EC O LO G I CA L E C O N O M I CS 6 0 ( 2 00 6 ) 4 9–6 4

a v a i l a b l e a t w w w. s c i e n c e d i r e c t . c o m

w w w. e l s e v i e r. c o m / l o c a t e / e c o l e c o n

SURVEY

Computable general equilibrium models for sustainability

impact assessment: Status quo and prospects

Christoph Böhringer a , Andreas Löschel b,⁎

a
Centre for European Economic Research (ZEW) and University of Heidelberg, P.O. Box 10 34 43, D-68034 Mannheim, Germany
b
Institute for Prospective Technological Studies (IPTS), DG Joint Research Centre, European Commission, Edificio Expo, C/ Inca Garcilaso,
s/n, E-41092 Seville, Spain

AR TIC LE I N FO ABS TR ACT

Article history: Sustainability Impact Assessment (SIA) of economic, environmental, and social effects
Received 8 August 2005 triggered by governmental policies has become a central requirement for policy design. The
Received in revised form three dimensions of SIA are inherently intertwined and subject to trade-offs. Quantification
22 February 2006 of trade-offs for policy decision support requires numerical models in order to assess
Accepted 4 March 2006 systematically the interference of complex interacting forces that affect economic
Available online 24 April 2006 performance, environmental quality, and social conditions. This paper investigates the
use of computable general equilibrium (CGE) models for measuring the impacts of policy
Keywords: interference on policy-relevant economic, environmental, and social (institutional)
Computable general equilibrium indicators. We find that operational CGE models used for energy–economy–environment
models (E3) analyses have a good coverage of central economic indicators. Environmental
Sustainability impact assessment indicators such as energy-related emissions with direct links to economic activities are
Sustainable development indicators widely covered, whereas indicators with complex natural science background such as
water stress or biodiversity loss are hardly represented. Social indicators stand out for very
weak coverage, mainly because they are vaguely defined or incommensurable. Our analysis
identifies prospects for future modeling in the field of integrated assessment that link
standard E3-CGE-models to themespecific complementary models with environmental and
social focus.
© 2006 Elsevier B.V. All rights reserved.

1. Introduction future generations” (UNCED, 1992a, Principle 3). SD has

meanwhile become one of the most prominent catchwords
In 1987, the report of the World Commission on Environment on the world's policy agenda. Nearly all governments and
and Development (WCED or Brundtland Commission) defined multinational firms have committed themselves to the overall
Sustainable Development (hereafter: SD) as “development, which concept of SD.
meets the needs of the present without compromising the The ubiquity of SD as a yardstick for human activities is
ability of future generations to meet their own needs” (WCED, reflected in the growing importance of Sustainability Impact
1987). In June 1992, the Rio Earth Summit concluded that “the Assessment (hereafter: SIA) of governmental policies. Initially,
right to development must be fulfilled so as to equitably meet the assessment of SD impacts concentrated on trade policy
developmental and environmental needs of present and reforms (see, e.g. Kirkpatrick and Lee, 1999, for the SIA of

⁎ Corresponding author. Tel.: +34 95 448 8438; fax: +34 95 448 8279.
E-mail address: andreas.loeschel@cec.eu.int (A. Löschel).

0921-8009/$ - see front matter © 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.ecolecon.2006.03.006
50 E CO L O G I CA L E CO N O MI CS 60 ( 20 0 6 ) 4 9–6 4

WTO's Millennium Round proposal). More recently, SIA has sustainable development rather than on fuzzy or contradic-
been extended to other policy areas. Taking a lead role, the tory hunches.
European Union (EU) meanwhile requires “careful assessment A major challenge in building quantitative SIA tools is the
of the full effects of a policy proposal [that] must include policy makers' demand for comprehensive coverage of
estimates of its economic, environmental and social impacts potentially important policy impacts. SIA tools must identify
inside and outside the EU” (EC, 2001). The argument behind is “the chain of significant cause-effect links from the … [policy]
that SIA can improve the SD coherence of policy initiatives measures … through to any sustainability impact” and
across various areas by identifying spillovers and inter- produce “comparable indicators of the magnitude and dimen-
linkages. However, SD, which is not just about environment, sions of each sustainability impact” (EC, 2003a) as an input
but also about economy and society, has proven hard to define into policy formulation and implementation. Obviously,
and rather susceptible for ambiguities. One reason for this is quantitative SIA does not only require an adequate reduction
that SD explicitly incorporates an (normative) equity dimen- of complex real-world relationships but — as a pre-requisite —
sion, which may be “so hopelessly subjective that it cannot be the translation of potentially vague policy proposals into a
analyzed scientifically” (Young, 1994). Another reason is that concrete policy impetus that can be “processed” within an
the scope of the concept seems prohibitively comprehensive analytical model.
and therefore complex to make it operational in concrete There is a wide range of quantitative models for assessing
practice. the causal chains between a proposed policy change and its
Acknowledging the huge inherent difficulties to come up potential economic, environmental, and social impacts.
with sound approaches to the concept of SD and the need for Models mainly differ with respect to the emphasis placed on
SIA, the scientific community has focused in a pragmatic first (i) sectoral details versus economy-wide scope, (ii) economet-
step on the identification of appropriate indicators. These ric foundation of functional relationships, and (iii) the
efforts included the development of qualitative tools (e.g. richness of behavioral assumptions for economic agents.
electronic checklists such as IAPLUS, see http://iaplus.jrc.es) Referring to criterion (i), there is a widespread distinction
that can provide useful orientation for policy decision between bottom-up sector-level models and top-down mac-
makers. Yet, qualitative approaches are unable to commen- roeconomic models. Referring to criterion (ii), models can be
surate different impacts. This constitutes a major shortcom- classified as either econometrically estimated when driving
ing, since the three dimensions of SD, i.e. environmental equations are based on econometric techniques using mostly
quality, economic performance (gross efficiency not includ- timeseries data or as calibrated when parameters of function-
ing environmental-related impacts), and equity concerns are al forms are simply selected to fit a single empirical
inherently intertwined and subject to trade-offs. Accom- observation. Referring to criterion (iii), models may be
plishing one objective frequently means backpedaling on classified into micro–/macro-founded approaches and simple
another. Therefore, research activities on SIA increasingly accounting frameworks.
aim at developing quantitative tools for trade-off-analysis While such taxonomy of models can be useful, it has its
along the SD dimensions. Since economics is the study of limits. For example, the common dichotomy between top–
trade-offs, there is plenty for economists to contribute in down economy-wide models and bottom-up sectoral models
order to make the concept of SD operational. One important is in general not of theoretical nature (i.e. due to controversial
contribution of (environmental) economics over the last theoretical underpinnings) but simply relate to the level of
decade has been the thorough assessment of external costs, aggregation and the scope of ceteris paribus assumptions. In
in particular for energy transformation and transport fact, there have been various model developments merging
activities, as a prerequisite towards “getting the prices bottom–up approaches and top-down models within one
right” (see, e.g. EC, 1999; Friedrich and Bickel, 2001). Given consistent framework (see, e.g. Böhringer, 1998). Further-
external cost estimates, two aspects of SD, namely economic more, among top–down models there is often an exaggerated
performance (gross efficiency) and environmental quality, divide between econometric demand-driven Keynesian mod-
can be merged to a comprehensive net efficiency dimension. els and computable general equilibrium (CGE) models.
Furthermore, while economics has little to say on equity per Popular arguments against the informational value of CGE
se, the sound economic quantification of distributional models include that these models must be calibrated (and
effects for different agents and trade-offs between equity thus lack empirical evidence) and can neither reflect dis-
and efficiency objectives are a prerequisite to any rational equilibria (such as unemployment or under-utilization of
policy debate. production capacities) nor transitional dynamics. In turn,
The quantification of trade-offs calls for the use of econometric Keynesian models are often accused of a lack of
numerical model techniques in order to assess systemati- micro-foundation. These claims ignore substantial develop-
cally and rigorously the interference of the many forces that ments during the last two decades to overcome such policy-
interact in the economy thereby affecting potential SD relevant shortcomings.
indicators. Compared to stylized analytical models, the In general, there is no specific model, which fits all
numerical approach facilitates the analysis of complex requirements for comprehensive SIA, but rather a package of
non-linear system interactions and the impact assessment models or methods depending on the policy measure or
of structural policy changes. In the end, the decisions how issue to be assessed and the availability of data. We argue
to resolve potential trade-offs must be taken on the basis of that CGE models can incorporate several key sustainability
societal values and political decisions. Model-based analysis (meta-) indicators in a single micro-consistent framework,
puts decision making on an informed basis concerning thereby allowing for a systematic quantitative trade-off
 EC O LO G I CA L E C O N O M I CS 6 0 ( 2 00 6 ) 4 9–6 4 51

analysis between environmental quality, economic perfor- able units of information on economic, environmental, and
mance and income distribution. Furthermore, the CGE social (including institutional) conditions. The crucial role of
approach constitutes an open framework for linkages to SD indicators has been prominently emphasized by the United
models from other disciplines thereby accommodating the Nations Conference on Environment and Development (UNCED),
integrated assessment of sustainability issues. A recent held in Rio de Janeiro in 1992, that calls on individual countries
example is the integrated assessment of costs and benefits as well as international governmental and non-governmental
from climate change policies (Böhringer et al., in press-a, in organizations to “develop and identify indicators of SD in
press-b). order to improve the information basis for decision-making at
In this paper, we investigate the use of CGE models for all levels” (UNCED, 1992b, Agenda 21, Chapter 40). Since the
measuring the impacts of policy interference on policy- early 90s a multitude of indicator lists has been developed; the
relevant economic, environmental, and social (institutional) Compendium of Sustainable Development Indicator Initiatives lists
indicators. We do not cover competing or complementary more than 500 sustainable indicator efforts (Parris and Kates,
quantitative modeling approaches to SIA, i.e. we do not 2003).
provide a cross-comparison of model-specific strengths and The United Nations Commission on Sustainable Development
weaknesses. The CGE models analyzed may be all classified (CSD), established in 1992 to ensure effective follow-up of the
as energy–economy–environment (E3) models since they are UNCED, focused its work to date mainly on the development
all concerned with linkages between economic activities, and testing of indicators that could be readily used in
energy transformation, and associated environmental planning, policy formulation, and evaluation at the national
impacts. We find that operational versions of E3–CGE- level. The initial work program on Indicators of Sustainable
models have a good coverage of central economic indicators. Development resulted in a list of 134 indicators, which covers
Environmental indicators such as energyrelated emissions social, environmental, economic, and institutional aspects of
with simple direct links to economic activities are widely SD. After voluntary national testing (within 22 countries) and
covered, whereas indicators with complex natural science expert group consultation, a reduced and revised set of 58
background such as water stress or biodiversity loss are “core indicators” categorized within 15 themes and 38 sub-
hardly represented. Social indicators stand out for very weak themes for monitoring the progress towards SD was released
coverage, mainly because they are vaguely defined or (UN, 2001).
incommensurable. Our analysis identifies prospects for Efforts by the European Community to integrate environ-
future modeling in the field of integrated assessment that mental objectives into the different fields of policy-making
link standard E3-models to theme-specific complementary date back to the early 1970s as manifested e.g. within the
models with environmental and social focus. first Environmental Action Plan (EAP, 1973). The Amsterdam
The structure of the paper is as follows. Section 2 Treaty, signed in 1997, codified environmental policy inte-
addresses the definition of SD indicators as a prerequisite gration as a central EU policy element within Article 6: “…
for SIA. We present two central, policy-relevant indicator environmental protection requirements must be integrated
lists (Eurostat, 2004; EC, 2003b). Section 3 provides a non- into the definition and implementation of the Community
technical introduction into a standard multi-sector, multi- policies … in particular with a view to promoting SD” and
region CGE model of global trade and energy use that we furthermore re-enforced Article 2, which defines SD as a
consider as a possible backbone for (more) comprehensive fundamental objective for the European Community. The
quantitative SIA and illustrates some selected extensions of Gothenburg Summit in 2001 came up with the definition of
this core modeling framework that allow for appropriate an European Union Strategy for Sustainable Development (SDS)
treatment of potentially important impacts triggered by that combines the commitment to improved environmental
specific policy proposals. Section 4 presents results from a performance (Helsinki European Council 1999) with the
questionnaire on the coverage of SD indicators in 18 objective “to become the most competitive and dynamic
established E3–CGE models distinguishing between (i) indi- knowledge-based economy in the world capable of sustain-
cators that are covered by the respective model, (ii) able economic growth with more and better jobs and greater
indicators that are in the scope of more or less straightfor- social cohesion” (Lisbon European Council 2000). An annual
ward extensions of the model, and (iii) indicators that are stocktaking of the progress towards SD — due at each spring
rather difficult to address in quantitative (CGE) analysis. We summit — was agreed upon (see EC, 2001).
thereby evaluate to which extent key indicators for SD are The European Union Strategy for Sustainable Development
currently captured or subject to feasible extensions in focuses on six themes which are enhanced by four other
operational E3–CGE models. Section 5 illustrates the idea of themes derived from further discussion on sustainability by
linkages between E3-models and complementary specific the EU, UN, etc. (Eurostat, 2004). In Table 1, within each
bio-physical or socio-economic approaches in order to widen theme, a number of sub-themes and “areas to be
the scope for quantitative impact assessment. Section 6 addressed” have been identified. The sub-themes encom-
concludes. pass the relevant SD issues addressed in the basic policy
documents. Generally, the sub-themes are closely linked to
the headline objectives, which are also reflected in the
2. Indicators for sustainable development labeling of the sub-themes. The “areas to be addressed”,
which can be considered as an interface between indicators
Monitoring progress towards SD requires in the first place the and policies, are closely linked to the measures announced
identification of operational indicators that provide manage- in the European Council's Communications (EC, 2001).
52 E CO L O G I CA L E CO N O MI CS 60 ( 20 0 6 ) 4 9–6 4

Table 1 – Theme, sub-themes and ‘areas to be addressed’ Table 1 (continued)

(Eurostat, 2004) Theme Sub-theme Areas to be addressed
Theme Sub-theme Areas to be addressed
growth and generation of
Economic Investment 1. Investment in R&D wastes
development 2. Investment in env. Agriculture 45. Pesticides use
friendly technologies 46. Nitrogen balances
3. Consumption and 47. Environmentally-
inflation friendly farming
4. Saving and borrowing Corporate 48. Triple bottom line
Competitiveness 5. Labor productivity responsibility
6. Unit labor costs Consumer 49. Consumer
7. Life-long learning awareness information
Employment 8. Employment rate Management Biodiversity 50. Protection of habitats
9. Unemployment rate of natural and natural systems and
Poverty and Monetary poverty 10. Income inequality resources biodiversity
social exclusion 11. Non-monetary 51. Maintaining the
deprivation carrying capacity
Access to labor 12. Poverty-in-work Marine ecosystems 52. Over-fishing
market Fresh water 53. Water extraction and
Other aspect of 13. Access to education resources use
social exclusion 14. Access to health care 54. Protection of surface
15. Access to housing and ground water
16. Social participation resources
Aging society Pensions adequacy 17. Income of elder Land use 55. Land use change
generations 56. Soil degradation
Demographic 18. Life expectancy 57. Forests
changes 19. Fertility Transport Transport growth 58. Decoupling of
20. Migrations economic and transport
Financial 21. Age of withdrawal growth
sustainability from Labor Market 59. Road to rail, water and
22. Pension expenditures public transport
Public health Human health 23. Financial 60. Land use by transport
protection and life sustainability systems
styles 24. Disability-free life Env. impact of 61. Air pollutants
expectancy transport act.
25. Premature mortality Good Policy coherence 62. Citizen's adherence
26. Life styles governance and support to EU actions
27. Health and safety at 63. Sustainability of EU
work actions and measures
28. Infectious diseases 64. Legislative compliance
and resistance to Public participation 65. Communication and
antibiotics mobilization
Food safety and 29. Pesticide residues Global Globalization of 66. Market access for least
quality 30. Microbiological partnership trade developed countries (LDC)
contamination Financing for SD 67. Foreign direct
31. Drinking water quality investments to
Chemicals 32. Chemicals production developing countries
management and consumption 68. Official Development
33. Exposure to chemicals Assistance (ODA)
Health risks due to 34. Air quality 69. Other official
environmental 35. Noise exposure financing
conditions Resource 70. Resource
Climate change Climate change 36. GHG emission
and energy reduction
Energy 37. Energy taxes
38. Energy efficiency However, a certain amount of balancing has been done
39. Renewable energy
between the sub-themes and the areas to be addressed to
resources
take into account different levels of importance of the
40. Management of
nuclear waste issues and to ensure consistency. Additional sub-themes
41. Air pollution from have sometimes been included in order to increase the
energy use visual clarity of the overall framework and to group the
Production and Eco-efficiency 42. Decoupling economic areas to be addressed.
consumption growth and resource use Since the large number of indicators made it rather difficult
patterns 43. Decoupling economic
to draw a clear picture on SD progress (EC, 2003b) the Spring
growth and emissions
44. Decoupling economic
Report 2004 has been scheduled to report on only 14 structural
indicators thereby balancing the importance of employment,
 EC O LO G I CA L E C O N O M I CS 6 0 ( 2 00 6 ) 4 9–6 4 53

Table 2 – European Commission (EC) structural indicators nevertheless be affected through international spillovers. In
(EC, 2003b) addition to the consistent representation of trade links, the
I. GDP per capita detailed representation of energy flows captures a major
II. Labor productivity segment of the environmental SD dimension, i.e. energy
III. Employment rate usage and air quality. Combustion of fossil fuels is a driving
IV. Employment rate of older workers
force of global warming through the release of CO2 and
V. Spending on human resources (public exp. on education)
causes serious regional and transboundary pollution through
VI. Research and Development expenditure
VII. Information Technology expenditure emissions/imissions of SOx and NOx. The comprehensive
VIII. Financial market integration (conv. of bank lending rates) scope of multi-region, multi-sector CGE models explains why
IX. At risk-of-poverty rate such models play an important role in the assessment of
X. Long-term unemployment trade policy impacts (see, e.g. Lee and Kirkpatrick, 2001;
XI. Dispersion of regional employment rates Francois and Reinert, 1997) and climate policy analysis (see,
XII. Greenhouse gases emissions
e.g. Böhringer and Löschel, 2002). In Section 3.2, we describe
XIII. Energy intensity of the economy
extensions of the core model that widen its scope and policy
XIV. Volume of transport
relevance with respect to SIA. In Section 3.3, we sketch the
central steps involved in applied CGE analysis for SIA of
policy reforms and we address the issue of model
parameterization.
innovation and research, economic reform, social cohesion,
and the environment (see Table 2). 3.1. Core CGE Model

Fig. 1 provides a diagrammatic structure of the standard

3. The CGE back-bone to (sustainability) multi-sector multi-region trade model with energy flows as
impact assessment used for comparative-static impact analysis of trade and
environmental or energy policies (Böhringer and Löschel,
Among numerical methods, computable general equilibrium 2002, 2004). Primary factors of a region r include labor ¯¯¯L r,
(CGE) models are widely employed by various national and capital ¯K ¯¯ ff,r of fossil fuels ff (crude oil, coal,
¯¯ r and resources Q̄
international organizations (IMF, World Bank, OECD, etc.),
the European Commission, research centers, and universi-
ties for economic policy analysis at the sector-level as well
as the economy-wide level. CGE models build upon general Indicator
equilibrium theory that combines behavioral assumptions
on rational economic agents with the analysis of equilibri-
um conditions. They provide counterfactual ex ante com-
parisons, assessing the outcomes with a reform in place
with what would have happened had it not been undertak-
Regions
Other

en. The main virtue of the CGE approach is its comprehen-

sive micro-consistent representation of price-dependent
Mir

market interactions. Besides, the simultaneous explanation

Region r

of the origin and spending of the agents' income makes it

possible to address both economywide efficiency as well as
distributional impacts of policy interference. Therefore, CGE
models have become a standard tool for the quantitative Core
fuels ELE EIS Sectors
Other

analysis of policy interference in many domains including CGE

fiscal policy, trade policy, and environmental policy. For Model
survey articles see, e.g. Bhattacharyya (1996), Bergman
(1990), Borges (1986), Conrad (1999, 2001), Klepper et al.
A ir

Y ir

(1994), or Shoven and Whalley (1992).

In Section 3.1, we lay out the generic structure of a multi-
L r, Kr, Q ff ,r
Fossil-

sector, multi-region CGE framework of global trade and

energy use. A multi-region framework seems indispensable
for SIA of major policy initiatives in a world that is
increasingly integrated through trade: Policy interference in
RA r
Cir

open economies not only causes adjustment of domestic

production and consumption patterns but also influences
international prices via changes in exports and imports. The
changes in international prices, i.e. the terms of trade, imply
a secondary benefit or burden, which can significantly alter Instrument
the primary domestic impacts of the policy. Likewise,
countries, which do not undertake policy reforms will Fig. 1 – CGE core model for SIA.
54 E CO L O G I CA L E CO N O MI CS 60 ( 20 0 6 ) 4 9–6 4

and gas). Production Yir of commodities i in region r is captured tion (Eurostat 3; EC I) that results from the new distortions in
by aggregate production functions which characterize tech- intermediate and final consumption.
nology through substitution possibilities between various
inputs. Nested constant elasticity of substitution (CES) cost 3.2. Core CGE model extensions
functions with several levels are employed to specify the
KLEM substitution possibilities in domestic production sectors The core CGE model covers only a few indicators for SIA. In this
between capital (K), labor (L), energy (E) and non-energy section, we illustrate some extensions of the core model for
intermediate inputs, i.e. material (M). Depending on data SIA. There are, however, various other developments of the
availability, the economy can be disaggregated into as many CGE methodology targeted to specific aspects for SIA that are
as several hundred producing sectors (see, e.g. U.S. Depart- not covered here.
ment of Commerce, 1993).
Final demand Cir in each region is determined by utility 3.3. Air (Eurostat 36, 61, 71; EC XII)
maximization of a representative agent RAr subject to a budget
constraint. Total income of the representative agent consists As to environmental impacts, our core model focuses on
of his factor income. Final demand is given as a CES composite carbon dioxide from fossil fuel use, since it constitutes the
which combines consumption of an energy aggregate with a largest contribution to global warming. However, there are
non-energy consumption bundle. The substitution patterns potentially important non-CO2 greenhouse gases (GHG) as
within the non-energy consumption bundle as well as the well as other detrimental emissions that should be included in
energy aggregate are described by nested CES functions. a broader SIA. In the core model carbon emissions are directly
Emissions are associated with fossil fuel consumption in linked to fossil fuel inputs in production or consumption.
production, investment, and final consumption. All goods Carbon emission abatement can take place either through
used on the domestic market in intermediate and final reduction of output or substitution of non-polluting inputs for
demand correspond to a CES composite Air of the domestically polluting inputs (fuel savings and fuel switching). However,
produced variety and a CES import aggregate Mir of the same other GHG emissions, ozone depleting substances or air
variety from the other regions, the so-called Armington good pollutants cannot be directly linked in fixed proportions to
(Armington, 1969). Domestic production either enters the input or output activities in economic sectors.
formation of the Armington good or is exported to satisfy Approaches to endogenize non-CO2 pollution control in
the import demand of other regions. The balance of payment CGE models include: (i) the creation of clean-up sectors (“end-
constraint, which is warranted through flexible exchange of-pipe” abatement activities) separately from the technology
rates, incorporates the benchmark trade deficit or surplus for associated with the production of output that use capital,
each region. labor and other inputs to provide abatement services which
Our core (generic) CGE model can be applied for quantita- are demanded by emitting sectors as an additional input
tive trade-off analysis along the three dimensions of sustain- (Conrad and Schröder, 1991) and (ii) modeling the pollutant
ability since it allows the representation of the entire chain directly as an input into production. To introduce emission
from policy interference to implied changes in major SD control, the production function is parameterized in consis-
indicators (predominantly economic and environmental tence with technology-based marginal abatement cost curves
impacts). A wide range of policy measures that are subject to of control options (Hyman et al., 2003).
SIA is readily available in the core modeling framework such
as environmental policies (e.g. emission permits and taxes), 3.4. Initial market distortions (Eurostat 8–9; EC III, IV)
fiscal measures (e.g. structural adjustments, tax reforms) or
trade policies (e.g. trade liberalization). The CGE model focuses Real world economies are characterized by initial market
on traditional economic performance indicators (Eurostat 3, 5– distortions such as taxes (subsidies) and imperfections on
6, 8–9, 37–38; EC I–III, XIII) and environmental impacts in terms goods and factor markets. The existence of market distortions
of emissions from fossil fuel combustion, most notably CO2 can substantially alter the impacts of policy interference.
(Eurostat 36, 41, 43; EC XII). The analytical chain from In our core model, revenues are recycled lump-sum to the
instruments to impacts on SD indicators can be illustrated representative agent in each region and initial tax distortions are
for the imposition of carbon taxes. Carbon (or energy) taxes not explicitly considered. It is well known that the way
(Eurostat 37) raise marginal costs of production due to revenues from environmental regulation (e.g. emission taxes
abatement expenditure and tax payments, which leads to or auctioned tradable permits) are used has major impacts on
higher market prices. The higher domestic price attracts the social costs of the environmental policies (see, e.g.
imports and lowers exports of energy-intensive goods (Euro- Goulder, 1995). When revenues are employed to reduce
stat 71). Firms substitute labor, material and capital for the existing tax distortions, environmental regulation presents
taxed energy input to keep adjustment costs low. Consumers an opportunity to simultaneously improve environmental
reduce their consumption of energy alike. The reduced energy quality and offset at least part of the adjustment burden by
consumption (Eurostat 3, 38; EC XIII) and fossil fuel use result reducing the costs of the tax system. It is straightforward to
in reductions in CO2 emissions, the main greenhouse gas incorporate a governmental sector that collects taxes (e.g.
(Eurostat 36, 41; EC XII). Labor demand benefits from the production taxes or subsidies, intermediate input taxes,
positive substitution effect (Eurostat 8, 9; EC II, III). Material use consumption taxes, tariffs), which are used to finance the
also tends to increase. However, there is also a negative output public good provision and public transfers. Additional income
effect due to increased prices and reduced domestic produc- from environmental taxes or emission levies such as carbon
 EC O LO G I CA L E C O N O M I CS 6 0 ( 2 00 6 ) 4 9–6 4 55

taxes on intermediate and final fossil fuel use may then be production side of the economy there are also myopic
used within a revenue-neutral (equal-yield) tax reform (see, expectations assumed. Equilibrium ensures the saving–
e.g. Goulder, 1995; Böhringer et al., 1997). investment equality. The path for the economy is a set of
Persistent involutary unemployment (Eurostat 8–9; EC III, IV) at connected equilibria where the current period's saving
high levels is a central impediment to SD in many countries. augments capital in the next period. Capital stocks are
Thus, a major requirement to new policy initiatives is that updated as an intermediate calculation between periods.
unemployment problems will at least not be worsened. A Following the intertemporal approach the static model is
convenient shortcut to replace the competitive labor market in cast into an intertemporal setting where consumption and
our core model is the specification of a “wage curve” (Blanch- investment decisions are based on rational expectations of
flower and Oswald, 1994). The wage curve reflects empirical future prices (Lau et al., 2002). The intertemporal framework
evidence on the inverse relationship between the level of reveals in a consistent manner the effects of policy changes
wages and the rate of unemployment. In such a model, the on intertemporal consumption and investment (savings)
wage curve, together with labor demand, determines the level decisions, permits measurement of transition costs (inter-
of involuntary unemployment (see, e.g. Böhringer et al., sectoral adjustments) and rates of resource depletion as well
2003a). as long-term growth effects.
The core model is based on perfectly competitive goods
markets. However, there is a widespread suspicion that such a 3.6. Technological change (Eurostat 1, 2 38, 42–44, 58; EC
setting misses important industrial organization effects of V–VII)
policy interference in imperfectly competitive goods markets such
as changes in economies of scale or price mark-ups. In order to For the measurement of sustainability, an appropriate incor-
account for these effects, the core model can be extended to poration of technological change may be of paramount
feature imperfectly competitive supply behavior and increas- importance (see EMF, 1996). In our core model (as in most
ing returns to scale (see Böhringer and Löschel, 2004). Relaxing existing CGE models), technological change is considered to be
the assumption of perfect competition, allocation effects an exogenous variable. Changes in technologies are solely the
emerge from increased competition and the exploitation of result of price substitution among different given production
scale economies (so-called pro-competitive effects): Market techniques. Existing technologies are gradually replaced in
enlargement increases competition between firms which in CGE models as relative prices of alternative technologies
turn enforces lower prices. It can also lead to a higher level of change.
production and the use of economies of scale. Increased Only more recently, CGE models took into account the
competition from the greater substitutability of products empirical evidence that technological change is to an
within the enlarged market is another source for rationaliza- important degree endogenous, i.e. responding to socio-
tion gains under imperfect competition. economic (policy) variables like prices, investment in R&D,
or cumulative production (see Löschel, 2002 for an overview).
3.5. Dynamic specification (Eurostat 2, 4) Induced technological change alters the character of pro-
duction and involves invention, innovation, development,
Key issues in SD policy involve interference over longer time and diffusion of new technologies. As, e.g. environmental
periods. Examples include GHG abatement strategies to cope policy implicitly or explicitly increases the price of energy,
with global warming or direct regulation of energy technolo- firms invest in R&D with the intention of producing
gies such as an administered phase-out of nuclear power or a profitable new (energy efficient) products and processes.
phase-in of renewable energies. A dynamic framework is Goulder and Schneider (1999) construct a dynamic CGE
essential to capture the adjustment path of physical and model in which firms in each sector employ physical capital
human capital stocks during the transition towards some SD and knowledge capital to produce output. Knowledge
targets in the future for such exogenous policy changes. accumulation (expenditure on R&D activities) reduces the
Furthermore, it permits addressing issues of resource deple- input requirements for the industries. But the accumulation
tion, stock pollution and economic growth which are central of knowledge is costly. In addition, the investment in R&D
to the SD debate. may provide spillovers, or positive technological externali-
On the consumption side, dynamics involve the repre- ties (see, e.g. Otto et al., 2005). In the same manner,
sentation of the savings behavior of households. On the spillovers from, e.g. information technology expenditures
production side, dynamics involve the description of invest- could be modeled.
ment decisions (including resource exploration and extrac-
tion strategies) of firms. There are two basic approaches to 3.7. Equity (Eurostat 10, 12, 17; EC III, IV, IX–XI)
handle dynamics: (i) the dynamic-recursive framework
based on myopic expectations, and (ii) the fully intertem- Our core model considers distributional analysis across
poral setting with perfect foresight. Adopting a dynamic- regions but lacks a disaggregation of the representative
recursive approach the static core model is solved for a agent into heterogeneous households or different generations.
sequence of temporary equilibria with consumers allocating As mentioned before, the quantification of social aspects in
income between present and future consumption (through CGE models featuring a single representative household per
savings) at each point in time. Savings is based on the region is limited. To assess the distributional impacts of
expected return assuming myopic predictions (i.e. the policies, a disaggregation of the household sector into several
households assume prices to remain constant). On the types of households is required. Jorgenson and Wilcoxen
56 E CO L O G I CA L E CO N O MI CS 60 ( 20 0 6 ) 4 9–6 4

(1993) subdivided the household sector into demographic 3.8. Central steps in CGE-based SIA
groups that differ by characteristics such as family size, age of
the head of the household, region of residence, race, and The use of quantitative models for measuring SD impacts of
urban or rural location. 1344 different household types were policy reforms requires the specification of indicators, instru-
considered based on these criteria. ments, and analytical chains. First, measurable sustainability
From a dynamic perspective the analysis of distribution indicators covering the SD concerns of interest have to be
issues may require an extension towards overlapping genera- defined and instruments that may promote sustainability in
tions (OLG) models, where finite lived agents of different different domains have to be identified. Then, the complete
generations are alive each period and consume and produce path from the application of the instrument to the impact on
simultaneously. There are various examples of single-country the sustainability indicator has to be modeled. Finally, policy
CGE models with overlapping generations. Keuschnigg and explorations have to be carried out and implications on the
Kohler (1994) use a Blanchard type approach where each sustainability indicators have to be assessed.
generation has the same constant death probability indepen- The central steps involved in constructing and using CGE
dent of age (perpetual youth approach) to study intergenera- models for policy impact analysis are summarized in Fig. 2.
tional effects of fiscal policy. Rasmussen and Rutherford (2004) Initially, the policy issue must be carefully studied to decide
employ a multisector OLG model of the Auerbach–Kotlikoff on the appropriate model design as well as the required data.
type to investigate the impacts of an environmental tax The second step involves the use of economic theory (at best,
reform at the single country-level. However, an OLG frame- the draft of a simple analytical maquette model) in order to lay
work with multiple regions, sectors and households (genera- out key economic mechanisms that may drive the results in
tions) still poses considerable computational challenge and the more complex numerical model (causal chain). Data work,
requires severe trade-offs with the level of remaining details model formulation and implementation then delivers the
that can be captured in the model. framework for numerical policy analysis. This step also
1. Issue

Policy background
2. Theory

Theoretical foundation of key

mechanisms (e.g. analytical
"maquette" of numerical model)

Construction of consistent Data:

input-output tables, national accounts,
Benchmark Equilibrium Data Set tax data, income and expenditure data
formulation
3. Model

Formulation and implementation of Choice of exogenous elasticities

numerical model (incl. choice of
functional forms); scenario definition (literature survey)

Calibration: Replication
Check
Calculation of parameter values
4. Computer

successful?
simulations

No
from benchmark data

Yes
Simulations:
Calculation of new policy Sensitivity analysis
equilibrium (counterfactual)
5. Interpretation

R eporting and economic Robust

interpretation of results results?
No

Yes

Conclusions and policy

recommendations

Fig. 2 – Steps in CGE analysis (Böhringer, 1996).

 EC O LO G I CA L E C O N O M I CS 6 0 ( 2 00 6 ) 4 9–6 4 57

involves the set-up of alternative policy instruments and Table 3 – Models in the expert poll
strategies that induce changes vis-à-vis the reference situa- Modell Institution Reference
tion (scenario definition). In determining results of policy
AMIGA (All-Modular Argonne National Hanson and
simulations, the choice and parameterization of functional
Integrated Growth Laboratory, US; Laitner
forms are crucial. The procedure most commonly used to
Assessment Environmental Protection (2004)
select parameter values is known as calibration (see Mansur Modeling System) Agency (EPA), US
and Whalley, 1984). Calibration of the free parameters of DART (Dynamic Applied Kiel Institute of World Klepper et
functional forms requires a consistent one year's data (or a Regional Trade Economics (IfW), Germany al. (2003)
single observation constructed from averaging over a number Model)
of years), together with exogenous elasticities that are usually DREAM (Danish DREAM CGE-Modelling Knudsen et
Rational Economic Group, Denmark al. (1998)
taken from literature surveys. Benchmark data is typically
Agents Model)
delivered in value terms. In order to obtain separate price and EDGE (Dynamic General Copenhagen Economics, Jensen and
quantity observations, the common convenient procedure is Equilibrium Model) Denmark Thelle
to choose units for goods and factors so that they have a price (2001)
of unity in the benchmark equilibrium. In a dynamic EPPA (Emissions Massachusetts Institute of Babiker et
framework, the baseline calibration encompass the calibra- Projection and Policy Technology (MIT), US al. (2001)
Analysis Model)
tion of the model to long-term projections on economic
GEM-E3 (General Catholic University of Capros et al.
growth and energy use. The calibration is a deterministic
Equilibrium Model forLeuven, Belgium; National (1998)
procedure and does not allow for statistical test of the model Energy–Economy– Technical University of
specification. The one consistency check that must necessar- Environment Athens, Greece; Center for
ily hold before one can proceed with policy analysis is the Interactions) European Economic
replication of the initial benchmark: the calibrated model Research (ZEW), Germany
must be capable of generating the base-year (benchmark) GEMINI–E3 (General Ministry of Equipment, Bernard et
Equilibrium Model of France; Atomic Energy al. (in press)
equilibrium as a model solution without computational work.
International Agency, France; University
Within the policy simulations single parameters or exogenous
National Interaction of Geneva, Switzerland
variables are changed and a new (counterfactual) equilibrium for Economy–Energy–
is computed. Comparison of the counterfactual and the Environment)
benchmark equilibrium then provides information on the G-Cubed (Global Australian National Wilcoxen
policyinduced changes of economic variables such as employ- General Eq. Growth University, Australia; and
ment, production, consumption, relative prices, etc. Finally, Model) Syracuse University, US McKibbin
(1999)
the model results must be interpreted based on sound
GRAPE (Global Institute of Applied Energy, Kurosawa
economic theory. In that, theoretical analysis and numerical Relationship Japan et al. (1999)
work are complementary. As theoretical models must be Assessment to
highly stylized to keep analytical tractability their direct Protect the
contribution to actual policy analysis remains limited. Nu- Environment)
merical methods are required to account for policyrelevant GTAP (Global Trade Purdue University, US Hertel and
complexities but must be accompanied by theoretical analysis Anal. Project Model) McDougall
(2003)
to detect potential (programming) errors and to reduce the
GTEM (Global Trade and Australian Bureau of Tulpule et
black-box character of quantitative simulations. Environment Model) Agriculture and Resources al. (1999)
The extent to which policy instruments alter sustainable Economics (ABARE),
development indicators depends crucially on the responsive- Australia
ness of supply and demand with respect to price changes (i.e. IGEM (Intertemporal Harvard University, US; Jorgenson
elasticities). Due to the reliance on exogenous elasticity values General Equilibrium Syracuse University, US and
Model) Wilcoxen
and a single base-year observation, comprehensive sensitivity
(1993)
analysis on key elasticities and possibly alternative assump-
MIRAGE (Modelling Centre d'Etudes Bchir et al.
tions on economic incentives should be performed before Intern. Relationships Prospectives et (2002)
concrete policy recommendations are derived. A deliberate in Applied General d'Informations
sensitivity analysis helps to identify robust insights on the Equilibrium) Internationales (CEPII),
complex relationships between assumptions (inputs) and France
results (outputs), i.e. sort out the relative importance of a MONASH Monash University, Dixon and
Australia Rimmer
priori uncertainties.
(2002)
PACE (Policy Analysis Center for European Böhringer
Based on Computable Economic Research (ZEW), et al.,
4. E3–CGE models for SIA: evidence from an Equilibrium Model) Mannheim (in press-b)
expert poll SGM (Second Pacific Northwest National MacCracken
Generation Model) Laboratory (PNNL), US et al. (1999)
In order to assess the state-of-the-art in SIA based on CGE (continued on next page)
analysis, we asked different CGE modeling groups to what
extent key indicators for SD as listed in Tables 1 and 2 are
currently captured or subject to feasible extensions in their
58 E CO L O G I CA L E CO N O MI CS 60 ( 20 0 6 ) 4 9–6 4

Table 3 (continued) combustion of fossil fuels. Non-CO2 greenhouse gases such as

Modell Institution Reference CH4 or N2O that cannot be directly linked in fixed proportions
to input or output activities in economic sectors are only
WIAGEM (World German Institute for Kemfert
weakly represented.
Integrated Economic Research (DIW), (2002)
Assessment General Germany Tables 5a b c d — which features slight re-arrangements of
Equilibrium Model) the EUROSTAT indicator list — reveals in much more detail
WORLDSCAN Central Planning Bureau Bollen et al. the shortcomings of existing E3–CGE-models with respect to
(CPB), Netherlands (1999) the extensive wish-list for impact assessment.
Whereas the theme “economic development” (see Table
5a) with central indicators including consumption, saving,
operational E3–CGE-models. The modeling groups had to fill labor productivity, labor costs, or employment rates, are
out the Tables 1 and 2, thereby marking with an “x” whether broadly represented, social aspects (see Table 5b) addressed
the indicators are currently measurable within their model in the themes “poverty and social exclusion”, “aging society”,
(column “CORE”) or could be measured after appropriate or “public health” are more or less missing in current E3–CGE
extensions (column “EXTENDED”). Blank spaces indicate the models; insightful in this respect is that nearly all social
modelers' assessment that the specific indicator is considered indicators are considered to be out of scope of feasible model
to be out of the current and extended model's scope. In total, extensions. On the one hand, these deficiencies can be
our SIA questionnaire has been filled out by 18 modeling explained by the explicit focus of E3–CGE-models on non-
teams (see Table 3). social dimensions. On the other hand, various indicators —
Tables 4 and 5 provide a descriptive evaluation of the poll e.g. social participation — are very vaguely defined and
results. We start with the assessment of responses on the difficult if not impossible to measure quantitatively.
coverage of the 14 EU structural indicators listed in the Spring Within the environmental dimension of SIA (see Table
Report on Sustainable Development (Table 4). Almost all 5c), the thematic field “climate change and energy” is widely
operational versions of E3-models can quantify central covered which reflects the policy demand and research
economic indicators such as GDP per capita, labor produc- focus in applied environmental modeling during the last
tivity, or the level of employment. Indirect measures of decades. In fact, concerns on climate change and energy use
economic potential such as R&D, information technology have driven to a larger extent the development of E3-
expenditure, or spending on human resources are hardly models. Other central aspects of sustainable environmental
covered in most of the operational E3-models and also management including consumer awareness, biodiversity,
viewed with reserve as feasible extensions. Social indicators water and land use are considered out of the scope of most
such as at-risk-of-poverty rates are not at all present in current E3-models.
current model versions and not regarded as feasible future Finally, there are various indicators to characterize good
extensions within most E3-models. governance that are absent from the investigated E3-models
With respect to the environmental dimension, the EC (Table 5d), again owing to the specific objectives of E3-models
structural indicators only cover greenhouse gas emissions, but also due to inherent measurement problems.
energy intensity, and transport volumes. As E3-models are
deliberately designed to link economic and energy flows, the
comprehensive coverage of energy intensity and energy- 5. Model extensions and model linkages
related emissions is not surprising. Yet, it should be made
clear that — at second glance — incorporation of greenhouse The E3–CGE models evaluated in our poll cover only few
gas emissions mainly refers to carbon dioxide emissions from indicators for SIA. Larger gaps exist on the field of

Table44– –Coverage
Table Coverage of structural
of structural indicators
indicators (EC, by
(EC, 2003b) 2003b) by E3–CGE-models
E3–CGE-models (Total: 18) (Total: 18)

Indicator
I. GDP per capita 17 1
II. Labor productivity 16 2
III. Employment rate 10 6
IV. Employment rate of older workers 1 3
V. Spending on human resources 3 6
VI. Research and Development expenditure 4 9
VII. Information Technology expenditure 2 5
VIII. Financial market integration 1 1
IX. At risk-of-poverty rate 1 3
X. Long-term unemployment 3 4
XI. Dispersion of regional employment rates 2 3
XII. Greenhouse gases emissions 16 2
XIII. Energy intensity of the economy 17 1
XIV. Volume of transport 10 5

CORE: Indicator is represented in the operational.

EXTENDED: Indicator is viewed as feasible extension.
 EC O LO G I CA L E C O N O M I CS 6 0 ( 2 00 6 ) 4 9–6 4 59

Table 5a– –Coverage

Table 5a Coverage of economic
of economic indicators
indicators (Eurostat,
(Eurostat, 2004) by2004) by E3–CGE-models
E3–CGE-models (Total: 18) (Total: 18)

Theme Sub-theme Areas to be addressed

Economic Investment 1. Investment in R&D 6 7
development 2. Investment in EFT 7 6
3. Consumption and inflation 16
4. Saving and borrowing 15 1
Competitiveness 5. Labour productivity 15 2
6. Unit labour costs 12 4
7. Life-long Learning 3
Employment 8. Employment rate 9 5
9. Unemployment rate 6 6

CORE: Indicator is represented in the operational model.

EXTENDED: Indicator is viewed as feasible extension.

environmental impact analysis and in particular with respect loosely termed as soft-link vis-à-vis hardlink. Roughly speak-
to social impact assessment. Inherently, the strength of rather ing, the soft-link approach involves combination of two or
aggregate, economy-wide CGE models in capturing sustain- more models that have been developed independently from
ability effects of policy initiatives at the level of different another and can be run stand-alone. Due to the heterogeneity
regions, sectors and households cause deficiencies when it in complexity and accounting methods across different
comes to more narrow, specific or small-scale impact assess- models, the soft-link approach stands out for substantial
ment. There are many complementary quantitative models problems in achieving overall consistency and convergence of
that feature substantially more details of technological con- iterative solution approaches. On the other hand, it allows for
ditions (e.g. engineering bottom-up energy system models), maintaining detailed information embodied within the vari-
socio-economic household behavior (e.g. micro-simulation ous — often interdisciplinary — models without requiring
models), or natural science relationships (e.g. climate models, comprehensive expertise. Furthermore, linkages can be based
water stress models, land-use models). on established models rather than requiring modeling work
This raises the question to which extent and in which from scratch. These rather pragmatic advantages may out-
manner different models can be linked towards a more weigh to some degree impending deficiencies in overall
comprehensive coverage of SIA requirements. In principle, consistency. The hard-link approach puts strong emphasis
there are two basic approaches for model linkages which are on internal consistency and therefore makes use of a single

Table 5b– –Coverage

Table 5b Coverage of social
of social indicators
indicators (Eurostat,
(Eurostat, 2004) by2004) by E3–CGE-models
E3–CGE-models (Total: 18) (Total: 18)

Theme Sub-theme Areas to be addressed

Poverty and Monetary poverty 10. Income inequality 4 5
social exclusion 11. Non-monetary deprivation 1
Access to lab. mark. 12. Poverty-in-work 2
Other aspect of 13. Access to education
social exclusion 14. Access to health care
15. Access to housing
16. Social participation
Aging society Pensions adequacy 17. Income of elder generations 1 4
Demographic chang. 18. Life expectancy 3 3
19. Fertility 3 3
20. Migrations 1 9
Financial 21. Age of withdr. from labour market 2 3
sustainability 22. Pension expenditures 2 3
Public health Human health 23. Financial sustainability 1 3
protection and 24. Disability-free life expectancy 2
life styles 25. Premature mortality 2
26. Life styles
27. Health and safety at work 1
28. Infectious dis. & resistance antibio. 1
Food safety and 29. Pesticide residues 3
quality 30. Microbiological contamination 2
31. Drinking water quality 2
Chemicals managem. 32. Chemicals production and cons. 5 1
33. Exposure to chemicals 2
Health risks due 34. Air quality 4 1
to env. conditions 35. Noise exposure 2

CORE: Indicator is represented in the operational model.

EXTENDED: Indicator is viewed as feasible extension.
60 E CO L O G I CA L E CO N O MI CS 60 ( 20 0 6 ) 4 9–6 4

Table 5c– –Coverage

Table 5c Coverage of environmental
of environmental indicators
indicators (Eurostat,
(Eurostat, 2004) by 2004) by E3–CGE-models
E3–CGE-models

Theme Sub-theme Areas to be addressed

Climate change Climate change 36. GHG emission reduction 16 2
and energy Energy 37. Energy taxes 17 1
38. Energy efficiency 16 2
39. Renewable energy resources 10 6
40. Management of nuclear waste 1 4
41. Air pollution from energy use 5 8
Production and Eco-efficiency 42. Decoupling econ. growth & re. use 10 3
consumption 43. Decoupling econ. growth & emiss. 14 3
patterns 44. Decoupling econ. growth and waste 4 2
Agriculture 45. Pesticides use 2 2
46. Nitrogen balances 4
47. Environmentally-friendly farming 2
Corporate responsib. 48. Triple bottom line 1
Consumer awareness 49. Consumer information 1
Management of Biodiversity 50. Protection of water resources 1
natural 51. Maintaining the carrying capacity
resources Marine ecosystems 52. Over-fishing 2
Fresh water resources 53. Water extraction and use 1 2
54. Protection of water resources
Land use 55. Land use change 4 4
56. Soil degradation 4
57. Forests 3 6
Transport Transport growth 58. Decoupling of econ. & trans. growth 8 5
59. Road to rail, water and public 3 5
transport
60. Land use by transporrt systems 3
Env. imp. transp. act. 61. Air pollutants 4 7
[Global Resource 70. Resource consumption 12 2
partnership] management 71. Air emissions & energy 11 3
72. Water 1 3
73. Waste 3

CORE: Indicator is represented in the operational model.

EXTENDED: Indicator is viewed as feasible extension.

integrated modeling framework, e.g. our core CGE model stat 37–41; EC XIII) and transport (Eurostat 58–61; EC XIV) that
presented in Section 3. Information from other models is may be soft-linked to CGE models to cover a wider range of
directly fed into the core model. This means that data and sustainability indicators such as PRIMES (EC, 1995) and POLES
functional relationships from other models must be con- (EC, 1996), or TREMOVE (Van Herbruggen, 2002). Beyond soft-
densed and synthesized in a way compatible to the structure linked energyeconomy model systems, integrated assessment
of the core model. models (IAM) seek to combine knowledge from multiple
In practice, there have been several examples of soft-links disciplines in an analytic framework to assess the effects of
between macro-economic CGE models and energy system different policy options. The IAM framework typically features
models in order to enrich macroeconomic analysis of energy broad system linkages and feedbacks, particularly between
or environmental policies with bottom-up technological socio-economic and biophysical processes. For example,
details (see, e.g. Bergman and Lundgren, 1990). There are within the IMAGE model system (IMAGE-Team, 2001), a
various large-scale detailed sectoral models for energy (Euro- macroeconomic model and a population model feed basic

Table 5d– –Coverage

Table 5d Coverage of institutional
of institutional indicators
indicators (Eurostat,
(Eurostat, 2004) by 2004) by E3–CGE-models
E3–CGE-models (Total: 18) (Total: 18)

Theme Sub-theme Areas to be addressed

Good Policy coherence 62. Citizen’s adh. & support to EU actions
governance 63. Sust. of EU actions and measures 1
64. Legislative compliance
65. ommunication and mobilization
Public participation 66. Market access for LDC 7 3
Global Global. of trade 67. FDI to developing countries 4 5
partnership Financing for SD 68. Official Development Assistance 2 4
69. Other official financing 1 3

CORE: Indicator is represented in the operational model.

EXTENDED: Indicator is viewed as feasible extension.
 EC O LO G I CA L E C O N O M I CS 6 0 ( 2 00 6 ) 4 9–6 4 61

information on economic and demographic developments for 34, 53–54, 71–72). Berck et al. (1991) provide an overview of the
several world regions into other linked sub-models such as a use of CGE models to assess water regulation. A major
land-cover model which calculates global landuse and land- modeling challenge in the field of water policy analysis
cover changes including changes in agricultural land, forests, concerns the appropriate representation of water supply and
and desertification (Eurostat 45–47, 55–57). Another example is demand (Hertel, 1999). Decaluwe et al. (1997) have addressed
the MIT Integrated Global System Model (IGSM) consisting of a this issue in the context of a CGE model of the Moroccan
set of coupled sub-models of economic development and economy in which they investigate the implications of water
associated emissions, natural biogeochemical cycles, climate, pricing policies. Supply responses of groundwater and surface
and natural ecosystems (Prinn et al., 1998). There, a macro- water (collected by dams) are modeled stochastically. Robin-
economic model is applied to “predict” emissions used son and Gehlhar (1995) developed a CGE model for Egypt in
subsequently as an input in the atmospheric chemistry which land and water are combined in a linear fashion in the
model and the climate model. To date, most of the potential sectoral production function.
feedbacks between the socio-economic and biophysical sys- Similar hard-links may substantially improve the applica-
tems are not formally modeled owing to uncertainty on bility of the E3–CGE-model family for problem-tailored SIA in
concrete causal chains or commensurability problems. In- various policy fields such as land use, desertification or
stead, the sub-models use the results of the economic model agriculture. Difficulties might arise in the reconciliation of
as exogenous parameters. In other words, there is only a one- top–down and bottom–up data stemming from different data
way soft link between economic variables and their relation- sources. Due to different accounting methods (e.g. different
ship with biophysical variables. Exceptions with two-way depreciation rules) substantial data adjustments may be
links where biophysical variables (such as air quality) affect necessary before a consistent data base for the hard-linked
consumer welfare, labor productivity or capital depreciation model is available.
include Nordhaus (1994) and Vennemo (1997).
Hard linkages stand out for the consistent decomposition
of a single framework into various model segments. As
6. Concluding remarks
illustrated initially by Böhringer (1998) in a static stylized
macroeconomic model, the detailed bottom-up representa-
The objective of SD needs a comprehensive methodology to
tion of certain segments of the economy within an otherwise
perform SIA quantitatively. An issue that cannot be clearly
aggregate model is straightforward: Practical applications to
measured will be difficult to improve. In this paper, we have
energy regulation (e.g. Böhringer et al., 2003b) testify that such
investigated the use of energy–economy–environment (E3)
hybrid models can enhance the transparency and “credibility”
CGE models for measuring the impacts of policy interference
of simulated technological responses. Recent examples of
on policy-relevant economic, environmental, and social
hard-linkages between socio-economic and bio-physical
(institutional) indicators: Operational versions of E3–CGE
model include integrated assessment of the costs and benefits
models have a good coverage of central economic indicators,
from climate change policies (Böhringer et al., in press-a,in
whereas environmental indicators with complex natural
press-b). There, complex relationships in the climate system
science background and — in particular — social indicators
have first been simplified through appropriate aggregation, i.e.
are hardly represented. Our cross-model evaluation confirms
reduced forms of more elaborated climate models (Nordhaus
the need for future modeling activities in the field of
and Yang, 1996; Nordhaus and Boyer, 2000).
integrated assessment that link standard E3–CGE models to
Bottom-up indicators may also be directly incorporated
theme-specific complementary models with environmental
into multi-sector, multi-region CGE models through exoge-
and social focus.
nous coefficients or estimated “meta”-functions. A prominent
Our exclusive focus on quantitative (CGE-based) analysis
example is the representation of complex abatement options
should not exaggerate the role numerical approaches can play
for non-CO2 greenhouse gases which are not modeled in
in SIA. Policy decisions are the outcome of a broader
detail. Instead, exogenous marginal abatement cost curves
participatory process where stakeholders and other interested
for non-CO2 greenhouse gases based on sophisticated bot-
parties communicate a wide range of values, perceptions and
tom-up analysis are employed (Hayhoe et al., 1999; Reilly et
judgements to policy makers (Tamborra, 2002). Quantitative
al., 1999). Fæhn and Holmøy (2003) link consumption of
analysis — if available at all — can at best strengthen or
material goods to solid waste generation for deposition
weaken policy arguments, putting decision making on a more
(Eurostat 44, 73). Xie and Saltzman (2000) use an environ-
informed basis.
mentally extended social accounting matrix to identify three
general types of pollution (waste water, smog dust, and solid
waste) and include the respective pollution-abatement sec-
tors (Eurostat 31, 34, 44, 71–73). Strutt and Anderson (2000) use Acknowledgements
a comprehensive environmental input–output data set com-
plemented by case studies to project anticipated changes in We want to thank all respondents of our expert poll for
technology in order to assemble a matrix of environmental participation and the anonymous referees for critical and
coefficients over time. Based on these coefficients, they helpful comments. Financial support by the European Com-
estimate the environmental impact per unit of economic mission (DG Research) under the projects Methodologies for
activity in each sector and project environmental outcomes Integrating Impact Assessment in the Field of Sustainable Develop-
for water use, water pollution, and air pollution (Eurostat 31, ment (MinimaSud), Transition to Sustainable Economic Structures
62 E CO L O G I CA L E CO N O MI CS 60 ( 20 0 6 ) 4 9–6 4

(TranSust) and Indicators and Quantitative Tools for Improving the Böhringer, C., Müller, A., Wickart, M., 2003b. Economic impacts of a
Process of Sustainability Impact Assessment (I.Q. TOOLS) is premature nuclear phase-out in Switzerland. Swiss Journal of
Economics and Statistics 139 (4), 461–505.
gratefully acknowledged. Löschel acknowledges support
Böhringer, C, Löschel, A., Rutherford, T.F., in press-a. Decomposing
from the Fritz Thyssen Stiftung for a research stay at Stanford
the Integrated Assessment of Climate Change. Journal of
University. Regarding any remaining inadequacies, the usual Economic Dynamics and Control.
caveat applies. The views expressed in this paper belong to the Böhringer, C, Löschel, A., Rutherford, T.F., in press-b. Efficiency
authors and should not be attributed to the European gains from “what” — flexibility in climate policy — an
Commission or its services. integrated CGE assessment. Energy Journal, Special Issue.
Bollen, J., Gielen, A., Timmer, H., 1999. Clubs, ceilings and CDM:
macroeconomics of compliance with the Kyoto Protocol.
REFERENCES Energy Journal, Special Issue: The Costs of the Kyoto Protocol: A
Multi-Model Evaluation, pp. 177–206.
Borges, A., 1986. Applied general equilibrium models: an assess-
Armington, P.A., 1969. A theory of demand for products ment of their usefulness for policy analysis. OECD Economic
distinguished by place of production. IMF Staff Papers 16 (1), Studies 7, 7–43.
159–178. Capros, P., Georgakopoulos, T., Filippoupolitis, A., Kotsomiti, S.,
Bhattacharyya, S.C., 1996. Applied general equilibrium models for Atsaves, G., Proost, S., van Regemorter, D., Conrad, K.,
energy studies: a survey. Energy Economics 18, 145–164. Schmidt, T.F.N., 1998. The GEM-E3 Model Reference Manual.
Babiker, M.H., Reilly, J.M., Mayer, M., Eckaus, R.S., Sue Wing, I., National Technical University, Athens.
Hyman, R.C., 2001. The MIT Emissions Prediction and Policy Conrad, K., 1999. Computable general equilibrium models for
Analysis (EPPA) model: revisions, sensitivities, and compar- environmental economics and policy analysis.
isons of results. MIT Joint Program on the Science and Policy of In: van den Bergh, J.C.J.M. (Ed.), Handbook of Environmental
Global Change. . Report, vol. 71. Massachusetts Institute of and Resource Economics. Edward Elgar, Cheltenham,
Technology, Cambridge, MA. pp. 1061–1087.
Bchir, M.H., Decreux, Y., Guérin, J.-L., Jean, S., 2002. MIRAGE, Conrad, K., 2001. Computable general equilibrium models in
a Computable General Equilibrium Model for Trade Policy environmental and resource economics. In: Tietenberg, T.,
Analysis. CEPII, Working Paper No 2002-17. Centre Folmer, H. (Eds.), The International Yearbook of Environmental
d'etudes prospectives et d'informations internationales, and Resource Economics 2002/2003, pp. 66–114.
Paris. Conrad, K., Schröder, M., 1991. An evaluation of taxes on air
Berck, P., Robinson, S., Goldman, G., 1991. The Use of Computable pollutant emissions: an applied general equilibrium approach.
General Equilibrium Models to Assess Water Policies. In: Dinar, Swiss Journal of Economics and Statistics 127 (2), 199–224.
A., Zilberman, D. (Eds.), The Economics and Management of Decaluwe, B., Patry, A., Savard, L., 1997. When water is no longer a
Water and Drainage in Agriculture. Kluwer Academic Publish- gift from heaven: a CGE model of the Moroccan economy. Paper
ers, Boston, MA. presented at the DIAL/PARADI Conference on CGE Modeling in
Bergman, L., 1990. The Development of Computable General Developing Economies, Paris, 4.–5.9.1997.
Equilibrium Modeling. In: Bergman, L., Jorgenson, D.W., Dixon, P.B., Rimmer, M.T., 2002. Dynamic, general equilibrium
Zalai, E. (Eds.), General Equilibrium Modeling and Economic modelling for forecasting and policy: a practical guide and
Policy Analysis. Basil Blackwell, Cambrigde, pp. 3–30. documentation of MONASH. Contributions to Economic
Bergman, L., Lundgren, S., 1990. General Equilibrium Approaches Analysis, vol. 256. North-Holland Publishing Company,
to Energy Policy Analysis in Sweden. In: Bergman, L., Amsterdam.
Jorgenson, D.W., Zalai, E. (Eds.), General Equilibrium Modeling EAP (Environmental Action Programme), 1973. First Environmen-
and Economic Policy Analysis. Basil Blackwell, Oxford, pp. tal Action Plan, 1973–1977. Published in OJ C 112, 20.12.1973.
351–382. EC (European Commission), 1995. PRIMES. DG XII, EUR 16713,
Bernard, A., Vielle, M., Viguier, L., in press. Burden Sharing Within Brussels.
a Multi-Gas Strategy. Energy Journal, Special Issue. EC (European Commission), 1996. POLES 2.2. DG XII, EUR 17358,
Blanchflower, D.G., Oswald, A.J., 1994. The Wage Curve. MIT Press, Brussels.
Cambridge, MA. EC (European Commission), 1999. EXTERNE: Externalities of
Böhringer, C., 1996. Allgemeine Gleichgewichtsmodelle als In- Energy. Vol. 1–10, Brussels 1995/99. http://www.ExternE.
strument der energie — und umweltpolitischen Analyse. info1999.
Theoretische Grundlagen und Empirische Anwendung. Peter EC (European Commission), 2001. A sustainable Europe for a better
Lang, Frankfurt. world: a European union strategy for sustainable development.
Böhringer, C., 1998. The synthesis of bottom–up and top–down in Commission's Proposal to the Gothenburg European Council,
energy policy modeling. Energy Economics 20 (3), 233–248. COM(2001)264 Final, Brussels, 15.5.2001.
Böhringer, C., Löschel, A., 2002. Economic impacts of carbon EC (European Commission), 2003a. Sustainability impact assess-
abatement strategies. In: Böhringer, C., Finus, M., Vogt, C. ment of trade agreements — making trade sustainable?
(Eds.), Controlling Gobal Warming, Perspectives from Eco- Background Paper, DG Trade Seminar, Brussels, 6– 7.2.2003.
nomics, Game Theory and Public Choice. Edward Elgar, EC (European Commission), 2003b. Structural Indicators. Com-
Cheltenham, pp. 98–172. munication from the Commission, COM(2003)585 final, Brus-
Böhringer, C., Löschel, A. (Eds.), 2004. Climate Change Policy and sels, 8.10.2003.
Global Trade. ZEW Economic Studies, Physica. Heidelberg. EMF (Energy Modeling Forum), 1996. Markets for energy efficiency.
Böhringer, C., Rutherford, T.F., Pahlke, A., Fahl, U., Voβ, A., 1997. EMF Report 13, Volume I. Stanford University, Stanford, CA.
Volkswirtschaftliche Effekte einer Umstrukturierung des Eurostat, 2004. The EU sustainable development strategy: a
deutschen Steuersystems unter besonderer Berücksichtigung framework for indicators. Doc. SDI/TF/30/4 rev. 5 (2004),
von Umweltsteuern. IER, Stuttgart. Brussels.
Böhringer, C., Conrad, K., Löschel, A., 2003a. Carbon taxes and Fæhn, T., Holmøy, E., 2003. Trade liberalisation and effects on
joint implementation — an applied CGE analysis for Germany pollutive emissions to air and deposits of solid waste. A general
and India. Environmental and Resource Economics 24 (1), equilibrium assessment for Norway. Economic Modelling 20,
49–76. 703–727.
 EC O LO G I CA L E C O N O M I CS 6 0 ( 2 00 6 ) 4 9–6 4 63

Francois, J.F., Reinert, K.A. (Eds.), 1997. Applied Methods for Trade Lee, N., Kirkpatrick, C., 2001. Methodologies for sustainability
Policy Analysis: A Handbook. Cambridge University Press, New impact assessments of proposals for new trade agreements.
York. Journal of Environmental Assessment Policy and Management
Friedrich, R., Bickel, P. (Eds.), 2001. Environmental External Costs of 3 (3), 395–412.
Transport. Springer, Berlin. Löschel, A., 2002. Technological change in economic models of
Goulder, L.H., 1995. Effects of carbon taxes in an economy with environmental policy: a survey. Ecological Economics 43,
prior tax distortions: an intertemporal general equilibrium 105–126.
analysis. Journal of Environmental Economics and Manage- MacCracken, C.N., Edmonds, J.A., Kim, S.H., Sands, R.D., 1999. The
ment 29, 271–297. economics of the Kyoto Protocol. Energy Journal, Special
Goulder, L.H., Schneider, S., 1999. Induced technological change, Issue: The Costs of the Kyoto Protocol: A Multi-Model
crowding out, and the attractiveness of CO2 emissions Evaluation, pp. 25–72.
abatement. Resource and Environmental Economics 21 (3–4), Mansur, A., Whalley, J., 1984. Numerical specification of
211–253. applied general equilibrium models: estimation, calibration,
Hanson, D., Laitner, J.A., 2004. An integrated analysis of and data. In: Scarf, H.E., Shoven, J.B. (Eds.), Applied General
policies that increase investments in advanced energy- Equilibrium Analysis. Cambridge University Press, New
efficient/low-carbon technologies. Energy Economics 26, York, pp. 69–127.
739–755. Nordhaus, W.D., 1994. Rolling the “DICE”: an optimal transition
Hayhoe, K., Jain, A., Pitcher, H., MacCracken, C., Gibbs, M., path for controlling greenhouse gases. Resource and Energy
Wuebbles, D., Harvey, R., Kruger, D., 1999. Costs of multi- Economics 15, 27–50.
greenhouse gas reduction targets for the USA. Science 286, Nordhaus, W.D., Boyer, J., 2000. Warming the World: Economic
905–906. Models of Global Warming. MIT Press, Boston, MA.
Hertel, T.W., 1999. Applied general equilibrium analysis of Nordhaus, W.D., Yang, Z., 1996. A regional dynamic general-
agricultural and resource policies. Staff Paper 99-2. Department equilibrium model of alternative climate-change strategies.
of Agricultural Economics, Purdue University. American Economic Review 86, 741–765.
Hertel, T., McDougall, R., 2003. GTAP Model Version 6.2. Center for Otto, V., Löschel, A., Dellink, R., 2005. Energy biased technical
Global Trade Analysis. Purdue University, West Lafayette. change — a CGE analysis. Fondazione Eni Enrico Mattei (FEEM)
Hyman, R.C., Reilly, J.M., Babiker, M.H., De Masin, A., Jacoby, H.D., Nota di Lavoro 90.2005, Milano.
2003. Modeling non-CO2 greenhouse gas abatement. Environ- Parris, T.M., Kates, R.W., 2003. Characterizing and measuring
mental Modeling and Assessment 8 (3), 175–186. sustainable development. Annual Review of Environment and
IMAGE-Team, 2001. The IMAGE 2.2 implementation of the SRES Resources 28, 559–586.
scenarios. A comprehensive analysis of emissions, climate Prinn, R., Jacoby, H., Sokolov, A., Wang, C., Xiao, X., Yang, Z.,
change and impacts in the 21st. Century. RIVM CD-ROM Eckaus, R., Stone, P., Ellerman, D., Melillo, J., Fitzmaurice, J.,
publication 481508018, National Institute for Public Health and Kicklighter, D., Holian, G., Liu May, Y., 1998. Integrated
the Environment, Bilthoven, The Netherlands. global system model for climate policy assessment: feed-
Jensen, J., Thelle, M.H., 2001. What are the Gains from Multi-Gas backs and sensitivity studies. Climatic Change 41 (3/4),
Strategy? Fondazione Eni Enrico Mattei (FEEM), Nota di Lavoro 469–546.
23.2002, Milano. Rasmussen, T.N., Rutherford, T.F., 2004. Modeling overlapping
Jorgenson, D.W., Wilcoxen, P.J., 1993. Reducing US carbon emis- generations in a complementarity format. Journal of Economic
sions: an econometric general equilibrium assessment. Re- Dynamics and Control 28 (7), 1383–1409.
source and Energy Economics 15 (1), 7–26. Reilly, J., Prinn, R., Harmisch, J., Fitzmaurice, J., Jacoby, H.,
Kemfert, C., 2002. An integrated assessment model of economy– Kicklighter, D., Melillo, J., Stone, P., Sokolov, A., Wang, C., 1999.
energy–climate — the model WIAGEM. Integrated Assess- Multiple gas assessment of the Kyoto Protocol. Nature 401,
ment–An International Journal 3 (4), 281–298. 549–555.
Keuschnigg, C., Kohler, W., 1994. Modeling intertemporal general Robinson, S., Gehlhar, C., 1995. Land, Water and Agriculture in
equilibrium: an application to Austrian commercial policy. Egypt: the Economywide Impact of Policy Reform. IFPRI TMD
Empirical Economics 19, 131–164. Discussion Paper, vol. 1. Washington, D.C.
Kirkpatrick, C., Lee, N., 1999. WTO New Round Sustainability Shoven, J.B., Whalley, J., 1992. Applying General Equilibrium.
Assessment Study, Phase Two Main Report (Manchester Cambridge University Press, Cambridge.
Study). Institute for Development Policy and Management and Strutt, A., Anderson, K., 2000. Will trade liberalization harm the
Environmental Impact Assessment Centre, University of environment? The case of Indonesia to 2020. Environmental
Manchester. http://www.idpm.man.ac.uk/sia-trade. and Resource Economics 17, 203–232.
Klepper, G., Lorz, J.-O., Stähler, F., Thiele, R., Wiebelt, M., 1994. Tamborra, M., 2002. Socio-Economic Tools for Sustainability
Empirische allgemeine Gleichgewichtsmodelle — Struktur und Impact Assessment — the Contribution of EU Research to
Anwendungsmöglichkeiten. Jahrbücher für Nationalökonomie Sustainable Development, European Communities. Office for
und Statistik 213, 513–544. Official Publications of the European Communities,
Klepper, G., Peterson, S., Springer, K., 2003. DART 97. Kiel Working Luxembourg.
Papers, vol. 1149. Kiel Institute of World Economics (IfW), Kiel. Tulpule, V., Brown, S., Lim, J., Polidano, C., Pant, H., Fisher, B.S.,
Knudsen, M.B., Petersen, L.H., Petersen, T.W., Stephensen, P., 1999. The Kyoto Protocol: an economic analysis using GTEM.
Trier, P., 1998. Danish Rational Economic Agents Model– Energy Journal, Special Issue: The Costs of the Kyoto Protocol: A
DREAM, Version 1.2. Statistics Denmark, Copenhagen. Multi-Model Evaluation, pp. 257–286.
Kurosawa, A., Yagita, H., Zhou, W., Tokimatsu, K., Yanagisawa, Y., UN (United Nations), 2001. Indicators of sustainable development:
1999. Analysis of carbon emission stabilization targets and guidelines and methodologies. United Nations, New York.
adaptation by integrated assessment model. Energy Journal, UNCED (The United Nations Conference on Environment and
Special Issue: The Costs of the Kyoto Protocol: A Multi-Model Development, Earth Summit), 1992a. Rio declaration on
Evaluation, pp. 157–175. environment and development, vol. 1. Rio de Janiero, June 3–14,
Lau, M.I., Pahlke, A., Rutherford, T.F., 2002. Approximating 1992. A/Conf. 151/26.
infinite-horizon models in a complementarity format: a primer UNCED (The United Nations Conference on Environment and
in dynamic general equilibrium analysis. Journal of Economic Development, Earth Summit), 1992b. Agenda 21. Rio de Janeiro,
Dynamics and Control 26, 577–609. June 3–14, 1992. http://www.un.org/esa/sustdev/agenda21.
64 E CO L O G I CA L E CO N O MI CS 60 ( 20 0 6 ) 4 9–6 4

U.S. Department of Commerce, 1993. Input–Output Table of the U.S. Wilcoxen, P.J., McKibbin, W.J., 1999. The theoretical and empirical
Economy, 1987. Bureau of Economic Analysis, Washington, DC. structure of the GCubed Model. Economic Modelling 16,
Van Herbruggen, B., 2002. Depth Description of the TREMOVE 123–148.
Model. Transport and Mobility Leuven, March 2002, Leuven. Xie, J., Saltzman, S., 2000. Environmental policy analysis: an
Vennemo, H., 1997. A dynamic applied general equilibrium model environmental computable general-equilibrium approach for
with environmental feedbacks. Economic Modeling 14, 99–154. developing countries. Journal of Policy Modeling 22 (4),
WCED (World Commission on Environment and Development, 453–489.
The Brundtland Commission), 1987. Our Common Future. Young, H.P., 1994. Equity in Theory and Practice. Princeton
Oxford University Press, Oxford. University Press, Princeton, N.J.