
income inequality in Ethiopia from (1988 to 2018).

1.3.2 Specific objective

 To analyze the trend of income inequality at national level

 To analyze determinants of income inequality in study area
 To identify the main determinants of income inequality
1.4 Significance of the Study

The study was providing essential information to bring sustainable and fair distribution of
income among residents. Identifying its determinants of income inequality is necessary for a
number of reasons. It is important from the view point of fair distribution of income and to
alleviate the poverty from the society. In addition the identification of the key determinants of
income inequality helps policy makers with appropriate ways of intervention for controlling
income inequality and this research was provide a conclusion and policy recommendations and
advice the responsible body who formulate economic policy to give a heavy attention to the
determinants that affect income inequality the most

4
 1.5 Scope and Limitations of the Study
Income inequality is abroad topic. However, the scope of this study would be to examine trends
and determinants of income inequality using the data for the period 1988 to 2018. Years of 1988
to 2018 ware selected based on data availability. The researcher faced obstacle which is difficult
to accomplish the paper successfully. The limitation of this study was the one associated with
data availability. There are shortages of data, particularly, on Gini coefficient and trade
openness, specially, for the early period. The main aim of this study is to analyze the
macroeconomic determinants of income inequality and its trends. However, there are also
another factors that affecting income inequality like high political stability, governance
effectiveness, rules of economic regulation (monitoring and fiscal policy), and rules of law
(property right) structure of the economy, government expenditure, external debit and financial
aid, foreign reserve and exchange rate, growth of population, privatization and level of tax, are
not addressed here and might be consider other limitations of this study. Despite the above
difficulty, the researcher uses maximum effort to accomplish the research paper
comprehensively.

1.6 Organization of the Thesis

The study was organized in to five chapters. Chapter one covered the introduction, statement of
the problem, objective of the study, scope of the study, significance of the study , limitation of
the study and organization of the study, second chapter contained review of literature, the third
chapter included methodology of the study and chapter four provided descriptive and
econometrics results. Finally, the last chapter included conclusion and recommendation of the
study.

5
 CHAPTER TWO

LITERATURES REVIEW
2.1 Theoretical Literature
2.1.1 Income Inequality and Its Measurement

Income inequality: is the unequal distribution of household or individual income across the
various participants in an economy. Income inequality is presented as the percentage of income
to a percentage of population. For example, a statistics may indicate that 70% of a country‟s
income is controlled by 20% of that country‟s residents. It is associated with the idea of income
“fairness”. It is generally considered “unfair” if the rich have a disproportionally larger portion
of a country‟s income compared to their population. Income inequality is the state of an economy
in which the share of total income earned by the rich and the poor are highly unequal
the distribution. Economic policy makers can face a tradeoff between promoting equality
and economic growth. As income shares become more equal, the incentive for individuals
to accumulate skills, work hard, and take risks might become smaller, thus shrinking the size of
the economy.

Income Inequality and Lorenz Curve: It is a graphical representation of the distribution of

income or of wealth. It was developed by Max O. Lorenz in 1905 for representing inequality of
the wealth distribution. The curve is a graph showing the proportion of overall income or wealth
assumed by the bottom x% of the people. It is often used to represent income distribution, where
it shows for the bottom x% of households, what percentage (y %) of the total income they have.
The percentage of households is plotted on the x-axis, the percentage of income on the y-axis. It
can also be used to show distribution of assets. In such use, many economists consider it to be a
measure of social inequality. If all individuals are the same size, the Lorenz curve is a straight
diagonal line, called the line of equality. If there is any inequality can be summarized by the Gini
coefficient (also called the Gini ratio), which is the ratio between the area enclosed by the line of
equality and the Lorenz curve, and the total triangular area under the line of equality. The degree

6
of asymmetry around the axis of symmetry is measured by the so-called Lorenz asymmetry
coefficient.

Income inequality and Gini Coefficient: It is one of the most widely used measures of inequality
and its measures the extent to which the Lorenz curve departs from the line of equality. It is
defined as a ratio with values between 0 and 1: the numerator is the area between the Lorenz
curve of the distribution and the uniform distribution line; the denominator is the area under the
uniform distribution line. It was developed by the Italian statistician Corrado Gini and published
in his 1912 paper “Variabilita e mutabilita” (“Variability and Mutability”). The Gini coefficient
is often used to measure income inequality. Here, 0 corresponds to perfect income inequality
(everyone has the same income) and 1 corresponds to perfect income inequality (i.e. one person
has all the income, while everyone else has zero income). The Gini coefficient can also use to
measure wealth inequality. It is also commonly used for the measurement of discriminatory
power of rating systems in the credit risk management. The Gini Index is the Gini coefficient
expressed as a percentage, and is equal to the Gini coefficient multiplied by 100. The Gini
coefficient is equal to half of the relative mean difference (www.Gini coefficient .org ).

7
Figure 2-1: Gini coefficient
Gini coefficient is defined as a ratio of areas on the Lorenz curve diagram, i.e.

Gini coefficient =

If the Lorenz curve is represented by function Y =L(x), then the value /area of under the Lorenz
curve can be found withe integration i.e.

Gini coefficient =1-2∫ ( ) -------------------------------------------------------(2.1)

Income Inequality and Gini index: It measures the extent to which the distribution of income
among individuals or households within an economy deviates from a perfectly equal distribution.
The Gini index measures the area between the Lorenz curve and a hypothetical line of absolute
equality, expressed as a percentage of the maximum area under the line. Thus a Gini index of 0
represents perfect equality, while an index of 100 implies perfect inequality. Gini index is a
multiple of Gini index by 100. (Data. world bank. Org 2017)

Income Inequality and Coefficient of variation: To find out the imbalance between the different
states in terms of per capita incomes, this measure is given by Yotopoulous and Lau. Coefficient
of variation is based on mean and dispersion. It is an average index of inequality for all regions.
It measures the variation of observation from the mean. If its value is positive, it means
observations are more than mean value as well as, if its value is high, it means that distances
from the mean value is high. Disparity occurs when its value is positive and increases during the
time period.

Income Inequality and Theil Index: This measure was developed by Theil (1967) and used by
Cuadrardo, Dehesa and Precedo (1993) to study income inequalities among the member states of
the European Economic Community. Under this measure relative inequality among the regions,
in economic indicators such as income, is best explained by a simple ratio which compares
shares of the states in that indicator (say, income) with their respective shares in population. By
comparing the ratios it provides a good description of inequality among regions. For example; if
we take income indicators, we can compare the ratios Yi/Pi across regions, where Yi and Pi is
respectively the ith share in total income and the region‟s population. The regions which have

8
Yi/Pi above unity are better off states and the regions which have Yi/Pi below unity are not
doing well.

2.1.2 Income Inequality and Its Determinants

Economic growth: There is a conceptual difference between economic growth and economic
development. A country‟s economic growth is usually indicated by an increase in that country‟s
gross domestic product, or GDP. Gross domestic product is an economic model that reflects the
value of a country‟s output. In other words, a country‟s GDP is the total monetary value of the
goods and services produced by that country over a specific period of time. Economic growth is
the increase in the inflation- adjusted market value of the goods and services produced by an
economy. It is measured as the percent rate of increase in real gross domestic product, or real
GDP, the growth of the ratio of GDP to population (GDP per capita, which is called per capita
income. An increase in growth caused by more efficient use of inputs (such as labor, physical
capital, energy as materials) is referred to as intensive growth. GDP growth caused only by
increase in the amount of inputs available for use (increased population, new territory) is called
extensive growth. While, economic development is a process where low income national
economies are transformed in to modern industrial economies. It involves qualitative and
quantitative improvements in a countries economy. Political and social transformations are also
included in the concept of economic development in addition to economic change. A country‟s
economic development is usually indicated by an increase in citizen‟s quality of life. „Quality of
life‟ is often measured using the Human Development Index, which is an economic model that
considers intrinsic personal factors not considered in economic growth, such as literacy rates, life
expectancy and poverty and poverty rates. (Study.com 2014-2016)

There have been attempt to establish links between GDP per capita and economic growth on one
side and inequality on the other since the mid -1950‟s. Kuznets hypothesis (1963) postulates that
in the early stages of development, both a country‟s economic growth and its inequality increase.
A country grows and develops the income gap between the rich and the poor should decrease.
Indeed, according to Kuznets, there is a gradual shift from a low-inequality, low-income,
agricultural economy, towards a high income and medium- inequality economy characterized by
industrial production. These shifts would lead to the inverted u- shaped relationship between real

9
GDP per capita and inequality. Kuznets argues that in the initial period, agriculture represents the
bulk of a country‟s economy, which is also characterized by low levels of inequality. A shift
towards the secondary and the tertiary sectors has two effects in the short run. The first effect is
that it can accelerate economic growth leading to higher levels of GDP per capita. The second
and most dramatic effect is that this increases the level of inequality. Consequently, in the initial
stages of economic development, the level of GDP per capita and inequality are positively
correlated. As countries develop they shift more and more resources from agriculture to industry
(and later to service), and this will in time decrease the income gap between the industry and
agriculture simply because there will be more and more workers working in the industrial sector.
Consequently, the long run relationship between inequality and GDP per capita is negative.
Kuznets‟ hypothesis examined in terms of quadratic equation which is used to test this
hypothesis. The Kuznets curve is a curve that graphs economic inequality (represented by Gini
coefficient) against income per capita over the course of economic development (which was
presumed to correlate with time). This curve is meant to illustrate economist Simon Kuznets
hypothesis about the behavior and relationship of these two variables as an economy develops
from a primary rural agriculture society to an industrialized urban economy. The curve implies
that as a nation undergoes industrialization (especially the mechanization of agriculture) the
center of the nation‟s economy will shift to the cities. As internal migration by farmers looking
for better- paying jobs in urban hubs causes a significant rural urban inequality gap (the owners
of firms would be profiting, while laborers from those industries would see their incomes at a
much slower rate and agricultural works would possibly see their income decrease), rural
populations decrease as urban populations increase. Inequality is then expected to decrease
when a certain level of average income is reached and the processes of industrialization-
democratization and the rise of welfare state – allowing for the trickle-down of the benefits from
rapid growth, and increase the per capita income. In general, Kuznets believed that inequality
would follow an inverted “U” shape as it rises and then fails again with the increase of income
per capita (www.Kuznetcurve )

10
Figure 2-2: Kuznets curve

There have been a number of critics for Kuznets curve. The first arguments of critics are based
upon the countries used in Kuznets‟ data set. Critics say that Kuznets curve does not reflect an
average progression of economic development for an individual country, but rather it is
representation of historical differences in economic development and inequality between
countries in the data set. The critics hold that when controlling for this variable, the inverted U–
shape of the Kuznets curve begins to diminish. Other criticisms have come to light over time as
more economists have developed hypothesis with more dimensions and more countries had
undergone rapid economic growth that did not necessarily follow Kuznets‟ hypothesized pattern.

Level of Unemployment: the countries employment level is determined by the performance of

institutions and the policies of the country. The role of institutions and labor market polices, such
as the union coverage and the setting of the minimum wage has a great role for the distribution of
income. For example, the effect of changing employment protection legislation (EPL) on income
inequality is ambiguous. This is because the overall effect is dependent on two contrasting
channels: (1) the effect of EPL on the employment rate and (2) the effect of EPL on the earnings
inequality among the entire working age population (i.e. inequality between workers and non-
workers or between full time and part- time workers). A weakening of employment protection, in
particular the liberalization of temporary contracts is expected to contribute to higher wage
inequality. To generalized, employment protection legislation (EPL) is expected to exert a
positive impact on income inequality when the effect of EPL on wages dispersion prevails,

11
Whereas is expected to exert a negative impact when the corresponding effect on the
unemployment rate appears to be larger.

Education: Another important factor in the creation of inequality is variation in individuals‟

access to education. Education, especially in an area where there is a high demand for workers,
creates a high wages for those with this education, however, increase in education first increase
and then decrease growth as well as income inequality. As a result, those who are unable to
afford an education, or choose not to pursue optional education, generally receive much lower
wages. The justification for this is that a lack of education leads directly to lower incomes, and
thus lower aggregate savings and investment. Conversely, education raises incomes and
promotes growth because it helps to increase the productive potential of the poor. Education is
very important for the growth of the economy; however educational inequality in gender also
influence towards the economy. Gender inequality in education can result to slow economic
growth, and continued gender inequality in education, thus creating a poverty trap. It is
suggested that a large gap in male and female may indicate backwardness and so may be
associated with lower economic growth, which can explain why there is economic inequality
between countries. In addition, labor market success is linked to schooling achievement, the
consequences of widening disparities in schooling is likely to be further increases in earnings
inequality. By increasing the efficiency of the labor force education create better conditions for
good governance, improving health and enhancing equality. (Abdul j. and others 2011)

Trade openness: A simple Hecksher-Ohlin model says that countries export those factors (in
goods bundles) that they are relatively well endowed with. This increases the demand for their
abundant factors and through that raises relative prices of these factors. In general, developed
countries can be said to be well endowed with capital and developing countries with unskilled
labor. From this theoretical standpoint we can predict that openness would benefit unskilled
labors in developing countries and capital owners in developed countries. If more factor of
production and more countries than in the simple two goods- two factor- two countries- model
are included comparative advantages become more complicated. Depending on the distribution
of factor of production between countries we may define different hypothesis from this setting.
(Satheesh A. and others).

12
Other determinants: The country‟s natural resource, democratic system, tax system,
globalization, technological change and other demographic, political, historical, cultural and
natural factors determine the distribution of income of that country. But, such factors have little
power to determine income inequality.

2.1.3 Effects of Income Inequality on Economic Growth

Higher inequality lowers growth by depriving the ability of lower income households to stay
healthy and accumulate physical and human capital. Increasing concentration of incomes could
also reduce aggregate demand and undermine growth, because the wealthy spend a lower
fraction of their incomes than middle – and lower – income groups. Therefore, inequality affects
growth drivers (IMF, 2015). For instance, it can lead to underinvestment in education as poor
children n ends up in lower quality schools and are less able to go on to college. As a result,
labor productivity could be lower than it would have been in a more equitable world (Stiglitz,
2012). In the same vein, Corak (2013) finds that countries with higher levels of income
inequality tend to have lower level of mobility between generations, with parents earning being a
more important determinant of children‟s earnings. So, it also affects the future growth
perspective.

2.1.4 Effects of Income Inequality on Human Capital

Income inequality affects human capital negatively. It decreases the efficiency of people in
different ways. If children‟s are less effective at school, they are less likely to become highly
skilled workers. Their productive capacity and the productive capacity of the economy are
diminished. Inequality reduces performance because of its segregation effects. If schools are
segregated, children‟s from socioeconomically disadvantaged households mix with other
disadvantage children who do not perform well at school. Segregation with more likely in an
unequal society, the negative effects of poor children associated with less gifted are greater than
any positive effects of poor children associating with more gifted children. So, inequality may
cause a net reduction in education attainment.

The other way is higher rates of health and social problems (obesity, mental illness, homicides,
teenage births, incarceration, child conflict, drug use, and lower rates of social goods ( life

13
expectancy by country, educational performance, trust among strangers, women‟s status, social
mobility, even numbers of parents issued) in countries and states with higher inequality.
inequality and social stratification leads to higher levels of psychologically stress and status
anxiety. which can lead to depression, chemical dependency, less community life, parenting
problems amid stress related diseases. And, if people are not healthy, they will not work to their
full productive capacity.

High and persistent unemployment, in which inequality increases, has a negative effect on
subsequent long run economic growth. Unemployment can harm growth not only because it is a
waste of resources, but also because it generates redistributive pressures and subsequent
distortions, drives people to poverty, constrains liquidity limiting labor mobility, and erodes self-
esteem promoting social dislocation, unrest and conflict.

2.1.5 Effects of Income Inequality on Policies of the Country

Inequality can lead to policies that hurt growth. For example, it can lead to a backlash against
growth, enhancing economic liberalization and fuel protectionist pressure against globalization
and market oriented reforms (Claessens and Perotti, 2007). At the same time, enhancing power
by the elite could result in a more limited provision of public goods that boost productivity and
growth, and which disproportionately benefit the poor (Bourguignon and Dessus, 2009). In
addition, the policies may not be based on poverty reduction. Growth is less efficient in lowering
poverty in countries with high initial levels of inequality or in which the distributional pattern of
growth favors the non-poor. Moreover, to the extent that economies are periodically subject to
shocks of various kinds that undermine growth, higher inequality makes a greater population
vulnerable to poverty.

Higher income inequality led to less of all forms of social, cultural, and civil participation among
the less wealthy. When inequality is higher the poor do not shift to less expensive forms of
participation. Following the utilitarian principle of seeking the greatest good for the greatest
number of economic inequality is problematic. An additional dollar spent by a poor person will
go to things providing a great deal of utility to that person, such as basic necessities like food,
water and health care; while, an additional dollar spent by a much richer person will very likely
go to luxury items providing relatively less utility to that person. Thus, the marginal utility of

14
wealth per person (“the additional dollar”) decrease as a person becomes richer. From this stand
point, for any given amount of wealth in society, a society with more equality will have higher
aggregate utility.

2.1.6 Effects of Income Inequality on the Market Structure

Grater income inequality can lead to monopolization of the labor force, resulting in fewer
employers requiring fewer workers. Remaining employment can consolidate and take advantage
of the relative lack of competition, leading to less consumer choice, market abuses, and relatively
higher real prices. A number of economists have argued that inequality leads to economic
instability. One mechanism by which this happens is that the rich consume a smaller proportion
of their income than the poor. They save money which people on lower incomes would spend.
This leads to a reduction in aggregate demand. This in turn leads to unemployment. In response,
governments take measures to stimulate demand, such as lowering interest rates. This feeds into
asset bubbles -for example unsustainably high housing prices. (IMF, 2015).

2.1.7 Effects of Income Inequality on Environment

The smaller the economic inequality, the more waste and pollution is created, resulting in many
cases, in more environmental degradation. This can be explained by the fact that as the poor
people in the society become wealthier, it increases their yearly carbon emissions. This relation
is expressed by the environmental Kuznets curve (EKC). It should be noted here; however that in
certain case, with great economic inequality, there is nonetheless not more waste and pollution
created or the waste/ pollution is cleaned up better after words (water treatments, filtering…).
Also note that the whole of the increase in environmental degradation is the result of the increase
of emissions per person being multiplied by a multiplier. If there were fewer people however,
this multiplier would be lower and thus the amount of environmental degradation would be lower
as well. As such, the current high level of population has a large impact on this as well. If
population levels would start to drop to a sustainable level, human inequality can be addressed
/correlated, while still not resulting in an increase of environmental damage (IMF, 2015).

15
 2.1.8 Effects of Income Inequality on National Stability and Social Cohesion

Extreme inequality may damage trust and social cohesion and thus is also associated with
conflicts, which discourage investment. Conflicts are particularly prevalent in the management
of common resources whose; for example, inequality makes resolving disputes more difficult.
More broadly, inequality affects the economies of conflict, as it may intensify the grievances felt
by certain groups or can reduce the opportunity costs of initiating and joining a violent conflict
(Lichbach, 1989). Researchers have shown an inverse relationship between income inequality
and social cohesion. In more equal societies, people are much more likely to trust each other,
measures of social capital (the benefits of goodwill, fellowship, mutual sympathy and social
connectedness among groups who makes up social units) suggest greater community
involvement, and homicide rate are consistently lower.

2.1.9 Effects of Income Inequality on Finance

A prolonged period of higher inequality in advanced economies was associated with the global
financial crisis by intensifying leverage, overextension of credit, and a relaxation in mortgage,
underwriting standards, and allowing lobbyists to push for financial deregulation (Acemoglu,
2010).

2.2 Empirical Literatures Review

Getasew Alemu (2014): this paper examines the trend and extent of income inequality at rural,
urban, and national level and economic growth, with a focus on trends of income inequality at a
national level and the relationship between income inequality and economic growth. The result
showed that there is a slight increment in income inequality since 2004/05 up to 2010/11; urban
inequality is higher than rural and national income inequality. There is a negative relationship
between income inequality and economic growth.

Abdurahman Hassen (2014): the paper investigates the relationship between economic growth
and income inequality in case of Ethiopia for the period 1996-2011. It also tries to examine the
determinants of income inequality and applicability of Kuznets inverted U-hypothesis in the
Ethiopian case. The study employs OLS estimation techniques and to overcome statistical

16
problems of these method different techniques was employed. The findings of the study reveal
that economic growth; trade openness and general government expenditure has negative impact.

Jesper R., Jonas V. and Daniel W. (2009): this paper studies determinants of income inequality
using a newly assembled panel of 16 countries over the entire 20th century. It focuses on three
groups of income earners: the rich, the upper middle class and the rest of population. The result
shows that periods of high economic growth disproportionately increases the top percentile
income share at the expense of the rest of top deciles. Financial development is also pro-rich and
the out breaking crises are associated with reduced income shares of the rich. Trade openness has
no clear distributional impact (if anything openness reduces top shares). Government spending,
however, is negatively for the upper class and positive for the nine lowest deciles but does not
seem to affect the rich. Finally, tax progressivity reduces top income shares and when accounting
for real dynamic effects the impact can be important over time.

Zlatko N. (2006); used panel dataset covering 81 countries. The paper analyze and empirically
testes the relationship between inequality and its political and economic determinants. The study
find strong statically evidence that natural resource abundance/measured through the oil and gas
production as well as ores and metal exports/ is associated with increase inequality. There were
also strong statistical evidences that industrialization and economic growth are associated with
lower inequality. The study also shows weak evidence for the existence of the Kuznets curve.

Antonio A., Ludger S. and Vito T., (2008): this paper examines empirically the role and
efficiency of public spending policies in affecting income distribution from cross-country
perspective. This study first discuss conceptually the determinants of income inequality: initial
conditions and public policies affect income distribution directly, it then studies the relation
between distribution indicators on the one hand and public spending (except pensions) and
education performance have a significance effect on income distribution as reflected in stylized
facts and in the regression analysis. Results for the role of institutions and personal income taxes
point in the right direction but are not robust while more open countries do not have less equal
income distributions.

Abebe Fenta (2006): the main focus of this paper is to analysis the determinants of income
inequality among sampled households who find themselves at the bottom and top of the income

17
/consumption distribution in urban centers in South Wollo Administrative Zone, Ethiopia. The
result of this thesis indicates that the role of education in consumption expenditures is strongly
significant. The result of OLS and quantile regression analysis also shows that the household,
adult equivalent family size, household head main employment status or income sources, quality
of houses, household energy sources, durable goods/assets, wastes and sanitation and place of
residence are the main determinants of expenditure/income inequality of per adult equivalent
consumption expenditure across all quantiles distribution.

Inflation has an effect on income distribution due to its outcome on economic growth. Tobin-
Sidrauski portfolio shift model shows that inflation increases capital accumulation or diminish
capital accumulation. Tyson (1998) concluded that income of the poor, decline with the increase
of inflation which grinds down real minimum wages. There for the income distribution increase
while inflation increase. Some of the researcher agreed with that negative relationship exists
between Gini co-efficient and Inflation, on the contrary some agree with the negative relation of
Gini co-efficient and inflation while some examined with the aim of there is no alliance between
income inequality and inflation but in our the country the relationship of inflation and income
inequality is expected to positive relationship.

2.3 General Conceptual Framework

Income Inequality is the degree to which distribution of economic welfare generated in an

economy differs from that of equal shares among its nations that means income inequality is the
unequal distribution of total income among the population (Sid, 2014, Gehring and Kulkarni
(2016).Reconciling the three objectives of income inequality i. e, trends of income inequality,
determinants of income inequality and identifying of main determinants of income inequality in
less developed countries by using macro variables are challenging target. It is generally believed
that income inequality is world problem. Conceptual frame work shows how the dependent
and independent variables are related ; accordingly, this research have one dependent and
five independent variables ,where income inequality is dependent variable ,where as real GDP
capital squared, unemployment rate, inflation rate , trade openness, education, are the
independent variables .The dependent and independent variables are related in the following
way shown on the conceptual framework (Figure 2-3).

18
Unemployment Rate
(UEMRt

Degree of Tread Openness

measured by

Income Inequality
Education measured by Net
Primary School Enrollment
Rate

Inflation rate

Economic growth which

measured real GDP

Figure 2-3: Conceptual frame work of the study

Source: Own construction based on literature

CHAPTER THREE

RESEARCH METHODOLOGY
This part of the study gives details on how the research activities would carried out. Therefore,
the researcher concentrates on the methods that were adopt throughout the study to accomplish
the research objectives. It includes the research design, the type of data and source of data would
use the model specifications, types of model, estimation techniques, data analysis and
methodology.

3.1 Research Approach and Design

The research was use a quantitative research approach to analysis trends and determinants of
income inequality case of Ethiopia. Furthermore, the study was employing an explanatory

19
research design in order to achieve its objectives. It is the most appropriate design for identifying
the relationships between income inequality and its determinants by using macro variables.

3.2 Data Source and Types of Data

The study employs secondary data that were collect (1988 to 2018) from World Bank
Indicter(WBI),World Bank(WB),National bank of Ethiopia (NBE) Ethiopia Economic
Association (EEA),Central Statistical Agency (CSA), Ministry of Finance and Economic
Development (MOFED), International Monetary Fund (IMF), dataset, the United Nation
Conference on Trade and Development (UNCTAD) dataset websites.

3.3 Econometrics Model Specification

There are a lot of factors that affect income inequality. Such factors have been studied by many
researchers from different countries. Because of difference in the levels of economic
development and characteristics of the economic system, the determinants of income inequality
are not the same from one country to another even within the country. The most common
determinants are GDP per capita, the technological progress, financial development, openness to
trade, education, unemployment, inflation, urbanization, structure of the economy, government
expenditure, external debit and financial aid, foreign reserve and exchange rate, growth of
population, privatization and level of tax etc .In fact there are many determinants, these paper
select five of them based on their relevance for developing countries like Ethiopia. With this
framework the mathematically relationship between income inequality and its major
macroeconomic determinant are expressed as follows:

GINIt = ƒ (Yt2, PSERt, TOt, UEMRt, INFt)……………………………………..(3.1)

Whereas GINIt – Income Inequality, Yt2 - Real GDP per capita squared according to many
studies on the same study area, there is a non-liner relationship between income inequality and
economic growth (like Kuznets).Based on this, this paper expects a non-liner relationship
between them and economic growth of Ethiopia is not reach at maximum .So, it takes the
squared real GDP per capita variable. PSERt- education, TOt Degree of Tread Openness, UEMRt

20
UEMRt Unemployment rate and INFRT Inflation rate. Thus, an explicit estimable econometric
model is formulated as follows;

LnGini=β0+β1lnY2t+β2lnPSERt+β3lnTOt+ β4 lnUEMRt + β3 lnINFT + et…………………(3.2)

Researcher transformed all the variables into Log data to convert nonlinear to linear and avoid
hetroscedasticity (Gujarati, 2004) and to show elasticity of the variables. Where all variables are
defending previously except, et, white noise process/marginal errors and t, time. Log
transformation can reduce the problem of heteroscedasticity because it compresses the scale in
which the variables are measured; thereby reducing a tenfold difference between two values to a
twofold difference (Gujarati, 2004).It is important to note that the model is a multiplicative one
where all parameters (coefficients) represent constant Elasticties.

3.4 Estimation Procedure

To test the long run relationship between dependent variable (income inequality ) and
independent variables (Real GDP per capita squared, , education, Degree of Tread Openness,
Unemployment Rate and Inflation Rate). The study was first investigating the time series
properties of our data / unit root tests of our data /by using Augmented Dickey-Fuller (ADF) and
Philip-Perron (PP) tests. After testing unites root test researcher was use ARDL model based on
unit root test result.

3.5 Model Specification

To time series data we have three main types of models, Vector Error Correction (VECM)
model, Auto Regressive Distributed Lag model (ARDLM) model and Vector Auto Regressive
(VAR) model. All the variables in a VAR model are endogenous, there is no exogenous variable.
Based on data researcher was use one of among models. Researcher was chosen model after
testing of data. The variables were integrated of different order, that is a model having
combination of variable with I(0) and I(1) order of integration, due to this reason researcher
was used ARDL model .ARDL model uses a combination of endogenous and exogenous
variables ,unlike a VAR model thet‟s strictly for endogenous variables ,from the boundl test
of the result .Because of the variables are integrated of different order ,that is a model having

21
combination of variable with I(0) and I(1) order of integration ,which are not integrate order
two and co integrated ,researcher was apply both long run (ARDL) and short run (VECM)
models. ARDL model is relatively more efficient in the case of small and finite sample data
sizes.

According to Gujarati, Fourth Edition (2004), the ARDL modeling of unrestricted error
correction model using Ordinary Least Square (OLS) can be representing as follows.

∆Yt=β0+∑ ∆Xt ∑ ( )

Where ∆ denotes for first difference operation, Yt is for a vector of dependent variables, Xt is a
vector of independent variables ,p is optimal leg length, ut is the residual term which is assumed
to be white noise.

In order to test the existence of long-term relationship among the variables, the following
equation will estimate by applying OLS.

∆GINt=βo+ ∑ ∑ ∑
∑ ∑
( )

Where asGINIt- Income Inequality Yt2-Real GDP per capita squared, PSERt-education, TOt-
Degree of Tread Openness ,UEMRt –Unemployment Rate and General inflation rate-INFt ,ut is
the residual term, which is assumed to be white noise, p is the optimal lag length and ln
is natural logarithm. To test the significance of lagged level of the variables under
consideration, the appropriate statistic is F or Wald test as Pesaranet al. (2001) proposed
for bound test approach was applied. The bounds test is mainly based on the joint Wald
test or F- test which its asymptotic distribution is non-standard under the null hypothesis of no
co integration. The null hypothesis for no co-integration in the long-run among the variables in
equation [3.4] is:-

Ho =θo=θ1=θ2=θ3=θ4=θ5=0 (meaning no long run relationship among the variables) against the
alternative one:

H1#θo#θ1# θ2 # θ3 # θ4# θ5# 0The F-test has no standard distribution which

22
depends on (i) whether the variables include in the model are I(0), or I(1), (ii) the numbers of
regressors, and (iii) whether the model contains an intercept and/or a trend (Nara yan, 2008).
According to Gujarati: (2004), there are two sets of critical value bounds for all
classifications of regressors‟ namely upper critical bound value and lower critical bound value.
The critical values for I (1) series are referred to as upper bound critical values; while the critical
values for I (0) series are referred to as lower bound critical values. If the calculated F statistic is
greater than the upper bound critical values, we reject the null hypothesis of no long run
relationship among the variables. If the calculated F statistic is less than the lower bound critical
values, we can‟t reject the null hypothesis rather accept the null hypothesis of no co integration
among the variables. However, if the calculated F statistic is between the upper and lower bound
critical values, inference is inconclusive and we need to have knowledge on the order of
integration of underling variables before we made conclusive inference (Gujarati,
2004).However, in this study we are not going to follow the bound critical value developed by
Pesaran because of the computed critical values are based on large sample size (500 and
more). Rather, a relatively small sample size in this study of 31 years observations, researcher
was use the critical values which are get from Eview‟s result .There is an evidence of long-
run relationship (co integration) and the variables are integrated of different order, the
following Long-run ARDL(P1,P2,P3,P4,,P5 ) model would be estimated.

∑ ∑ ∑
∑ ∑ ∑
( )

Here all variables are as previously defined. The orders of the lags in the ARDL Model is
selected by the Akaike Information Criterion (AIC) .Researcher was use the Akaike
Information Criterion (AIC) in lag selection because of its advantages for small sample
size (Tsadkan, 2017). Determination of the optimal lag length is two, so it is crucial in ARDL
model, because of it helps us to address the issue of over parameterizations and to save the
degree of freedom (Taban, 2010) as cited in Tsadkan (2013). For annual data, Pesaran and
Shin(1999) recommend choosing a maximum of 2 lags. From this, the lag length that
minimizes Akaike Information Criterion (AIC) is selected. In the presence of co integration,

23
short-run elasticity‟s/dynamics/ can also be derived by constructing an Error Correction Model
of the following Form:

; ∑ ∑ ∑
∑ ∑ ∑
( )

Where ECTtis the Error Correction Term, defined a

: ( ∑ ∑ ∑

) ( )

Here ∆ is the first difference operator; β’s are the coefficients relating to the short -run dynamics
of the model's convergence to equilibrium, and Y measures the speed of adjustment.

3.6 Unit Root Test

It is fundamental to test for the statistical properties of variables when dealing with time series
data. Time series data are rarely stationary in level forms. Regression involving non-stationary
(I.e., variables that have no clear tendency to return to a constant value or linear trend)
time series often lead to the problem of spurious regression. This occurs when the regression
results reveal a high and significant relationship among variables when in fact, no
relationship exist. Moreover, Stock and Watson (1988) have also shown that the usual test
statistics (t, F, DW, and R2) will not possess standard distributions if some of the variables in the
model have unit roots. The other necessary condition for testing unit root test when we
applying ARDL model is to check whether the variables enter in the regression are not
order two (I.e. I(2)), which is precondition in ARDL model. Therefore, it is necessary to test
for time series variables before running any sort of regression analysis. Non-stationarity can be
tested using Augmented Dickey-Fuller (ADF) test, Phillips Perron (PP) test and Kwiatkowski-
Phillips-Schmidt-Shin (KPSS) test. However, to ensure reliable result of test for stationarity,
the study employs both Augmented Dickey-Fuller (ADF) test and Philip Perron (PP) tests.

The testing procedure for the ADF unit root test is specified as follows:

24
 ∑ ( )

Where Yt is a time series variables under consideration in this model at time t is a time trend

Variable, λ speed of adjustment, Δ denotes the first difference operator; εt is the error term; p is
the optimal lag length of each variable chosen such that first -differenced terms make a white
noise. Thus, the ADF test the null hypothesis of no unit root (stationary).

That is: H0: = 0; H1: # 0

If the t value or t -statistic is more negative than the critical values, the null hypothesis (I.e. H0)
is rejected and the conclusion is that the series is stationary. Conversely, if the t -statistic is
less negative than the critical values, the null hypothesis is accepted and the conclusion is that
the series is non-stationary.

3.7 Description of variables

The dependent variable is income inequality. There are many types of measurements that
measures income inequality in the global, country and regional level. Gini coefficient is the most
common or popular measures of income inequality in the world duo this it used. The model
includes five explanatory variables. One of the independent variable is economic growth. This
variable is measured by real GDP per capita. GDP per capital is growth domestic product
products divided by midyear population. According to many studies on the same study area,
there is a non-liner relationship between income inequality and economic growth (like Kuznets).
Based on this, this paper expects a non-liner relationship between them. So, it takes the squared
real GDP per capita variable. For the case of Ethiopia, it expects a positive relationship between
them. Second independent variable of this study is education. When we take education for the
purpose of this study, it can be measured by many measurements. Primary school enrollment rate
is the most common measurement of the countries education level. Net primary school
enrollment rate is defined as the number of children enrolled in primary school that belongs to
the age group that officially corresponds to primary schooling, divided by the total population of
the same age group. Education creates a high wages for those with this education, and then it
leads to higher competition in the labor market. Thus, uneducated peoples will be unemployed

25
and they can‟t generate income. Finally, the income gap between the educated and uneducated
increased. Therefore, net primary school enrollment rate expected to affect income inequality
negatively. Third independent variable of this study is Trade openness. Trade openness is a
measure of economic policies that either restrict or invite trade between countries .It can be
calculated as the simple average of total trade (i.e. the sum of exports and imports of goods and
services) relative to GDP. According to Hecksher- Ohlin model, developing countries are
thought to have more unskilled labor relative to skilled labor (and/or relative to capital) is
assumed to be unequally distributed across the population and the increase in the relative demand
for skilled labor (capital) in developed countries as a result of trade the distribution of income
between rich and poor are not equal .But, within one developing country trade used to efficiently
utilizing the hidden resource and the poor‟s with unskilled labor start generate a better income.
So it is expected to get a negative relationship between income inequality and Trade openness.
The fourth independent variable of this study is Unemployment Rate. Unemployment occurs
when people who are without work are actively seeking work. The most frequently used measure
of unemployment is the unemployment rate. The unemployment rate is a measure of the
prevalence of unemployment and it is calculated as a percentage by dividing the number of
unemployed individuals by all individuals currently in the labor force. The rise in unemployment
rate results high dependency ratio and lower per capita GDP. In one family, if the number of
unemployed members is larger than the employed, the overall income of that family will be
lower when we compare it with the family with most of the family members with job. As a
result, the gap between the rich with a job (employed) and the poor without job (unemployed)
widen with increased unemployment rate. So, it is expected to get a positive relationship between
income inequality and unemployment rate. The least independent variable of this study is general
Inflation (INF), Inflation is defined as an increase in the overall price level in a country and
measured in percent (CPI). Therefore to analyze its effect on income inequality, it is the other
interest of the researcher‟s, which is included in this study as independent variable. The
coefficient of this variable would be expected a positive sign. Inflation is measured in percent
(CPI). Inflation reduces the purchasing power of individual as a result demand of goods
produced by individuals will significantly increase. This implies that income inequality is
increase. Therefore, positive sign is expecting for the estimated coefficient of the inflation
variable in the regressions.

26
 Table 3-1: Description of variables

Variables Short name Expected sign Results of Description

sign
Income Inequality GINIt Dependent Dependent Measured by GINI coefficient
Variable Variable
Economic growth Yt2 + + Measured Real GDP per capita
squared
Education PSERt - - Measured by net primary school
enrollment Rate
Tread Openness TOt - - Measured by (Tread/GDP

Unemployment UEMRt + + Measured by unemployment/total

Rate labor force in the market

Inflation rate INFt + or - + Measured by CPI

3.8 Methods of Data Analysis

The study was using both the descriptive and econometric methods of data analysis. To analyze
the data, in descriptive part, the researcher was use tables; figures and trend of graphs to describe
the given data. On the other hand standard econometrical technique would apply to analyze the
major determinants income inequality under the study period. In econometric part researcher was
use the following multivariate models i.e, Auto Regressive Distributed Lag model (ARDL)
model and Vector Error Correction (VEC) model. Finally, Eview 10.0 versions have been used
as statistical software package for the entire analyze running this study.

27
 CHAPTER FOUR

DATA ANALYSIS AND INTERPRETATION

The main aim of this section is to provide results of the study as well as to discuss each one of
them rigorously. All the variables were tested using descriptive and econometric methods of data
analysis. In addition in line with the objectives of the paper secondary data from National bank
of Ethiopia (NBE), World Bank (WB), Ethiopia Economic Association (EEA), Central Statistical
Agency (CSA), Ministry of Finance and Economic Development (MOFED), International
Monetary Fund (IMF), dataset, the United Nation Conference on Trade and Development
(UNCTAD) dataset, etc.as used to study the trends of income inequality in Ethiopia and its
determinants by using macro variables. This section begins with the discussion of recent trends
in Ethiopia„s income inequality. Eventually this section presents the final results of the study
supported by econometric analysis. This chapter contains both the descriptive and econometrics
analysis. Under the descriptive statistics the trends and overall performances of the variables of
interest are presented. The statistical tools such as tables and graphs are used to describe
the variables used in the model. The econometric analysis begins by testing the
necessary tests such as stationary tests, diagnostic tests and bound test. After passed the
necessary testes both the long run and short run model are estimated using ARDL model and
VECM model respectably. Because of the variables are integrated of different order , that is a
model having combination of variable with I(0) and I(1) order of integration and two
variables are co-integrate. Estimation has been made the interpretation and discussions are
continued based on the models result.

4.1 Descriptive analysis

4.1.1 Trend of Income Inequality

According to World Bank report Gini coefficient of Ethiopia was 0.35 in 2015. This records an
increase from the previous numbers of 0.33 for 2010. The GINI coefficient 2001 and 2002, 1988
and 1989, 1990 and 1991, 1992 and 1993 are 0.30, 0.37, 0.38and 0.39 respectably. The trend of

28
income inequality from 2003 to 2018 is fluctuating year to year.The minimum and maximum
value of Gini is 0.29 in 2000 and 0.44 in 1995 respectably.

According to Todaro (2012) The Gini coefficient of countries with highly unequal income
distribution typically lies between 0.50 and 0.70, relatively equal distributions, it lies between
0.20 and 0.35 and it is approximately 0.44 for a relatively unequal distribution . The average
(mean) value of Gini in Ethiopia which is 0.34 lies between 0.20 and 0.35, so, this represents
there is relatively equal distribution. A trend of income inequality in Ethiopia from 1988 to 2018
is given graphically below.

48

44
Gini coefficient

40

36

32

28
88 90 92 94 96 98 00 02 04 06 08 10 12 14 16 18

Years

Figure 4-1: Trends of income inequality in Ethiopia from 1988 to 2018

Source: Computed based on WBI data (2020)].

4.1.2 Trends of GDP per capital or Economic growth

According to World Bank , the real GDP per capital of Ethiopia was 504.533 million birr in
1988 and it reach 20,143.0933 million in 2018 . Figure 4.1.2 below showed that from 1988 to
2018 the sharply up wards that indicates higher rate is attributed due to a combination of
pro poor growth policy (since 2003 on wards ) and state led development program (since
2005 on wards ) and the present government implementing a development program aimed
at poverty reduction through rapid economic growth and macroeconomic

29
stability(Zerayehun,,2013) .This indicates that the probability of achieving high future real GDP
growth rate is high. A trend of GDP per capital in Ethiopia from 1988 to 2018 is given
graphically below.

24,000
GDP per capital in Million Birr

20,000

16,000

12,000

8,000

4,000

0
88 90 92 94 96 98 00 02 04 06 08 10 12 14 16 18
Years

Figure 4-2: Trends of GDP per capital in Ethiopia from 1988 to 2018
Source: Computed based on WB data (2020)].

4.1.3 Trend of net primary school enrollment rate or Education

The countries level of education measured by primary school enrollment rate shows ups and
downs from 1988 up to 2015.The net primary school enrollment rate shows a good improvement,
it increases dramatically from 2016 to 2018.According to World Bank and UNESCO, School
enrollment, primary (% net) in Ethiopia was 29.97% in 1988. The average value of Ethiopia
during period was 52.50 percent with minimum of 19.18 percent in 1994 and a maximum
of 89.45 percent in 2018 .This means at this time most of the children‟s get a primary school
education. A trend of School Enrollments rate, Primary Ethiopia from 1988 to 2018 is given
graphically below.

30
 100
Net Primary School enrollment rate

90

80

70

60

50

40

30

20

10
88 90 92 94 96 98 00 02 04 06 08 10 12 14 16 18

Years
Figure 3-3: Trends of Net Primary School Enrollments rate in Ethiopia from 1988 to 2018
Source: Computed based on WBI data (2020)].

4.1.4 Trend of Trade openness

Trade openness refers to the outward or inward orientation of a given country‟s

economy. Outward orientation refers to economies that take significant advantage
of the opportunities to trade with other countries. Inward orientation refers to
economies that overlook taking or are unable to take advantage of the opportunities
to trade with other countries. Trade openness is exports plus imports as percentage of GDP. For
that variable, researcher provides data of Ethiopia from 1988 to 2918. The average value of a
data was 35.70 percent with a minimum of 23.45 percent in1994 and a maximum of 59.532
percent in 2018.This shows that the sum of the export and imports of goods and services relative
to GDP is ups and downs from the 1988 up to 2001. Some of the trade policy decisions made by
countries that empower out ward or inward orientation are trade, barriers, import, export,
infrastructure technologies, scale economies and market competitiveness.

31
The trends of the graph indicates that the transaction of import-export between
the years 1988 to 2001‟s show ups and downs, it had a little fluctuation; however, since the
beginning of the year 2002 up to 2006 was increasing. Starting from 2007 up to 2013 its up‟s
and down‟s, finally starting from 2014 up to 2018 is significantly increasing. A trend of trade
openness in Ethiopia from 1988 to 2018 is given graphically below.

60

55

50
Trade Openness

45

40

35

30

25

20
88 90 92 94 96 98 00 02 04 06 08 10 12 14 16 18

Years

Figure 4-4: Trends of Trade openness in Ethiopia from 1988 to 2018

[Source: Computed based on WBI data (2020)].

4.1.5 Trend of Unemployment rate

In 2018, unemployment rate of Ethiopia was 2.1%. Unemployment rate of Ethiopia was fell
gradually from 5.3% in 1990 to 2.1 % in 2018 .Ethiopia unemployment rate for 2019 was 2.08,
a 0.01 increasing from 2018 and a 0.04 decline from 2017 . Ethiopia unemployment rate in
2017 was 2.12 %, a 0.05 decline from 2016. Ethiopia unemployment rate in 2016 was 2.17%, a
0.03 % decline from2015. The average value of a data was 2.9 percent with a minimum of 2.1 percent
in 2018 and a maximum of 5.3 percent in 2011. A trend of unemployment rate in Ethiopia from 1988
to 2018 is given graphically below.

32
 5.5

5.0

4.5
Unemployment rate

4.0

3.5

3.0

2.5

2.0
88 90 92 94 96 98 00 02 04 06 08 10 12 14 16 18

Years

Figure 4-5: Trends of unemployment rate in Ethiopia from 1988 to 2018

:
Sources https://www.macrotrend.net/countries/Ethiopia/unemlyomentrate

http://www.statista.com/statistics/unemployment rate in Ethiopia

https://tradingeconomic.com/Ethiopia/unemploymenrtrate and
http://Knoma.com/atlas/ethiopa/unemployment rate
4.1.6 Trend of Inflation in Ethiopia

Trends of inflation show the change in the inflation over the years. Inflation remained at a
reasonable low level rate before 2000/03. However, post 2003/04 sow sharp increase despite
rapid economic growth during the same period (Alemayehu and Kibrom, 2008). According to
Alemayehu and Kibrom (2008), the sharp increasing of general inflation was caused primarily by
food inflation, which is the effect of food demand triggered and international food price hike.

The official headline inflation during 2008 stood at about 33 percent with food inflation being
about 44.4 percent. This was huge macroeconomic shock in the history of Ethiopia for the last
five decades and until 2003, was below 5 percent per annum (Ibid). This high rate of inflation
continued until 2017/18.

33
In 2018, inflation rate of Ethiopia was 13.8%. Though Ethiopia inflation rate fluctuated
substantially in years, it tended ups and downs, through 1988 to 2018 period ending 13.8% in
2018. The inflation rate of consumer price in Ethiopia moved over the 31 years between 2.2 % in
1988 and 44.4 in 2008.The average value of a data was 10 .38 percent with a minimum of -8.2
percent in 2001 and a maximum of 44.4 percent in 2008. A trend of inflation rate in Ethiopia
from 1988 to 2018 is given graphically below.

50

40

30
Infaltion Rate

20

10

-10
88 90 92 94 96 98 00 02 04 06 08 10 12 14 16 18

Years

Figure 4-6: Trends of Inflation rate in Ethiopia from 1988 to 2018

[Source: Computed based on WBI data (2020)].

4.2 Econometric Analysis

4.2.1The Unit Root Test Analysis

In order to determine the degree of integration, a unit root test is carried out using the standard
Augmented Dickey-Fuller (ADF) and Phillips-Person test statistic (PP) test. Moreover in
applying ARDL model all the variables entered in the regression should not be integrated of
order two. To check these conditions, unit root test is conducted before any sort of action taken.
Even though the ARDL framework does not require per-testing variables to be done, the unit
root test could convenience us whether or not the ARDL model should be used. The result in
Table 4-1 shows that there is a mixture of I (0) and I (1) but not any order two.

34
 Table 4-1: Unit root test (Augmented Dickey-Fuller test

Augmented Dickey-Fuller test statistics (ADF-Test )

Variables With Intercept Trend and Intercept

At At first Order [ ] At Level At first Order ( )
Level difference differen
ce
Gini I[0]at1,5 and 10 I[0]at1,5 and 10 %
Coefficient(LGini) -1.79 -6.177 % -1.86 -6.088

I[0] at 1, 5 and I[0]at1,5 and 10 %

GDP Per Capital 0.747 3.11 10 % -1.581 -3.085
(LY)
Net Primary School I[0] at1, 5 and I[0] at1, 5 and 10 %
Enrollment -0.1144 -4.248 10 % -2.07 -4.1257
Rate(LPSER)
Unemployment Rate I[0] at1, 5 and I[1]at1,5 and 10 %
(LUEMR) -1.680 -4.422 10 % -3.83 -2.895
I[0] at1, 5 and I[1]at1,5 and 10 %
Inflation Rate -1.887 -7.598 10 % -3.35 -3.083
(LINFR)
I[1] at 1, 5 and I[0]at1,5 and 10 %
Trade -3.48 -2.065 10 % -2.546 -3.2205
Openness(LTO)
MacKinnon (1996) with constant, no trend with constant and trend

Test critical values:1% level = -3.67 Test critical values:

5% level = -2.96 1% level = --4.2967
10% level = -2.62 5% level = -3.5683
10% level = -3.2183
Note: If absolute value of t- Statistics is less than Test of Note: If absolute value of t- Statistics is less
critical values then the data is stationery or if probability is than Test of critical values then the data is
greater than 5% then data is stationary i.e we accept null stationery or if probability is greater than
hypothesis 5% then data is stationary i.e we accept null
hypothesis

Source: Eview 10.0 results

As we have seen form Table 4-1. Gini coefficient, real GDP per capital, primary school of
enrolment rate, inflation rate, and unemployment rate are integrated of order zero (I.e. I(0)) while
trade openness is integrated of order one (I(1)). Meaning Gini coefficient, real GDP per capital,
primary school of enrolment rate, inflation rate and unemployment rate are stationary in level

35
 where as trade openness is stationary in first difference (with intercept). However, with trend and
Intercept, except unemployment rate and inflation rate, all the variables are stationary in level.

Table 4-2: Unit root test (Phillips-Perron test statistic test)

Phillips-Perron test statistic (PP Test)

Variables With Intercept Trend and Intercept

At Level At first Order [ ] At At first Order [ ]
differen Level differen
ce ce
I[0]at1,5 and 10 % I[0]at1,5 and 10 %
Gini Cofficient -1.83 -6.165 -1.98 -6.0844
(LGINI)
I[1] at 1, 5 and 10 % I[1]at1,5 and 10 %
GDP per 11.09 0.516 6.32 -1.789
Capital (LY)
Net Primary I[0] at1, 5 and 10 % I[0] at1, 5 and 10 %
School -0.2722 -4.25 -2.08 -4.134
Enrollment
rate (LPSER)
Unemployment I[0] at1, 5 and 10 % I[1]at1,5 and 10 %
Rate -1.517 -7.09 - -2.923
(LUEMR) 3.624
Inflation Rate I[0] at1, 5 and 10 % I[0]at1,5 and 10 %
(LINFR) -1.887 -7.598 - -3.783
3.082
I[0] at 1, 5 and 10 % I[0]at1,5 and 10 %
Trade -1.338 -4.894 - -4.91949
Openness 0.496
(LTO)
MacKinnon (1996) with constant, no trend with constant and trend
Test critical values:1% level = --4.31
Test critical values:1% level = -3.679 5% level = -3.57
5% level = -2.967 10% level = -3.22
10% level = -2.622
Note: If absolute value of t - Statistics is less than Test of critical Note: If absolute value of t - Statistics
values then the data is stationery or if probability is greater than is less than Test of critical values then
5% then data is stationary i.e we accept null hypothesis the data is stationery or if probability
is greater than 5% then data is
stationary i.e we accept null hypothesis

Source: Eview 10.0 results

36
Similarly, the PP test shows that there is a mixture of integration order zero and order one. That
is, Gini coefficient , primary school of enrollment rate , unemployment rate , inflation rate and
trade openness are stationary in level while real GDP per capital is e stationary in first difference
(with intercept only). However, except unemployment rate and real GDP per capital all the
variables are stationary at level with intercept and trend. Form table 4-1 and 4-2 we can conclude
that none of the variables entered in the regression are order two, which are not desire in
applying ARDL model. So ARDL co integration technique proposed by Pesaranet al. (2001) is
the most appropriate method for estimation or to check the long run relationship among the
variables.

4.3 Model Stability and Diagnostic Test

To check the verifiability of the estimated long run model, some diagnostic tests are undertaken.
Priority in doing any analysis, researcher required to check the standard property of the model. In
this study researcher carried a number of model stability and diagnostic checking, which includes
Functional form (Ramsey‟s RESET) test Normality (Jaque-Bera test), Multicolinearity (Variance
Inflation Factor test), Autocorrelation test (Durbin-Watson test) and Hetroscedasticity (Breusch-
Pagan-Godfrey test .) In addition to the above diagnostic tests, the stability of long run estimates
has been tested by applying the cumulative sum of recursive residuals (CUSUM) and the
cumulative sum of squares of recursive residuals (CUSUMSQ) test. Such tests are recommended
by Pesaranet al. (2001). In order to reject or accept the null hypothesis, we can decide by looking
the p-values associated with the test statistics. That is the null hypothesis is rejected when the p-
value are smaller than the standard significance level (I.e. 5%).

4.3.1 Test of Multicolinearity

Multicolinearity refers to the condition that variable are correlated and it‟s the features of sample
for the population. The classical linear regression models assume that there is no multicollearnity
among the explanatory variables. If the perfect multicolinearity exist, the regression coefficient
of the explanatory variable are indeterminate and there standard errors are infinite cannot be
estimated the coefficient with greater accuracy. In order to test multicolinearty, the paper used
Variance Inflation Factor (VIF). The larger the mean value of VIF, the more some variable

37
occurred. As the rule, if the mean of VIF is greater than 5 (VIF>5), that variable is highly
collinear between explanatory variable (Gujarati, 2004).

Table 4-3: VIF test for Multicolinearity

Variables VIF 1/VIF

Constant 11.49 0.0870

GDP per capital (LY) 7.56 0.1323
Net Primary School Enrollment Rate (LPSER) 2.74 0.3651
Unemployment rate (LUEMR) 2.0 0.4897
Inflation rate (LINF) 1.12 0.8890
Trade Openness (LTO) 1.38 0.7270

Mean of VIF 4.39

Source: Eview 10.0 results

From the above table the mean of VIF shows that there is no a problem of multicollinearty or
linear relationship between a given explanatory variables. If the mean value of VIF greater than
5, then we would say there is a problem of multicolinearity. However, it is far less than 5
implying there is no the problem of multicollinearity.( Gujarati,2004).

4.3.2 Functional form (Ramsey RESET test)

Ramsey RESET test is stands for regression specification error test and was proposed by Ramsey
(1969). The Ramsey Regression Equation Specification Error Test (RESET) test is a
general specification test for the linear regression model. More specifically, it tests whether non-
linear combinations of the fitted values help explain the response variable. The intuition behind
the test is that if non-linear combinations of the explanatory variables have any power in
explaining the response variable, the model is misspecified in the sense that the data generating
process might be better approximated by a polynomial or another non-linear functional form so,
when we test the specification of the functional form the following result was obtained.

38
Table 4-4: Functional form (Ramsey RESET Test)

Value Df Probability

t- statistics 1.8326 24 0.0724

3.3584 (1,24) 0.0724

F- statistics

Likelihood ratio 3.4995 1 0.0614

Note: Decision criteria of RESET test ,if t- statistics ,F- statistics and likelihood ratio are not
significant since the probability value are greater than 0.05. It means the estimated model is free
from specification errors.

Source: Eview‟s 10.0 results

We could not reject the null hypothesis test for Ramsey‟s RESET test, which tests whether the
model suffers from omitted variable bias or not. As the test result indicates above we can‟t reject
the Ramsey‟s test, which means that the model is correctly specified.

4.3.3 Test of Hetroscedasticity

To test Hetrscdasticity, the Breusch-pagenGodfrey test is used. The result shows as follows; as
an important assumption of the classical linear regression model is that the disturbance μi
appearing in the population regression function is homoskedastici.e. They all have the same
variance but when there is exist an outlying observation in relation to the observation in the
sample the assumption of constant variance is violated. This violation refers to as
hetroscedastisticity which leads to estimator to be inefficient and, estimated variance to be
biased.

Table 4-5: Test of Hetroscedasticity

Heteroskedasticity Test: Breusch-Pagan-Godfrey

F-statistic 0.4783 Prob. F(8,20) 0.8571

Obs*R-squared 4.6580 Prob. Chi-Square(2) 0.7934
Scaled explained SS 3.1214 Prob. Chi-Square(8) 0.9265

Source: Eview 10.0 results

39
As we have seen from the above table, we can reject alternative hypothesis at 5% significant
level due to its p-value associated with the test statistics are greater than standard significance
level( I.e. 0.7934> 0.05).From the above result the prob (chi (2)>5% level of significant that is
accept the null hypothesis so, the error term is not hetroscedasticity that means there is the
problem of homoscedastic.

4.3.4 Tests for Autocorrelation

The disturbance term of any observation is not influence by the disturbance term of any other
observations. However, if there is such dependence there is autocorrelation. The simplest and
widely used model is one where the error term μt and μt-1 have correlation p. For this model one
can testing hypothesis about p based on estimated correlation coefficient between the residuals.
A common used statistic for this purpose is the Durbin-Watson (DW) denoted by DW. When the
DW statistic is zero DW=0, there is a series positive autocorrelation. When the Durbin-Watson
statistic (DW) = (1.5<DW<2.5), there is no autocorrelation problem. If the DW closes to 4, there
is a series negative autocorrelation. In addition to this, to test a correlation R can be used. If R is
greater than Durbin- watson statistic, there is a series problem of autocorrelation. From the
regression result DW=1.98 it is found between 1.5 and 2.5 (1.5<1.98<2.5) so, there is no the
problem of autocorrelation.

4.3.5 Test for normality

The model assumes that the random variable u has a normally distributed. Symbolically:
( ), which reads as: u is normally distributed around zero mean and constant variance
This means that small values of u‟s have a higher probability to observed than large values.
This assumption is necessary for constructing confidence intervals. If the assumption of
normality is violated, the estimates of parameters are still unbiased but the statistical reliability
by the classical tests of significance of the parameters cannot be assessed because these tests are
based on the assumption of normal distribution of the u. The null hypothesis is that has normal
distribution against the alternative hypothesis that the u is not normally distributed.

40
9
Series: Residuals
8
Sample 1990 2018
7 Observations 29

6 Mean 1.90e-15
5
Median 0.256655
Maximum 5.228063
4 Minimum -5.567476
Std. Dev. 2.138817
3
Skewness -0.162783
2 Kurtosis 3.817873

1
Jarque-Bera 0.936350
0 Probability 0.626144
-6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6

Figure 4-7: Normality Test

Source: Eviews 10.0 results

As the result indicates that we could not reject the null hypothesis which says that the residuals
are normally distributed, for the reason, that the p-value associated with the Jaque-Berra
normality test is larger than the standard significance level (I.e. 0.937>0.05) ,then error term is
normally distributed. Moreover, the stability of the model for long run and short run relationship
is detected by using the cumulative sum of recursive residuals (CUSUM) and the cumulative
sum of squares of recursive residuals (CUSUMSQ) tests. The test finds serious parameter
instability if the cumulative sum goes outside the area (never returns back) between the two
critical lines.

41
15

10

-5

-10

-15
2000 2002 2004 2006 2008 2010 2012 2014 2016 2018

CUSUM 5% Significance

Figure 4-8: Plot of Cumulative Sum of Recursive Residuals

Source: Eview‟s 10.0 results

The straight lines represent critical bounds at 5% significance level.

As can be seen from the above figure, the plot of CUSUM test did not cross the critical limits.

So, we can conclude that long run estimates are stable.

1.6

1.2

0.8

0.4

0.0

-0.4
2000 2002 2004 2006 2008 2010 2012 2014 2016 2018

CUSUM of Squares 5% Significance

Figure 4-9: Plot of Cumulative Sum of Squares of Recursive Residuals

Source: Eview‟s 10.0 results

42
The straight lines represent critical bounds at 5% significance level.

As can be seen from the first figure, the plot of CUSUM test did not cross the critical limits.
Similarly, the CUSUM of squares test shows that the graphs do not cross the lower and upper
critical limits. So, we can conclude that long run estimates are stable and there is no any
structural break. In addition to the model stability 75.6 percent of the model has been explained
by the regressors. Hence the results of the estimated model are reliable and efficient.

4.4 Long Run ARDL Bounds Tests for Co-integration

Since researcher determined the stationary nature of the variables, the next task is the bounds
test approach of co-integration is estimating the ARDL model specified in equation (3.5) using
the appropriate lag-length selection criterion. According to Pesaran and Shine (1999), as cited in
Narayan (2004) for the annual data are recommended to choose a maximum of two lag lengths.
From this, a lag length that minimize AIC is 2. In addition to this, researcher have also used AIC
to determine the optimal lag because it is a better choice for smaller sample size data as this
study. Apart from this, AIC found to produce the least probability of under estimation among all
criteria available (Liewet al., 2004) as cited in Tsadkan (2014). As we discus in the third part of
this study, the F-test through the Wald-test (bound test) is performed to check the joint
significance of the coefficients specified in equation (3.5). The Wald test is conducted by
imposing restrictions on the estimated long-run coefficients of Gini coefficient, real GDP per
capital, primary school of enrollment rate, unemployment rate, trade openness and inflation rate.
The computed F-statistic value is compared with the lower bound and upper bound critical
values provided byEview‟s 10.0 result

43
 Table 4-6: F-Bounds test

F-Bounds test Lag length Critical value Lower Bound Upper Bound
statistics value Or I(0) Or I(1)

6.100904 2 1 percent 3.06 4.15

5 percent 2.39 3.38

10 percent 2.08 3
Note : Decision criteria for Bounds test , If the calculated F-statistics is greater than the critical values for
upper bound I(1), than we can conclude that there is co integration. That is along run relationship. Reject
the null hypothesis. Estimate the long run model which is the error correlation model (ECM ).

If the calculated F- statistics is lower than the critical value for lower bound I(0), then we conclude that
there is no co-integration, hence no long run relationship. Do not reject the null hypothesis. Estimate the
short run model which is Autoregressive Distribute Lag (ARDL) model.

If the F-statistics falls between the lower bound I(0) and the upper bound I(1).the test is considered
inclusive.

Source: Eview‟s 10.0 results

As it is depicted in Table 4.6 above, with an intercept and trend, the calculated F statistics
(6.100904) is higher than upper bound critical values at 1% 5% and 10% level of significance.
This implies that the null hypothesis of no long-run relationship is rejected; rather accept the
alternative hypothesis (there is long-run relationship) based on the above critical values at 1%, 5
% and 10% level of significance. Therefore, there is co integration relationship among the
variables in long run. Then researcher must estimate the short run model which is the Error
Correlation Model (ECM).

4.5 ARDL Model Estimation

After confirming the existence of long-run co-integration relationship among the variables, the
next step is running the appropriate ARDL model to find out the long run coefficients and ECM
model to find out short - run coefficients, which are reported the following tables below.

44
 Table 4-7: Estimated long run Coefficients

Dependent variable is Gini Coefficient

Regressors Coefficient Standard Error T- Ratio [ Prob ]

GDP Per Capital [LY] 0.2998*** 0.0780 3.8435[0.0001]

Net Primary School enrollment rate [LPSER] -0.0840*** 0.0230 -3.6521[0.0013]

Unemployment rate [LUEMR] 0.2579* 0.0696 3.7054[0.0116]

Inflation rate [LINFR] 0.0830 0.0567 1.4638[0.1590]

Trade openness [LTO] -0.1291 0.1973 -0.6538[0.5201]

Constant [C] 0.0103*** 0.0015 6.8666[0.0000]

R-Squared =0.7560 Adjusted R- Squared =0.6590

Durbin –Watson statistics =1.9819 P (F- Statistics) =0.0009

NOTE: Decision criteria for significance , If the Absolut value of t- ratio or t-critical is greater than t-
statistics , for some chosen level of significance( Usually 1,5 or 10% ) then the null hypothesis is can be
rejected and variables are significant .

Source: Eview‟s 10.0 results

Note: *, **, *** indicate significance at the level 10%, 5% and 1%, respectively.

From chapter three the model has the following specification.

LnGini = βo + β1 lnYt2+β2 lnPsert+β3 lnTot+ β4 lnUemrt+ β5 lnIfrt + ei . From the above ARDL
estimation result the following regression model is obtained.

45
Gini =0.0103 + 0.2998Yt2 – 0.084Psert – 0.1291Tot+ 0.2579Uemr + 0.0830Infrt

SE = (0.0015) (0.0780) (0.0230) (0.1973) (0.0696) (0.0567)

t = (6.6500) (3.7452) (-3.4525.) (-0.7546) (3.6018) (1.8731)

But researcher was put only significant variables as follows.

LNGini = 0.0103 + 0.2998Yt2+ 0.2579Uemrt - 0.0840Psert

SE = (0.0015) (0.0780) (0.0696) (0.0230)

t = (6.6500) (3.7062) (3.6018) (-3.4525)

4.5.1 Interpretation of the ARDL model estimation coefficients

As the ARDL model estimation shows, all the variables have a sign as expected by the paper.
Real GDP per capita, unemployment rate, inflation rate and constant term have a positive sign.
When the variables unit increased the GINI coefficient also increased, vice versa. On the other
hand, primary school enrollment rate and trade openness a has a negative sign. This means, when
this variables unit increased the GINI coefficient decreased, it changed in the opposite direction.
As we have discussed in the theoretical and empirical literature parts, Real GDP per capita,
unemployment rate, and inflation rate have positive impact on income inequality while primary
school enrollment rate and trade openness have an inverse impact on income inequality
regardless of significant. As the ARDL model estimated result of the above table showed,
unemployment rate have a positive impact on income inequality and statistically significant at 10
% percent level of significance. Holding other things constant, the GINI coefficient will be
increased by 0.2579 when unemployment rate increased by 1%. The real GDP per capita
coefficient, which is 0.2998, has a positive value and it is statically significant at 1%, 5% and
10% percent significant level. Holding other variables constant, the GINI coefficient will be
increased by 0.2998, when the real GDP per capita increased by 1birr. This result supports the
Kuznets hypothesis. This hypothesis says that in the initial stages of development income
inequality and real GDP per capita increases in the same direction. After achieving maximum
stages of economic growth income inequality reaches its maximum point and starts to decline
with a high economic growth. Ethiopia is one of the least developed countries. Then, based on
this hypothesis the paper gets a positive relationship between them. Finally, the results of the
paper show that the Kuznets hypothesis is applicable for Ethiopia. The thread significant variable

46
is primary school enrollment rate. The coefficient of primary school enrollment rate, which is
0.0840, has a negative sign and it is statically significant at 1%, 5% and 10% level of significant.
Other things remains constant, if the proportion of the number of children enrolled in primary
school that belongs to the age group that officially corresponds to primary schooling to the total
population of the same age group increased by 1%, the GINI coefficient will decrease by
0.0840.R-squaredis 0.7568: This implies that75.68 % of the income inequality function is
explained by the selected explanatory variables. In other words, 75.68 % of variation of the
dependent variable is due to the variation of the independent variables which included in the
model and the remaining variation 24.32% is explained by the variables which are not included
the model. If the value of R-Squared is higher, than model is the greatest the goodness of fit.
There for, is R- Squared in the regression model reveals that there is good fitness of value for a
given result. The overall model is statistically significant because of P (F- Statistics) is 0.0009,
which is less than 5% percent. Real GDP per capital and unemployment rate are the main factors
that determine the income inequality this because of coefficient is high and also statistically
significant and the result support kunzites hypothesis.

4.6 Short-Run Error Correction Model (ECM)

After the acceptance of long-run coefficients of the growth equation, the short-run ECM model is
estimated. The error correction term (ECM), as we discussed in chapter three, indicates the speed
of adjustment to restore equilibrium in the dynamic model. It is a one lagged period residual
obtained from the estimated dynamic long run model. The coefficient of the error correction term
indicates how quickly variables converge to equilibrium. In short run there may be
disequilibrium even if there is a long-term equilibrium relationship between the dependent
variable and the independent variable means that there is co-integration. In order to correct this
disequilibrium and to determine the short run relationship between variables researcher use the
Vector Error Correction Model because of data is co-integration. The dynamic short run
equilibrium is obtained by regressing the first difference of the dependent variable with the first
difference of the explanatory variable and one period lagged error term to capture the adjustment
towards the long run equilibrium. The coefficient of the error correction term indicates how
quickly variables converge to equilibrium. Moreover, it should have a negative sign and
statistically significant at a standard significant level (i.e. p-value should be less than 0.05).

47
 Table 4-8: Error Correction Representation for the Selected ARDL

Dependent variable is First Difference of Gini coefficient [DLGini]

Regressors Coefficient Standard T-Ratio [Prob]
Error
Difference of Constant [DCONS] 1.9414*** 1.0902 -1.78067[0.000]

The Error Correlation Coefficient [ECM-1) ] -0.8427** 0.4232 -1.9912[0.008]

Difference of GDP Per Capital [D(LNY) ]

0.0033*** 0.0016 2.0625 [0.003]

Difference of Unemployment Rate [D(LNUEM) ] 2.3549 1.2256 -1.9214[0.569]

Difference of Net Primary School Enrollment rate -0.1083** 0.1603 -0.6756[0.056]

[ D(LNPSER) ]

Difference of Trade Openness [D(LNTO) ] -0.1974 0.1825 -1.0816[0.281]

Difference of Inflation rate 0.0138** 0.0415 -0.3325[0.047]

[ D(LNINFR) ]

R-Squared = 0.6647 Adjusted R- Squared =0.6227

Durbin –Watson statistics = 2.0800 P (F- Statistics) =0.0032

Source: Eview‟s 10.0 results

*, **, *** indicate significance at the level 10%, 5% and 1%, respectively

From the above table, similar to the log run result, real GDP per capital, unemployment rate and
inflation rate have positive impact on income inequality.Net primary school enrollment rate and
trade openness have negative impact on income inequality in Ethiopia. The short run impact of
unemployment rate on income inequality in Ethiopia is positive but insignificant

The error correction coefficient, estimated at -0.8427 is highly significant, has the correct
negative sign, and imply a very high speed of adjustment to equilibrium. According to Bannerjee
et al. (2003) as cited in Kidanemarim (2014), the highly significant error correction term further
confirms the existence of a stable long-run relationship. Moreover, the coefficient of the error

48
term (ECM-1) implies that the deviation from long run equilibrium level of income inequality in
the current period is corrected by 84.27% in the next period to bring back equilibrium when there
is a shock to a steady state relationship. The short run coefficients of real GDP per capital
indicate a positive and significant effect on income inequality, at 1, 5 and 10 percent significant
level. That is when real GDP capital increase by one unites or one birr, income inequality is
increase by 0.0033. As one can understand form the above tables,( 4-7) and (5-8) trade openness
is not significantly affect income inequality during the study period, despite their relationship is
negative both short run and long run. From this we can understand that under the study period,
both in the long run and in the short run, trade openness, does not have significant effect on
income inequality. Unlike the long run, the inflation rate variable significantly affects income
inequality in the short run at 5 and 10 percent significance level. Even though, the sign is
positive. The constant term is positive, which is 1.9414. This indicates, if all variables are zero at
the same time, the GINI coefficient becomes 1.9414. The short run R-squared is 0.6647. This
implies that real GDP per capita, net primary school enrollment rate, unemployment rate, trade
openness and inflation rate explained 66.47 % of variations on GINI coefficient. The overall
model is statistically significant in the short run because of P (F- Statistics) is 0.0032, which is
less than 5% percent. As the result indicates, the error correction term is statistically significant.
Therefore, there is adjustment in the short run.

4.7 Summary of results

The findings of econometrics analysis summarized as follows:

 Real GDP per capita which is a measure of economic growth has a positive effect on
income inequality both in long run and short run. It is statically significant in both the
long run and short run
 In both analyses, unemployment rate has a positive effect. But, it is statically significant
in the long run and statically insignificant in the short run.
 Education which is measured by net primary school enrollment rate has a negative and
statically significant impact on income inequality both in short run and long run.
 Trade openness calculated as the proportion of total value of trade that a country transact
with the rest of the world in a year to annual GDP of the country has a negative effect and
it is statically insignificant in both analysis.
 Inflation rate measured consumer price index is found to have a positive impact on
income inequality and it is statically insignificant in the long run and statically significant
in the short run.

49
 CHAPTER FIVE

CONCLUSION AND RECOMMENDATIONS

5.1 Conclusion

The main objective of this study is to analyze income inequality and its determinants by using
macro variables during the specified period. As the descriptive analysis shows the trend of
income inequality measured by GINI coefficient and its determinants are shows some
fluctuations. All determinants have a sign as expected by this paper based on theoretical
framework. To determine the long run and short run relationship among the variables,
Autoregressive Distributed Lag (ARDL) and ECM model were applied. Before applying the
ARDL model, all the variables are tested for their time series properties (stationariety properties)
using the ADF and PP tests. As a result, Gini coefficient, real GDP per capital, primary school of
enrolment rate, inflation rate and unemployment rate are stationary in level where as trade
openness is stationary in first difference (with intercept). However, with trend and Intercept,
except unemployment rate and inflation rate, all the variables are stationary in level. Next to
testing for time series property, the model stability was done by testing the diagonal testing
techniques. The result revealed that, no functional form problem (the model is correctly
specified), the residual is normally distributed, no multicolinearity, no autocorrelation and
hetroscedasticity problem. The dependent variable that was the being income inequality was
regressed against five explanatory variables. As discussed above, this study applied the
methodological approach called ARDL model also known as bound test approach. As the result
indicted the calculated F-statistics is greater than the critical values for upper bound I(1), than
we can conclude that there is co integration. That is along run relationship between income
inequality and its determinants (real GDP per capital, school of enrollment rate, unemployment
rate, trade openness and inflation rate in long run during the study).As we have discussed in the
theoretical and empirical literature parts, Real GDP per capita, unemployment rate, and inflation
rate have positive impact on income inequality while primary school enrollment rate and trade
openness have an inverse impact on income inequality. In the long run unemployment rate, have
a positive impact on income inequality and statistically significant at 10 % percent significance

50
level. The empirical result showed that unemployment rate, inflation rate and real GDP per
capita are found to have positive impact on income inequality during the study period
.Unemployment rate have a positive impact on income inequality and statistically significant at
10 % percent significance level. A one percent increase in unemployment rate results in 0.2580
and 2.3550 percent increase in income inequality in long run and short run, respectively.
Likewise, a one percent increase in real GDP per capital will result in 0.2998 and 0.0035 percent
increase in real GDP in long run and short run, respectively. According to the result, economic
growth measured by real GDP per capita and unemployment rate are the major determinant of
income inequality. In the long run, a coefficient of real GDP per capita is 0.2998, it is also
statistically significant and it affects it positively as expects .In the short run, like in the long run
it has a positive effect. Ethiopia is at initial level of economic development, so according to
Kuznets hypothesis it is expects to have a positive relationship between them. Therefore, the
result supports Kuznets hypothesis. Primary school enrollment rate and trade openness also has
negative impact in income inequality during the study period in both long run and short run. A
one percent increase in primary school enrollment rate will result in 0.0840 and 0.1083 percent
decline in income inequality in long run and short run, respectively. It is statically significant at
1%, 5% and 10% percent level of significant in the long run and it is statically significant at
10% percent level of significant in the short run. However, the study found out trade openness
has statistically insignificant impact on income inequality with negative sign in the both long run
and short run. Inflation rate has statistically insignificant impact on income inequality in the long
run but it is statistically significant impact on income inequality in the short run at 10 % percent
level of significance.

5.2 Recommendations

Based on the finding of the Study the Following Recommendations are forwarded.

 Though inflation is one a problem in income inequality, the federal government should
work to reduce the inflation rate if possible; otherwise, it should sustain the existing
inflation rate by financing of budget deficit from non-inflationary sources and
implementation of price stabilization program by subsiding basic food items and by
controlling money supply.

51
 Education creates a high wages for those with this education, and then it leads to higher
competition in the labor market. Thus, uneducated peoples will be unemployed and they
can‟t generate income. Then, educational level was negative influence income inequality.
These clearly indicate that when education increases income inequality is decrease, so to
reduce income inequality, responsible body gives more attention for expansion of
education and the responsible bodies should have provide more equal access to basic
education (by spending on public education that benefits the poor) to reduce inequality by
facilitating the accumulation of human capital and making educational opportunities less
dependent on socio economic circumstances and have to provide better job related training
and education for low- skilled workers (on- the job- training).
 As the paper result indicates, real GDP per capital had positively and a highly significant
effect on income inequality of the country. Based on Kuznets hypothesis after some high
economic development level the relationship changes inversely (when the economy grows
the income gap diminish). So, to reduce income inequality the country must grow very fast
to reach at that high economic development level. In order to grow very fast, the
government should implement some policies like, Pro poor growth strategy to participate
all people from the benefits of growth, Well- targeted income support policies and Policies
that encourage innovations, skill- intensive production techniques, and formulate a better
market that initiate competition, technology diffusion and create a good chain to products
movement.
 When the unemployment rate decrease the income gap also decrease. If the country aims at
decreasing income inequality the government should, create accessible, productive and
rewarding jobs, facilitate and encourage access to employment by formulating a policy that
reduces market imperfection and institutional failure. For instance; minimum wage,
spending on well- designed active labor market policies aimed at supporting job searching
people, reducing the gap in employment protection like permanent and temporary workers,
legalizing informal workers by giving some training and expanding formal sectarian
employments by reducing tax, financial and regulatory constraints.
 This research can be used as a bench mark for further researches, therefore, anyone who are
interested can assess the effect through adding additional variables which could be
considered as a determinants of income inequality. Further studies should be conducted
with a wider coverage as this study only confined 31 years data.

52
 REFERENCE
Abdul, Jamale. (2011). A increasing the efficiency of labor force in education. Cerates better conditions for good
governance (Master's thesis). Improving health and enhancing equality.

Abdurahman, Hassen .(2014). There is Negative relationship between income inequality and economic growth:
Unpublished Master thesis, Ethiopia.
Abebe, Fenta. (2016). Determinants of income inequality. South Wollo Administration zone: Amhara National
Regional State, Ethiopi
Alemayehu, G.Maryem ., &Kibrom, D, L .(2019). The sharp increasing of general inflation. Journalof food
inflation, 90(6), 75–85.
Alemayehu, G.Maryem .&, Addis , & Vito, J. (2014). The relationship between growth, poverty and inequality. the
period from 2000-2013;unequal distribution and economic issue 48(2010,(pp.184-193).
Alemayehu,G. Maryem and Addis, Y.(2014).The relationship between growth, Poverty and Inequality. Ethiopia
for the period 2000-2013: unequal distribution and economics, Issue 48(2010),pp.184-193).
Alemayehu, G.Maryem 2009).The correlation between economic growth and income Inequality. Published
Master thesis. Addis Ababa University, Ethiopia.
Antonio, A., Luder,S., & Vito, J. (2008). The income distribution determinants and Public Assessment , A chapter
in Book to be published by Forum for social studies ,FSS, Cambridge university Press ,Co integration
Analysis . S. Strom, Econometrics and economic theory in the 20th co integration.
Antonio, A., Luder,S., & Vito, J. (2008). The public spending can affect income distribution. Journal of ncome
distribution, 11(1), 75–85. Cross countries perspective.
Classens, .T & Perortis , L. (2007). Income inequality can led Political that hurt growth.

Era, D., Norris,H., &Kalpana, K. (2015). Economic growth and unemployment have a positive impact on income
inequality. Journal of economic growth and unemployment: India economy (pp.180-195).
Engle, F.,& Granger, J. (2008). Co-Integration and Error correlation: Representations, Estimation and
testing.Journals of Econometrics, Vol 55, No 2: 251-276.
Field, G (1980). Poverty, Inequality and Development: Cambridge University press,46-56.
Galor, O (2011). Inequality, Human capital formation and the process of development: Brown University working
papers, 2011-56.
Getasew, Alemu (2011).Trends and Extents of income inequality with respect of economic growth. Master’s thesis,
Addis AbabaUniversity, Ethiopia
Gujarati, D (2004).Log transformation can reduce the problem of hetrocedesticity (4rd.ed). By reducing a tenfold
difference between two values(pp.345-348).
Gujarati, D (2004).If calculated F- statistics is less than the lower bound critical values (4rd.ed). Hypothesis is no
integration (pp.434-47).
Gujarati, D (2004). If the mean of Vif is greater than 5(vif>5) (4rd.ed) that variable is highly collinear (pp.345-
52).
https://openknwledge.worldbank,org/handle/10986/21334/2020/12.

53
Jesper, R., Jonas ,V., &Danial, W. (2009). Determinants of income inequality. Journal of income inequality:
assembled panel of data 16 countries, 20the century.
IMF .(2017). Income inequality affects economic growth (Report No.IMF-567-79,2017).
IMF .(2015).Grater income inequality can lead to monopolization of labor force (Report No.IMF-344 -79, 2015).
Kuznets, S. (2004). Economic growthand income inequality. The American Economic Review 45(1),1-28.45(1),18.
Khan, K., Mohsin,S., & Abdelhak, S and Senhedji, H. (2000). Positive relationship between inflation and
incomeineualit. IMF staff paper vol.48,No.1 and June,2000.
Kidanemarim, M. (2014). The highly significant error correction term. Confirms the existence of astable long run
relationship :Error correlation coefficient ,econometrics .
Liew, V., &Khimsen, A. (2004). Lag length selection criteria. Economic Bulletin, Vol.3 No. 33.1-9.
LIchbach, T. (2017).Extreme inequality damage trust and social cohesion.
Narayan, K. (2014). Reformulating Critical Values for the bounds F- statistics approach.
Nasiru , I. (2012). Government expenditure and economic growth in Nigeria.Conintegration analysis is and
Causality testing: Journal of Academic International, Vol.2 No.2,pp.718-28.
Obama, B.(2014). Income inequality is challenging time of mobility between generations.Challenging income
inequality, paper vol. 48, No1 and June,2000.
Oxfam. (2014). Working for the few .Political capture and Economic inequality, Oxfam briefing paper, oxford.
Papers No 02, MonashUniversity, Victoria, Australia
Oxford, D. (2017). Income inequality: unfair, unjust distribution of money and future prospects. Washington, DC:
From investing or from business (pp.345-49).
Pesaran, M. H., Shin,Y., &R.J, Smith . (2001). Bounds Testing Approach: Analysis of long run relationship with
correlation. Journal of Applied econometrics ,Vo1.16pp 289-326.
Pesaran, M. H., &Shin,Y.(1999). An Autoregressive Distributed Lag Modeling approach to long run relationship:
Journal of Applied Econometrics, Vo1.16pp 289-326: South Wollo Administrative Zone, Amhara, And
National Regional State.
Pesaranetal , Y. (2001 ). ARDL Co integration: Most method for estimating or to check long run relationship
.Journal of applied economics, 16,289-326.
Pesaran, etal. (2004).The stability of long run can estimate the CUSUM and CUSUMSQ .Journal of Applied
econometrics .16, 289-326.
Stieglitz, P. (2017). Labor productivity is lower than it would have been in more equitable else in business.
Journal of human capital.
Sid, H. (2014).,Gehring&Kulkarni, C. (2016). Income inequality is unequal distribution of income: A case study
examining Maya perspectives on the Indigenous inequality, and informed consent. Wealth & pay
inequality s, 27, 231–248
Taban, S. (2010).An examination of government spending.Nexus for income inequality and economic growth:
Journal of effects economic growth on inequality, 29(2), 83-87.
Tessew, Malaku (2009). Income inequality in urban and rural. Income inequality in rural is unchanged:
Unpublished Master’s thesis, Addis Ababa university.
Theil, I. (1993). Measurement of income inequality. Among the members of states, European, CPA measurements
of inequality,77(6), 66–71.

54
Todaro. (2012). Gini coefficient; Turkey the bounds test approach. International research journal income
inequality measurement.
Tsadkan, T. S. (2013). The nexus between income inequality and economic growth in empirical investigation .
International Journal of Applied economics, 2(2), 38-48.
Tsadkan, T.(2017). Akaike information criterion (AIC)in lag ;Advantages for small sample size and leg.
Unpublished Master’s thesis, Addis Ababa University.
Tyson, G. (1998).Relationship between income inequality and inflation .American Journal of inflation, 76(4), 482-
488. doi:10.1037/0002-9432.76.4.482.
Vein, C.(2008). Higher level of income; higher income inequality tends to have lower level of economic growth.
College inequality and economic growth, 29, 263-268.
Vsalian, P. &Gopakumar, K. S. (2008).Inflation and income inequality. E-Journal of Applied economics of,
Indian Economic Service, New Delhik. 2(2), 38-48.
Wooldridge, M (2000).Introductory Economics;(2rd.ed), A modern approach (pp.345-348).
World, B .(2020). World Bank: World Bank Data best updated, annual review and summary of world data .
World ,B.,G .(2018). Ethiopia Poverty assessment Technical report of the world bank.
Zlako, N. J. (2006 ). Income inequality and political economy;relationship. Determinants of income inequality;
Journal of political economy among 81 countries
Zerayehu, S.(2013). Economic Transformation and Sect oral Growth Options in Ethiopia;
Considerations for Achieving Middle Income Country Status: University of Nairobi, Nairobi.

55
APPENDIX:I

Coefficient Std. Error t-Statistic Prob.

C(1) -0.084277 0.042322 -1.991318 0.0486

C(2) -0.118967 0.202228 -0.588285 0.5574
C(3) 0.003359 0.001601 2.098419 0.0379
C(4) -2.354937 1.225658 -1.921366 0.0569
C(5) -0.108326 0.160373 -0.675465 0.5006
C(6) -0.197443 0.182547 -1.081600 0.2815
C(7) -0.018474 0.041545 -0.444672 0.6573
C(8) -1.941432 1.090280 -1.780672 0.0774
C(9) -2.68321 5.454382 3.792035 0.0002
C(10) -1.356159 26.06262 -0.052035 0.9586
C(11) 0.259887 0.206289 1.259821 0.2101
C(12) 259.3838 157.9600 1.642086 0.1031
C(13) -15.64007 20.66849 -0.756711 0.4506
C(14) 9.150366 23.52627 0.388942 0.6980
C(15) 12.62289 5.354262 2.357540 0.0199
C(16) 564.7923 140.5128 4.019507 0.0001
C(17) -0.015195 0.006557 2.317350 0.0221
C(18) -0.021639 0.031331 -0.690662 0.4910
C(19) -0.000323 0.000248 -1.302377 0.1952
C(20) 0.136249 0.189891 0.717511 0.4744
C(21) -0.011639 0.024847 -0.468440 0.6403
C(22) 0.095895 0.028282 3.390650 0.0009
C(23) -0.001225 0.006437 -0.190345 0.8493
C(24) 0.116162 0.168917 0.687684 0.4929
C(25) -0.017358 0.056746 0.305883 0.0602
C(26) 0.085330 0.271149 0.314696 0.7535
C(27) -0.002353 0.002146 -1.096416 0.2750
C(28) 3.301241 1.643377 2.008816 0.0467
C(29) 0.479128 0.215030 2.228192 0.0276
C(30) -0.144779 0.244762 -0.591512 0.5552
C(31) -0.062448 0.055704 -1.121059 0.2644
C(32) 2.497750 1.461861 1.708610 0.0900
C(33) -0.081845 0.049952 1.638486 0.0038
C(34) -0.320226 0.238684 -1.341634 0.1821
C(35) -0.005039 0.001889 -2.667165 0.0087
C(36) 0.093798 1.446610 0.064840 0.9484
C(37) 0.040709 0.189284 0.215068 0.8301
C(38) -0.122932 0.215456 -0.570566 0.5693
C(39) 0.019439 0.049035 0.396427 0.6925
C(40) 2.848239 1.286828 2.213380 0.0287
C(41) -0.215771 0.208257 1.036079 0.0021
C(42) -0.203278 0.995113 -0.204277 0.8385
C(43) -0.012874 0.007876 -1.634551 0.1046
C(44) 0.552109 6.031169 0.091543 0.9272
C(45) 0.398218 0.789157 0.504612 0.6147
C(46) -1.146753 0.898271 -1.276622 0.2041
C(47) -0.284169 0.204434 -1.390025 0.1670
C(48) 7.140692 5.365009 1.330975 0.1856

56
APPENDIX: II
Variables Obs Mean Std.Dev Minimum Maximum

Gini- index 31 0.34245 .7640982 29.0 44.6

GDP per capital
(current LCU ) 31 4174.447 994.4169 504.533 20,1430933

Inflation 31 10.38 1.920533 -8.2 44.4

Unemployment rate 31 2.9067 .1356572 2.1 5.3

Trade openness 31 35.7 1.631436 23.448 48.23

Primary school of 31 52.5 4.290362 19.188 85.10

enrollment rate

Source: Own estimation based on World Bank data sets (2020)

APPENDIX: III

. mean lngini lny lnpser lnto lnuemr lninfr

Mean estimation Number of obs = 31

Mean Std. Err. [95% Conf. Interval]

lngini 34.24516 .7640982 32.68466 35.80566

lny 4174.447 994.4169 2143.577 6205.317
lnpser 52.48199 4.290362 43.7199 61.24408
lnto 35.69174 1.631436 32.35991 39.02358
lnuemr 2.906774 .1356572 2.629725 3.183823
lninfr 10.37742 1.920533 6.455169 14.29967

57
 APPENDIX: IV

Data of dependent variable and independent variables from 1988 to 2018

Year Gini Real GDP School of Trade Unemployment Inflation rate in

coefficients per capital enrollments, openness rate in (%) percentage (%)
(%)(GINI) (Yt) In primary(%) net at constant (UERT) (INFt)
million Birr (PSERt ) price, percent
1988 37.0 504.533 29.974 26.21986 3.7 2.2

1989 37.0 513.407 28.703 31.08399 4.5 9.6

1990 38.0 526.284 25.256 23.38600 5.3 5.2

1991 38.0 561.786 31.556 27.58489 3.0 20.9

1992 39.0 571.851 25.281 27.84407 3.0 21.0

1993 39.0 707.792 21.708 28.69366 3.1 10.0

1994 39.0 726.091 19.188 23.44896 3.1 1.2

1995 44.6 840.073 21.947 24.20362 3.2 13.4

1996 42.5 917.208 26.192 23.42700 3.3 0.9

1997 36.0 919.58 28.123 25.41995 3.4 -6.4

1998 37.5 860.774 33.6 29.34442 3.6 3.6

1999 30.0 898.935 35.684 31.98112 3.71 7.9

2000 29.0 1,014.119 40.172 33.04811 3.5 0.7

2001 30.0 1,005.713 43.938 31.47233 3.3 -8.2

2002 30.0 956.161 46.064 34.11079 3.1 1.7

2003 29.0 1,025.287 46.784 37.77908 2.9 17.8

2004 29.8 1,176.271 49.578 45.99167 2.7 3.2

2005 30.0 1,405.303 60.386 49.91167 2.5 11.7

2006 29.9 1,690.062 65.014 49.86843 2.4 13.6

58
2007 30.2 2,148.251 70.0713 46.95410 2.3 17.2

2008 31.3 3,017.582 75.78 47.8065 2.2 44.4

2009 32.3 3,965.149 75.362 48.44308 2.3 8.5

2010 33.2 4,402.974 73.352 46.20343 2.3 8.1

2011 33.0 5,714.211 77.234 48.2375 2.3 33.2

2012 29.3 8,059.43 79.25 48.4087 2.3 24.1

2013 33.6 9,088.577 78.242 45.47974 8.1

2.3
2014 31.2 10,814.236 85.10 49.0450 2.2 7.4

2015 35.0 12,872.074 86.620 51.6740 9.6

2.2
2016 37.0 15,135.56 85.564 54.0240 2.2 6.6

2017 17,225.48 88.098 57.3451 2.1 10.7

36.7
2018 33.8 20,143.098 89.453 59.5320 2.1 13.8

World FRED, World Bank , CSA,WB

Sources World World development Global World
indicator ,
Bank, Bank UNESCO/
economy.com (WB),Develop
World ,World http:/uis.unesco ,World Bank ment Indicter
Devlopme Devlopme .org ,University of (WDI)
ntIndicater ntIndicate Pennsylvania
r(WDI )

59