The Scottish Government

« Previous | Contents | Next »

Listen

TECHNICAL APPENDIX 1. Review of literature on applying the CGE modelling approach to the problem of climate change mitigation and other environmental issues

A1.1 Introduction

Our objective in this section of the report is to identify the general issues that are likely to be important in applying a environmental computable general equilibrium ( CGE) modelling approach to the problem of climate change mitigation for the specific case of Scotland. A comprehensive review of the literature, examining all issues that are relevant to a CGE modelling project, is out with the scope of this report. Instead, we focus on issues that are specific, or at least especially relevant to the case of environmentalCGE modelling. This involves reviewing the current literature in order to examine how existing CGE models deal with questions of resource use, (particularly energy use) and pollution generation in production and consumption.

A basic introduction to CGE modelling

There is a wide range of general equilibrium model types. They share their roots in Walrasian general equilibrium theory though few CGE models nowadays assume full market clearing or universal perfect competition. The schematic of a typical CGE is reminiscent of the simplest macro-economic circular flow diagram, and is often characterised by disaggregation of households and industries/commodities. The requirement of general equilibrium is a simultaneous equilibrium in commodity and factor markets at an identifiable set of relative prices. However, within this framework the variation amongst model types reflects heterogeneity of views with respect to how markets function and the particular model focus (target economy(s), types of markets and transactors, macroeconomic closures/assumptions, dynamics etc).

Typically, the core database is a social accounting matrix ( SAM), which describes the structure of the economy in terms of flows of income and expenditure between each production and consumption activity, and transfers of income to and from local and external transactors. ⁸ The parameters of a CGE model are generally determined in three ways. First, structural parameters ( e.g. industry cost structures) are given by the base-year ( SAM) data. Second, key parameters ( e.g. substitution elasticities) are identified by econometric estimation and/or informed judgment involving literature review. These parameters are imposed in the model but may be subject to sensitivity analysis. Third, all the remaining parameters are determined through calibration to the base-year dataset. This involves assuming that the base-year SAM reflects a long-run equilibrium so that running the model with no change in exogenous variables will reproduce this equilibrium.

An excellent non-technical introduction to CGE modelling is given in Greenaway et al (1993). Here we focus on key factors that should be considered in the application of CGE models to environmental issues.

Environmental CGE modelling

The application of CGE models to examine environmental issues has grown substantially over the last two decades (the key milestone in the wider policy arena is likely have been the Rio Earth Summit in 1992). Environmental CGE models tend to fall into one of two categories (Bergman, 2005). The smaller of these is models designed to examine issues relating to the management of natural resources and is mainly relevant at a global level. However, the larger category of environmental CGE models focuses on the external effects of production and consumption, primarily through the emission of greenhouse gases. Within this there are two sub-categories. First, there are a number of global and multi-country models, which are mainly used to examine trans-boundary pollution problems and multi-lateral policies. Second, there are single country models (and, to a more limited extent, single region models), which tend to be more detailed at the sectoral level and have been developed for the purpose of analysing country-specific policy issues and proposals.

Bergman (2005) provides the most recent overview of the current 'state of play' in the field, identifying the different categories of environmental CGE model outlined above, identifying the main model specification issues that should be considered in constructing different models for different purposes, and considering the types of analysis that are possible within an environmental CGE framework. An important point to make at the outset is that the case for using a CGE model for policy analysis can be made where proposed policy measures (or other potential or expected changes in economic conditions) are likely to have economy-wide (general equilibrium) effects. However, not all policy measures related to environmental issues are likely to have such effects. Some environmental problems are local and site-specific, or relate to specific substances or processes. While the resolution of such problems may be costly for some firms and households, the repercussions for the rest of the economy may be very small or non-existent.

Overview

The first issue addressed in this chapter, in Section 2, is a key one in terms of modelling climate change and broader 'sustainable development' policies: geographical/spatial focus and the 'small open economy' assumption. In Section 3 we examine the issues associated with introducing natural resource inputs, particularly energy, on the production side of the economy, alongside the standard capital, labour and material inputs. In Section 4 we examine the main issues involved in modelling the generation of pollution as a result of economic activity in general, and natural resource use in particular. We turn our attention to issues involved in modelling consumer behaviour in an economic-environmental context in Section 5. Finally we provide a summary and conclusions in Section 6.

In this review we have identified a number of studies that may provide a useful insight in considering how an environmental CGE approach may be useful in the case of Scotland. All of the studies identified in this review are summarised in Tables 1 and 2 at the end of the paper. The studies in Table 1 all fall under the global or multi-country category identified above, while the studies in Table 2 are all national or regional level applications. Where relevant, studies from these tables are cited in the text below, but full bibliographic details are provided so that individual studies of interest can be followed up independently. In the process of conducting our own review, we have also identified several review articles, which may be particularly useful to read alongside the current paper. The most recent of these is Bergman (2005), which provides an overview of the current 'state of play' in the field, but does not attempt to conduct an exhaustive survey of applications. For this purpose, Bergman (2005) refers the reader to Conrad's (1999) excellent survey article. More recently Conrad (2001) provides another review (but this latter study is more technical than the 1999 article). Wajsman (1995) provides an earlier review of environmental CGE studies. Reviews with a more specific focus on energy (rather than pollution generation) can be found in Bergman (1988) and Bhattacharyya (1996), and LÅ'schel (2002) provides a review that focuses on modelling technological change.

A1.2 Geographical/spatial focus and types of policies modelled

Global and multi-country models

As noted above, a number of environmental CGE models are global and/or multi-country models. Use of environmental CGE models has become a popular approach to examining the economic causes of the global climate change (global warming) problem. This is illustrated by the prevalence of CGE models in the OECD comparative modelling project (reported in Dean & Hoeller, 1992) on assessing the costs of reducing CO ₂ emissions. Three of the global CGE models in Table 1 were included in this project: the GREEN model (Burniaux et al, 1992a,b), the Carbon Rights Trade Model, or CRTM (Rutherford, 1992), and the Whalley-Wigle Model (Whalley & Wigle, 1992). The approach of these global CGE models is to divide the world into a number of regions (they are essentially inter-regional CGE models), with groups of countries being treated as regions of the global economy.

Since the OECD project a number of global and sub-global multi-country CGE models ( e.g. see Gottinger, 1998, Petersen, 2003 and Klepper and Petersen, unpublished, which all focus at the EU level) have been developed. Perhaps the most notable development in international CGE analysis has been through the modelling framework developed by the Global Trade Analysis Project ( GTAP), based at Purdue University in Indiana in the United States. ⁹ The GTAP project invites contributions of input-output tables for individual countries around the world and GTAP network subscribers can access and use the international CGE modelling framework developed around this database (where countries and regions can be aggregated as desired for specific applications). In recent years, a key focus of the GTAP project has been extension of the database and modelling framework for analysis of environmental issues and an increasing number of input-output and CGE modelling analyses based on the GTAP framework can be found in the literature ( GTAP also holds an annual conference where delegates present their work using the framework).

National/regional models and the small open economy assumption

However, given the specific focus of the current project on modelling the impact on the Scottish economy of reducing greenhouse gas emissions in Scotland, the multi-country focus of the contributors listed in Table 1 is of more limited relevance. Rather, the national, regional and interregional national/sub-national applications identified in Table 2 are of more direct interest. In particular, it is from studies where small open economies are modelled that we can learn the most direct lessons for Scotland. Of particular importance is the application of the 'small open economy assumption'. When this assumption is made it implies that the rest of the world is taken to be exogenous, so that only the economy under consideration is modelled. The 'small open economy' does interact with the rest of the world, via trade, but its activity is assumed to have no significant impact on the rest of the world. This means that any spillover effects from the target economy to the rest of the world are trivial, given the scale of the rest of the world, which in turn implies that there are no significant feedback effects from the rest of the world to the target economy.

Of course, it may not be appropriate to assume that the whole of the rest of the world is exogenous. For example, in the case of Scotland, it may be appropriate to conduct an inter-regional analysis of the UK economy so that policies implemented at the national level can be considered, and feedback effects between UK regions can be modelled. This will be true of any sub-global or sub-national area where interregional interactions are thought to be important. In the case of environmental problems, political boundaries may be less important than ecological ones. For example, Abler et al (2000) conduct an interregional analysis of the Susquehanna River Basin area in the US, which includes areas of 3 states, to examine forest resource issues.

However, the key point of making the small open economy assumption in the context of environmental modelling is that the global environmental impact of the target region's economic activity is taken to be trivial. Thus, it can be assumed that there are no feedback effects from the global environment to the local economy arising from disturbances in the latter. In other words, the contribution from a small open economy like Scotland to, for example, the problem of climate change, is taken to be so trivial that we would not expect economic activity in Scotland to lead to significant climate change effects in Scotland through this mechanism. Such environmental feedback effects cannot be handled in a model of a small individual country or region (any environmental impact from the rest of the world on the local economy would have to be handled exogenously). The implication of this is that while a single country/region model can be used to assess the costs to the local economy of tackling global environmental problems, any resultant benefits of an improved global environment for that economy cannot be assessed. In effect we are assuming that there will be no environmental benefits to the local economy from its individual policy actions aimed at global problems such as climate change. Considering that any beneficial feedback effects would be expected to be very small, this is may be consistent with concerns that may be expressed locally regarding the costs of unilateral action to address the climate change problem (this is the public goods problem inherent in tackling global environmental problems). ¹⁰

Types of policies commonly modelled in single country/region models

The studies in Table 2 that focus on other small open economies, and policies/disturbances that are more likely to occur, are perhaps of more direct interest in a Scottish context. Nonetheless, it is possible to draw lessons from global models and other international or multi-country (as well as from models of larger economies that may not be treated using the small open economy assumption, such as the US) in terms of how to incorporate resource use and pollution into models of economic systems. However, such models would be expected to focus on different types of environmental issues than those addressed by single country/region models. The models in Table 1 tend to focus on the problem of global climate change in the context of trans-boundary pollution problems and multi-lateral policies. Greenhouse gas emissions resulting from economic activity are also a common concern of more locally focussed national and regional models. However, even in the case of modelling larger countries or groups of countries, this is often set in the context of constraints on economic activity resulting from commitments by individual economies to limit their share of global emissions; or, put another way, the local costs of pursuing global sustainability policies.

Similarly, global models like GREEN and CRTM may focus on the problem of depleting non-renewable natural resources, through inclusion of sub-models of resource depletion, specifically fossil fuels, as part of their dynamic structure. Again, fossil fuel depletion, like global warming, is a global sustainability concern, and one that has little meaning at the level of individual countries and regions, since such resources can be imported from elsewhere in the world. The issue of endogenous fossil fuel depletion is only likely to be a concern in non-global models if the country in question is a "producer" of fossil fuels (for non-producers, the effects of depletion will be transmitted through changes in exogenous prices). It may also be an issue in the form of a self-imposed constraint as individual regions or countries attempt to 'do their bit' in conserving global resources, and/or to limit the emission of pollutants resulting from their use. More locally-focussed models may, however, be more directly concerned with questions of resource depletion where the natural resources in question are of the type that cannot easily be imported, such as fish stocks (if the economy in question is significantly dependent on fishing industries) or land, though this is not apparent in the applications reviewed here.

The range of policies and other disturbances examined at a national (or regional) level using environmental CGE models can be seen in Table 2. A few of the papers - e.g. Adams et al (2003), BÅ'hringer and LÅ'schel (2006), and Fergusson et al (2004) - focus on methodological issues such as developing an environmental component on an existing CGE model, building sustainability indicators into environmental CGE models etc but most of these studies focus on particular disturbances. The most prevalent broad policy instrument modelled is energy or carbon taxation - e.g. BÅ'hringer et al (2001), who model a unilateral national carbon tax in Germany; BÅ'hringer and Rutherford (2007), also focussing on Germany, but extending for issues such as industry exemptions on carbon taxes; SÅ'derhom (2007), carbon taxes for Sweden; Boyd and Uri (1991) fuel taxes in the US; Gottinger (1998), GHG taxes in the Netherlands; and O'Ryan et al (2003, 2005), taxes on PM10 emissions and fuel use. While Scotland's devolved fiscal powers do not extend to taxation beyond the limited income tax varying power, the applications in this area that may be of most interest are Wissema and Delink's (2007) analysis of a (national) carbon tax in the Irish economy and Li and Rose's (1995) analysis of a regional (state) carbon tax in Pennsylvania.

There are also a number of applications in Table 2 that focus on modelling the impacts of increased energy efficiency in production and/or consumption, with particular attention to the issue of economy-wide 'rebound' effects (where reductions in energy consumption from increased efficiency are partially or wholly offset due to the effects of reductions in effective energy prices) - e.g. Semboja (1994); Dufournaud et al (1994); Grepperud and Rasmussen (2004); Glomsrød and Taojuan (2005); Hanley et al (2006); Allan et al (2007a); and Barker et al (2007). Modelling rebound effects in an environmental CGE framework has actually been the focus of a programme of research directly commissioned in the UK by DEFRA, the results of which are reported in Allan et al (2006), and Herring (2006). Another study resulting from environmental CGE work commissioned by a government body that may be of interest in a Scottish context is Learmonth et al's (2007) work on modelling the economic and environmental (local pollution) impacts of changes in population in the economy of Jersey (a UK crown-dependency). This is an example of examining the impacts of implementing a non-environmental, but sustainability-related, policy on other sustainability indicator variables.

However, the common problem for all models, whatever their geographical focus, is how to incorporate an economy's consumption of resources, particularly energy, and the resulting generation of pollutants, into a CGE model of an economic system. In this sense it is useful to draw on a wide variety of models, and this is the purpose of the following two sections on modelling energy use (Sections 3) and pollution generation (Section 4).

A1.3 Issues involved in modelling energy use in production

In a survey of general equilibrium approaches to energy modelling, Bergman (1988) identifies the crucial factor in determining the system-wide effects of changes in energy supply and demand as the elasticity of substitution between energy and other factors of production. He goes on to argue that this finding suggests that the representation of the substitutability of energy between other factors of production is one of the most basic issues that has to be addressed in energy/environmental CGE modelling. Our review of the literature on modelling energy before and since Bergman's (1988) survey demonstrates that this continues to be the key issue. A particular source of controversy is whether energy and other factors of production are in fact substitutes or complements, and whether this varies between the short and long run. It is outwith the scope of the current (non-technical) review to examine these issues in detail; here, we limit the discussion to identifying the key questions that would arise and to summarise what would seem to be the key points of debate in the literature.

The main questions would seem to be as follows:

• What is the best way to model the production structure so that, where appropriate, it can cope with the reality of multiple inputs and elasticities of substitution that differ over types of input, as well as between different sectors and over time?
• What types of energy input should be modelled to capture a full range of general equilibrium effects, including the pollution effects of different input choices and/or the response to environmental policies to reduce pollution? (since some inputs will be more polluting than others).
• How important is the country-specific context and institutional setting for which the model is being built? The economic, social and policy conditions present in the particular country/region being modelled will influence how the model is built.

Modelling the production structure

The main source of controversy lies in how different studies address the first of these questions. The most common approach to modelling production relationships in CGE models generally involves using nested production functions involving a hierarchy of Constant Elasticity of Substitution ( CES), as well Cobb-Douglas ( CD) and/or Leontief (where appropriate), relationships between different inputs. The motivation for using CES forms is that they permit more flexibility than CD, where the elasticity of substitution must be equal to unity, and Leontief, where the elasticity of substitution is equal to zero ( i.e. no substitutability at all - fixed factor proportions). However the flexibility offered by the CES functional form is limited:

The name 'constant elasticity of substitution' derives from the fact that using CES means employing a production function where the elasticity of substitution is the same regardless of factor proportions and scale.

For any one CES relationship, the elasticity of substitution must be the same among all factors. This problem can be overcome by nesting a series of CES relationships within a hierarchical structure. However this requires imposing a set of separability assumptions.

The models reviewed here demonstrate alternative ways of dealing with substitution between energy (E) and the other factors of production (mainly capital (K), labour (L), and intermediate materials (M) - known as KLEM production functions). Most of the studies employ nested production functions of one (or more, if different sectors are treated in different ways ¹¹) of the forms shown in Figure 1 below. This means that they employ separability assumptions among the KLEM inputs. Generally energy inputs (making up the energy composite, E) are produced by the energy sectors of the economy in question or imported, but are treated in a different manner to the non-energy (produced) intermediates.

Figure A1.1: Alternative specifications for production functions involving energy

For exposition, it should be noted that each combination of two goods is normally termed a "composite" good, e.g. a energy-capital composite substitutes with labour in function C. The letters s in the figure above correspond to the elasticity of substitution between the energy good and the other good with which it substitutes. In some models, the function shows above only corresponds to a section of the overall production structure.

However, while this is the most common approach, it can be argued that this type of nested production structure is still too inflexible because of the imposition of separability among the inputs. To avoid this problem, Hertel & Mount (1985), Depotakis & Fisher (1988) and Li & Rose (1995) adopt some type of flexible functional form ( FFF) production function (using the dual to the production function, the cost function). The idea is to make the production function as flexible possible by minimising the number of prior assumptions about its form. In practice, however, this argument over whether to use CES or FFF is likely to boil down to a trade off between flexibility and tractability. In a model with a highly detailed treatment of energy, Naqvi (1998) argues that separability assumptions are necessary from a practical point of view, where there are multiple inputs and/or multiple sectoral outputs. Indeed, Hertel & Mount (1985), Depotakis & Fisher (1988) and Li & Rose (1995) all choose to employ two levels of cost functions, with substitution between KLEM inputs on the first level, then within the energy and/or materials aggregates on the second level. Thus, while advocating the employment of flexible functional forms to reduce the number of restrictions, including separability, that are imposed on the production function by use of nested CES functions, they are in fact prepared to accept some separability assumptions.

This highlights an important issue: why is separability acceptable in one case but not the other? None of the models reviewed here report on testing different separability assumptions ( i.e. carrying out model specification tests). Therefore we can only presume that the decision over separability is based on some prior belief or judgement as to the appropriate assumptions to adopt. We would argue, however, that this decision would be better made as the result of actually testing alternative assumptions, where it is practical to do so, an issue that we are exploring in current research for Scotland and the UK. ¹²:

The second main area of debate in, but not limited to, the environmental CGE literature is what has become known as the 'energy-capital complementarity controversy'. This problem has been a major issue and the source of considerable controversy in energy-economy modelling ever since research began in this area. Field & Berndt (1981) explain how the controversy arose from disparity in the estimates of the elasticity of substitution between energy and capital in early empirical studies. A detailed consideration of this debate is out with the scope of this review. The main point is that issues such as the non-homogeneity of capital, and different vintages/ages of machinery will be important, as will how capital and energy enter the production function (see the discussion above), differences in usages of capital and energy across different sectors and the time-frame under consideration. Once again, these are empirical questions that need to be addressed on a study-specific basis. However, as with issue of specifying production functions, addressing major model specification issues of this type, and estimation and sensitivity analysis of key parameter values, is generally more suited to longer term research programmes, such as the current ESRC and EPSRC-funded work being carried out at the Fraser of Allander Institute for the Scottish and UK economies.

Different types of energy input modelled

A variety of different types of energy input are modelled in the studies reviewed here. The most commonly identified energy types are oil, gas, coal (the three basic fossil fuel types) and electricity. However, the decision over what energy/natural resource types are modelled is largely dependent on the actual nature of energy/resource demand and supply in the economy in question (as well as the availability of energy use data in an appropriate format - i.e. that is consistent and compatible with the input-output tables, which form the core database for any CGE model).

The different types of energy identified would also seem to reflect the problem that the modeller is attempting to examine with respect to pollution. Models that focus mainly on CO2 emissions, and particularly policies to affect these using carbon taxes, appear to pay most attention to the relative carbon content of different types of fossil fuels. This is because CO2 emissions are primarily dependent on the fuel properties, with generation resulting from any economic activity relating directly to the amount and carbon content of fuel combusted. However, where models focus on a wider range of pollutants, the generation of emissions from fuel use depends on a many more factors than carbon content. This is perhaps why Beauséjour et al (1994, 1995) emphasise the distinction between 'motive' and 'non-motive' fuels as the generation of non-CO2 greenhouse gases like methane (CH4) and nitrous oxide (N2O) depends heavily on combustion conditions and technology as well as on fuel types.

Prior to the possibilities for substitution between different types of energy, most of the models reviewed here first model the choice between domestic and imported sources for each energy type considered (and this is true in terms of consumption as well as production decisions). As is the case in CGE models in general, the 'Armington Assumption' is often employed for tradable commodities, in circumstances where foreign and domestically produced goods are imperfect substitutes for one another, allowing domestic (endogenous) and world (exogenous) prices to differ

A1.4 Modelling pollution generation

Modelling pollution generation is a more recent innovation in CGE modelling than modelling energy use and not all the models that reviewed here incorporate this element. Even though pollution generation is generally associated with energy use in an economy, a number of the energy-economy CGE models reviewed here appear to have been built solely for the purpose of analysing the economic effects of energy-use issues, such as changes in energy prices and taxes. Bergman (1988) explains that interest in energy-economy modelling arose in the 1970s and 1980s in response to public concern about the economic impact of changing energy supply conditions. This is an important research area in itself, and the focus of these models does not extend to how these effects feed through to the environment, or more generally the environmental effects of economic activity.

However, as international concerns grow over the problem of global climate change, and people become more concerned over local environmental quality, CGE modelling efforts have increasingly begun to focus on the problem of pollution generated during economic activity. Most of the models reviewed here focus on the generation of greenhouse gases, and the economic and welfare effects of environmental policies employed to combat this problem, rather than on the question of local environmental quality. Another general point is that, as in modelling energy use, the specific context for which the model is being built is important. That is to say, the nature of pollution problems differs across economies with different production and consumption patterns, as do the objectives of policymakers.

As well as focussing on the generation of pollutants, a few studies also give attention incorporation of 'end-of-pipe' pollution abatement technologies into the general equilibrium system. However, this topic is not well-advanced in environmental CGE modelling and treated fairly simply where it is considered. For example, Bergman (1990, 1991) and Beauséjour et al (1994, 1995) model pollution abatement services simply as a distinct form of capital services. In both these models, since capital is assumed to be the sole factor input to abatement activities, the marginal cost of abatement is taken to be equal to the cost of capital services for this purpose. Since there is no private return on undertaking abatement activities, there is no private demand for abatement capital in the absence of pollution policy. Where pollution policy does exist, firms will choose to undertake abatement activities if the marginal cost of an additional unit of abatement is less than or equal to the cost incurred in emitting an additional unit of pollution. For example, in Bergman's (1990, 1991) model pollution policy takes the form of an upper limit on total emissions, and government operates a market for tradable emissions permits to meet the target levels for each type of emission. This means that firms will only demand abatement if the marginal cost of emissions abatement is less than or equal to the market price for emissions permits.

Pollutants modelled

The main factor determining the choice of individual pollutants to be incorporated is the purpose that the model has been set up for. For example, the focus of global models like GREEN tends to be on global sustainability questions like the emissions of greenhouse gases, in particular CO2 (which is the biggest contributor to the global warming effect). Therefore Burniaux et al (1992) (GREEN) only models this single pollutant (the focus of this application of GREEN is to quantify the costs of curbing CO ₂ emissions). Modelling of multiple pollutants seems to be more common in the case of individual region/country or sub-global inter-regional/inter-country models, most likely reflecting the different type of environmental concerns that exist at local (national/regional) levels. For example, Beghin et al (1995), a single country model of Mexico, models thirteen water, air and soil pollutants, and Lee & Roland-Holst (1997), a two-country CGE model of Indonesia and Japan, models eleven air, water and toxic pollutants. O'Ryan et al's (2005) analysis of Chile focuses on local environmental and social policy issues, including consideration of local air quality and models emissions of PM10, SO2 and NO2.

Whether pollution is related to inputs or outputs

It should be noted that a number of the papers reviewed here do not make explicit just how pollution is modelled. However, there are two broad approaches - linking emissions to inputs or outputs. The simplest way to model pollution as a result of economic activity is through linear output-pollution coefficients, representing the amount of pollution per unit of output for each of the different sectors modelled. This approach was one of the earliest steps in environmental input-output modelling (Leontief, 1970), the most straightforward (but very restrictive) variant of CGE analysis. However it can still be observed in environmental CGE modelling thirty years on.

For example in Lee & Roland-Holst's (1997) model, each sector has pollution coefficients that are linear in output for seven air, two water, and two toxic pollutants. This means that pollution is proportionately related to sectoral outputs only. However the authors point out that a major limitation of this approach is that there is no scope for technical substitution within sectors, meaning that emissions are proportional to sectoral output regardless of relative prices and factors such as differential pollution taxes. Roland-Holst addresses this limitation in other work: in a co-authored paper, Beghin et al (1995), the point is made that if pollution coefficients are output-based and/or only pure Leontief technology is modelled, then the only way to reduce emissions within any sector is to reduce that sector's output. Beghin et al (1995) go on to identify three underlying components of changes in emissions levels over time:

(1) Composition - change in pollution induced by a change in the commodity composition of aggregate production (more or less dirty/clean goods)

(2) Technology - evolving cleaner technologies (which usually result in a change in the input mix)

(3) Scale - increase/decrease in pollution attributable to an increase in aggregate economic activity

Where modelling of pollution involves simply relating emissions of pollution to sectoral outputs, only the composition and scale effects will be captured. The easiest way of modelling the technology effect will involve linking pollution emissions to production techniques through input-based pollution coefficients. However, it may be more useful to split Beghin et al's (1995) 'Technology' effect into two parts:

(a) Technology - evolving cleaner technologies, independent of the input mix ( e.g. installing catalytic converters in cars - this would mean a change in the emissions factor applied to the combustion of petrol in cars).

(b) Input substitution - changing the input mix towards cleaner types of energy/fuel ( e.g. changing from regular to low sulphur petrol) or towards non-energy inputs ( e.g. reducing the amount of energy used per unit of existing capital).

Of course, there may be instances where both (a) and (b) would occur together - for example, in switching from oil to gas powered heating systems. However, it is useful to make the distinction because the manner in which (a) and (b) are captured in a CGE modelling framework differs. Input substitution, i.e. factor (b), will be captured endogenously in a production structure with fixed input-pollution coefficients and appropriate possibilities for input substitutions. Such input substitutions would typically occur in response to a change in relative prices. However changes in technology ( i.e. case (a) above) are likely to involve adjustment of relevant input-pollution coefficients and/or changes to the production structure to reflect differing technical relationships in sectors and/or particular input mixes where adjustments have occurred.

Since this time, the input-pollution method has become common in environmental CGE models, to the extent that it is rarely explicitly discussed in applications. An explicit consideration of linking emissions of pollutants to input use can be found in Beghin et al (1995). The approach adopted in this model involves drawing on work reported in Dessus et al (1994) on econometrically estimating the relationship between the production of each type of pollution and the level of intermediate consumption of each type of input. However all this work seems to involve is using a basic estimation model in which the emissions of each pollutant are simply regressed on each of the intermediate material inputs used (including fuels and industrial chemicals), with no apparent investigation of causality.

A more common approach to input-pollution modelling involves using information on emissions factors associated with different types of fuel use that have been calculated elsewhere. That is to say, pollution coefficients tend to be based on actual technical relationships, rather than econometrically estimated ones. The application of emissions factors would appear to be the most straightforward in the case of CO2 emissions, as these are primarily dependent on the fuel properties rather than combustion conditions and/or technology. Most of the models reviewed here that adopt an input-pollution approach do tend to focus solely or primarily on CO2 emissions. For example, Burniaux et al (1992) (GREEN), Barns et al (1992), Stephan et al (1992) and Böhringer & Rutherford (1997) all explain that they use CO ₂ emissions coefficients based on carbon content for the different fuel types modelled to model input-pollution relationships.

However, modelling input-pollution relationships becomes more complex when it comes to non-CO2 emissions. This is because non-CO2 emissions tend to be dependent not only on fuel type, but also combustion conditions and technology, with the implication that appropriate emissions factors are likely to be more difficult to identify and too numerous for models with a high level of sectoral detail. This may also be why some models follow Beaséjour et al (1994, 1995) in introducing the distinction between 'motive' and 'non-motive' fuels in their production structure (also note that the IPCC regard the distinction between stationary and non-stationary sources to be the key distinction in measuring emissions generated during any given activity).

Nonetheless, not all the models reviewed here adopt the input-pollution approach. However, this is likely to be related to data constraints and/or early stages of model development, rather than any specification debate. For example, in Ferguson et al's (2004) Scottish model, AMOSENVI, pollution is related to sectoral outputs. However, by the time of Hanley et al's (2006) study, the model specification has been developed so that CO2 emissions are related to energy use.

However, Beauséjour et al (1994, 1995) argue that there is a role for modelling both input-pollution relationships, and output-pollution relationships. Beauséjour et al's (1994, 1995) model uses output-pollution coefficients to deal with production processes that are inherently polluting, for reasons other than the combustion of fossil fuels, and where the only way to reduce pollution may be to reduce output. Therefore in this model emissions of (air) pollutants arise from (1) the combustion of fossil fuels in intermediate production and final demand, and (2) from some industrial processes that are inherently polluting (such as pulp and paper production) without actually burning fossil fuels.

1. In the case of burning fossil fuels, the model assumes that emissions are a linear function of the volume of fuel combustion.

2. In the case of polluting industrial processes, where emissions are not caused by burning fossil fuels, emissions are assumed to be a linear function of the level of output from polluting industrial processes. (The sectoral disaggregation used by Beauséjour et al (1994, 1995) separately identifies the most polluting industrial processes - e.g. 'pulp & paper', which generates CO2 pollution, and non-ferrous smelting, which generates SOX).

Bergman (1990, 1991) also makes this distinction, modelling emissions from combustion proportional to fuel use and emissions from polluting industrial processes proportional to output. Thus, industries can reduce emissions by:

(i) Altering use of inputs - substitution towards 'cleaner fuels' and/or reducing the overall energy-intensity of production

(ii) Reducing output levels - this is the only option where processes are inherently polluting ( i.e. where emissions are not the direct result of input choice)

(iii) Using emissions abatement technologies if they exist and are economically feasible. This will reduce emissions from any given input-output combination.

However, since sectoral output-pollution coefficients are often derived from fuel use data, it could be argued that in the case (ii) of inherently polluting processes it would still be reasonable to use input-pollution coefficients but set the substitutability between inputs to zero. In other words, if by 'inherently polluting' production processes we mean that there is only one way to produce output and the input mix will not affect the generation of pollution, fixing input decisions by Leontief technology means that the input-pollution coefficients effectively act as output-pollution coefficients. The two would be equivalent in terms of impact. In order to study the potential for reducing pollution in the case of such processes, one possibility would then be to switch between alternative production processes or techniques with differential pollution characteristics for key sectors.

A1.5 Modelling consumption behaviour

We have not carried out a great deal of work on modelling consumption beheaviour to date. Therefore, here we use Conrad's (1999) review as a guide to what type of issues should be considered.

Conrad (1999) explains that the usual approach in CGE models is to assume that consumers perform a multi-stage budgeting procedure:

1. At the first level inter-temporal consumer behaviour allocates a lifetime wealth endowment across consumption in different time periods.

2. Then at the second level there is an intra-temporal choice between leisure (supply of labour) and consumption.

3. At the final stage consumption is then allocated among a number of consumption goods/categories.

Within this general framework, Conrad's (1999) survey finds several different ways of specifying the consumption decision, the most frequent being the use of linear expenditure system ( LES), nested CES or translog demand functions. He also finds that most studies focus on efficiency issues, with all consumers aggregated into a single representative consumer. However some studies use household disaggregation, modelling several different types of consumer, in order to assess the distributional impacts of different environmental policy options.

The models reviewed in this chapter can generally be described within Conrad's (1999) framework, with some exceptions. Firstly, most of the models reviewed here are not inter-temporal optimisation models, so the first level described above tends to be a decision between consumption and saving, with savings generally being some fixed proportion of income in any given time period. Savings is generally listed as one of the available consumption goods or categories and its price relative to other (present) consumption choices is determined as the expected future return to capital. In static single- or multi-period models this is generally taken to be equal to the current return to capital.

However, the largest proportion of the models reviewed here simply model consumer preferences/utility using a nested CES/ CD/Leontief specification as in the case of production, with substitution possibilities between energy and non-energy consumption goods. As in the case of production, particular specifications depend on the economy and policy issue being examined (and most likely on data availability). Moreover, in general less attention seems to have been given to modelling consumption relative to production in environmental CGE studies (to date anyway). However, there are some interesting examples from which lessons can be drawn.

For example, Stephan et al (1992), focus on the main sources of CO2 emissions in private consumption, which they identify as traffic and space heating. Through CES sub-nests of their CD utility function, they explain that consumers can substitute between (a) public and private transportation (identified as distinct consumption goods), and (b) in the case of space heating, both between different fuels, and between conventional and electric heating systems.

Conrad & Schröder (1991, 1993) also seem to be alone in highlighting another important issue: modelling the effect of some durable and non-durable goods being complements in consumption. This issue would be important, for example, if taxes were imposed on CO2 emissions resulting from the combustion of fossil fuels by private road-users will affect the use and demand for cars. Similarly, emissions taxes on the generation of electricity would affect the price of electricity to consumers, and hence their use and purchase of electric appliances. Generally, Conrad & Schröder (1991) argue that, since only part of the consumption of a non-durable good may be regarded as essential for the operation of a non-durable good, it is necessary to improve on the type of demand structure commonly used in environmental CGE models.

The argument regarding the importance of recognising the link between the demands for durable and non-durable goods is an entirely valid one. However, it is not clear to why this could not be handled in a more standard consumption structure by extending the disaggregation of non-durable goods such as gasoline into different 'types' for essential and recreational (non-essential) purposes. Moreover, all the arguments concerning testing of model specification in production apply in the case of consumption also.

Distributional analysis

The final issue raised by Conrad (1999) in his review is whether studies focus on efficiency issues, with all consumers aggregated into a single representative consumer, or whether there is disaggregation into several different types of household in order to assess the distributional impacts of, for example, different environmental policy options. In addition to differential distribution effects, it is also likely to be the case that patterns of consumption will vary significantly across income groups, with the implication that, among other things, the environmental impact of the consumption activities of different households will vary. Household/consumer disaggregation allows analysis and identification of which types of household contribute most to environmental problems, as well as distributional analysis to identify which households suffer the greatest economic impact from environmental policy actions.

However, as found by Conrad (1999), the majority of the models reviewed here assume a single representative consumer/household. Nonetheless a growing number of environmental CGE models do incorporate household/consumer disaggregation in order to address the type of issues raised above, generally differentiating households by income, expenditure and/or demographic categories. All the models reviewed here that incorporate household disaggregation - Boyd & Uri (1991), Stephan et al (1992), Weise et al (1995), Naqvi (1998), Kamat et al (1999) and O'Ryan et al (2005) - do so by allocating households into income bands.

Classification of households by income categories would appear to be the most common method of household disaggregation. However, where data permit, households should ideally be disaggregated to identify different demographic rather than income groups. This is because the latter will be variable if income is determined endogenously. For example, Naqvi (1998) attempts to introduce non-income classifications, allocating households to groups relating to employment and regional location within the defined income bands. Weise et al (1995) go one step further, mapping individuals to households to account for multiple job-holders and multiple earners in any one household, arguing that this is important for proper in-depth distributional analysis.

Where households are disaggregated by income, Weise et al (1995) argue that this should be done using income bands ( e.g. £10,000 - £20,000, £20,000 - £30,000 and so on) rather than breaking the population into equal-sized groups ( e.g. into quintiles or deciles as in O'Ryan et al, 2005). They explain that this is because the latter method tends to result in over-aggregation of important groups (such as high earners). This of course will depend on how many households are likely to fall into different classifications. All of the models I review here that incorporate household disaggregation according to income do appear to do so by income bands. In fact, Weise et al's (1995) point is illustrated in Stephan et al's (1992) model of Switzerland, where 66% of the population are captured in the lowest income band, with the highest three income bands containing only 10% of Swiss households in total. Stephan et al (1992) are concerned with the effects of varying responses of different households to changes in relative prices brought about by a carbon tax, due to the varying inclination of different income groups to substitute away from private transport towards public transport. These are the type of effects that Weise et al (1995) argue may be hidden if distinct income groups are not identified in the disaggregation process.

The important issue, whether income or non-income criteria are used, would seem to be determination of the homogeneity of households that are classified together in one household group. The general approach observed here is to specify distinct preference structures/demand systems for each income band (meaning that preferences, and the timing of consumption choices, are taken to be function of income). Thus, in grouping households in a given demographic or income band, it will be important to attempt to ensure that consumption preferences are in fact fairly homogenous among the households contained therein. (in the models reviewed here, differences in preferences and demand across different income bands are generally modelled via differences in elasticity of substitution parameter values, rather than by having distinct structures for consumption decision process for each type of consumer.)

One of the key issues that Boyd & Uri (1991), Stephan et al (1992) and Naqvi (1998) focus on is whether the type of results found in partial equilibrium analysis of energy/fuel taxation - i.e. that such taxes are likely to be regressive - carry over to a general equilibrium analysis. This is one of the most important questions associated with distributional analysis and sustainability policy concerns - whether the burden of environmental taxes is unfairly borne by low income groups, since essential expenditure on energy/fuel is likely to account for a larger proportion of their income. However, in contrast to what is commonly reported from partial equilibrium analyses, the general equilibrium results of Boyd & Uri (1991), Stephan et al (1992) and Naqvi (1998) all suggest that the distributional effects of various energy taxation changes are minimal. Boyd & Uri (1991) explain that this does not imply that the direct effects of, in their case, an increase in fuel taxes are not regressive. Rather they explain that, mainly because of changes in relative prices, in a general equilibrium setting indirect effects are likely to mean that such a tax increase may not, on balance, be a regressive one in terms of the redistribution of the tax burden.

Weise et al (1995), whose paper is the only one of those reviewed here that focuses specifically on distributional issues (in the context of motor fuel taxes and household welfare), identify what appear to be the two most important influencing factors in a distributional analysis. The first is what happens to energy/fuel tax revenues - i.e. how these are redistributed/spent by government. Weise et al (1995) consider several different expenditure patterns for fuel tax revenues and find that the distributional effects are conditional on what government does with increased revenues. How the revenues are spent will impact on what Weise et al (1995) identify as the second key factor in determining distributional effects: the different sources of income available to different income groups. They find that the distributional impacts of energy/fuel taxes depend on whether the tax change and expenditure of revenue lead to increased or decreased rates of return to capital and skilled labour services, which are important sources of income for upper-middle and high income households. They report that, where the incomes of these households suffer from a decline in the rates of return to capital and/or skilled labour services, the distributional impacts of increased motor taxes tend to be smoothed out (or in some situations, are actually found to be progressive). Therefore, Weise et al (1995) conclude that to model distributional effects properly attention should be given to how environmental tax revenues are redistributed back to households. Specifically, they argue that the key issue is whether revenues are used to purchase goods and services, capital and labour, and to the different sources of income (and factor endowments) of different households.

A1.6 Summary and conclusions from literature review

The purpose of this review is to identify and consider issues that are likely to be important in conducting an environmental CGE analysis of the impacts on the Scottish economy of reducing greenhouse gas emissions in Scotland. Given that the current Scottish model ( AMOSENVI) is an environmental CGE model for a single small open economy, the first issue discussed is the nature of the effects that we should hope to be able to capture. In Section 2 we explained how in the case of a model based solely on data for a single small regional economy, it is necessary to adopt the 'small open economy assumption'. This means only attempting to model the target (local) economy with the rest of the world assumed to be exogenous. The key implication of this, in terms of economic-environmental modelling, is that it is not possible to take account of any feedback effects from the global environment to the local economy resulting from disturbances in the latter. In other words, while it is possible to assess the economic effects of global environmental policies on the local economy, it is not possible to assess any environmental feedback effects resulting from any consequent improvement in the global environment. However, by employing the small open economy assumption, what we are saying is that we assume that these feedback effects would be so small (because the target economy is so small relative to the rest of the world) that they can be considered negligible.

In Section A1.2 we also argued that another implication of the target economy being a small open economy is that issues like modelling resource depletion are not likely to be relevant. We argued that resource depletion will only be an issue in modelling a small open economy if the resource in question cannot be easily imported from elsewhere in the world. The main way in which natural resource issues will enter into such a model will be in addressing problems of resource constraints and/or the effects of resources prices that are set elsewhere in the world ( i.e. exogenously).

In Section A1.3 we went on to consider issues that have been identified in the literature as being important in modelling energy use, which is the main source of environmental problems that are attributable to economic activity. A number of important issues were discussed, including the question of whether energy is a complement or a substitute for other factors of production, and appropriate functional forms. However, our main conclusion is that, while many important arguments have been put forward for how production involving energy use should be modelled, none of the models reviewed here properly addressed the issue of model specification by actually testing a full range of alternatives. Sensitivity analyses are largely limited to parameter values, not to the production structure itself. This conclusion will also apply to the discussion of consumption structures in Section 5. Nonetheless, our arguments on this issue are qualified by the admission that such an ideal approach to model specification is likely to be highly resource intensive and often not feasible in practice. However, the problem of the judgmental nature of model specification in CGE applications (or any type of complex system-wide model), should be considered explicitly when assessing alternative specifications.

We then went on to discuss how the problem of modelling production with energy inputs has been approached in existing models of economic-environmental interactions. The main conclusion here is that the detail and nature of energy modelling is very much case dependent - i.e. the detail of energy modelling depends very much on the nature and structure of the specific economy being modelled, and most likely on the availability of appropriate data for this purpose.

In Section A1.4 we considered the question of modelling pollution, where many of the same type of issues arise. In terms of what pollutants are modelled, we again conclude that this tends to depend on the specific economy being modelled and the purpose of building the model - i.e. whether the focus is on global or local environmental concerns, on specific types of policy etc. We note that the main model specification issue is whether pollution is related to inputs to or outputs from production. If there are opportunities for reducing emissions through input substitution ( i.e. if production technology is sufficiently variable), it is necessary to model input-pollution relationships in order to quantify environmental impacts. For example, if we want to assess the implications of changing/imposing energy taxes such as carbon taxes, then it is necessary to directly model the relationship between input use and pollution. This is because the aim of this type of tax is to induce substitution away from polluting inputs (by affecting relative prices). Therefore, input-pollution coefficients will be necessary if this type of policy is being model if we want to capture any consequent change in the output of pollution. However, we also stress the importance of recognising that not all emissions in the economy in question will be directly related to input use, since many production processes involve non-combustion related emissions generation.

Section A1.5 focussed on how the consumption side of the economy is specified in the models reviewed. In terms of issues relating to the appropriate choice of functional forms and how energy use and pollution generation are modelled, we find that the same type of issues arise as in the case of production. We argued that the main consumption-specific issue, particularly if the model is intended to analyse sustainability questions, is the scope for distributional analysis. Our conclusion is that, while a number of the models reviewed do attempt to address distributional issues by incorporating household disaggregation according to income, clearly this is an area that has not yet received a great deal of in-depth attention in the environmental CGE literature. Nonetheless some important modelling issues are identified, such as the importance of the sources of income available to different households.

References (Literature Review)

Abler, D.G, J.K. Lazo, G. Oladosu, and J. Shortle (2000), 'Valuation Of Climate Change Impacts In Computable General Equilibrium Models', Paper to the European Association of Environmental and Resource Economists Annual Conference, 2000.

Adams, P., M. Horrdige and G. Wittwer (2003), ' MMRF-GREEN: A Dynamic Multi-Regional Applied General Equilibrium Model of the Australian Economy, Based on the MMR and MONASH Models', Centre of Policy Studies, General Working Paper No. G-140, October, Monash University.

Allan, G., N.D. Hanley, P.G. McGregor, J.K. Swales and K.R. Turner (2007), 'The impact of increased efficiency in the industrial use of energy: A computable general equilibrium analysis for the United Kingdom', Energy Economics, Vol. 29, pp. 779-798.

Allan, G., N.D. Hanley, P.G. McGregor, J.K. Swales and K.R. Turner (2006), 'The Macroeconomic Rebound Effect and the UK Economy', Final Report to DEFRA, May 2006.

Barker, T. and P. Ekins (2004), 'The Costs of Kyoto for the US Economy', The Energy Journal, Vol. 25, N ^o3, pp. 53-71.

Barker, T., P. Ekins and T. Foxon (2007), 'The macroeconomic rebound effect and the UK economy', Energy Policy, Vol. 35, No. 10, pp. 4935-4946.

Beauséjour, L., G. Lenjosek, and M. Smart (1995) 'A GCE Approach to Modelling Carbon Dioxide Emissions Control in Canada and the United States', The World Economy, Vol.18, pp.457-489.

Beauséjour, L., G. Lenjosek and M. Smart (1994) 'An Environmental CGE model of Canada and the United States', Department of Finance Canada, Working Paper No. 92-04.

Beghin, J., D. Roland-Holst and D. v. d. Mensbrugghe (1995) 'Trade Liberalisation and the Environment in the Pacific Basin: Coordinated Approaches to Mexican Trade and Environmental Policy', American Journal of Agricultural Economics, Vol.77, pp.778-785.

Bergman, L. (1988) 'Energy Policy Modelling: a Survey of General Equilibrium Approaches', Journal of Policy Modelling, Vol.10, No.3, pp.377-399.

Bergman, L. (1991) 'General Equilibrium Effects of Environmental Policy: a CGE Modelling Approach', Environmental and Resource Economics, Vol.1, pp.43-61.

Bergman, L. (1990) 'Energy and Environmental Constraints on Growth: a CGE Modelling Approach', Journal of Policy Modelling, Vol.12, No.4, pp.671-691.

Bergman, L. (1988) 'Energy Policy Modelling: a Survey of General Equilibrium Approaches', Journal of Policy Modelling, Vol.10, No.3, pp.377-399.

Bergman, L. (2005) ' CGE Modelling of Environmental Policy and Resource Management', Chapter 24 in Mäler and Vincent (eds) Handbook of Environmental Economics, Volume 3: Economywide and International Environmental Issues, Elsevier, North Holland.

Bhattacharyya, S.C. (1996), 'Applied general equilibrium models for energy studies: a survey', Energy Economics, Volume 18, Issue 3, July 1996, Pages 145-164.

Bohringer, C., H. Welsch, A. Loschel (2001), 'Environmental Taxation and Structural Change in an Open Economy - A CGE Analysis with Imperfect Competition and Free Entry', ZEW.

Böhringer, C. and A. Löschel (2006), 'Computable general equilibrium models for sustainability impact assessment: Status quo and prospects', Ecological Economics, Volume 60, Issue 1, 1 November 2006, Pages 49-64.

Böhringer, C. and T. Rutherford (1997) 'Carbon Taxes with Exemptions in an Open Economy: a General Equilibrium Analysis of the German Tax Initiative', Journal of Environmental Economics and Management, Vol.32, pp.189-203.

Boyd, R. and N.D. Uri (1991) 'An Assessment of the Impacts of Energy Taxes', Resources and Energy, Vol.13, pp.349-379.

Burniaux, J-M., G. Nicoletti and J. Oliveira-Martins (1992a), 'GREEN: a global model for quantifying the costs of policies to curb CO2 emissions', OECD Economic Studies, No. 19.

Burniaux, J-M., G. Nicoletti and J. Oliveira-Martins (1992b), 'The effect of existing distortions in energy markets on the costs of policies to reduce CO2 emissions: evidence from GREEN', OECD Economic Studies, No. 19.

Burniaux, J-M., J.P. Martin, G. Nicoletti and J.M. Martins (1992c) 'Green. A Multi-Sector, Multi-Region General Equilibrium Model for Quantifying the Costs of Curbing CO2 Emissions: a Technical Manual', OECD Economics Department Working Papers, No.116.

Burniaux, J-M., G. Nicoletti, J. Oliveira-Martins (1992d), 'The Cost of Reducing CO ₂ Emissions: A Comparison of Carbon Tax Curves with Green', OECD Economic Working Paper, No. 118.

Conrad, K. (1999) 'Computable General Equilibrium Models for Environmental Economics and Policy Analysis', in J.C.J.M. van den Bergh ed. Handbook of Environmental and Resource Economics, Edward Elgar Publishing Ltd, 1999.

Conrad, K. (2001), 'Computable General equilibrium Models in Environmental and Resource Economics', Institut für Volkswirtschaftslehre und Statistik, Paper No. 601-01.

Conrad, K. and M. Schröder (1991b) 'Demand for Durable and Nondurable Goods, Environmental Policy and Consumer Welfare', Journal of Applied Econometrics, Vol. 6, pp.271-286.

Conrad, K. and M. Schröder (1993) 'Choosing Environmental Policy Instruments Using General Equilibrium Models', Journal of Policy Modelling, Vol.15, pp.521-543.

Dean, A. and P. Hoeller (1992) 'Costs of Reducing CO2 Emissions: Evidence from Six Global Models', OECD Economic Studies, No.19, Winter 1992.

Dellink, R., M. Hofkes, E. Van Ierland and H. Verbruggen (2004), 'Dynamic modelling of pollution abatement in a CGE framework', Economic Modelling, Volume 21, Issue 6, December, Pp. 965-989.

Despotakis, K.A. and A.C. Fisher (1988) 'Energy in a Regional Economy: a Computable General Equilibrium Model for California', Journal of Environmental Economics and Management, Vol.15, pp.313-330.

Dessus, S., D. Roland-Holst and D.v.d. Mensbrugghe (1994) 'Input-Based Pollution Estimates for Environmental Assessment in Developing Countries', OECD Development Centre Technical Papers, No.101. OECD, Paris, 1994.

Dixon, P. and M.T. Rimmer (1998), 'Forecasting and Policy Analysis with a Dynamic CGE Model of Australia', Centre of Policy Studies, Preliminary Working Paper No. OP-90, June, Monash University.

Dufournaud, C.M., J.T. Quinn and J.J. Harrington (1994), 'An Applied General Equilibrium ( AGE) analysis of a policy designed to reduce the household consumption of wood in Sudan', Resource and Energy Economics, Vol. 16, pp. 67-90.

Ferguson, L., P.G. McGregor, J.K. Swales, K.R. Turner and Y. Ping Yin (2004), Incorporating Sustainability Indicators into a Computable General Equilibrium Model of the Scottish Economy', Strathclyde Discussion Papers in Economics, No. 04-13.

Field, B.C. and E.R. Berndt (1981) 'An Introductory Review of Research on the Economics of Natural Resource Substitution', in B.C. Field and E.R. Berndt eds. Modelling and Measuring Natural Resource Substitution,MIT Press, 1981.

Glomsrød, S and Taoyuan, W. (2005), 'Coal cleaning: a viable strategy for reduced carbon emissions and improved environment in China?', Energy Policy, Vol. 33, pp. 525-542.

Greenaway, D., S.J. Leyborne, G.V. Reed and J. Whalley (1993), 'Applied General Equilibrium Modelling: Applications, Limitations and Future Development', HMSO.

Grepperud, S. and I. Rasmussen (2004), 'A general equilibrium assessment of rebound effects', Energy Economics, Vol. 26, pp. 261-282.

Goulder, L.H. and W.A. Pizer (2006), 'The Economics of Climate Change', Resources for the Future, Discussion Paper 06-06.

Gottinger, H.W. (1998), 'Greenhouse Gas Economics and Computable General Equilibrium', Journal of Policy Modeling, Volume 20, Issue 5, October 1998, Pages 537-580.

Haan, M (2001) 'A structural decomposition analysis of pollution in the Netherlands', Economic Systems Research, Vol.13, No.2, pp.181-196.

Hanley, N.D., P.G. McGregor, J.K. Swales, K.R. Turner (2006), 'The impact of a stimulus to energy efficiency on the economy and the environment: A regional computable general equilibrium analysis', Renewable Energy, Vol. 31, pp. 161-171.

Herring, H. (2006), 'The Macro-Economic Rebound Effect and the UK Economy', Final Report to DEFRA, May 2006.

Hertel, T., H-L. Lee, S. Rose and B. Sohngen (2006), 'The Role of Global Land Use in Determining Greenhouse Gases Mitigation Costs', GTAP Working Paper, No. 36.

Hertel, T.W. (1988) 'General Equilibrium Incidence of Natural Resource Subsidies: the Three Factor Case', Journal of Environmental Economics and Management, Vol.15, pp.206-223.

Hertel, T.W. and T.D. Mount (1985) 'The Pricing of Natural Resources in a Regional Economy', Land Economics, Vol.61, No.3, pp.229-243.

Kamat, R., A. Rose and D. Abler (1999) 'The Impact of a Carbon Tax on the Susquehanna River Basin Economy', Energy Economics, Vol.21, pp.363-384.

Kemfert, C. and T. Truong (2005), 'Impact assessment of emissions stabilization scenarios with and without induced technological change', Energy Policy,
Volume 35, Issue 11, November, Pp. 5337-5345.

Klepper, G. and S. Peterson (unpublished), 'The EU Emissions Trading Scheme: Allowance Prices, Trade Flows, Competitiveness Effects', Kiel Institute for World Economics.

Lee, H., J. Oliveira-Martins, D. Van der Mensbrugghe (1994), 'The OECD GREEN Model: An Updated Overview', OECD Development Centre, Technical Paper No. 97.

Lee, H. and D.W. Roland-Holst (1997) 'Trade and the environment', in J.F. Francois and K.A, Reihert eds. Applied Methods for Trade Analysis: a Handbook, Cambridge University Press.

Learmonth, D., P.G. McGregor, J.K. Swales, K.R. Turner and Y.P. Yin (2007), 'The importance of the regional/local dimension of sustainable development: An illustrative Computable General Equilibrium analysis of the Jersey economy', Economic Modelling, Vol. 24, pp. 15-41.

Leontief, W. (1970) 'Environmental Repercussions and the Economic Structure: An Input-Output Approach', Review of Economic Statistics, Vol.52, pp.262-277.

Li P-C., C.H. Huang and S.H. Hsu (2000), 'Energy Substitution in CGE Modelling Using the "Technology Bundle" Approach: The Case of Taiwan', Conference paper for the Third Annual Conference on Global Economic Analysis 2000.

Li, P. and A. Rose (1995) 'Global Warming Policy and the Pennsylvania Economy: A Computable General Equilibrium Analysis', Economic Systems Research, Vol.7, No.2, pp.151-171.

Löschel, A (2002), 'Technological change in economic models of environmental policy: a survey', Ecological Economics, Volume 43, Issues 2-3, December 2002, Pp. 105-126.

Hondo, H. (2005), "Life cycle GHG emission analysis of power generation sector: Japanese case", Energy, 2005, 30, p. 2042-2056.

Naqvi, F. (1998) 'A Computable General Equilibrium Model of Energy, Economy and Equity Interactions in Pakistan', Energy Economics, Vol.20, pp.347-373.

Nordhaus, W.D. and Z. Yang (1996) 'A Regional Dynamic General-Equilibrium Model of Alternative Climate-Change Strategies', The American Economic Review, Vol. 86, No. 4. (Sep., 1996), pp. 741-765.

O'Ryan, R., S. Miller and C.J. De Miguel (2003), 'A CGE framework to evaluate policy options for reducing air pollution emissions in Chile', Environment and Development Economics, Cambridge University, Vol. 8, pp. 285-309.

O'Ryan, R., C.J. De Miguel, S. Miller and M. Munasinghe (2005), 'Computable general equilibrium model analysis of economy-wide cross effects of social and environmental policies in Chile', Ecological Economics, Volume 54, Issue 4, 15 September, Pp.447-472.

Otto, V.M., A. Löschel and J.M. Reilly (2006), ' Directed Technical Change and Climate Policy�, FEEM Working Paper, No. 81.

Palatnik, R. and M. Shechter (2008), 'Can Climate Change Mitigation Policy Benefit the Israel Economy? A Computable General Equilibrium Analysis', FEEM Working Paper, No. 166.

Peterson, S. (2003), 'The EU Emissions Trading Scheme and its Competitiveness Effects upon European Business Results from the CGE Model DART', Kiel Institute for World Economics.

Proops, J. L. R., Gay, P. W., Speck, S., and Schroder, T. (1996), "The lifetime pollution implications of various types of electricity generation", Energy Policy, 24(3), p. 229-237

Robaina Alves, M. and A. Marvão Pereira (2006), 'Survey of Applied General Equilibrium Models for Energy and Environmental Studies', (publisher unknown).

Rutherford, T. (1992) 'The Welfare Effects of Fossil Carbon Restrictions: Results from a Recursively Dynamic Trade Model', OECD Economics Department Working Papers, No.112 .

Semboja, H.H.H. (1994), 'The effects of an increase in energy efficiency on the Kenya economy', Energy Policy, Vol. 22, Issue 3, Pp. 217-225.

Söderholm, P. (2007), 'Modelling the Economic Costs of Climate Policy', 2007:14, Department of Business Administration and Social Sciences Economics Unit, Luleå University of Technology.

Stephan, G., R. Van Nieuwkoop and T. Wiedmer (1992), 'Social incidence and economic costs of carbon limits', Environmental and Resource Economics, Volume 2, Number 6 / November, pp. 569-591.

Van der Mensbrugghe, D. (1998), 'A (Preliminary) Analysis of the Kyoto Protocol: Using the OECD GREEN Model', OECD Development Centre.

Vaze, P. (1997) 'Environmental input-output tables for the United Kingdom', Economic Trends, No. 527, October 1997.

Vaze , P., (1999) 'A NAMEA for the UK', Structural Change and Economic Dynamics, Vol.10, pp15-37.

Vikstrom, P. (2004), 'Energy efficiency and energy demand: a historical CGE investigation on the rebound effect in the Swedish economy 1957', Paper presented at Input-Output and General Equilibrium Data, Modelling and Policy Analysis, Brussels, 2nd-4th September 2004.

Wajsman, N. (1995), 'The Use of Computable General Equilibrium Models in Evaluating Environmental Policy', Journal of Environmental Management,
Volume 44, Issue 2, June 1995, Pages 127-143.

Washida, T. (2004), 'Economy-wide Model of Rebound Effect for Environmental Efficiency', International workshop on Sustainable Consumption, University of Leeds, 5-6th March 2004.

Wiese, A.M., A. Rose and G. Schluter (1995) 'Motor-Fuel Taxes and Household Welfare: an Applied General Equilibrium Analysis', Land Economics, Vol.71, No.2, pp.229-243.

Whalley, J. and R. Wigle (1992) 'Results for the OECD Comparative Modelling Project from the Whalley-Wigle Model', OECD Economics Department Working Papers, No.121. OECD, Paris, 1992.

Wissema, W. and R. Dellink (2007), ' AGE analysis of the impact of a carbon energy tax on the Irish economy', Ecological Economics, Volume 61, Issue 4, 15 March 2007, Pages 671-683.

Xie, J. and S. Saltzman (2000), 'Environmental Policy Analysis An Environmental Computable General-Equilibrium Approach for Developing Countries', Journal of Policy Modeling, Volume 22, Issue 4, July 2000, Pages 453-489.

Yang, H-Y. and T-F. Wang (2002) 'Social incidence and economic costs of carbon limits: a computable general equilibrium analysis for Taiwan', Applied Economics Letters, Volume 9, Number 3, pp. 185-189.

Zhang, Z. (1998) 'Macro-economic and Sectoral Effects of Carbon Taxes: a General Equilibrium Analysis for China', Economic Systems Research, Vol.10, No.2, pp.135-159.

Zhang, Z. and H. Folmer (1998), 'Economic Modelling Approaches to Cost Estimates for the Control of Carbon Dioxide Emissions', Energy Economics, Vol. 20, No. 1, pp. 101-120.

TableA1.1: Global and multi-country applications

Table A1.2: National, regional and interregional national/sub-national applications

Table A1.3: Surveys of CGE Modelling and Environmental Policy

« Previous | Contents | Next »