Decomposing the Cost of Kyoto

Transcription

1 Discussion Pape No Decomposing the Cost of Kyoto Global CG nalysis of Multilateal Policy Impacts Chistoph Böhinge and Thomas F. Ruthefod

2 Non-Technical Summay In this pape we apply a geneic pocedue developed by Haison, Hoidge and Peason (1999) in ode to decompose the economic implications of the Kyoto Potocol at the coss-county level. The total economic impact fo each egion is split into contibutions fom its own emission abatement policy and those fom othe egions. Ou analysis which is based on a lage-scale computable geneal equilibium model fo the wold economy indicates that spillove effects ae an impotant consequence of multilateal cabon abatement policies. mission mitigation by individual developed egions may not only significantly affect economic development and pefomance in non-abating developing counties but may also cause lage changes in the economic costs of emission abatement fo othe industialized egions. nalyzing the individual contibutions acoss policy measues of abating counties, action on behalf of the United States poduces by fa the lagest spilloves to othe counties. Majo competitos such as uope and Japan benefit significantly fom abatement in the US wheeas the US - in tun - is hadly affected by abatement policies of these egions. We calculate a coss-county matix fo monetay tansfe payments which would have to be assigned on a bilateal basis in ode to povide compensation fo the Kyoto spilloves. We show that the contibutions of individual policy changes to the oveall effect depend on the policy instument. When we simply change fom a pice instument (taxes) to a quantity instument (emissions) in ode to meet the Kyoto commitments the quantitative decomposition esults change. Howeve, the qualitative esults emain obust suggesting that the decomposition pocedue povides a useful stating point fo bilateal negotiations on policy-elevant tansfe payments acoss paties of the Famewok Convention.

3 Decomposing the Cost of Kyoto Global CG nalysis of Multilateal Policy Impacts Chistoph Böhinge a and Thomas F. Ruthefod b bstact We decompose the economic implications of the Kyoto Potocol at the coss-county level, splitting the total economic impact fo each egion into contibutions fom its own emission abatement policy and those fom othe egions. Ou analysis which is based on a lage-scale computable geneal equilibium model fo the wold economy indicates that spillove effects ae an impotant consequence of multilateal cabon abatement policies.. We calculate a cosscounty matix fo monetay tansfe payments which would have to be assigned on a bilateal basis in ode to povide compensation fo the intenational spilloves associated with the implementation of the Kyoto Potocol. Zusammenfassung Im Rahmen eine angewandten Gleichgewichtsanalyse untesuchen w die ökonomischen uswkungen des Kioto-Potokolls auf bilateale bene. Hiezu zelegen w den gesamten Wohlfahtseffekt fü jede Region in einen nteil, de auf die eigene missionsmindeungspolitik zuückzufühen ist und die jeweiligen nteile (intenationale spilloves), die duch Klimaschutzpolitiken in andeen Regionen veusacht weden. s wd deutlich, daß intenationale spilloves einen wesentlichen Beitag zu den allgemeinen Gleichgewichtseffekten multilateale Politikmaßnahmen liefen. W beechnen eine Matix von bilatealen Tansfezahlungen, die als usgangspunkt fü politikelevante Vehandlungen übe Kompensationen fü intenationale spilloves im Zuge de Umsetzung des Kioto-Potokolls dienen könnte. JL classification: C63, C68, D58, F11, Q4 a Cente fo uopean conomic Reseach (ZW), Mannheim b Depatment of conomics at the Univesity of Coloado, Boulde Contact the authos at: boehinge@zew.de o uthefod@coloado.edu

4 1 Intoduction Unde the Kyoto Potocol, industialized counties and economies in tansition (nnex B counties), have committed themselves to educe the geenhouse gas emissions below 1990 levels in the peiod 2008 to Developing counties have agued that they cay only mino histoical esponsibility fo the incease in global CO2 concentations and theefoe have efused any abatement commitment so fa. t fst glance, abstinence fom domestic action seems to insue the developing counties against potential costs fom geenhouse gas abatement because the economic development will not be constained by domestic emission limits. Howeve, in a wold whee economies ae inceasingly linked though intenational tade, policy changes in one county cause spillove effects to all tading patnes via changes in the intenational pices, i.e. the tems of tade. In this context, developing counties ae concened that abatement stategies in nnex B counties negatively affect the economic development and welfae. The developed wold as a lage tading block might exploit intenational maket powe and influence intenational pices at the expense of its tading patnes in the developing wold, heeby passing on some faction of the costs of abatement ( begga-thy-neighbo policy). The Kyoto Potocol explicitly eflects these concens postulating that developed counties...shall stive to implement policies and measues... in such a way as to minimize advese... economic impacts on othe Paties, especially developing counties Paties... (UN 1997, ticle 2, paagaph 3). On the othe hand, the developed nnex B counties fea advese changes in the tems of tade fom unilateal abatement, because the enegy use will be taxed, while thee will be no taxes in the developing wold, hence they can expect to lose competitiveness in enegy-intensive goods. In the policy debate, the issue of induced changes in tems of tade is linked to demands fo adjustment mechanisms which povide moe o less explicit compensation fo "unfa" spilloves. The developing wold is guaanteed compensation by nnex B fo induced economic costs unde ticles 4.8 and 4.9 of the United Nations Famewok Convention on Climate Change (UNFCC 1992). In tun, the developed wold agues fo compensation 1 if unilateal abatement policies povide economic gains to non-abating non-nnex B counties. This 1 Compensating measues also include poposals fo bode taxes on enegy intensive impots and bode subsidies on enegy-intensive expots in ode to mitigate negative shifts in compaative advantage. 1

5 agument may be suppoted by fa division theoy stating that "no-one should benefit fom the emission abatement budens of othes" (Helm 1999, p.11). Moulin, fo example, agues that "fa division conveys the idea of no subsidization: the pesence of othe agents who ae willing to pay highe monetay tansfes than me fo consuming the esouces should not tun to my advantage" (Moulin 1992, p. 1333). 2 Of couse, the debate on compensation fo advese spilloves does not only take place between the developed and the developing wold, but also among counties of the espective goups (see, e.g., the bagaining pocess on buden shaing within the U; Böhinge, Haison and Ruthefod 1998). The compehensive policy evaluation of abatement stategies with espect to the distibution of potential costs and benefits acoss counties eques quantitative assessment of how economic pefomance in a specific county will not be only affected by its own action, but also by those of othe egions. Thee is a need fo uncoveing the sign and the magnitude of multilateal spilloves fom policy intefeence though intenational channels. In this pape we apply a geneic pocedue developed by Haison, Hoidge and Peason (1999) in ode to decompose the economic implications of the Kyoto Potocol at the cosscounty level. The total economic impact fo each egion is split into contibutions fom its own policy changes and those fom othe egions. pat fom qualitative insights into the souces of economic gains and losses acoss counties, the decomposition allows the measuement of the individual contibutions of changes in exogenous policies (in ou case: multilateal abatement measues) to the oveall change in endogenous economic vaiables. In pinciple, wee they able to agee on model stuctue and paametes, policy makes could use such infomation as a efeence point fo negotiating bilateal tansfe payments accoding to some ageed equity pinciples. Fo the implementation and application of the decomposition by Haison, Hoidge and Peason, we use a static lage-scale computable geneal equilibium model of the wold economy. Ou key findings can be summaized as follows: mission constaints as mandated unde the Kyoto Potocol induce significant spilloves fom abating nnex B counties to non-abating Non-nnex B egions. Compliance with the Kyoto 2 This contasts to the position of some authos explicitly aguing that climate change potection stategies should be designed such that they favo developing counties (Simonis 1996). 2

6 tagets though national abatement measues imposes welfae losses fo all OCD egions wheeas developing egions with exception of lage enegy expotes may incease welfae due to gains in compaative advantage. nalyzing the individual contibutions acoss policy measues of abating counties, action on behalf of the United States poduces by fa the lagest spilloves to othe counties. Majo competitos such as uope and Japan benefit significantly fom abatement in the United States wheeas the United States - in tun - is hadly affected by abatement policies of these egions. mong developing counties, Bazil and India gain fom action of all OCD counties, wheeas the evese is tue fo enegy expoting developing egions. lthough the specific quantitative values fo the decomposed bilateal spill-oves depend on the policy instument (hee: taxes vesus pemits) the qualitative esults emain obust. Tanslation of decomposed welfae effects into a matix of bilateal tansfe payments could theefoe povide a stating point fo bilateal negotiations on compensation negative intenational spilloves fom domestic abatement policies. The pape is oganized as follows: section 2 povides a desciption of the decomposition technique; section 3 lays out the model famewok in use fo the impact analysis of Kyoto; section 4 discusses the numeical esults; and section 5 concludes. 2 The Geneal quilibium Decomposition Technique Geneal equilibium povides an established mico-consistent appoach fo evaluating the impacts of public policy on esouce allocation (efficiency) and the associated changes in income fo economic agents ("equity"). It has been, and still is, widely used in analytical wok fo assessing policy measues, such as tax efoms, whee maket inteactions potentially play an impotant ole. Howeve, fo the sake of tactability, analytical appoaches ae typically athe simple and not sufficiently complex fo applied policy analysis. Theefoe, numeical models ae commonly used to accommodate the systematic analysis of economic poblems whee analytical solutions ae eithe not available o do not povide adequate infomation. 3 3 Fo suveys on the use of numeical models in diffeent fields, see e.g. Begmann 1990, Kehoe and Kehoe 1994, Shoven and Whalley 1992, Peea and Shoven

7 The main vtue of complex computational geneal equilibium (CG) models, i.e. the compehensive and consistent quantification of dect and indect policy impacts, constitutes also the majo challenge fo the use. s vaious patial effects, which may wok in opposite dections, contibute to the oveall effect, it can get vey difficult to explain in depth the aggegate policy outcome. Numeical applications inheit some ambiguity in the intepetation of the esults as long as it is not possible to make tanspaent the sign and the magnitude of individual effects. Theefoe, pocedues which allow the decomposition of geneal equilibium effects in a meaningful way ae vey helpful fo the undestanding and intepetation of policy simulations. delibeate decomposition not only facilitates analysis of the vaious souces of the total effects but also assues a moe igoous check fo the coect numeical implementation of policy questions. 4 In the context of multilateal policy appaisal, Böhinge and Ruthefod (1999) pesent a decomposition that splits the oveall economic effect into a domestic maket effect keeping intenational pices constant, and an intenational maket effect as a esult of changes in intenational pices (tems of tade effect). 5 In othe wods, the decomposition allows sepaation of the pimay effect of domestic policy action fom the seconday buden o benefit tansmitted via changes in intenational pices. Yet, the pocedue is not suited fo quantifying how much of the total economic impact fo one specific egion is due to its own action and what is contibuted by the individual actions of othe egions. Haison, Hoidge and Peason (1999, hencefoth "HHP") popose a linea decomposition methodology fo calculating the contibutions of multiple exogenous policy instuments to the esulting changes in individual endogenous vaiables. The HHP method may be best explained along a simplified example whee an endogenous vaiable Z can be expessed as an explicit function of a vecto X of exogenous vaiables (policy instuments): Z = F X ) = F( x, x,..., x ). ( 1 2 n 4 Typically, CG models ae calibated to a benchmak data set in economic flows fo given values of elasticities. The eplication check of the benchmak equilibium seves as a test fo the consistent integation of data, but does not assue "pope" economics. 5 The key idea is that each egion of a multi-egion model can be epesented as a small open economy in ode to sepaate the domestic policy effect unde fixed tems of tade. The changes in intenational pices can then be imposed paametically on the small open economy to yield the full policy as peviously calculated in the multegion famewok. 4

8 We conside the effects of a change in X which induces a change in Z, Z. One way of decomposing the total change Z in the endogenous vaiable with espect to the individual contibutions fom exogenous vaiables would be a sequential appoximation of the impacts of one exogenous vaiable while keeping all othes constant (see e.g. Huff and Hetel 1997). ssuming that F is diffeentiable, the contibution of a change in the exogenous vaiable x i which moves fom the initial value x i0 to the new value x i1 can then be computed as the line integal: x i1 F Z = xi dxi. x xio Fo the numeical computation, the total change in the exogenous vaiable i xi is divided into sufficiently small steps in ode to appoximate the line integal though lineaization. When F is nonlinea, the total change fom shocks in exogenous vaiables can not be decomposed in additive line-integals fo each exogenous vaiable stating fom the same efeence (initial) value Z 0. The impact of a change in an exogenous vaiable must be calculated, taking into account the contibutions of pevious changes in othe exogenous vaiables. This implies that the decomposition is potentially sensitive to the sequential odeing of changes in the exogenous policy vaiables. s thee ae n! ways of sequential odeing of n exogenous vaiables, one quickly ends up with a lage numbe of possible (diffeent) decompositions fo elatively small-scale policy expeiments. 6 Fo many policy packages no decomposition might be obviously moe plausible than the est. HHP theefoe suggest an ode-independent "natual " way of calculating contibutions. On the "natual" path, the exogenous vaiables move togethe towads the final value along a staight line between the stating values X 0 and the final values X 1. The staight line between these points is obtained by changing the elements of X as a diffeentiable function H of some paamete t holding the ate of change in the exogenous vaiables constant along the path (whee X 0 = H to ), X 1 = H ( t ) ). ( 1 Figue 1 illustates the diffeence between the sequential method of decomposition and the HHP appoach. 6 In ou policy simulation of Kyoto we assess the diffeential impacts of 6 emission abating counties, which yields 720 possible ways of decomposing the total effect. 5

9 X 3 H G X ρ 2 F D C B X ρ 1 X 2 X 1 Figue 1: Sequential odeing vesus "natual "path In contast to tavelling on altenative combinations along the edges of the policy cube, the HPP method follows a staight line between the pe- and post-simulation values. Fo n exogenous vaiables, multivaiate calculus tell us that the total change in the endogenous vaiable is equal to: Z = n t= t1 F x i= 1 t= to i dxi dt dt This concept is easily genealized to the case whee the elationship between exogenous and endogenous vaiables is implicit, which is typically the case fo applied geneal equilibium models. s HHP point out, it is possible to calculate numeical values fo the gadients all points of the "natual" path by solving a system of linea equations. The individual contibutions of changes in policy instuments x i can then be appoximated though lineaization of the espective line integal which involves solving a system of linea equations R times, whee F x i dx dt i at 6

10 R endes a sufficiently small step-size t / how the decomposition can be implemented numeically. R. 7 ppendix povides the concete desciption of 3 Model Specification and Baseline Calibation/Paametization Fo ou analysis we use a static 8-secto, 13-egion CG model of the wold economy. The choice of sectos captues key dimensions in the analysis of geenhouse gas abatement such as diffeences in cabon intensities and the scope fo substitutability acoss enegy goods and cabon-intensive non-enegy goods. The egional aggegation coves the nnex B paties as well as majo non-nnex-b egions which ae cental to the geenhouse gas issue. 8 Table 1 summaizes the sectos and egions incopoated in ou model. Table 1: Oveview of sectos and counties/egions Sectos Regions COL Coal CN Canada CRU Cude oil C Cental uopean ssociates GS Natual gas UR uope (U15 and FT) OIL Refined oil poducts FSU Fome Soviet Union (Russian Fedeation and Ukaine) L lecticity OO Othe OCD (ustalia and New Zealand) IS negy-intensive sectos JPN Japan Y Manufactues and sevices US United States CGD Savings good SI Othe sia (except fo China and India) BR Bazil CHN China IND India MPC Mexico and OPC ROW Rest of Wold 3.1 Nontechnical Model Oveview This section povides a non-technical summay of the model. The algebaic model documentation is given in ppendix B. Pimay factos include labo, capital and fossil-fuel esouces. Labo and capital ae intesectoally mobile within a egion but cannot move between egions. secto-specific 7 Note that the pespective on spilloves by HHP is diffeent fom Böhinge and Ruthefod (1999). In the HHP appoach, the individual contibutions fom exogenous policy changes include both changes in domestic as well as changes in intenational pices, wheeas Böhinge and Ruthefod sepaate domestic fom intenational makets. 8 The aggegation is based on the GTP- data base which econciles the GTP economic poduction and tade dataset (Mc Dougall 1997) with OCD/I enegy statistics (I 1996) fo 45 egions and 23 sectos (Babike and Ruthefod 1998). See ppendix C fo the mapping of GTP egions and sectos with espect to the definitions of Table 1. 7

11 esouce is used in the poduction of pimay fossil fuels (cude oil, coal and gas), esulting in upwad sloping supply schedules fo those goods. 9 Poduction of commodities othe than pimay fossil fuels is captued by aggegate poduction functions which chaacteize technology though substitution possibilities between vaious inputs. Nested constant elasticity of substitution (CS) cost functions with thee levels ae employed to specify the KLM substitution possibilities in domestic poduction between capital, labo, enegy and mateial (non-enegy) intemediate inputs. t the top level, non-enegy inputs ae employed in fixed popotions with an aggegate of enegy, capital and labo. The mateial input of good i in secto j coesponds to a CS mington aggegate of non-enegy inputs fom domestic poduction and impoted vaieties. t the second level, a CS function descibes the substitution possibilities between the enegy aggegate and the aggegate of labo and capital. Finally, at the thd level, capital and labo tade off with a constant elasticity of substitution. s to the fomation of the enegy aggegate, we allow sufficient levels of nesting to pemit substitution between pimay enegy types as well as substitution between a pimay enegy composite and seconday enegy, i.e. electicity. In the poduction of fossil fuels, labo, capital and fossil fuel inputs ae aggegated in fixed popotions at the lowe nest. t the top level, this aggegate tades off with the sectospecific fossil-fuel esouce at a constant elasticity of substitution. The latte is calibated in consistency with exogenously given pice elasticities of fossil fuel supplies. Final demand in each egion is detemined by a epesentative agent who maximizes his utility subject to a budget constaint with fixed investment (i.e. given demand fo the savings good). Total income of the epesentative household consists of facto income and taxes 10. Final demand of the epesentative agent is given as a CS composite which combines consumption of an enegy aggegate and a non-enegy consumption bundle. Substitution pattens within the nonenegy consumption bundle ae eflected via Cobb-Douglas functions with an mington aggegation of impots and domestic commodities. The enegy aggegate in final demand consists of the vaious enegy goods tading off at a constant elasticity of substitution. 9 This model does not account fo impefectly competitive behavio on the pat of oil expoting counties. 10 In the benchmak, the model includes taxes on output, intemediate inputs, expots and impots, as well as taxes on final demand. 8

12 ll commodities ae taded intenationally. Cude oil is impoted and expoted as a homogeneous poduct, subject to taiffs and expot taxes. Fo all othe commodities, we adopt the mington assumption of poduct diffeentiation. n impotant caveat with espect to the model fomulation adopted hee concens the epesentation of intetempoal issues. In the absence of an endogenous esponse of capital to changes in ates of etun, the static model may fail to account fo some impotant mechanisms though which cabon taxes affect the global economy. We leave to futue wok the application of the decomposition methodology illustated hee to a model with fowad-looking agents and endogenous capital stocks. 3.2 Baseline Calibation The economic effects of the Kyoto Potocol depend cucially on the extent to which quantified emission limitation and eduction objectives (QLROs) bind the economies in the budget peiod. In othe wods, the magnitude and distibution of costs associated with the implementation of futue emission constaints depend on the Business-as-Usual (BU) pojections fo GDP, fuel pices, enegy efficiency impovements, etc. In ou compaative-static famewok we infe the BU stuctue of the model s egions fo 2010 using the most ecent pojections on the economic development. We measue the economic effects associated with abatement measues elative to that baseline. s a stating point fo ou fowad pojection, we use the GTP- database (Babike and Ruthefod 1998) which econciles economic poduction and tade data (based on GTP4 - McDougall 1997) and OCD/I enegy statistics (I 1996) fo the most ecent yea fo which a complete set of statistics is available. We use this benchmak data to calibate paametes of the CS functional foms fom a given set of quantities and pices (given exogenous elasticities). In a second step, we do the fowad calibation of the 1995 economies to 2010 incopoating exogenous infomation by the U.S. Depatment of negy (DO 1998) fo GDP gowth, enegy demand and futue enegy pices. The fossil fuel poduction functions ae finally calibated so that they ae consistent with exogenous pice elasticities of supply. 9

13 4 Policy Simulations and Results The Kyoto Potocol commits nnex B counties to the eduction of the aggegate CO 2 equivalent emissions on aveage by 5.2% below 1990 levels in the peiod 2008 to Table 2 summaizes the individual commitments by the nnex B egions as incopoated in ou model. In ou simulation, we assume that nnex B counties apply domestic cabon taxes which ae high enough to meet the individual Kyoto commitments. Cabon tax evenues accue dectly to the epesentative agent in each egion. This tax policy setting is fomally equivalent to a pemit system whee the epesentative agent auctions the egion's Kyoto emission budget on domestic makets. The pemit pice then coincides with the cabon tax, both measuing the maginal costs of abatement. Table 2: Quantified emissions limits unde the Kyoto Potocol (Baon et al. 1999) Region Commitments in % of 1990 geenhouse gas emissions CN: Canada 94 C: Cental uopean ssociates 93 UR: uopean Union and uopean Fee Tade ea 91.5 FSU: Fome Soviet Union 100 JPN: Japan 94 OO: Othe OCD US: United States of meica ffective Reduction Requements and Maginal batement Costs It is impotant to notice that the effective emission constaint fo nnex B counties unde the Kyoto Potocol must be measued against the BU economic activity without abatement equements. 11 Because emissions of most nnex B counties gow significantly along the baseline as compaed to 1990 levels, the Kyoto tagets which ae stated with espect to 1990 as the base yea tanslate into much highe effective cabon equements with espect to BU emission levels in Fo example, OO is allowed to incease emissions unde the Kyoto Potocol by 7% ove 1990 levels, while it faces effectively the need fo a decease by moe than 15% fom BU emissions in The one outlie is FSU whose pojected emission levels fo 2010 ae below 1990 levels due to economic ecession and industial estuctuing between 1990 and See Böhinge, Jensen and Ruthefod (2000) fo the implications of altenative baseline pojections on the magnitude and distibution of emission abatement costs. 10

14 Table 3 epots the effective pecentage cutbacks joint with the maginal abatement costs acoss nnex B counties. The level of abatement is a majo deteminant of the maginal abatement costs. The futhe out we ae on the abatement cost cuve, the moe costly it is - ceteis paibus - to substitute away fom cabon in poduction and consumption. Obviously, the maginal abatement costs fo FSU ae zeo because its economic development will not be constained by the Kyoto commitment. C faces a athe weak eduction equement of only 2.3 % with espect to BU emissions in 2010 which explains the modeate cabon tax of 11 $US pe ton of cabon. Table 3: ffective eduction equements and maginal abatement costs Region % Reduction wt 1990 % Reduction wt 2010 Maginal batement Cost * CN C UR FSU JPN OO US *$US pe ton of cabon s we move towads highe eduction tagets the emission taxes incease significantly anging up to 455 $US pe ton of cabon fo Japan. Compaison of tax ates and eduction equements acoss counties show that the elative cutback equements ae only one deteminant of maginal abatement costs. The latte depend also on diffeences in cabon intensity fo diffeent sectos acoss counties. Fo example, JPN faces much highe cabon taxes compaed to US o CN, although its pecentage eduction taget is smalle. The eason is that JPN uses elatively little cabon in sectos with low-cost substitution possibilities, e. g. electicity geneation (due to nuclea powe). s a consequence, JPN has to cut back elatively moe emissions in othe sectos such as taffic whee abatement comes moe costly at the magin. 4.2 Total Welfae Costs Table 4 summaizes the welfae implications of Kyoto measued as pecentage change in eal consumption with espect to BU (positive numbes indicate welfae impovements). We see that abatement policies by nnex B egions poduce substantial spilloves to non- nnex B egions as well as between nnex B egions. pat fom MPC and ROW, all non- 11

15 nnex B counties gain a compaative advantage, i. e. impove the tems of tade (ToT). mong nnex B counties, FSU, which does not have to impose cabon taxes, faces nonnegligible welfae losses though the deteioation of ToT. On the othe hand, C expeiences a significant incease in welfae despite the application of domestic cabon taxes; ToT gains fo C ae obviously lage enough to offset the domestic costs of cabon abatement. Table 4: Total welfae impact of Kyoto Region % Change wt BU consumption CN: Canada -1.6 C: Cental uopean ssociates 0.5 UR: U15 and FT -0.2 FSU: Russian Fedeation and Ukaine -0.3 JPN: Japan -0.5 OO: ustalia and New Zealand -1.1 US: United States -0.9 SI: Othe sia 0.2 BR: Bazil 0.3 CHN: China 0.1 IND: India 0.3 MPC: Mexico and OPC -1.7 ROW: Rest of the Wold -0.4 Wold Total -0.5 Fo all OCD counties (US, CN, UR, JPN, OO), imposition of the Kyoto tagets leads to economic losses, which is not supising given the high tax ates. To undestand the welfae effects fo the vaious egions at the aggegate level, it is useful to distinguish whethe (i) egions must abate o not, and (ii) egions ae eithe fossil enegy expotes o impotes. Neglecting ToT effects, the economic impacts should be negative fo abating nnex B counties and zeo fo non-abating non-nnex B counties. Those nnex B egions which face binding cabon limits must substitute fossil fuels with moe expensive enegy souces (fuel switching) o employ moe expensive manufactuing and poduction techniques (enegy savings). On the consumption side, highe enegy pices imply a change in the consumption mix which esults in a loss of welfae (consume suplus). Fo a single abating egion, the magnitude of infamaginal welfae losses depends on the level of the maginal abatement costs. ccounting fo ToT, an abating egion might pass on some of the pice inceases in the poducts to othe counties. Note that buden shifting occus also among abating counties. Fo any paticula egion, the tems of tade effect is detemined by the composition of a scale effect and a substitution effect. Fo example, non-abating counties might gain expot shae fo enegy 12

16 intensive goods in abating counties; howeve, this might be (patially) offset by a eduction in the scale of expots to abating counties as final demand in those counties shinks with the ise in consume goods pices. negy expoting nations suffe fom nnex B abatement policies. The eason is that the cut in enegy demand by the industialized wold 12 leads to a dop in poduce pices fo fossil fuels. mong fossil enegy goods, the decline in pices is most ponounced fo cabon-intensive coal, followed by less cabon-intensive (cude) oil and gas. Based on the consideations above, we can give futhe explanation to Table 4. ll developing counties, except fo MPC and ROW, ae lage impotes of fossil enegies and benefit fom the fall in intenational enegy pices. MPC and ROW ae expotes of fossil enegies and face a loss of evenues fom enegy sales. With espect to tade in enegy-intensive goods, developing counties gain a compaative advantage as compaed to OCD counties with high enegy pices. Howeve, fo MPC and ROW, this benefit is not lage enough to offset the advese effect on fossil fuel makets. Like MPC, FSU, as a lage enegy expote of oil and gas, suffes fom the decease in fuel pices. mong abating OCD counties, C not only impoves welfae due to lowe enegy impot pices; it also impoves its tade position among the heavily integated OCD club because it faces by fa the lowest cabon taxes. In contast to all othe OCD counties, C does not lose, but gains intenational competitiveness in enegy intensive poduction. (The static natue of the model leads to an undeestimate of the impacts on global capital accumulation and gowth. The calculated welfae gains obseved in enegy-impoting developing counties would most likely be absent in a dynamic fomulation. See Benstein, Montgomey and Ruthefod 1999 o Böhinge and Ruthefod 2000). 4.3 Decomposition of Total Welfae Impact Wheeas Table 4 conveys infomation on the magnitude and distibution of costs and gains fom Kyoto fo individual egions 13 it does not eveal how the total welfae changes can be attibuted to the individual abatement policies acoss nnex B counties. These contibutions by abating egions povide futhe insights into the souces of geneal equilibium effects and ae potentially 12 The expansion in enegy use by some non-abating counties is by fa too small in ode to offset the educed demand by nnex-b counties. 13 Note that at the global level binding emission constaints induce a welfae loss as it esticts oveall ("fst-best") poduction and consumption possibilities (see the bottom ow "WORLD" of Table 4). 13

17 impotant fo the negotiation of tansfe payments at the bilateal level. pplication of the HHP decomposition with espect to cabon taxes in nnex B delive this infomation. The esults, pesented in Table 5, show the pecentage of the welfae cost fo each egion (ows) attibutable to cabon taxes in each of the nnex B egions (columns). These numbes ae obtained along a line integal in which we change the cabon taxes acoss abating egions at equal ates stating fom zeo and ending with the final cabon taxes as epoted in Table 3. Table 5: Pecentage of the welfae cost fo each egion (ows) attibutable to cabon taxes in each of the nnex B egions (columns) CN C UR JPN OO US CN C UR FSU JPN OO US SI BR CHN IND MPC ROW Note that thee is no column fo nnex B county FSU, simply because FSU does not have to undetake any action in ode to meet its Kyoto commitment. Futhemoe, a negative sign indicates that the effect of the abatement measue by the column egion is contay to the change in egional welfae of the ow egion. The diagonal elements fo abating egions indicate the pecentage of welfae cost which is due to the own action. Fst of all, we see that the contibution of a egion fom its own abatement measue is welfae deceasing. This esult confms basic economic intuition that emission constaints educe esouce poductivity and induce income losses. The "own" policy effect vaies substantially acoss abating egions. While own action is vey costly fo uope (moe than the double of the aggegate welfae loss) it has a negligible effect fo C. The latte can be explained by the vey modeate cabon taxes which ae necessay fo C in ode to fulfill its Kyoto commitment. The lage numbes fo all othe egions eflect much highe cabon tax ates. 14

18 Reading Table 5 by ows it follows that action by the US poduces by fa the lagest spilloves to othe counties. This finding is consistent with the fact that US plays a majo ole on intenational makets fo fossil fuels and must impose athe high cabon taxes to cope with its Kyoto taget. batement polices in OO and C on the othe hand have vey small impacts on othe egions which can be taced back to the modeate tax ates joint with small shaes in oveall tade volumes. With espect to spilloves fom othe egions' action we can split the OCD counties in two goups. Fo US and CN the total welfae impacts ae athe independent of the abatement activities in othe OCD counties. This indicates that action by each of the tading patnes in the developed wold poduce offsetting effects on intenational makets. On the othe hand C, UR, JPN and OO ae athe sensitive to abatement action by the abating tading patnes. The welfae implications ae howeve opposite fo C, UR and JPN as compaed to OO. While C, UR and JPN benefit fom actions of all othe OCD counties (in paticula fom abatement in the US) the evese is tue fo OO. FSU is negatively affected by spilloves fom all OCD counties except fo C. mong developing counties IND and BR gain fom action of all OCD counties, wheeas the opposite holds fo MPC and ROW. SI and CHN suffe fom high cabon taxes in JPN but benefit fom cabon taxes in all othe counties. We can tanslate the above pecentage changes in welfae fom individual policy action into monetay units. The esulting Table 6 could then be intepeted as a coss-county matix fo tansfe payments which have to be assigned on a bilateal basis in ode to povide compensation fo the Kyoto impacts. Table 6: Compensating tansfes fom egion (ows) to egion (column) in billion dollas annually between 2008 and 2012 (cabon tax policy) US CN UR JPN OO FSU CN UR JPN OO FSU C CHN IND BR SI MPC ROW

19 In ou exposition of compensating tansfes we have netted out payments between abating counties such that we deive net tansfes in pesent value tems. positive enty indicates compensation claims of the ow egion towads the column egion. Fo example, US should compensate the OPC counties plus Mexico (MPC) with oughly 1461 million dollas annually to offset the advese impact of its abatement policy on that egion. Likewise, uope has to tansfe nealy the same amount to the US to make up fo the benefits UR expeiences fom US action Despite the appaent tidiness of this calculation, thee emain substantial uncetainty egading the estimated values. One key uncetainty concens the decomposition pocedue itself because - as we will see below - the decomposition poposed by Haison, Hoidge and Peason does not esolve all ambiguities.. Complementay to the decomposition of the total welfae effect using a pice instument (emission taxes) we could just as easily pefom the decomposition using a quantity instument (emission pemits). Fo the latte, we educe the emission budgets of abating counties in equal popotions, stating fom BU emission levels towads the emission quantities as imposed by the Kyoto Potocol. The quantity instument poduce exactly the same aggegate changes in endogenous vaiables as the pice instument pocedue. Howeve, as indicated by compaing Tables 5 and 7 (o likewise Tables 6 and 8), thee ae diffeences in the quantitative estimates fo the contibutions by individual policy measues. Table 7: Pecentage of the welfae cost fo each egion (ows) attibutable to emission pemit systems in each of the nnex B egions (columns) CN C UR JPN OO US CN C UR FSU JPN OO US 2 98 SI BR CHN IND MPC ROW

20 The eason is that the two pocedues fo appoximate the line integals of individual policy changes follow a diffeent path though the same outcomes. Obviously, the decomposition appoach poposed by HHP does not esolve all ambiguities. In ou policy application, the qualitative esults ae lagely the same, howeve thee ae consideable diffeent magnitudes in welfae changes and associated potential tansfe payments. We theefoe conclude that while the HHP decomposition pocedue is insightful, it does not povide an unambiguous estimate of bilateal tansfe payments which would compensate fo spilloves fom nnex B action. Table 8: Compensating tansfes fom egion (ows) to egion (column) in billion dollas annually between 2008 and 2012 (cabon tax policy) US CN UR JPN OO FSU CN UR JPN OO FSU C CHN IND BR SI MPC ROW Conclusions In this pape we evaluated the welfae implications of the Kyoto Potocol fo the wold economy. We used a new decomposition pocedue developed by Haison, Hoidge and Peason to investigate the souces of welfae changes acoss egions induced by geenhouse gas emission constaints. Ou simulations povide inteesting insights into the qualitative intedependencies of multilateal abatement policies. We have shown that the contibutions of individual policy changes to the oveall effect depend on the policy instument. When we simply change fom a pice instument (taxes) to a quantity instument (emissions) in ode to meet the Kyoto commitments the quantitative decomposition esults change. Howeve, the qualitative esults emain obust suggesting that the decomposition pocedue povides a useful stating point fo bilateal negotiations on policy-elevant tansfe payments acoss paties of the Famewok Convention. 17

21 6 Refeences Babike, M. and T.F. Ruthefod (1998): GTP-: global negy-conomy Dataset, Univesity of Coloado, Boulde, Baon, R. et al. (1999), Peliminay nalysis of the U Poposals on the Kyoto Mechanisms, negy and nvonment Division, Intenational negy gency, eupo/eulong.html. Begmann, L. (1990), The Development of Computable Geneal equilibium Models, in: Begman, L., D.W. Jogenson and. Zalai (eds.): Geneal quilibium Modeling and conomic Policy nalysis, Cambidge, Benstein, P.M., Montgomey, W.D. and T.F. Ruthefod (1999), Global impacts of the Kyoto ageement: Results fom the MS-MRT model, Resouces and negy conomics; ugust Böhinge, C., Haison, G.W. and T.F. Ruthefod (1998), Shaing the Buden of Cabon batement in the uopen Union, Woking Pape 98-5, Univesity of Boulde. Böhinge, C. and T.F. Ruthefod (1999), Decomposing Geneal quilibium ffects of Policy Intevention into Open conomies, ZW discussion pape 99-36, Mannheim (evised fo: conomic Modelling). Böhinge, C.; Jensen, J. and T.F. Ruthefod (2000), negy Maket Pojections and Diffeentiated Cabon batement in the uopean Union, in: C. Caao (ed.): fficiency and quity of Climate Change Policy, Kluwe. Böhinge, C. and T.F. Ruthefod (2000), Wold conomic Impacts of the Kyoto Potocol, in Welfens, P.J.J, Hilleband, R. and. Ulph (ds.): Intenalization of the conomy, nvonmental Poblems and New Policy Options, Spinge, Heidelbeg/New Yok, fothcoming. DO (1998), Depatment of negy, nnual negy Outlook (O 1998), negy Infomation dministation Haison J., Hoidge, M. and K.R.Peason (1999), Decomposing Simulation Results with Respect to xogenous Shocks, woking pape IP-73, Cente of Policy Studies, Monash Univesity, fothcoming: Computational conomics. Helm, C. (1999), conomic Theoies of Intenational nvonmental Coopeation, ph. d. thesis, Potsdam Univesity. Huff, K.M. and T.W. Hetel (1997), Decomposing Welfae Changes in the GTP Model, GTP Technical Pape No. 5, Pudue Univesity, West Lafayette. I (1996), Intenational negy gency, negy Pices and Taxes / negy Balances of OCD and Non-OCD counties, Pais: I publications. Kehoe, P.J. and T. J. Kehoe (1994), Pime on Static pplied Geneal quilibium Models, Quately Review, Sping 1994, Fedeal Reseve Bank of Minneapolis. McDougall, R.. (1997), ed., Global Tade, ssistance and Potection: The GTP 4 Data Base, Cente fo Global Tade nalysis, Pudue Univesity, West Lafayette. Moulin (1992), n application of the shapely value to fa division with money, conometica 60 (6), Peea,.M and J.B. Shoven (1992), Suvey of Dynamic Computational Geneal quilibium Models fo Tax Policy valuation, Jounal of Policy Modeling 10, Shoven, J.B. and J. Whalley (1992), pplying Geneal quilibium, Cambidge Univesity Pess. Simonis, U.. (1996), Intenationally tadable emission cetificates: Linking envonmental potections and development. conomics 53, UN 1997, United Nations, Kyoto Potocol to the United Nations Famewok Convention on Climate Change, Kyoto, 10 Decembe UN-Doc FCCC/CP/1997, Geneva. UNFCCC (1992), United Nations Famewok Convention on Climate Change, New Yok, 9 May 1992, in foce 21 Mach

22 ppendix : lgebaic Model Summay This appendix povides an algebaic summay of the equilibium conditions fo ou compaativestatic model designed to investigate the economic implications of the Kyoto Potocol in 2010 as compaed to a Business-as-Usual economic development whee no cabon abatement policies apply. Befoe pesenting the algebaic exposition we state ou main assumptions and intoduce the notation. Nested sepaable constant elasticity of substitution (CS) functions chaacteize the use of inputs in poduction. ll poduction exhibits non-inceasing etuns to scale. Goods ae poduced with capital, labo, enegy and mateial (KLM). epesentative agent (R) in each egion is endowed with thee pimay factos: natual esouces (used fo fossil fuel poduction), labo and capital. The R maximizes utility fom consumption of a CS composite subject to a budget constaint with fixed investment (i.e. fixed demand fo the savings good). The aggegate consumption bundle combines demands fo fossil fuels, electicity and non-enegy commodities. Total income of the R consists of facto income and taxes (including cabon tax evenues). Supplies of labo, capital and natual esouces ae exogenous. Labo and capital ae mobile within domestic bodes but cannot move between egions; natual esouces ae secto specific. ll goods, except fo cude oil, ae diffeentiated by egion of oigin. Nested CS functions chaacteize the choice between impoted and domestic vaieties of the same good (mington). Cude oil is impoted and expoted as a homogeneous poduct. Two classes of conditions chaacteize the competitive equilibium fo ou model: zeo pofit conditions and maket cleaance conditions. The fome class detemines activity levels and the latte detemine pice levels. In ou algebaic exposition, the notation 19 z Π is used to denote the pofit function of secto j in egion whee z is the name assigned to the associated poduction activity. Diffeentiating the pofit function with espect to input and output pices povides compensated demand and supply coefficients (Shepad s lemma), which appea subsequently in the maket cleaance conditions. We use i (aliased with j) as index fo commodities (sectos), (aliased with s) as index fo egions and d as index fo the demand categoy (d=y: intemediate demand, d=c: pivate household demand, d=g: investment demand,

23 d=i: investment demand). The label G epesents the set of enegy goods and the label FF denotes the subset of fossil fuels. Tables.1.6 explain the notations fo vaiables and paametes employed within ou algebaic exposition..1 Zeo Pofit Conditions 1. Poduction of goods except fossil fuels: Π Y = p j G θ j p Yj θ KL 2. Poduction of fossil fuels: 1 1- KL 1 - σ σ KL 1-σ L K KL [ θ p +( 1 - ) α ( w j α v j ) ] = 0 i FF θ Π Y = p - Q θ q 1 σ Q θ 3. Secto-specific enegy aggegate: θ 1 σ Q 1 1 σ Q Q FF FF FF + ( 1 θ ) + + = L w K v Yj 0 j j θ p i FF Π = p - θ L p 1 σ L { Y, L, } + (1 θ L ) θ CO p 1 σ CO { Y, CO, } + (1 θ CO ) Π j LQ p β j Yj 1 σ CO 1 σ 1 σ L CO 1 1 σ L =0 4. mington aggegate: Π d = p d σ 1-σ 1-σ M CO2 CO2 ( θ p + (1 ) p ) + t a i = 0 d θ d 5. ggegate impots acoss impot egions: Π M M M 1- M = p - is p is σ θ s 1 1-σ M = 0 6. Household consumption demand: Π C C = p - θ C p 1-σ C C +(1-θ C ) i G p γ C ) 1-σ C 1 1-σ C = 0 20

24 7. Household enegy demand: Π C = p C 1 1-σ L, C 1-σ L, C σ, {,, }, (1 {,, }) σ 1-σ NL 1- L C 1- NL - θ + θ ) L C pl L C = 0 `{ } ( θ ic pc i G L 1.2 Maket Cleaance Conditions 8. Labo: L = i Y 9. Capital: K = i Y Π w Y Π v 10. Natual esouces: Q = Y Π q Y Y i FF 11. Good makets: Y Π p Y = dj dj Π p dj + s M is Π p M is 12. Secto specific enegy aggegate: =Y Π p Y 13. Impot aggegate: M = d d Π p d M 14. mington aggegate: d = j Y j Π p Y j d +C Π p C d 21

25 15. Household consumption: C p C =( w L + v K + j FF q j Q j t CO2 Y CO2 + + CGD, CGD, p + B 16. ggegate household enegy consumption: C = C Π p C C 17. Cabon emissions: CO2 = di d a CO2 i 22

26 Table.1: i j s G FF LQ d Sets Sectos and goods liased with i Regions liased with ll enegy goods: Coal, cude oil, efined oil, gas and electicity Pimay fossil fuels: Coal, cude oil and gas Liquid fuels: Cude oil, efined oil and gas Demand categoies: Y = intemediate, C = household and I = investment Table.2: Y M C d C ctivity vaiables Poduction in secto i and egion ggegate enegy input in secto i and egion ggegate impots of good i and egion mington aggegate fo demand categoy d of good i in egion ggegate household consumption in egion ggegate household enegy consumption in egion Table.3: Pice vaiables p p p M p d p C p C w v q CO2 t Output pice of good i poduced in egion fo domestic maket Pice of aggegate enegy in secto i and egion Impot pice aggegate fo good i impoted to egion Pice of mington aggegate fo demand categoy d of good i in egion Pice of aggegate household consumption in egion Pice of aggegate household enegy consumption in egion Wage ate in egion Pice of capital sevices in egion Rent to natual esouces in egion (i FF) CO 2 tax in egion 23

27 Table.4: Cost shaes θ j KL θ θ T α Q θ FF θt θ L θ CO β j θ M is θ d θ C γ θ L, C, θ ic Shae of intemediate good j in secto i and egion (i FF) Shae of KL aggegate in secto i and egion (i FF) Shae of enegy in the KL aggegate of secto i and egion (i FF) Shae of labo (T=L) o capital (T=K) in secto i and egion (i FF) Shae of natual esouces in secto i of egion (i FF) Shae of good i (T=i) o labo (T=L) o capital (T=K) in secto i and egion (i FF) Shae of electicity in enegy demand by secto i in egion (i FF) Shae of coal in fossil fuel demand by secto i in egion Shae of liquid fossil fuel j in liquid fossil fuel demand by secto i in egion (i FF, j LQ) Shae of impots of good i fom egion s to egion Shae of domestic vaiety i in mington aggegate fo demand categoy d in egion Shae of enegy in aggegate household consumption in egion Shae of non-enegy good i in non-enegy household consumption demand in egion Shae of electicity in aggegate household enegy consumption in egion Shae of non-electic enegy good i in the non-electic household enegy consumption in egion Table.5: ndowments and emissions coefficients L K Q ggegate labo endowment fo egion ggegate capital endowment fo egion ndowment of natual esouce i fo egion (i FF) B Balance of payment suplus in egion (note: B = 0 ) CO2 CO a 2 i ndowment of cabon emission ights in egion Cabon emissions coefficient fo fossil fuel i (i FF) in demand categoy d of egion 24

28 Table.6: σ KL σ FF lasticities Substitution between enegy and value-added in poduction (except fossil fuels) Substitution between natual esouces and othe inputs in fossil fuel poduction calibated consistently to exogenous supply elasticities µ. σ L Substitution between electic and non-electic enegy in poduction 0.1 σ CO Substitution between coal and liquid fossil fuel aggegate in poduction 0.5 σ M Substitution between impots fom diffeent egions 8 σ Substitution between the impot aggegate and the domestic input 4 σ Substitution between electic and non-electic enegy in household enegy 1 L,C consumption σ Inte-fuel substitution in final fossil fuel demand 2 NL FF 0.3 µ CO =0.5 µ CRU =1.0 µ GS =1.0 25