A partnership with academia

Building knowledge for trade and development

Vi Digital Library - Text Preview

Review of Analytical Tools for Assessing Trade and Climate Change Linkages

Working paper by Truong, Truong P./ ARTNeT, 2010

Download original document

This paper briefly refers to the essential elements underlying the theoretical linkages between trade, economic development, and climate change, and reviews the analytical tools which are used to describe these linkages. The author looks specifically at a particular type of analytical tool called computable general equilibrium (CGE) models, and considers their strengths and limitations when used as a tool for the analysis of these trade and climate change linkages. The paper finds that the tool have been more useful than ‘misused’, and this explains for the popularity of its use in the past.








Review of Analytical Tools for Assessing Trade and
Climate Change Linkages





By


Truong P. Truong








Truong P. Truong is Honorary Professor of Sustainable Transport Systems, The Institute of Transport and
Logistics Studies, The University of Sydney, Australia and an advisor of ARTNeT for work on Trade and
Climate Change Linkages. This paper was prepared as part of the Asia-Pacific Research and Training Network
on Trade (ARTNeT) initiative on building research capacity in the area of Trade and Climate Change Linkages.
The technical support of the United Nations Economic and Social Commission for Asia and the Pacific is
gratefully acknowledged. The opinions are the responsibility of the author and should not be considered as
reflecting the views or carrying the approval of the United Nations, ARTNeT, or the University of Sydney. Any
errors are the responsibility of the author, who can be contacted at T.Truong@itls.usyd.edu.au  




The Asia-Pacific Research and Training Network on Trade (ARTNeT) is aimed at building regional


trade policy and facilitation research capacity in developing countries. The ARTNeT Working Paper
Series disseminates the findings of work in progress to encourage the exchange of ideas about trade
issues. An objective of the series is to get the findings out quickly, even if the presentations are less than
fully polished. ARTNeT working papers are available online at www.artnetontrade.org. All material in
the working papers may be freely quoted or reprinted, but acknowledgment is requested, together with
a copy of the publication containing the quotation or reprint. The use of the working papers for any
commercial purpose, including resale, is prohibited.











Asia-Pacific Research and Training Network on Trade


Working Paper Series, No. 81, July 2010




  2


Table of Contents


Executive Summary ...................................................................................................................4


1. Introduction..........................................................................................................................5


2. Trade and Climate Change Linkages – Theoretical Hypothesis..........................................5


2.1 Impacts of Trade on Climate Change: Scale, Composition, Technique and Direct
effects..........................................................................................................................6


2.2 Impacts of Climate Change on Trade: Productivity changes, Changes in
Comparative Advantages ............................................................................................7


3. Trade and Climate Change Linkages – Empirical Analytical Tools ...................................8


3.1 Econometric techniques ..............................................................................................8


3.2 General equilibrium models........................................................................................9


4. The Use of Computable General Equilibrium (CGE) Models for Policy Analysis in the
Area of Climate Change and Trade Linkages............................................................................9


4.1 Why 'general equilibrium'? .........................................................................................9


4.2 What 'computable'? ...................................................................................................10


4.3 Uses and pitfalls of CGE models ..............................................................................13


5. Examples of CGE Models Used for the Study of Trade and Climate Change Linkages ..14


5.1 ORANI-G model.......................................................................................................14


5.2 GTAP model for the analysis of global trade issues.................................................23


6. Conclusions........................................................................................................................31


Appendices...............................................................................................................................32













  3


List of Figures


Figure 1: Trade, investment and climate change linkages.........................................................6


Figure 2: Comparative-static interpretation of results .............................................................11


Figure 3: ORANI Model – data base .......................................................................................15


Figure 4: ORANI Model – Theoretical structure associated with the use of intermediate
goods in current production ......................................................................................16


Figure 11: GTAP Model – Production structure......................................................................25


Figure 12: GTAP-E Model – Production structure..................................................................25


Figure 13a: GTAP-E Model – CO2 tax – first best .................................................................26


Figure 13b: GTAP-E Model – CO2 tax – second best .............................................................27 



List of Tables





Table 1: Percentage deviation of emissions from projected Business-as-Usual level for period
2005-2007according to the NAP(*)..........................................................................28


Table 2: Marginal Abatement Cost (1995US$/ton of CO2) when there is no emissions trade.
...................................................................................................................................28


Table 3: Marginal Abatement Cost (1995US$//ton of CO2) when there is domestic emissions
trading only. ..............................................................................................................29


Table 4: Marginal Abatement Cost (1995US$//ton of CO2) when there are domestic
emissions trading as well as regional emissions trading...........................................29


Table 5: Macroeconomic effects of Regional Emission Trading (*).......................................30













  4


Executive Summary


Trade and climate change are clearly among the most important economic and
political issues facing the global community. Although it is generally agreed that the two
areas are closely related, the nature and outcome of these linkages are still debatable. On the
one hand, there is a view that trade can contribute negatively to the problem of climate
change because of its impacts on the level of economic activities and the impact on
international transport. On the other hand, there is also the contrary view that trade is not
only helpful, but may even be necessary, for the development, diffusion and transfer of
technologies which can help in the combat against climate change. To assist in the
understanding of the nature of these complex interrelationships and to assess their overall
impacts on the economy and the environment, especially with respect to the problem of
climate change, it is important that we understand the theories behind these interrelationships
and use the practical models which are built to represent these linkages in the analysis of
climate change and trade policies. In this paper, we briefly refer to the essential elements
underlying the theoretical linkages between trade, economic development, and climate
change and review the analytical tools which are used to describe these linkages. We look
specifically at a particular type of analytical tool called computable general equilibrium
(CGE) models; consider their strengths and limitations when used as a tool for the analysis of
these trade and climate change linkages. The paper finds that the tool have been more useful
than ‘misused’, and this explains for the popularity of its use in the past. Looking to the
future, to increase the usefulness of the tool in the area of policy analysis, there will need to
be continuing training for the policy analysts in the modern and expanding techniques of
CGE modelling. Such training will include not only the surveying and reading of the
literature and understanding the basic theories but also ‘hands on’ experience on its practical
applications. This survey paper therefore is only an important first step towards that ultimate
direction.



Insofar as trade leads to growth, and growth leads to an increased willingness and
ability to pay for a cleaner environment, freer trade and investment flows will enable
countries to adapt better to any adverse effects of climate change and to mitigate
emissions. Sallie James (2009). p.14.


Globalization...has been a major driver behind global warming. This trade model
has promoted the production and consumption of goods regardless of their impact on
our environment, excessive and wasteful shipping of goods globally, depletion of
natural resources at a break-neck pace...Free trade has most significantly
contributed to global warming ...Sierra Club (2008) p. 2


Trade...can - at best - offer no more than part of the answer to climate change. It is
not in the WTO that a deal on climate change can be struck, but rather in an
environmental forum, such as the United Nations Framework Convention on Climate
Change. Pascal Lamy, WTO Director- General, Bali, December 2007














  5


1. Introduction


Trade and climate change are clearly among the most important economic and political issues
facing the global community. Although it is generally agreed that the two areas are closely related, the
nature and outcome of these linkages are still debatable. On the one hand, there is a view that trade
can contribute negatively to the problem of climate change because of its impacts on the level of
economic activities and the impact on international transport (Sierra Club, 2008). On the other hand,
there is also the contrary view that trade is not only helpful, but may even be necessary, for the
development, diffusion and transfer of technologies which can help in the combat against climate
change (see, for example, James (2009)). To assist in the understanding of the nature of these complex
interrelationships and to assess their overall impacts on the economy and the environment, especially
with respect to the problem of climate change, it is important that we understand the theories behind
these interrelationships and use the practical models which are built to represent these linkages in the
analysis of climate change and trade policies. In this paper, we briefly refer to the essential elements
underlying the theoretical linkages between trade, economic development, and climate change and
review the analytical tools which are used to describe these linkages. We look specifically at a
particular type of analytical tool called computable general equilibrium (CGE) models; consider their
strengths and limitations when used as a tool for the analysis of these trade and climate change
linkages.


The plan of the paper is as follows. Section 2 explains the theoretical linkages between trade
and climate change issues. Section 3 looks at the analytical tools used in the analysis of these
linkages. Section 4 looks more closely at a particular type of analytical tool: CGE models, and
assesses the strengths and limitations of this tool. Section 5 gives some examples of the use of CGE
models in the analysis of trade and climate change linkages. Section 6 concludes.




2. Trade and Climate Change Linkages – Theoretical


Hypothesis


Trade and climate change can be assumed to be linked in several ways. Figure 1 shows a
schematic diagram of these linkages.
























  6


Figure 1: Trade, investment and climate change linkages


Source: Cosbey (2007)




2.1 Impacts of Trade on Climate Change: Scale, Composition, Technique and Direct effects


The impacts of trade and investment policy on climate change can be summarised in terms of
four different components1: scale effects, composition effects, technique effects, and direct effects. In
practice, these different components are closely intertwined and it's hard to separate them out, but
from a theoretical viewpoint, it is useful to distinguish between these components so that we can have
a better understanding of the nature of the interrelationships.



• Scale effect: this is the effect that trade (and investment) policy can have on climate change via a


change in the scale of production and consumption activities. For example, if trade results in an
increase in the level of economic activities in certain sectors of an economy, and/or certain parts
of the world, and if these increased activities result in higher levels of GHGs emissions, then trade
can be said to have a negative impact on climate change. The scale effect is almost always
negative; therefore, criticisms of the current trading system often resort to this scale effect to point
to the negative impact of globalisation on the environment and especially on climate change.



• Composition effect: trade and climate change policies can also have impacts on the patterns of


production and consumption activities in different countries. For example, through trade opening,


                                                            
1 "Scale, composition, and technique effects" were first used by Grossman and Kruger (1991) and others to
describe the impacts of the North American Free Trade Agreement (NAFTA) on pollution levels in North
America. The precise definitions of these terms in the context of a general equilibrium model were subsequently
given in Copeland and Taylor (1994). 




  7


the income level of trading countries can increase and if we assume that the environment is a
normal good, then an increase in income level will lead to an increase in demand for this good.
The pressure of demand for more environmental good means the patterns of production and
consumption activities will have to change and shift gradually from a reliance on environmentally
'dirty' goods (such as steel, cement, and chemicals) towards 'cleaner' goods (such as electronics,
telecommunications, and other services)2. These composition effects can have a beneficial impact
on climate change. However, this depends also on other factors. For example, if climate change
regulations in rich countries are not matched by similar regulations in other poorer countries, then
the 'leakage effects' implies the beneficial composition effects in the former countries will be
offset by the negative composition effects in the latter countries.



• Technique effects. Trade liberalization (and investment agreements which may go with it) can


bring about changes in production techniques which are often more energy efficient, and hence
emit less GHGs per unit of output. The changes in production techniques can come about from
the autonomous pressure of competition but can also be induced by policies. For example, the
European Union climate policy of targeting the share of renewable energy in production and
consumption activities of the European Union in the year 2020 to a level of 20% may have the
effect of inducing climate friendly technological change in the European Union. Currently within
the Doha Round, there are discussions about how to use trade liberalisation in the area of so-
called environmental goods and services (EGS) to help in the diffusion and transfer of climate
friendly technologies between countries.3 The analysis of these climate and trade policies linkages
may require further research using tools which can capture the essential elements of these
linkages.



• Direct effects: free trade increases the demand for international transport of goods. Transport


currently uses fossil fuels and hence this will increase the overall emissions of GHGs. The direct
(negative) effects of trade and transport on the environment and climate change, however, must be
considered in the context of trade and transport are only a means to an end ('margin' commodities)
rather than an end in itself (i.e. final commodities). Therefore, although the direct effects of trade
and transport on the environment are always negative, this does not mean these activities are not
necessary or useful for other activities. Negative direct effects are only part of the overall scale,
composition and technique effects considered previously.



2.2 Impacts of Climate Change on Trade: Productivity changes, Changes in Comparative


Advantages


The impacts of climate change on trade can be summarised under two headings:(i) physical
impacts of climate change on the natural resource endowments of a particular country which then
affects the comparative advantage of the country in international trade, and (ii) policy impacts of
climate change policies on comparative advantage or competitiveness of firms in these countries.



• Physical impacts: with rising temperature, changing level of precipitation, increased level of CO2


concentration in the atmosphere, productivity of the agricultural sector may be affected. It has
been estimated (see Cline (2007) for example) that agricultural productivity in some regions such
as India, South East and South West Plains of the United States, Mexico, South Africa, Ethiopia
can be reduced by these aspects of climate change by as much as -20% to -30%. Some other
regions, however, may gain: For example, China, the United States (other than South East and


                                                            
2 This is also the main hypothesis underlying the so-called 'Environmental Kuznets Curve' (EKC) (see World
Bank (1992), Grossman and Kruger (1995)). It has been suggested (see WTO (2009, p.52)) that although the
hypothesis may work well for the case of a local environmental good attached to a specific country, it may not
apply well to the case of a global environment issue such as GHGs emissions because in this case the bulk of the
costs of GHGs emissions are borne by other countries and hence there is always very little incentive left for the
polluting country to reduces its own emissions even if its income are rising. 
3 See WTO (2009). 




  8


South West Plains), Canada, Germany, Spain, Russian Federation can gain in agricultural
productivity, and these gains can range from about 5% to 12%. The increase in temperature as
well as other aspects of climate change such as the bleaching of coral reefs, forest die-off, and
fundamental ecological changes can also affect other sectors of the economy such as tourism and
infrastructure (harbour, shipping docks, etc.).



• Policy impacts: climate change policies can affect the comparative advantage of a country and the


competitiveness of firms in various sectors of an economy. One of the principal concerns when
countries try to implement unilateral climate change policies is the fact that such policies may not
be effective from the global environmental viewpoint. This is because of the problem of so-called
'leakage': environmental goods in one country are offset by environmental bads in other countries
due to a lack of international policy co-ordination. Another important concern is the impacts of
such unilateral policies on the relative comparative advantages of a country in international trade,
and also the relative competitiveness of different firms in different sectors of the economy in
domestic trade. To deal with these concerns, there have been suggestions that some border tax
adjustment (BTA) measures such as environmental tariffs could be applied. However, the
effectiveness of such policy measures can be doubtful and the impacts of such measures on the
world economic and trading systems can also be unpredictable. Therefore, there is a need for
further research into these trade-climate change policy linkages before such policies measures
could be adopted.





3. Trade and Climate Change Linkages – Empirical Analytical
Tools


The most common tools which are used in applied analysis of trade-climate change linkages
are (i) econometric techniques, and (ii) applied (or computable) general equilibrium models4.



3.1 Econometric techniques


Generally, these are used to establish partial statistical relationships between certain
environmental or climate change variables (temperature, humidity, precipitation, wind velocity, etc.)
and some specific socio-economic variables. For example, in the study of the (partial or direct)
impacts of climate change on health issues, regression analysis can be used firstly to establish a
statistical relationship between morbidity or mortality rates (dependent variable) and maximum daily
or average weekly temperature, humidity ratio, wind velocity, etc. (independent variables). This
statistical relationship is then fed into some other micro-simulation or computable general equilibrium
models to estimate more generally the overall impact of a particular temperature (climate change)
scenario on the health condition of a particular region.5 Similarly, in the analysis of the impact of
climate change on the tourism industry, regression analysis can first be used to establish a relationship


                                                            
4  Some authors (e.g. Mitra-Kahn (2008)) distinguish between 'applied general equilibrium' (AGE) and
'computable general equilibrium' (CGE) models, citing mainly their differences in historical origins and 
computational methods as the reasons. However, as Hertel et al. (1991) pointed out, these differences are not
really the main issue because the two 'schools' have much in common in theory. Dixon (2006) also pointed out
that computational technique of the 'AGE School' (Scarf algorithm) was mainly inspirational rather than
practical or 'applied' and in fact, by the 1980s, it was completely abandoned by the 'AGE School' in favour of
more traditional techniques used by the 'CGE school' such as Newton-Ralphson and Euler algorithms.
Computational technique is also becoming less of an issue for economists using CGE models because now with
the advent of powerful modern computer and computational softwares (such as GEMPACK (Pearson (1988),
Harrison and Pearson (1996)) or GAMS (Kendrick et al. (1988)), economists can concentrate on the more
productive task of interpreting and explaining the results rather than the methods of computation behind these
results. 
5 See, for example, Kalkstein et al. (1987). 




  9


between the total number of tourist arrivals or departures from a particular region and the
environmental or climate change variables relating to this region. The relationship can then be used in
a more general micro or macro simulation model to estimate the overall impacts of some climate
change scenarios on tourism.6 Econometric techniques therefore, can be part of the set of analytical
tools which are used in the analysis of the impacts of climate change on economic activities or the
linkages between climate change and trade policies.

3.2 General equilibrium models


Linkages between climate change and the economy and international trade are much more
complex than a simple (and partial) econometric relationship can capture and hence these
relationships must be used in conjunction with a more general or comprehensive simulation model
(such as CGE model) to analyse the overall linkages between trade and climate change. In theory the
'general equilibrium' in a CGE model refers to the complete balance between supply and demand in
all markets described by the model. In practice, however, disequilibrium is also routinely allowed into
these models to capture the real situations in some specific markets. For example, if the labour market
is in a Keynesian-type disequilibrium situation because of the 'stickiness' of labour price, then one can
assume that this price is to be determined exogenously of the model (e.g. by union bargaining) and
then a 'slack' variable is introduced into the model which represents the excess demand or supply
which now characterise this market. The slack variable is now endogenously determined by the model
instead of the labour price variable. 'General equilibrium' is thus not a very accurate term to be used to
describe these models. A more accurate term would be 'general inter-connectedness' or 'general
linkages' model.7





4. The Use of Computable General Equilibrium (CGE) Models
for Policy Analysis in the Area of Climate Change and


Trade Linkages


In this section we look more closely at the CGE tools which are commonly used in the
analysis of trade-environment-climate change issues. We explain why these tools are used, what their
strong characteristics are and what are their limitations or inappropriate uses.

4.1 Why 'general equilibrium'?

An alternative to 'general equilibrium' is 'partial equilibrium' analysis using methods such as
regression analysis which is used to estimate parts of the linkages between trade and climate change
(see previous section 3.1). Partial equilibrium analysis is not comprehensive enough to account for all
the complex interrelationships between trade, economic activities and climate change issues. These
complex interrelationships involve the linkages between different sectors of an economy
(upstream/downstream; domestic/foreign, infrastructure/final production); different agents
(consumers, producers, governments, investors, savers, importers, exporters), different economies
(developing/developed), and also different generations (current/future).

• Different sectors of an economy may face with different choices or opportunities for reducing


GHGs emissions and the imposition of different climate change policies focusing on these
different sectors may also have different overall impacts on the economy and on the costs of these


                                                            
6 See, for example, Lise and Tol (2002), Hamilton et al. (2005). 
7 Apart from the word 'general' which is used only in a relative' sense, the word 'equilibrium' is abstract and can
be artificial. For example, in the analysis of trade-economic-environmental linkages, the idea of 'equilibrium'
between the trade-economic system and the natural environment is not something which is easy to define.
Therefore a more appropriate term to use is 'interconnectedness', or simply 'linkages'. 




  10


climate change policies. For example, if a carbon tax is imposed on producers of fossil fuels (i.e.
on upstream coal mining or petroleum refining sectors) instead of on the final consumers (down
stream electricity generation, or transport services sectors). this may save on transaction and
monitoring costs (because there are less number of establishments to monitors) but this can result
an opportunity for end-of-pipe emissions reduction method such as carbon capture and
sequestration to be utilized because there is now no incentive for the downstream polluters to
install them (see for example, Mansur, 2010).



• Different types of agents may face with different decisions in responding to climate change and


trade policies. For example, consumers may want to maximize their utilities or welfare, firms may
want to minimize their costs or maximize profits, government may have to optimize with respect
to their fiscal budget constraints and/or trade balances. Decisions by private individuals or firms
or different countries may also create 'externalities' which regulatory bodies or public
organisations (such as the UNFCCC or WTO may also want to take into account (for example,
when dealing with the issue of carbon leakage)



• Different types of economies (developing/developed) may follow different types of policies


because they face with different economic/environmental challenges (e.g. growth or poverty
reduction versus environmental protection).



• Finally, different generations (current/future) can also have different and sometimes conflicting


interests in the preservation of the environment or exploitation of natural resources.


Because of all these interconnecting differences, there is a need for an analytical framework
which can comprehensively describe and also measure up all the differences and consistently 'adding
up' all the results. For example, all income and expenditure flows of all economic activities from all
sectors of an economy (two sides of the same coin) must be balanced; total imports and total exports
of all countries must be checked to be equal; total world savings must balance total world
investments, etc. These requirements that the world economic and environmental systems to be in
some state of 'general balance' or 'equilibrium - if not for theoretical reasons then at least for
numerical consistency - is the main reason why (computable) general equilibrium' model is the
adopted tool for most trade-climate change linkage studies.



4.2 What 'computable'?


Computable general equilibrium (CGE) models can be used in different 'modes'. In the
traditional 'comparative statics' analysis, it is used to analyse the impacts of a particular policy on an
economy ('what if' or impact analysis). In a more dynamic approach, CGE models can be run in four
different modes: historical, decomposition, forecasting, and policy (or deviation) modes (see Dixon
and Rimmer (1998)).



• Comparative static analysis: in this approach, first, an economy is assumed to be in a state of


equilibrium at position A at time t=0 (see Figure 2). Assume there are 'business as usual' shocks to
some exogenous variables but no policy is applied to the economy at this point, the economy will
move away from position A but after a period of time T it will settle back again to a new
equilibrium position at point B. Now, assume that a policy shock had also been applied to the
economy at time t=0 (in addition to all the BaU shocks). The new equilibrium position at time t=T
may be at point C rather than point B. The 'distance' BC can be said as resulting from the impacts
of the policy.




  11


In comparative static analysis, the implicit assumption is that the distance BC due to the impact of
the policy can be measured independently of the time path of the economy from t=0 to t=T.8
However, as Dixon and Rimmer (1998) discovered, this is not to be the case. The policy impact
BC can depend on assumptions made regarding the time path of the economy from t=0 to t=T in
the 'base case' (i.e. when no policy was applied).9 This means even for policy impact analysis,
comparative static analysis is not sufficient and a more dynamic analysis may be necessary.


          Figure 2: Comparative-static interpretation of results
























Source: Horridge (2003).


• Dynamic analyses: in dynamic analysis, a CGE model can be run in four different 'modes':
historical, decomposition, forecast, and policy (or deviation) modes.



o Historical mode: in this mode a CGE model is run to retrace the history of the economy


over a specific period of time in the past to derive information about certain key variables
which are normally assumed to be exogenous in a comparative static analysis: changes in
technology, consumer preferences, positions of foreign demand curves for domestic
products and numerous other naturally exogenous trade variables. To set up the CGE for
this type of run, first, certain variables which are often assumed to be endogenous in a
CGE model for a comparative static run are now assumed to be exogenous: investment,
domestic output, export, imports, etc. Next, using published information about these
variables (also called 'observables') for the historical period involved, the variables can
then be shocked by their actual past values. Now the variables which are normally


                                                            
8 This assumption is necessary because in fact, for comparative analysis, the model contains no time variable
and the analysis is actually 'timeless', i.e. there is no time path involved but the only information available are
about the positions A, B, and C. 
9  Critics of the CGE model have always pointed out this fact, but this criticism only refers to the fact that
without a time path, the costs of adjustment to the new equilibrium cannot be estimated. Here, however, Dixon
and Rimmer (1998) refers to, not just the adjustment costs (which depends on the 'path' from A to B), but also
the distance BC itself. 


Employment


0 T


Change


A


years


B


C




  12


assumed to be exogenous in a comparative static run (technology, tastes, etc.) can now be
set as endogenous, in exchange for the above 'observable' variables which are now
defined as exogenous. Hence, the results of the historical run can be used to estimate the
historical values of these technological and taste change variables for the historical period
involved. In addition to these technological and taste change variables (often referred to
as 'shift' variables in a traditional CGE model because they refer to the changes in demand
which involve only a 'shift' in the curve - i.e. keeping the relative ratios of the various
component demands unchanged), other new and important variables can also be defined
and estimated using the historical run. For example, in an analogous manner to the 'shift'
variables, some 'twist' variables can also be defined which refer not to the shift in the
absolute levels of demand, but to the relative changes in demand ratios (imports to
domestic, capital to labour, motor vehicle relative to all other goods, etc.).10



o Decomposition mode: given the important results obtained from the historical run, the


model can now be rerun in a 'decomposition' mode. Here, for example, the 'endogenous'
(historically estimated) variables representing technology and taste changes (both 'shift'
and 'twist' variables) can be reset as 'exogenous'. However, with their values having been
estimated from the historical run, these values can be used as shocks in a decomposition
run. This enables us to estimate the results for other variables in the case 'with' and the
case 'without' technological change, and hence isolate or decompose the results into pure
technological change effects, and the effect due to other variables. Similar exercises can
be repeated for other variables which are assumed to be exogenous in a decomposition
run, but allowed to be endogenously estimated in a historical run. In this way, the
combined historical and decomposition runs will enable a complete decomposition of
many historical changes into various components (technological change component,
export shift component, employment change component, etc.).



o Forecasting mode: Forecasting mode is similar to historical mode, except that instead of


'exogenizing everything that we know about the past', here in a forecasting mode, we
'exogenize everything that we think we know about the future' (Dixon and Rimmer, 1998,
p. 7). Opinions from experts about future technological developments, tastes changes,
growth of the economies, changes in the natural environment and resources, etc., can be
incorporated as 'exogenous shocks' in a forecasting run. It is quite naturally expected,
however, that what we think we know about the future is much less than what we know
about the past, hence in most cases, past information is used in combination with expert
opinions about the future to provide some 'guesswork' about the future (e.g. on
technological developments) for the exogenous variables.



o Policy mode: As forecasting mode is similar to historical mode, policy mode is similar to


the decomposition mode. If historical simulation provides information on certain
exogenous variables for the decomposition mode (such as changes in technologies and
tastes), then similarly, forecasting mode provides information on certain exogenous
variables for the policy mode - except of course for those exogenous variables which are
now controlled or determined by the policy itself. While historical and decomposition
simulations should provide the same results for all variables (except for the switch in
exogenous/endogenous definitions) - this is because they are both defined by the known
past - results for a forecast (of a base case scenario) are quite naturally not the same as the
results for a policy scenario, due to differences in the values of the policy variables in the
two scenarios.




                                                            
10 See Dixon and Rimmer (1998). Traditional neoclassical analysis assumes that changes in the ratio of demands
can only come from changes in the relative (i.e. ratio of) prices. Here, the twist variables allow for this relative
demand change even if the relative prices remain unchanged. 




  13


4.3 Uses and pitfalls of CGE models
As Dixon (2006) pointed out, the 1960s witnessed "the development of large-scale, economy-


wide econometric models (e.g. the Wharton, DRI, MPS, St Louis, Michigan and Brookings models)"
and interests in CGE models were given a boost only after the 1973 oil price shock. This event was an
important turning point for CGE modelling because up to that point, applied economists were
attracted mostly to the underlying philosophy of econometric modelling of "letting the data speak"
and hence CGE models, being based on optimisation theories and input-output data rather than time-
series, were not a contender in this respect. With the oil price shock, however, the situation changed,
since time-series data do not have much to "speak" about significant oil price changes of such
magnitude as the oil price shock. Therefore, economists turned their attention to CGE models because
at least with a reliance on optimisation theories rather than merely historical data, the models can
provide some foundation on which the future can plausibly be assessed. During the 1990s, the
problem of climate change and trade added yet another challenge for econometric modelling because
historical experience with respect to these problems were limited or non-existent, hence, to predict the
future, reliance on theories must take an equally important role as data. CGE modelling became a
popular tool because of its reliance on theories and of its flexible ability to incorporate many different
kinds of theories into the model (for example, theories about induced technical change and technology
diffusion).


In summary, as Mattoo (2009, p. 3) pointed out: "An econometric approach seems
handicapped by the absence of past events and our inability to construct experiments that are
comparable with the policy changes of greatest interest". A CGE model can help to bridge these gaps
with its reliance on theoretical insights and 'expert' opinions to look into the future rather than just
reliance on historical data to look at the past. A CGE model also has the flexibility of design to allow
various theories and types of policy experiments to be simulated. These are the strengths of the CGE
modelling approach and which explains for its popularity.


On the weaknesses side, because of its complexity, a CGE model is often considered to be a
'black box' and this can lead to some misuses. McDougall (1993) considered the following examples
where CGE models can be misused:


• Misinterpretation of results: The structure of a CGE model is often rich in inter-connections
between different markets and different effects; therefore, these effects can be mixed up or
confused. A CGE model which is used to study the effects of trade liberalisation on the economy
can contain rich details on both domestic labour market and international trade. A close
connection between these two areas means that if trade liberalisation impacts on the levels of real
wages and employment, the main welfare gains from such a policy may actually come from
labour market deregulation rather than from the theories of comparative advantage and the 'gains
from trade'.



• 'Confabulation': Related to the case of misinterpretation of the results is the case of so-called


'confabulation', meaning confused interpretation of the results of a CGE model to such an extent
that the interpretation may give some explanations which makes sense from a theoretical
viewpoint but which has nothing to do with the actual structure of the model. An example is the
case of a study looking at the effect of growth in East Asian countries on the Australian economy.
The study claimed adverse welfare effects on the Australian economy due to the outflow of
foreign capital from Australia, but the actual structure of the model contained no government
taxes on foreign-owned capital and therefore the welfare effects could not have come from the
effects of foreign capital outflow.


Other weaknesses of a CGE modelling approach includes:


• Absence of money or financial assets in most CGE models: it is not easy to incorporate money and
financial assets into a CGE model, therefore, most CGE are used only for study of the 'real'
economy where, only relative prices matter. Most CGE models therefore, cannot be used for the
analysis of monetary or financial policies.




  14


• Consistent and up-to-date data for CGE modeling is not easily constructed or readily available:
All CGE models require a data set which must be constructed consistently from input-output
tables, social accounting matrices (SAMs), trade tables, and, in the case of a global economy-
energy-environment CGE models, also data on energy and environmental variables such as GHG
emissions which must be linked to the economic data. The availability of such a consistent data
set requires considerable efforts which go beyond the resources of any single individual CGE
modeler, and therefore, the existence of a consortium to compile such a database as GTAP
(Narayanan and Walmsley (2008)) is quite unprecedented and unique.




5. Examples of CGE Models Used for the Study of Trade and
Climate Change Linkages


In this section we look at the structures of some publicly available CGE models which have
been used for the analysis of trade and climate change linkages.11 The objective is to give some
insights into the workings of a CGE model, how it is constructed, what type of database it requires,
and how the model can be used for the analysis of trade and climate change issues. Emphasis is on the
hands-on techniques for the practical implementation of a CGE approach rather than discussing the
theoretical issues - which have already been covered in sections 2-4, hence we concentrate on just one
type of CGE models, namely those related to 'GTAP-family' of models. Even here, limited space
allows us only a brief overview of the main points rather than all the necessary details which are
required by a CGE specialist. For the latter, the readers are referred to practical training courses for
these models,


5.1 ORANI-G model


ORANI (Dixon et al., 1982) was a CGE model of Australia built during the late 1970s for the
analysis of trade issues in Australia. The model has been extended in several ways and the current
version, ORANI-G, has been used as a ‘template’ for the creation of many other CGE models: for
South Africa, Vietnam, Indonesia, Republic of Korea, Thailand, the Philippines, Pakistan, Denmark,
China, Fiji and Taiwan, Province of China. Apart from being among the very first CGE models built
to look at trade issues (tariff reform) for a national economy (Australia), ORANI was pioneering in
the development of techniques for data, theory, and software construction for use in the building of a
CGE model. These techniques were later used in the construction of other (global) CGE models such
as GTAP. This section briefly describes the techniques of the ORANI-type model.


5.1.1 Data base


The data base for ORANI-G is shown in Figure 3. The basic information used for the
construction of this database are firstly, input-output data on flows of intermediate goods c ={1,...,C}
from sources s ={dom, imp} to industries i ={1,...,I} used in current production (i.e. production in
current period) (matrix V1BAS); margins services (whole sale and retail trade, transport) associated
with these flows (matrix V1MAR):; commodity taxes on these intermediate goods (matrix V1TAX)
including import duty on imported goods (matrix V0TAX); primary factors (labour, capital, land)
used in current production (V1LAB, V1CAP, V1LND); production tax or subsidy (matrix V1PTX);
                                                            
11 Although all CGE models have similar theoretical approaches, their practical implementation in terms of
computer software and mathematical techniques used in solving the equations of the model can be different. To
enable us to look more closely at the practical implementation of CGE models rather than just theoretical issues
- which have already been covered in sections 2-4, we concentrate on just one type of CGE models, namely
those related to 'GTAP-family' of models. Even here, the emphasis is on a brief overview rather than details
which are necessary for a CGE specialist. For the latter, the readers are referred to practical training courses for
these models. 




  15


'other costs ticket' which stands for miscellaneous items such as municipal taxes or charges (V1OCT).
IF we sum up the columns of these matrices, we get the values of production in these industries.
These values defined the 'basic price' of commodities produced in these industries. To allow for
multiple-product industries (mainly in agricultural sector), we define a 'MAKE" matrix which shows
which commodity is produced by which industry. Next, in addition to usage (as intermediate gods) in
current production, each commodity can also have other uses: as investment goods for next period
production (V2BAS), as final goods used for household and government consumption (V3BAS,
V4BAS), as export goods (V5BAS) and as addition t inventory stock (V6BAS). There are margin
services (V2MAR, V3MAR, V4MAR, V5MAR) and commodity taxes (V2TAX, V3TAX, V4TAX,
V5TAX) associated with all these uses (except inventory), but there are no primary factor inputs in
these final consumption activities.


Figure 3: ORANI Model – data base




Source: Horridge (2003).


5.1.2 Theoretical structure


The theoretical structure for ORANI-G is shown in Figures 4-8. First, Figures 4-6 highlights
the structure associated with current production. The structure of the demand for intermediate goods
(V1BAS) is highlighted in Figure 4. Here, each intermediate good is seen to be a 'composite' of goods
produced domestically and imported goods. The composition is represented by a CES (Constant
Elasticity of Substitution) production function which allows for domestic and imported goods to be
treated as though imperfectly substitutable goods (the Armington approach, see Armington (1969,
1970)). Figure 5 highlights the structure associated with the use of primary factors (V1LAB, V1CAP,
V1LND). Here, aggregate "Labour" is assumed to be a composite of different types of labours in




  16


different occupations which are combined via a CES function. Aggregate labour is then combined
with capital and land via a CES function to produce the aggregate 'primary factors' input. This is then
combined with composite intermediate inputs and other costs via a Leontief production function (i.e.
no substitution) to produce the final output for each industry. To allow for multiple-product industries,
the 'output' of these industries are represented in terms of an 'activity level', which is then
'transformed' into outputs of different commodities via a constant elasticity of transformation (CET)
production function. The theoretical structure associated with investment and household consumption
activities are shown in Figures 7-8. Investment activity is seen to be simply a Leontief function of
various commodity inputs, while household final consumption activity is assumed to follow a Klein-
Rubin utility function (linear expenditure system of demand). Export demand is defined by a
downward-sloping demand function with a constant elasticity of demand for each commodity, and the
demand for inventory is assumed to be proportion to the level of output (with changes to be
represented by exogenous shocks).


Figure 4: ORANI Model – Theoretical structure associated with the use of intermediate
goods in current production




Source: Horridge (2003).




  17


Figure 5: ORANI Model – Theoretical structure associated with the use of primary
factors in current production


Source: Horridge (2003).





























  18


Figure 6: ORANI Model – Theoretical structure for the make matrix



Source: Horridge (2003).




  19


Figure 7: ORANI Model – Theoretical structure of investment activity



Source: Horridge (2003).








  20


Figure 8: ORANI Model – Theoretical structure of household consumption activity



Source: Horridge (2003).








  21


5.1.3 Details on the basic building block


As can be seen from Figures 4-8, the structure of a CGE model such as ORANI-G can be said
to consist of many different 'branches' joined together to form a 'tree' structure which describes either
production, consumption, or other activities. The basic 'building block' in this tree structure is either a
CES function (or its variants such as Leontief function -when substitution elasticity is zero, or CET -
when substitution elasticity is negative i.e. transformation elasticity is positive) or other functional
forms such as Klein-Rubin, CDE (constant distance of elasticities).12 The consumer or producer in a
CGE model is assumed to optimize (maximize utility or profit, minimize cost) with respect to this
function subject to a (budget, production, or cost) constraint to derive the demand functions (for final
consumption goods), or producer's demand function (for inputs into production). To get familiar with
the structure of most CGE models, therefore, it is important to get familiar with this basic 'building
block'. Appendix 2 gives a brief description of how to derive demand functions from a CES
(production, cost, or utility) function, in level form, or in percentage change form. Here it can be
summarized that, if given a basic activity structure such as given in Figure 9, and assuming that the
level of Y (production output, utility level) is related to the levels of X1 and X2 (production inputs, or
consumption levels) according to a CES function with substitution elasticity δ, i.e.:
ρρρ δδα /12211 }{ −−− += XXY (1)
where ρ=(1−σ)/σ ; α is the scale parameter; δ1, δ2 are distribution parameters; then, maximize Y with
respect to δ1, δ2 and subject to a (cost, budget) constraint:
2211 XPXPM +≥ (2)
where P, P2 are the prices of X, X2 respectively, and M is the production cost or consumer budget
level, then the optimal levels of demand for X, X2 are given by:


.2,1;}/{ == − iPPYX aveiii σσδ (3)
where
σρσρσσρσ δδ /12211 }{ PPPave += (4)

Equations (1) - (3) are all in level form. If we now convert all variables into percentage change form,
and using lower case to denote percentage change, we have:
.2,1;}{ =−−= ippyx aveii σ (5)
}{ 2211 pSpSpave += (6)



2211 xSxSy += (7)


5.1.4 Calibration


A CES function such as given in equation (1) has 4 parameters:  ,  and (or ). The
scale parameter can be eliminated by defining the variables relative to the initial (base year) values,
i.e.:


ρρρ δδ /1022201110 })/()/({/ −−− += XXXXYY (8)


                                                            
12 Hanoch (1975). CDE function is used in other models, such as GTAP (see section 5.2 below). 




  22


where the superscript '0' denotes initial value. The distribution parameters can be shown to be related
to the value share of the inputs (see Appendix 2), i.e.:


.2,1;}/{ 2211 =+= iXPXPXP iiiδ (9)
There remains only one parameter, (or ) to be determined, and this parameter is normally
assumed to be given exogenously (i.e. determined independently of the CGE model).


5.1.5 Closure


Typically a CGE model has more variables than independent equations. Equations are derived
from theories of consumer and producer behaviour, accounting identities, and various other
constraints such as zero profit conditions, market clearing, etc. Each equation can be used to estimate
the value of one (endogenous) variable. The total number of endogenous variables thus equal the
number of equations, and the rest of the variables must be regarded as exogenous, i.e. determined
outside of the model. The split between endogenous and exogenous variables for a particular
simulation is referred to as a 'closure'. The choice of closure depends on the specific objective of a
simulation. For example, in simulations which seek to estimate the impact of trade or climate change
policies on the economy in the short run, capital stock and real wages are often assumed to be
exogenous while the rates of return on capital and employment level are endogenous. In contrast, in
long run simulations, capital stock and real wages are to be determined endogenously while rates of
return to capital and employment levels are given exogenously. In some simulations, the trade balance
is allowed to change and hence specified as endogenous, while in others, it is constrained to some
exogenous value and hence will be specified as exogenous. If trade balance is exogenous, some other
variable (such as government spending, or tax rate) must be 'swapped' with trade balance, i.e.
specified as endogenous, and conversely. In Section 4.2, we have discussed the use of CGE model for
comparative static analysis as well as for historical, decomposition, forecasting and (dynamic) policy
analysis. To enable this flexibility in the use of a CGE model for various purposes, the choice of
closure must also be flexible. This flexibility in the choice of a closure to fit in with different policy
experiment is one of the advantages of a CGE modeling approach. On the other hand, because of this
flexibility in the specification of a closure, it may require some skills and experience in CGE
modeling before a user of a CGE model can independently design an appropriate closure to fit in with
a particular experiment using a large CGE model. Without this skill, the user may run into difficulties
such as non-convergence of a solution (due to numerical errors which arise from an inappropriate
choice of closure), or - even when a solution is obtained - the solution may not make much sense from
a theoretical viewpoint due to a poor choice of closure.


To illustrate the issue of closure, we can use the simple economic system of equations (5)-(7)
even if this represents only the basic building block of a CGE model rather than any substantial part
of an actual CGE model. Here, we notice that the system consists of six variables (x1, x2, p1, p2, y, pave)
and four independent equations. This means to 'close' the system, two variables must be treated as
exogenous. There are, however, 30 different combinations of endogenous/exogenous variables for this
simple system that highlights the difficult task of choosing an appropriate closure to ensure the system
has a solution and a meaningful one. As a first choice, we can select the input prices (p1, p2) to be
exogenous and let the system to solve for (x1, x2, y, pave). This closure resembles the traditional
assumptions in partial analysis of production or consumption activities where input prices are
assumed to be given exogenously and input quantities are then determined by relative input prices.
Output quantity then follows from input quantities via the production or utility function, and output
price is determined by input prices (assuming zero profit condition). Next, we can also assume (x1, x2)
to be exogenous (for example, when they are determined by a lower 'nest' where (x1, x2) are linked to
other variables). In this case, (p1, p2, y, pave) can be endogenous. Finally, we can also assume (y, pave)
as exogenous (e.g. determined by an upper nest), then (x1, x2, p1, p2) in this case will be endogenously.
The choice between these closures depends on a particular policy experiment that we are conducting.
For example, if x1 is energy usage and x2 is the aggregate primary factor input (such as labour) and y is
output activity of a particular sector (such as electricity generation), then to simulate an experiment




  23


where carbon tax is to be imposed on x1 to ensure some target GHG emissions (from the use of x1) is
met, and also to achieve the policy objective of keeping the level of employment unchanged in this
sector, then we can set x1 and x2 to be exogenous (x1 determines the level of GHG emissions and x2
determines the level of employment) while letting (p1, p2, y, pave) to be determined endogenously. A
solution for p1 will determine the level of carbon tax, p2 gives the nominal wage of labour, y is the
level of electricity output, and pave is the price of electricity. The impact of this particular policy
therefore may be a change in the level of nominal wage to keep employment constant, while letting
the price of electricity to increase and perhaps also the output to be reduced.


Figure 9: Basic building block in a CGE model - CES function






5.2 GTAP model for the analysis of global trade issues


GTAP (Hertel, 1997) is a global trade CGE model developed at the Center for Global Trade
Analysis in the Department of Agricultural Economics at Purdue university. The model has been
widely used for the analysis of global trade issues and also for the study of the linkages between trade
policies and climate change.


5.2.1 Data base of the GTAP model


The data base for the GTAP model is similar to that of ORANI-G except the it has an extra
dimension to denote 'region' since this is a multi-regional rather than national data base (see Figure
10). In addition, GTAP data base also contains information on international trade and transport
margins. In some other respects, GTAP has fewer details than ORANI-G. For example, in the
standard GTAP data base and model, there is no information on domestic margin13, all industries in
GTAP are single-product industries, labour is split into skilled and unskilled rather than into
occupations, and generally there are fewer tax details in GTAP than in ORANI-G.


5.2.2 Theoretical structure of the GTAP model


The production structure in GTAP (see Figure 11) is similar to that in ORANI-G, except that
in GTAP, an imported good can be sourced from many different regions, reflecting the multi-regional
character of the GTAP model. The consumption structure in GTAP, however, is different from that of
ORANI-G: GTAP uses a CDE (constant difference of elasticity) consumption function whereas
ORANI-G uses a Linear Expenditure System (LES) of demand based on the Klein-Rubin utility
function. With respect to the investment decision, GTAP has a different structure due to the multi-
regional characteristic of the model (see Hertel, 1999, Chapter 2). Firstly, total international
investment is balanced by total international savings. The allocation of the total volume of global
investment to different regions is decided according to two options: (i) regional and global net
                                                            
13 Although a special model called GTAP-M has been built to include domestic margins information into the
standard GTAP data base and model (see Petersen (2006)) 




  24


investments move in the same proportion so as to preserve the initial composition of capital stocks
across regions, (ii) net investment to each region is decided on the basis of the regional rates of returns
using a 'theory' which is similar to ORANI-G's theory of investment allocation across different
sectors).


5.2.3 GTAP-E extension for the analysis of global trade-energy-environmental issues


The production structure in GTAP has one limitation which makes it unsuitable for the
analysis of energy-environmental issues: it treats energy commodities (coal, oil, gas, refined
petroleum products, electricity) on the same basis as other intermediate inputs into production, hence,
given the standard GTAP production structure as shown in Figure 11, this implies there is no
substitution between energy commodities, nor substitution between energy and other factors of
production like capital and labour. This is a rigid assumption because it implies that in all production
activities, energy and emission intensities are constant and cannot be changed in response to climate
change policies. The only way to reduce total emissions, therefore, is by reducing total production
level and/or by substitution between activities. This is clearly an unrealistic assumption which does
not give technologies any significant role to play in responding to environmental and climate change
policies. As a result of this limitation, to enable the GTAP model to be used for the analysis of
energy-environmental and climate change policies, the assumptions on structure of production need to
be changed. One of the possible modifications to this structure is suggested in the model GTAP-E
(Burniaux and Truong, 2002). This modification is shown in Figure 12.


Figure 10: GTAP Model – data base




  25


Figure 5: GTAP Model – Production structure








Figure 6: GTAP-E Model – Production structure











  26


5.3.5 An example of the use of GTAP-E for the analysis of trade-climate change linkages


The multi-regional characteristic of the GTAP family of models (i.e. standard GTAP and GTAP-E
extension as well as others using the standard GTAP as the starting point) makes these models
especially appropriate as analytical tools for use in the analysis of global trade issues. Issues such as
the economic and/or environmental impacts of trade liberalization policies on the world economies
are easily handled and within these models. With an extension such as GTAP-E where the focus on
energy commodities is given enhanced attention, the issues of energy usage, technologies, taxation, as
well as issues of Greenhouse Gas (GHG) emissions and taxation from the use of these energy
commodities are also given detailed attention. A model such as GTAP-E, therefore, is most suitable
for use in the analysis of trade and climate change policy issues and their linkages. In this section, we
look at an example where the GTAP-E model has been used to analyse such policy issues.


• Introducing carbon-tax or Emissions Trading Scheme into the GTAP-E model.
To facilitate the analysis of climate change policies, the concept of a carbon tax and/or emissions
trading scheme need to be introduced into the model. If a tax is levied on the use of an energy
commodity according to the carbon emissions associated with such use, this is called a carbon tax. A
carbon tax imposed on an energy commodity usage will result in the cost of such commodity being
increased to account for (‘internalise’) the environmental costs of such usage. This will result in the
use of the commodity being reduced; hence GHG emissions will also be reduced. Conversely, if an
emissions trading scheme put a constraint on the GHG emission level (and hence on the use of such
energy commodities), this automatically imposes a ‘shadow’ price on GHG emissions, and this acts as
though a carbon tax.14 In a model such as GTAP-E, a carbon tax can be imposed either as a ‘first best’
or ‘second best’ option (see Figures 13a and 13b). In the first best case, the tax is imposed on the
energy commodity before any other distortionary taxes are imposed, whereas in the second best case,
it is imposed on top of pre-existing taxes. The difference between the two cases is that in the later
case, a given absolute level of carbon tax will result in a lower tax rate than in the former case.


Figure 7a: GTAP-E Model – CO2 tax – first best






                                                            
14 In principle, therefore, emissions trading scheme is equivalent to a carbon tax system (except for differences
with respect to the definition of endogenous and exogenous variables, i.e. closure in a CGE model experiment).
In most cases and under the condition of perfect certainty, they yield the same results, and only give different
results under condition of uncertainty (with respect to the parameters of the production and consumption
functions), (see, for example, Low (2009)).  




  27


Figure 8b: GTAP-E Model – CO2 tax – second best




• A European climate policy study using the GTAP-E model
In 2005, the EU introduced an emissions trading system (EU-ETS) in order to pursue its


Kyoto obligations. Under this scheme, major emitters from the power and manufacturing sectors are
first allocated with a given amount of CO2 emission permits. The permit allows the emitters to emit a
certain amount of CO2 in any particular year, or it can be traded in an emissions trading market. The
trading scheme allows firms to undertake the most efficient means to reduce their CO2 emissions and
therefore allows the EU meet its Kyoto obligation with a minimum amount of cost. Under this
scheme, at equilibrium, the price of the emissions permit will indicate the minimum marginal cost of
CO2 emissions abatement by the most efficient means. To estimate this minimum marginal abatement
cost (MAC) for Germany, Kemfert et al. (2006) used a version of the GTAP-E model.


First, an estimate of the likely percentage reduction of CO2 emissions from a ‘Business-as-
Usual’ or Reference scenario to meet with the quotas under the national allocation plan (NAPs) for
various regions and sectors in the EU are made (see Table 1). These emissions reductions are then
imposed on the model (emissions) variables as exogenous shocks. The model then gives an estimate
of the equilibrium values of the MACs or CO2 taxes to achieve these emissions reduction targets
under various trading scenarios (see Tables 2-4). It can be seen from these tables that if there is no
emissions trade (Table 2), the MACs for different sectors and regions will differ substantially, ranging
from a low of 0.1 $/t CO2 in the ‘paper’ sector in Denmark (which has to cut down CO2 emissions by
-7.1% relative to the BaU case – see Table 1) to a high of 163.1 $/t CO2 in the refined oil sector in
Sweden (which has to cut down CO2 emissions by -13.9% relative to the BaU case). When there is
domestic emissions trading between different sectors of a region but not across different regions, the
MAC for each region will be uniform, but this will vary across different regions, and they can range
from a low of 0.8 $/t CO2 for the UK (which has to cut down CO2 emissions by -5% relative to the
BaU case – see Table 1) to a high of 8.4 $/t CO2 for Sweden (which has to cut down CO2 emissions by
–6.3% relative to the BaU case). Finally, when there are emissions trading not only between sectors
but also across regions (Table 4), the MAC for the EU as a whole is at 2 $/t CO2 (with a total
emissions reduction of -2.2%).


Table 5 shows the macroeconomic effects of emissions trading such as changes in the GDP
level (column 2), trade balance (column 3), and the welfare effects due to CO2 tax (column 4) and due
to terms of trade effects (column 6). As expected, emissions trading will result in improvements in
GDP level, but not necessarily for trade balance for some regions (those regions which have higher
MACs than other (see Table 3) and thus have to buy emissions permits from other regions, such as




  28


Austria, Belgium, Denmark, Sweden). In general, however, most regions will gain in welfare from
emissions trading with only a few exceptions such as Italy and the Netherlands (see the last column of
Table 5) where the negative terms of trade effects is part of the reasons.


Table 1: Percentage deviation of emissions from projected Business-as-Usual level for
period 2005-2007according to the NAP(*)


Region\\
Sectors Elec


tricity
Refined
Oils Metals


Mineral
Products Paper


Motor
Equip
ment


Constr
uction Textile


Other
Indus
tries



Total


Austria -8.9 -7.9 -3.5 -4.3 -3.6 -4.9 -4.6 -5.9 -4.3
Belgium -27.4 -5.3 -5.3 -5.3 -5.3 -5.3 -5.3 -5.3 -5.3 -8.5
Denmark -26.2 -7.1 -7.1 -7.1 -7.1 -7.1 -7.1 -7.1 -7.1 -15.1
Finland -12.5 -6.2
France -0.4 -2.8 -10.3 -8.1 -1.5
Germany -3.1 -2.6 -0.5 -0.4 -1 -2.2 -2.2 -2.2 -2.2 -1.8
Greece -6.5 -16.8 -6.6 -3.4
UK -8.7 -0.9 -18.4 -5.7 -3.3 -3.3 -2.9 -2.5 -5.0
Italy -5.5 -4.2 -1.7 -3.4 -2.5
Netherlands -7.8 -7.8 -7.8 -7.8 -7.8 -7.8 -7.8 -7.8 -7.8 -3.7
Portugal -6.2 -1.2 -2.5
Spain -6.5 -3.6 -2.9 -5.4 -4.5 -3.2
Sweden -13.9 -13.9 -13.9 -13.9 -13.9 -13.9 -13.9 -13.9 -13.9 -6.3
Czech Rep. -4.5 -4.3 -4.6 -4.5 -4.1 -3.5
Hungary -3.1 -5.1 -5.1 -5.1 -5.1 -2.4
Poland -9.3 -3.8 -10.3 -2 -7.5 -6.4


(*) (Allocated emissions – Projected Emissions)/(Projected Emissions) * 100
 


Table 2: Marginal Abatement Cost (1995US$/ton of CO2) when there is no emissions
trade.


Region\\
Sectors Elec


tricity
Refined
Oils Metals


Mineral
Products Paper


Motor
Equip
ment


Constr
uction Textile


Other
Indus
tries


Austria 3.8 42.2 1.6 3.0 1.0 2.0 3.2 2.0
Belgium 11.5 32.3 1.6 4.4 3.2 5.7 6.4 3.5 7.7
Denmark 7.5 50.5 0.2 1.1 0.1 0.0 9.4 0.0 0.1
Finland 8.0
France 0.5 17.3 4.1 11.6
Germany 1.6 22.5 0.7 0.5 0.7 1.5 2.4 1.4 1.8
Greece 2.8 137.0 0.7
UK 2.2 13.3 0.4 0.3 0.0 0.0 0.2 0.0
Italy 2.8 0.0 2.1 2.6 1.7
Netherlands 3.8 30.7 1.8 8.7 1.1 0.2 0.0 0.4 13.9
Portugal 2.0 1.6
Spain 2.3 19.2 1.3 6.8 3.4
Sweden 5.1 163.1 10.7 16.2 15.0 13.4 26.9 18.7 8.8
Czech Rep. 1.1 53.8 1.3 2.5 1.4
Hungary 1.0 28.6 1.9 3.9 1.9
Poland 2.3 45.8 2.9 1.3 2.6




  29


Table 3: Marginal Abatement Cost (1995US$//ton of CO2) when there is domestic
emissions trading only.


Region\\
Sectors Elec


tricity
Refined
Oils Metals


Mineral
Products Paper


Motor
Equip
ment


Constr
uction Textile


Other
Indus
tries


Austria 3.7 3.7 3.7 3.7 3.7 3.7 3.7 3.7
Belgium 8.0 8.0 8.0 8.0 8.0 8.0 8.0 8.0 8.0
Denmark 6.1 6.1 6.1 6.1 6.1 6.1 6.1 6.1 6.1
Finland 8.0
France 2.0 2.0 2.0 2.0
Germany 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5 1.5
Greece 3.5 3.5 3.5
UK 0.8 0.8 0.8 0.8 0.8 0.8 0.8 0.8
Italy 2.6 2.6 2.6 2.6
Netherlands 3.5 3.5 3.5 3.5 3.5 3.5 3.5 3.5 3.5
Portugal 2.0 2.0
Spain 2.8 2.8 2.8 2.8 2.8
Sweden 8.4 8.4 8.4 8.4 8.4 8.4 8.4 8.4 8.4
Czech Rep. 1.2 1.2 1.2 1.2 1.2
Hungary 1.3 1.3 1.3 1.3 1.3
Poland 2.2 2.2 2.2 2.2 2.2




Table 4: Marginal Abatement Cost (1995US$//ton of CO2) when there are domestic
emissions trading as well as regional emissions trading.


Region\\
Sectors Elec


tricity
Refined
Oils Metals


Mineral
Products Paper


Motor
Equip
ment


Constr
uction Textile


Other
Indus
tries


Austria 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0
Belgium 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0
Denmark 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0
Finland 2.0
France 2.0 2.0 2.0 2.0
Germany 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0
Greece 2.0 2.0 2.0
UK 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0
Italy 2.0 0.0 2.0 2.0 2.0
Netherlands 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0
Portugal 2.0 2.0
Spain 2.0 2.0 2.0 2.0 2.0
Sweden 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0
Czech Rep. 2.0 2.0 2.0 2.0 2.0
Hungary 2.0 2.0 2.0 2.0 2.0
Poland 2.0 2.0 2.0 2.0 2.0





  30


Table 5: Macroeconomic effects of Regional Emission Trading (*)


Welfare Decomposition:
Equivalent Variation (EV)
due to various components


($Millions)



Region


Real GDP
change


(%)


Trade Balance
due


to Emission
Trading


($Millions)


Allocative
effects


due to CO2
Tax


Other
Allocative


effects
Terms of


Trade effects Total(**)
Austria 0.10 -2.4 10.5 181.7 -20.4 171.9
Belgium 0.11 -13.4 43.2 209.5 -127.7 125.2
Denmark 0.12 -11.7 33.8 163.8 -10.0 187.1
Finland 0.04 -5.7 14.2 35.2 -7.7 37.9
France 0.05 0.0 31.9 660.2 -85.3 606.0
Germany 0.06 11.6 39.7 1155.7 118.0 1312.4
Greece 0.33 -2.4 30.6 357.2 -99.5 286.1
UK 0.02 34.6 -7.7 249.4 14.4 259.1
Italy 0.00 -4.1 3.5 -40.8 -13.1 -49.8
Netherlands 0.05 -5.9 21.8 183.5 -377.2 -171.8
Portugal 0.00 0.1 0.6 1.1 9.2 10.2
Spain 0.06 -3.7 9.1 334.6 -68.7 275.9
Sweden 0.13 -4.4 28.3 264.0 -122.2 174.1
Czech Rep. 0.05 8.5 -4.3 31.3 -6.9 19.4
Hungary 0.10 1.5 3.0 49.7 -5.6 47.0
Poland 0.02 -2.7 6.9 30.3 11.9 50.1


(*) The values shown in this Table are changes from the case of ‘No Emissions Trading’ (Table 2) to the case of ‘Regional
Emission Trading’ (Table 4). All values are in 1995US$.
(**) Including a small effect due to changes in the price of capital goods.





  31


6. Conclusions


Climate change is becoming increasingly important as a scientific and economic issue facing
the world community in the same way that trade liberalisation and economic growth have always been
and continuing to be the most important issues for the world economies. The new challenge is in
understanding the close linkages between these two important areas of interests. To face with this new
challenge, one needs to build up a body of analytical tools, not only in the scientific area to
understand the physical and environmental aspects of climate change, but also in the economic and
trade area for the analysis of the linkages between these physical and environmental aspects and their
economic impacts. Econometric techniques and computable general equilibrium models are the two
most important analytical tools that have often been used in these areas. In this paper, we attempted to
explain briefly why this has been the case. One the one hand, there is a need for a comprehensive
analytical tool to analyse the complex relationships between trade and climate change issues. The tool
must also be able to produce quantitative assessments and predictions based on current knowledge.
The theoretical underpinnings of the tool, even though deeply rooted in the economic theories of
human behaviour and the economic system, cannot be confined just to these areas but must also be
able to accommodate knowledge or assumptions about other physical and environmental systems. In
other words, it is more like an analytical and computational ‘platform’ on which to build up and link
the basic pieces of information and knowledge about economic, trade, and climate change issues
rather than a single overarching ‘theory’ about the whole complex interrelationships. This has been
the main characteristic of the use of computable general equilibrium (CGE) models (supplemented
with the use of partial econometric techniques) to understand the linkages between trade and climate
change issues. This has also been the main advantage of this approach. On the other hand, because of
the complex and ambitious task, the analytical tool used has often been difficult for a beginning
analyst to understand easily and/or the lay person to appreciate the usefulness of the results from such
tools. Therefore, the tools have often been labelled as ‘black boxes’. This is its main disadvantage, in
addition to the fact that there are still many areas of interest that the tool will need to incorporate into
its ‘platform’ in the future, such as realistic theories about the ‘pure’ financial aspects of the economic
and trade systems, about future technological development, etc.


All in all, however, the tools have been more useful than ‘misused’, and this explains for the
popularity of its use in the past. Looking to the future, to increase the usefulness of the tool in the area
of policy analysis, there will need to be continuing training for the policy analysts in the modern and
expanding techniques of CGE modelling. Such training will include not only the surveying and
reading of the literature and understanding the basic theories but also ‘hands on’ experience on its
practical applications. This survey paper therefore is only an important first step towards that ultimate
direction.




  32


Appendices


Appendix 1


Linearization – Some Useful Formulae


Let Z = X.Y W= X + Y


dZ = XdY + Y dX dW=dX +dY


dZ/Z = dY/Y + dX/X dW/W = [(X)/(X + Y)]dX/(X) + [(Y)/(X + Y)]dY/(Y)


d lnZ = d lnX + d lnY d lnW = Sxd lnX + Sy d lnY


Using lower case to denote ‚‘dln‘ or percentage change:


z = x + y w = Sx x + Sy y




Appendix 2


Basic Equations for the CES building block


Maximize the level of Y (production output, utility level) which is of the CES form:


ρρρ δδα /12211 }{ −−− += XXY (A1)
subject to a (cost, budget) constraint:


2211 XPXPM +≥ (A2)
where α is a scale parameter; δ1, δ2 are distribution parameters; X1 and X2 are production inputs, or
consumption levels; P1, P2 are the prices of X1, X2 respectively, and M is the production cost or
consumer budget level. Forming the Lagrangian:


][}{ 2211
/1


2211 XPXPMXXL −−++= −−− λδδα ρρρ (A3)
The first order conditions for optimality are:


.2,1;0]}{/ 1)/1(2211
1 ==−+=∂∂ −−−−−− iPXXXXL iiii λδδαδ ρρρρ (A4)


from which we can derive:


.2,1;}{ /)1(2211
)1( =+= +−−−+− iXXXP iii ρρρρρ δδαδλ (A5)


MY /=λ (A6)
.2,1;]/[/ 2211 =+== −−− iXXXMXPS iiiii ρρρ δδδ (A7)
.2,1;]/[ )1()1(1 == ++−− iYXP iii ρρ ααδλ (A8)




  33


.2,1;)]/()[/( )1/(1 == + iPYX iii ρλαδα (A9)
In percentage change terms (using lower case to denote percentage change, i.e. x = dX/X):




)ln()ln(
)ln()ln()/1(
]ln[)/1(ln


}{


2211


222111


2211


/1
2211


XdSXdS
XdSXdS
XXdyd


XXY


+=
+−=


+−=⇒
+=


−−


−−


−−−


ρρ


ρρ


ρρρ


δδρ
δδρ


δδα


or:


avei xxSxSy =+= 221 (A9)
aveave


i
ii


i
ii


i
ii xpxSpSmXPM +=+=⇒= ∑∑∑ (A10)


avepmy =−=λ (A11)
.2,1;)1/(1 where][][ =+=−−=−−= ippypyx aveiii ρσσλσ (A12)




  34


References
Armington, P.S. (1969). The Geographic Pattern of Trade and the Effects of Price Changes, IMF Staff


Papers, XVI, July, pp 176-199.
Armington, P.S. (1970) Adjustment of Trade Balances: Some Experiments with a Model of Trade


Among Many Countries, IMF Staff Papers, XVII, November, pp 488-523.
Brooke, A., Kendrick, D., and Meeraus, A. (1988) GAMS: A User's Guide , Redwood City,


California: Scientific Press).
Burniaux J.-M., Truong, T. (2002). GTAP-E: An Energy-Environmental Version of the GTAP Model,


GTAP Technical Paper No. 16 (revised version), Center for Global Trade Analysis, Purdue
University, West Lafayette.
https://www.gtap.agecon.purdue.edu/resources/res_display.asp?RecordID=923


Cline, W. R. (2007). Global Warming and Agriculture: Impact Estimates by Country," The Peterson
Institute for International Economics.


Cosbey, A. (2007), "Trade and Climate Change Linkages", International Institute for Sustainable
Development (IISD).


Dixon P.B., Parmenter, B.R., Sutton, J.M., and Vincent, D.P. (1982) ORANI: A Multisectoral Model
of the Australian Economy, Amsterdam: North-Holland.


Dixon, P.B., and Rimmer, M. (1998), "Forecasting and Policy Analysis with a Dynamic CGE Model
of Australia", Preliminary Working Paper No. OP-90, Centre of Policy Studies and the
IMPACT Project, Monash University, Australia.


Dixon, P.B. (2006), “Evidence-based Trade Policy Decision Making in Australia and the
Development of Computable General Equilibrium Modelling", General Working Paper No.
G-163, Centre of Policy Studies and the IMPACT Project, Monash University, Australia.


Eboli, F., Parrado, R., and Roson, R. (2010) " Climate Change Feedback on Economic Growth:
Explorations with a Dynamic General Equilibrium Model", IEFE ‐ The Center for Research
on Energy and Environmental Economics and Policy at Bocconi University, Working Paper
No. 29.


Grossman, G. and Krueger, A. (1991)," Environmental Impacts of a North American Free Trade
Agreement" NBER Working Paper No. 3914, November.


Grossman, G. and Krueger, A. (1995)," Economic growth and the environment," Quarterly Journal of
Economics, 110, 353–77.


Hamilton, J., Maddison, D., Tol, R. (2005). Climate change and international tourism: A simulation
study. Global Environmental Change — Human and Policy Dimensions 15(3), 253–266.


Hanoch, G. (1975). "Production and Demand Models with Direct or Indirect Implicit Additivity,"
Econometrica 43:395-419.


Harrison, W.J., and Pearson, K.R. (1994). "Computing Solutions for Large General Equilibrium
Models Using GEMPACK", Computational Economics, Vol. 9, pp. 83-127.


Hertel, T. (1997), Global Trade Analysis. Modelling and Applications, Cambridge University Press.
Horridge, M. (2003). “ORANI-G: A Generic Single-Country Computable General Equilibrium


Model” Notes prepared for the Practical GE Modelling Course June 23-27, 2003. Centre of
Policy Studies and Impact Project, Monash University, Australia.


Hufbauer, G. C. and Kim, J. (2009). The World Trade Organization and Climate Change: Challenges
and Options. Peterson Institute for International Economics.
http://www.iie.com/publications/wp/wp09-9.pdf


International Food Policy Research Institute (IFRI) (2009), Climate Change Impact on Agriculture
and Costs of Adaptation, Food Policy Report.


James, Sallie (2009) "A Harsh Climate for Trade How Climate Change Proposals Threaten Global
Commerce" Trade Policy Analysis no. 41 (September 9), Center for Trade Policy Studies,
Cato Institute.


Kalkstein, L. S., and K. M. Valimont. 1987. Climate effects on human health. In Potential effects of
future climate changes on forests and vegetation, agriculture, water resources, and human




  35


health. EPA Science and Advisory Committee Monograph no. 25389, 122-52. Washington,
D.C.: U.S. Environmental Protection Agency.


Kemfert, C., Kohlhaas, M., Truong P. T., Protsenko, A. (2006), ‘The Environmental and Economic
Effects of European Emissions Trading’, Climate Policy 6(4), S. 441-455.


Lise, W., and Tol, R. (2002). "Impact of Climate on Tourist Demand", Climate Change , 55, 429-449.
Low, P. (2009). "Climate Change and Trade: the Challenges of International Cooperation”


ESCAP/WTO Fifth ARTNeT Capacity‐Building for Trade Research, 22‐26 June 2009,
Bangkok, Thailand, http://www.unescap.org/tid/artnet/mtg/cbtr_lowday2.pdf.


Mansur, E.T. (2010) "Upstream versus Downstream Implementation of Climate Policy" National
Bureau of Economic Research, Working Paper 16116.


Mattoo, A., Subramanian, A., van der Mensbrugghe, D., and He, J. (2009). "Reconciling Climate
Change and Trade Policy", Peterson Institute for International Economics, Working Papers
WP-09-15, (December).
http://www.petersoninstitute.org/publications/wp/wp09-15.pdf


McDougall, R.A. (1993). Uses and Abuses of AGE Models, GTAP Short Course Training document,
https://www.gtap.agecon.purdue.edu/resources/download/21.pdf.


McDougall, R.A., and Golub, A. (2007). GTAP-E: A Revised Energy-Environmental Version of the
GTAP Model. GTAP Research Memorandum No. 15, October.
https://www.gtap.agecon.purdue.edu/resources/res_display.asp?RecordID=2959


Mitra-Kahn, Benjamin H. (2008). "Debunking the Myths of Computable General Equilibrium
Models", SCEPA Working Paper 01-2008.


Narayanan, Badri G. and Terrie L. Walmsley (Editors) (2008), Global Trade, Assistance, and
Production: The GTAP 7 Data Base, Center for Global Trade Analysis, Purdue University
Pearson, K.R. (1988). "Automating the Computation of Solutions of Large Economic
Models", Economic Modelling, Vol. 7, pp. 385-395.


Piermartini, R. and Teh, R (2005). Demystifying Modelling Methods for Trade Policy. World Trade
Organization Discussion Paper No. 10.
http://www.wto.org/english/res_e/booksp_e/discussion_papers10_e.pdf


Petersen, E. (2006). GTAP-M: A GTAP Model and Data Base that Incorporates Domestic Margins,
GTAP Technical Paper 26, Center for Global Trade Analysis, Purdue University.


Sierra Club (2008) " Trade and Global Warming - What are the Connections?".
Smith, K. (1990). "Tourism and Climate Change" Land Use Policy , Vol 7, Issue 2, pp. 176-180,


April.
World Bank. (1992), World Development Report 1992, New York: Oxford University Press.


http://www.mse.ac.in/trade/World%20Bank%20(2007).pdf.
UNEP and WTO (2009). Trade and Climate Change. A Report the United Nations Environment


Programme and the World Trade Organization.
http://www.wto.org/english/res_e/booksp_e/trade_climate_change_e.pdf




Login