Developing Countries and the UNFCCC Process: Some Simulations From an Armington Extended Climate Model

Developng Countres and the UNFCCC Process: Some Smulatons From an Armngton Extended Clmate Model TIAN Hufang (IWEP, CASS), John Whalley (CIGI) MDB and the G20 Development Agenda Shangha T20 Conference

Developng Countres and The UNFCCC Process: Some Smulatons From an Armngton Extended Clmate Model Developng Countres and The UNFCCC Process: Some Smulatons From an Armngton Extended Clmate Model 1 TIAN Hufang, John Whalley 2 ABSTRACT We report smulaton results for alternatve multlateral emssons cuts and accompanyng polces whch could come under renewed reconsderaton for the process to follow the Durban UNFCCC negotatons. The model s an Armngton type trade model extended to capture clmate change. We calbrate the model to alternatve BAU damage scenaros followng the Stern report and the lterature that has followed. We consder dfferent depth, forms, and tmeframes for emssons reductons by Chna, Inda, Russa, Brazl, US, EU, Japan and a resdual Row both ontly and block wse. We assume regonally unform percentage both clmate change and damages by regon, whch are relax later n senstvty analyss. The welfare mpacts of both emsson reductons and accompanyng measures are computed n Hcksan money metrc equvalent form over 3 alternatve potental commtment perods: 2012-2020, 2012-2030, and 2012-2050. Our multyear multcounty global modelng framework captures the beneft of emssons mtgaton through preferences ncorporatng temperature change. Countres are lnked not only through shared welfare mpacts of global temperature change but also through trade among country subscrpted goods. These trade mpacts nfluence net country benefts from alternatve emssons reducton agreements. We also evaluate the potental mpacts of potental accompanyng mechansms ncludng funds/transfers, border adustments, and tarffs. 1 1 We are grateful to the Academc Development Fund (ADF) at the Unversty of Western Ontaro, and to the Ontaro Research Fund (ORF-R3) for fnancal support. We also acknowledge support from the Centre for Internatonal Governance Innovaton (CIGI). Hufang Tan s also grateful to the Natonal Socal Scence Foundaton of Chna (Proect No. 14ZDA081), and to Renmn Unversty of Chna. 2 Tan Hufang s an assocate professor of IWEP, CASS; John Whalley s a CIGI Dstngushed Fellow. Ths artcle was publshed n Clmate Change Economcs, Vol. 6, No. 4 (2015) Workng Paper 2015-06

Developng Countres and The UNFCCC Process: Some Smulatons From an Armngton Extended Clmate Model I. Global Polcy Context and Introducton Negotatons on clmate change arrangements under the Unted Natons Framework Conventon on Clmate Change (UNFCCC) as part of the shapng of a post Kyoto/ post 2012 regme have moved from a negotaton ntated n Bal n late 2007 through Copenhagen meetngs n December 2009 to an nconclusve Durban meetng n late 2011. Dscussons are to contnue through untl 2015, and the role of large developng countres (Chna, Inda, and Brazl) s the key as they are more rapdly growng. Here, we use a numercal general equlbrum clmate model ncorporatng Armngton type features. It captures the beneft sde of emssons reductons n preferences n cooperatng temperature change and s used to evaluate the mpacts of possble ont actons on emssons, transfers, and other actons by developed and developng countres post Durban. We adopt an assumpton of unform percentage both clmate change and damages by regon whch we relax n later senstvty analyss. In the negotaton process thus far, there has been lttle by way of clear agreement. The notfcaton and verfcaton of the unlateral reductons tabled n Copenhagen as part of the Copenhagen Accords s the man content of what has been acheved even though the post 2012 process has attempted to acheve much broadened ssue coverage to also nclude adaptaton, mtgaton, and fnance. It has also amed to fully nclude the large populaton, rapdly growng economes of Chna, Inda, and Brazl n the global process, who unlke n Kyoto have been asked to take on commtments. These countres, n turn, have consstently argued both that as rapd growers and countres relatvely new to ndustralzaton they should be treated dfferently from more mature OECD countres. Furthermore, they argue that ths was commtted to under the prncple of Common but Dfferentated Responsbltes (CBDR) n the Unted Natons Framework Conventon on Clmate Change (UNFCCC). 2 A range of contentous ssues have emerged n the negotatons as to what form any specal developng country treatment should take as far as emssons reductons are concerned. One s whether emssons reductons targets by countres should be based on a sngle global target (and target date) wth country targets allocated to countres on a cumulatve bass rather than the Kyoto annual emssons bass. The developng country argument has been that emssons n the upper atmosphere have mostly orgnated from OECD countres over many years, and that emssons targets allocated by country should reflect ths hstorcal responsblty. Another s whether emssons reducton targets should focus on reducng emssons ntenstes (emssons/dollar of GDP) rather than emssons levels, so as to allow more room for growth by rapdly growng developng countres. Yet another s that emssons targets should be based on the carbon content of consumpton of goods n countres, rather than geographcal locaton of producton. Fnally, some developng countres have argued that the prncple of CBDR mples preferental and lower emssons reducton targets for them relatve to OECD countres. But other ssues have also arsen. One s the choce of base date for calculatng reductons wth, for example, 1990 (the Kyoto data) beng strongly preferred by Russa (and the EU) due to negatve (slow) growth between 1990 and today, and 1990 beng strongly ressted by Chna for Workng Paper 2015-06

Developng Countres and The UNFCCC Process: Some Smulatons From an Armngton Extended Clmate Model the opposte reason. And detals wthn broad ssues, such as wth the use of ntensty targets, how GDP s calculated and used n settng ntensty targets (usng market exchange rates or PPP for converson nto US$) also enter. Fnally, there has been dscusson of the use of accompanyng fnancal arrangements through the Adaptaton and Innovaton Funds, as well as possble trade measures aganst nonpartcpants snce these affect country outcomes. Developng countres, represented by the G77, argued n both Bal and Copenhagen for a large fund to help them adapt to clmate change, ntally proposed at the top end of $300 bllon per year by 2013. The Copenhagen Accords adopted language of workng towards a clmate change fund for developng countres of $100 bllon per year by 2020. And on the trade front, both the US and the EU have made proposals for the use of carbon emsson based tarffs and export rebates affectng trade wth nonpartcpant countres. There s nether lttle nor no quanttatve model based evaluaton work ether on the potental welfare and other mpacts of ether dfferent concrete proposal for global mtgaton and/or the accompanyng fnancal or trade mechansms. Earler work (such as Goulder (2000),Veenendaal (2008)), has rather focused on mpacts of across the board percentage cuts wthout relatng these to the global negotatng process. Here we use an extended verson of the N country N good Armngton type clmate modelng framework developed by Ca, Rezman, and Whalley (CRW) (2012) and buldng on Uzawa (2004). The orgnal use of the CRW framework was to explore whether nternatonal trade makes partcpaton n clmate change negotatons more lkely. In ths framework, temperature change drectly enters preferences and countres can set asde part of ther potental consumpton avalable under a no mtgaton busness as usual (BAU) scenaro to meet clmate polcy commtments and lower global temperatures. Ths gves them a utlty gan, but at a utlty cost n terms of foregone consumpton. CRW then specfy a temperature change functon lnkng global temperature change to emssons, and an abatement cost functon whch captures the margnal cost of mtgaton whch, n turn, mples country resource or abatement costs of emsson reducton. Here we also extend t to capture dfferent mtgaton targets and potental accompanyng fnancal and trade related mechansms. 3 Our formulaton treats commtment perods of several decades as a sngle perod and compares a BAU scenaro to model outcomes under alternatve emssons reductons and use of accompanyng mechansms. We do not explctly model any mult perod dynamcs. Explct ntertemporal allocaton ssues such as savngs and nvestment mpacts are not at the centre of clmate polcy debate and n our vew are secondary. We use calbratons to alternatve busness as usual (BAU) scenaros for 3 dfferent potental commtment perods out to 2020, 2030 and 2050. We frst use annual data for 2006 whch we proect to a 2012 base data set usng 2000-2006 country growth rates. We then calbrate a temperature change functon to Stern lke BAU damage estmates of both damage and Workng Paper 2015-06

Developng Countres and The UNFCCC Process: Some Smulatons From an Armngton Extended Clmate Model temperature change out to 2050. These damage estmates exceed those of the bounds of the IPCC-AR4 and AR5 confdence ranges, and we perform senstvty analyss. Our results produce a wde range of potental outcomes dependng on the formulae used to allocate cuts to countres, and hence a large barganng set for the post 2012 process, whch n turn suggests t may be dffcult to conclude. Our results suggest that gven the central case damage estmates from clmate change we use n model calbratons, the countres we consder wll both ndvdually and collectvely lose from clmate reducton ntatves proposed both n post 2012 process f there are no accompanyng mechansms. The ssue for them becomes the form of mtgaton whch mnmzes country losses. But f we calbrate to larger damage cost estmates the pattern changes to ont gans, whle the relatve pcture across countres s much the same. The dfferences between country emssons reductons based on cumulatve and annual emssons are especally large. Other ssues such as consumpton or producton as a bass for cuts or the use of ntensty targets have smaller but stll pronounced cross country mpacts. The same s true of a 1990 versus a 2012 base date for certan countres (Russa (1990), Chna (2012), EU (1990)). Our results also ndcate that each percentage pont dfferentaton n cuts between developng and developed countres sgnfcantly benefts developng countres. Results on accompanyng mechansms suggest that the sze of accompanyng funds s crtcal for developng countres. In the model, an Adaptaton Fund of $100 bllon a year s nsuffcent to nduce developng country partcpaton, whle $200 bllon/year s suffcent. The costs to the developed countres are, however, large. Border tax adustments emerge as quanttatvely relatvely less sgnfcant n mpact. Here, effects depend on the sze of border adustment and who undertakes them. Fnally, trade sanctons (tarffs) can also have sgnfcant effects but typcally need to be large to convert loses from partcpaton nto country gans from avodng the sancton. 4 II. An Armngton Type Extended Clmate Modelng Framework for the Evaluaton of Possble UNFCCC Emssons Reducton Arrangements Model Summary To analyze the potental mpacts on large developng countres of the varous possbltes for UNFCCC agreements we set out n the precedng secton, we have used a modelng framework due to Ca, Rezman, and Whalley (2012) and Tan and Whalley (2010) (and buldng on that set out n Uzawa (2004)) to analyze lnks between trade, trade polcy and clmate change polcy arrangements whch varous proposed emsson reductons and fnancal arrangements mply. We evaluate the welfare mpacts of alternatve emssons reducton mechansms such as usng a cumulatve or annual emsson bass, emsson ntensty or emsson level targets, carbon content of consumpton or geographcal locaton of producton. We also explore whether border Workng Paper 2015-06

Developng Countres and The UNFCCC Process: Some Smulatons From an Armngton Extended Clmate Model taxes, tarffs, and/or transfers and at what level make partcpaton n specfc clmate change arrangements more lkely. The emphass s on net country welfare benefts from partcpaton n ontly agreed packages of emssons reductons and accompanyng measures (transfers, border taxes). We focus the analyss on the large populaton rapdly growng developng economes of Chna, Inda, Russa and Brazl. We use a sngle perod model coverng a number of years durng whch each natonal economy s assumed to grow at a compoundng constant rate 3. As such t s a statc model wth a multyear sngle perod of analyss. Explct dynamc effects on captal accumulaton are excluded to avod excessve model complexty as they are not central to the global clmate polcy debate, although they also arse wth ssues of the speed of emssons slowng as noted by Nordhaus (1990). Because the model uses a sngle perod, dscountng does not formally enter the analytc structure. Dscountng does, however, arse wth the use of a dscount rate n calculatng the dscounted present value of GDP over the model perod. We consder cases n senstvty analyss wth a common dscount rate of 1% across all countres, snce the growth rates of key OECD countres (EU, Japan) are low. We adopt an assumpton of unform percentage both clmate change and damages by regon n our central case results, whch we relax n later senstvty analyss In the model, each country s able to consume or export one country specfc heterogeneous good n the perod whose potental consumpton (or use) grows at the rate set n the base case. We assume that consumpton of the good ether by the country drectly, or by others through trade, generates emssons of carbon whch, n turn, rase global temperatures. Countres receve postve utlty from consumpton, but negatve utlty from temperature change. Countres export ther own good and mport other country goods. If countres are small, ther own actons have lttle or no effect on temperature change. Countres effectvely face an upper bound on the use of ther own good reflectng a Busness as Usual (BAU) scenaro snce a BAU represents zero emssons reducton, and f they use (consume or export) less than the upper bound they experence less temperature change, as do all other countres. The amount of resources needed to be put asde to acheve gven reductons reflects abatement cost estmates. 5 As we later work wth the mpacts of agreements to reduce carbon emssons over dfferent perods of tme, we take the sngle perod to cover alternatve horzons from 2012 out to 2020, 2030 or 2050. These reflect dfferent commtment perods for possble UNFCCC agreements on a post 2012 regme. In ths mult year perod, we focus on changes n consumpton (use of own and foregn goods) and utlty, and measure changes n these varables relatve to the outcome of zero growth over the perod. We report changes n utlty n money metrc (Hcksan) form n US$ amounts. The model dffers from recent Integrated Assessment Models (IAM) of clmate change n beng a sngle nternally consstent Arrow-Debreu type model n whch welfare analyss can be 3 See also the dscusson of dscountng and clmate change polcy n Wetzman (2007) and Dasgupta (2008), and the key role dscountng plays n the conclusons of the Stern (2006) report. Workng Paper 2015-06

Developng Countres and The UNFCCC Process: Some Smulatons From an Armngton Extended Clmate Model conducted. IAMs nvolve modelng structures wth less well defned welfare metrcs. 4 Preferences The preferences of each country over the perod are reflected n a utlty functon wth arguments gven by ts own composte consumpton as well as temperature change. The utlty functon s thus effectvely defned over multyear both consumpton and temperature change. The potental use of each country good reflects potental output from the economy over the same perod. We frst analyze a busness as usual (BAU) scenaro whch reflects current observed growth rates remanng unchanged over the model perod, temperature change as proected wth no global or sngle country emssons lmtaton ntatves n place. We assume the utlty functon for each country has the form H T U U( RC, T ) RC* ( ) H In ths specfcaton, 5 RC represents the change n consumpton of a composte of ther own good and other country s goods whch they acqure by mportng other country s goods and exportng ther own good for each country over the perod. Ths structure provdes the lnk between trade, tarffs and sanctons and emsson reducton ncentves used to explore the possble mpacts of accompanyng measures n possble UNFCCC post 2012 packages and s that used n Armngton type trade models. ΔT s temperature change n perod and H s an assumed upper bound global temperature change at whch all economc actvty ceases (say 20 o C). As ΔT approaches H, utlty goes to zero. If ΔT goes to zero, there s no welfare mpact of temperature change. Utlty over any model perod thus ncreases as temperature change falls. (1) 6 The share parameter determnes the severty of damage (n utlty terms) from any gven temperature change. We later calbrate the model to varous damage estmates from busness as usual global temperature change estmates reported by Stern (2006) and Mendelsohn (2006), and ths procedure determnes. For smplcty, we assume β s the same value across countres. Ths s a strong assumpton snce avalable evdence ndcates heterogenety n clmate change damage across countres, and we later relax ths assumpton n senstvty analyss. Temperature Change Functon Global temperature change s determned n the model by the change n carbon emssons over the perod across all countres. We frst adopt a smple sngle global temperature change 4 See, for example the World Integrated Assessment General Equlbrum Model (WIAGEM; Kemfert, 2002), the World Induced Techncal Change Hybrd (WITC H) model (Bosett et al., 2006), the Dynamc Integrated model of Clmate and the Economy (DICE; Nordhaus and Boyer, 2000), the Integrated Assessment Model for Global Clmate Change (MERGE; Manne and Rchels, 2004), the Global C hange Assessment Model (GCAM; Km et al., 2006), and the Integrated Model to Assess the Global Envronment (IMAGE; Bouwman, 2006). 5 Wetzman (2007) suggested that wrtng down preferences to adequately capture temperature change mpacts on welfare was an unsolved problem. The formulaton (1), frst used by CRW, seems a reasonable specfcaton for ths purpose. Workng Paper 2015-06

Developng Countres and The UNFCCC Process: Some Smulatons From an Armngton Extended Clmate Model functon whch apples to all regons and assume that emssons by each country equal the change n consumpton tmes country emssons ntensty (emssons/gdp) so as to allow for dfferng emssons ntenstes by country. In later senstvty analyss we relax ths strong assumpton. Defnng the emssons ntensty of country as e, we use the power functon (2) to represent global temperature change due to changes n emssons by all countres over the model perod. (2) T g( e RS ) a( e RS ) b c where ΔRS represents the change n the use (consumpton plus exports) of the own good for each country. 6 ΔRS and RC thus dffer. In the central case formulaton of the model, e s exogenous and fxed at ts 2006 base case levels. Consumpton of each country good by all countres s less than or equal to ΔRS; and ΔRS s less than or equal to the upper bound RS assocated wth the base case scenaro snce countres can choose to partcpate n emsson reductons ntatves and reduce the use of ther own good. The typcal scenaro we consder s where countres n the model can commt to emsson reductons whch are a gven percentage of ther own RS. We also conduct senstvty analyses n whch the e change over tme to reflect ncreased effcency of energy use over tme. When we consder accompanyng trade and fnance mechansms along wth emssons reductons, developng countres then have the opton of onng wth the same or dfferentally negotated percentage reducton (and also possbly recevng transfers) or not onng (and possbly facng border adustments and/or tarffs). 7 Composte Consumpton by Country We model the composte consumpton good RC as a CES functon of domestc and mported consumpton goods, smlar to that used n nested CES Armngton trade models (see Whalley (1985)). The model s thus effectvely an Armngton N good N country pure trade economy n whch the endowment s varable and temperature change enters preferences. In ths structure, a carbon reducton commtment by a sngle country mples a reducton n compostes of consumpton n all countres. Ths has both negatve and postve effects on utlty for all countres over the model perod. On the one hand, a reducton n consumpton lowers utlty for the consumng country, but on the other hand, country consumpton reductons lower global emssons and hence world temperature change, and ncrease the utlty both of the country reducng emssons and all other countres. For each country, the RC are determned by solvng the country optmzaton problems. 6 Ideally, ths power functon should have the property that there s ncreasng margnal mpact on temperature change for progressve ncreases n consumpton,.e., b > 1. However, we calbrate ths functon to estmates of temperature change of 3 o C by 2030 and 5 o C by 2050 gven n the Stern (2006) report, whch ontly mples b < 1. Workng Paper 2015-06

Developng Countres and The UNFCCC Process: Some Smulatons From an Armngton Extended Clmate Model Max 1 1 1 1 1 (, ) (( 1) ( 2) ) RC RC D M D M (3) s.t. p w D p M I p RS (4) m w where D and good respectvely, wth M, n turn, represent consumpton of the domestc and a composte mported w p and m p as ther prces, 1 as ncome, and as the substtuton elastcty 7. and 2 as the consumpton shares, I Demands for domestc consumpton goods and mported composte consumpton goods are: M D I ( ) ( ( ) ( ) ) 2 m w (1 ) m (1 ) p 1 p 2 p I ( ) ( ( ) ( ) ) 1 w w (1 ) m (1 ) p 1 p 2 p ( = 1 N) (5) ( = 1 N) (6) Where I s country ncome and s gven by sales of own good w RS p at the world prce. Unlke n a conventonal Armngton trade model, RS s endogenous and also the outcome of a dscrete choce optmzaton problem nvolvng partcpaton or non partcpaton n any proposed UNFCCC clmate change arrangement. The composton of the M s determned 8 by a thrd level of nestng n the model, and equaton (9)). Compostes of Imported Goods The CES mport compostes m p s a prce ndex of seller s prces w p (see M are modeled as compostes of mported goods from each supplyng country. Gven that each country has one good t can sell, but N-1 goods t mports, the CES composte of other goods defne the mport composte, and are the outcome of a subutlty maxmzaton problem Max 1 m1 m m m m 1 ( 1, 2,..., 1, 1,..., N ) ( ( ) ( ) ) M H R R R R R R (7) d m m s.t. p R I p M (8) where m R p s the country good mported by country, s the composte mport prce for 7 We use the same central case settngs of elastctes as Ca, Rezman and Whalley (2012) 0.5 and m 0.9 Ca et al provde. lterature based dscusson of these values, whch we later vary n senstvty analyss. Workng Paper 2015-06

Developng Countres and The UNFCCC Process: Some Smulatons From an Armngton Extended Clmate Model country, m s the consumpton share and m s the second level substtuton elastcty. I s the ncome devoted to expendtures on mports (from (6)). These CES sub-utlty maxmzatons gve: p 1 m d 1 1 m [ ( ) m p ] (9) R p M ( p ) M m m m d m d 1 m d m ( p ) ( p ) ( p ) (10) Costs of Mtgaton The fnal element n the model s the cost of mtgatng damage from clmate change through emssons reducton, or abatement costs. We use a smple mtgaton cost functon where country mtgaton costs are a constant margnal cost functon of use of own good ( RS or consumpton plus export). Stern (2006) places these costs for a 50% reducton n emssons by 2050 at 1% of GDP wth a varaton of ±3%. We use a central case estmate of 2.5% and then use senstvty ranges around ths value. Later estmates summarzed n Clarke et al (2009) report dfferences n abatement cost schedules across countres due to dfferences n energy sources and producton patterns. These are not captured here but could be modelled n senstvty analyss. The mtgaton (abatement) cost functon we use s: ( E E) MC RS E (12) 9 where MC are the mtgaton costs of country for a change n emssons gven by ( E (base case) - E (new emssons)). E E are the emssons along the BAU path and are the emssons mpled by the emssons reducton. emssons. ( E E ) E s the proportonal change n s the emsson reducton cost factor lnkng the proportonal change n emssons to own resources RS. We set analyses wth values of 0.01 and 0.04. Equlbrum n the Model Gven values of equal to 0.025 n the base case, and conduct the senstvty RS (whch mply emsson reductons), an equlbrum for the model s gven w by prces p,..., 1 p for whch global markets n all N country goods clear,.e. w N R D RS ( 1... N) (11) Workng Paper 2015-06

Developng Countres and The UNFCCC Process: Some Smulatons From an Armngton Extended Clmate Model In ths structure, when countres partcpate n a global clmate agreement, f they reduce emssons there wll be general equlbrum mplcatons for all prces and quanttes. Importantly, f there are accompanyng mechansms, tarffs used aganst exportng countres wll cause the prce of ther own good to fall gvng a terms of trade loss for the country not makng the emssons reducton. As emphaszed by CRW, ths wll, n turn, ncrease the wllngness of countres to partcpate n global emsson reductons negotatons. Transfers do not exert ths drect term of trade effect through a relatve prce nterventon, but as countres recevng transfers spend most of ther ncome on ther own good, n the calbrated Armngton structure a terms of trade effect wll come nto play through ncome effects. The RS are the values counterfactual analyses. RS n the base case. They then take on one of two values n RS captures the mpled reducton n emssons for countres partcpatng n UNFCCC emssons reducton arrangements. Alternatvely, RS for non OECD countres f they do not partcpate. RS s equal to Model Extensons We can use the model to analyze dfferent counterfactuals relatve to the BAU scenaro and for emssons reductons of varous forms we can compute counterfactual equlbra for the chosen perod. There are some experments whch we conduct wth the smulaton structure whch requre extensons to the basc model form. One s where we evaluate the mpacts of usng ntensty targets. For ths we use a smple uncertanty extenson of the model, snce n the certanty case the two nstruments are typcally equvalent n mpact. The model captures uncertanty n a smple way by consderng three alternatve growth scenaros: hgh growth, low growth and BAU growth and we compute a dfferent base case (no emsson reducton) scenaro for each. For each scenaro we frst compute utlty change and consumpton of goods by regon. We then ntroduce dfferent level and ntensty emsson targets for the varous growth scenaros. We frst treat an emsson level target case as a gven percentage reducton n use of own good n the country makng the emsson reducton, and then compute an equvalent country emsson ntensty reducton whch gves the same expected emssons reducton under the emssons level target, gven the BAU output of each country. We can then compute the model utlty change under hgh, low and BAU growth scenaros respectvely for each of the emssons targets, and compare expected utlty for hgh and low growth scenaros across the two targets to assess the mpact of usng ntensty targets. Ths extenson allows us to analyze the relatve country attractveness of ntensty versus level targets for emssons reducton, gven that n the certanty case they are equvalent. 10 We also ncorporate trade polces and transfers to evaluate the possble mpacts of accompanyng trade and/or fnance mechansms and modfy the model approprately. For ths purpose, the model s extended to capture border tax adustments, tarffs, and fnancal transfers as penaltes or nducements to partcpate n negotatons. The sze of transfers, ether as a percentage of recpent country GDP or of donatng country GDP, or as an amount n $ transferred from developed countres s treated as exogenous, but can be vared n Workng Paper 2015-06

Developng Countres and The UNFCCC Process: Some Smulatons From an Armngton Extended Clmate Model counterfactual analyses. Tarffs and border adustments apply to the prces of goods crossng natonal borders and generate revenues. Trade mbalances (ncludng transfers) are exogenous n the model. III. Data and Model Calbraton We construct a BAU growth profle usng forward proectons of 2006 data, and model calbraton to ths profle determnes key model parameters. In ths, we use varyng estmates of assocated damage over the ranges reported by Stern (2006) and Mendelsohn (2006) and abatement cost estmates as n Stern (2006). We use an 8 country groupng, of Brazl, Russa, Inda, Chna, US, EU, Japan, and the Rest of the World (ROW). We use calbraton to a temperature change functon for prospectve changes n temperature under busness as usual scenaros, out to 2020, 2030 and 2050. These correspond to possble commtment perods n a UNFCCC post 2012 arrangement. Data Sources We use GDP growth as the measure of potental change n consumpton by country over the perod of analyss. Because of our analyss of ntensty as well as level targets we use three growth scenaros: hgh, BAU and low growth rates. We frst assume that under the dfferent (BAU, hgh, low) growth scenaros, country growth rates n the perod 2006-2050 reman unchanged over the whole perod. Data for 2012 are forward proected based on data for 2006 and provde the reference base case. We use averaged data between 2000 and 2006 as country growth rates. We have three components n our BAU data for each growth scenaro: proected base case data for 2012, cumulatve data for 2020, 2030, 2050 gven hgh, BAU and low growth rates, and cumulatve data over the perod relatve to the base year for the same three growth scenaros. 11 We assume Chna, Inda, Russa, Brazl, USA, EU, Japan and the Rest of the World (Row) have BAU growth rates of 0.09, 0.07, 0.07, 0.032 0.026, 0.020, 0.17, and 0.30 respectvely, gven by average growth rates between 2000 and 2006 (data from World Bank webste). We then use the BAU growth path data to calbrate the temperature change functon usng estmated BAU temperature change over the perod drawng on key lterature sources, ncludng Stern (2006) and Mendelsohn (2006). Ths mples that n hgh growth scenaros emssons are larger and also temperature change s hgher. Table 1 reports the 2006 output and emssons data used n our proectons, and the growth rates used. Calbraton of Model Parameters We use data on consumpton and trade for both OECD and larger developng (BRIC) economes and along wth country growth profles to yeld busness as usual (BAU) scenaros under varous damage and temperature change assumptons. Preferences towards goods and temperature change are determned for each country usng alternatve damage estmates from the same sources. We undertake numercal nvestgaton wth our analytcal structure usng calbraton to determne model parameters values followed by counterfactual analyses of varous forms. The base data are for 3 dfferent perods 2012-2020, 2012-2030 and 2012-2050 wth assumed yearly Workng Paper 2015-06

Developng Countres and The UNFCCC Process: Some Smulatons From an Armngton Extended Clmate Model growth rates over the perod. Table 1: BAU Total Output, Emssons, and Emssons Intensty and Growth Rates Proected to 2020, 2030 and 2050 Chna Inda Russa Brazl U. S E.U. Japan ROW Output n 2006, trll$ 2.65 0.91 0.99 1.07 13.16 10.64 4.37 14.68 Emsson ntensty 2006 2.22 2.01 2.58 0.50 0.52 0.29 0.27 0.98 Emsson n 2006, bmt 5.88 1.83 2.54 0.53 6.81 3.13 1.19 14.37 Cumulatve 1900-2012, bmt Proected emssons emssons from 2006 to 2012, bmt Proected emssons from 2012 to 2020, bmt Proected emssons from 2012 to 2030, bmt Proected emssons from 2012 to 2050, bmt 165.06 45.67 117.89 14.41 385.11 354.93 54.88 321.50 62.61 16.75 22.72 3.91 51.30 23.43 8.67 111.00 176.94 41.49 56.35 8.66 108.44 47.90 17.57 243.83 731.944 152.110 206.782 28.380 341.927 147.587 54.113 798.795 6111.658 946.295 1289.959 132.775 1454.634 606.404 230.369 3848.604 Note: The hgh/low growth specfcaton s where all rates are averages of country growth rates above/below average BAU growth rates for 2000-2006. We frst dscuss the calbraton of preference parameters. Accordng to the Stern Revew (2006), Mendelssohn (2006) and other lterature, damage costs from emssons on BAU paths ranges from 1 to 20% of GDP out to 2050. We treat damage from clmate change n the model as a utlty change of the same proporton over the same tme perod and use t to calbrate the preference parameters n the model. 12 Wthout temperature change, the utlty functon s: U * And wth damage we have: RC (13) H T * U / U ( ) (14) H Wth temperature change, there wll thus be a utlty loss from damage. We can thus calbrate usng equaton (14) above for gven dfferent values of H. For llustratve purposes, n Table 2 we report calbrated values for a tme perod of 50 years as the base case. In our smulaton analyss, we use H=10 as the base case, and perform senstvty analyss wth H=20 and H=30. The temperature change functon (2) s wrtten as a functon of emsson changes over the same perod, and we treat t as a power functon of total emsson (not output) change for the world. Workng Paper 2015-06

Developng Countres and The UNFCCC Process: Some Smulatons From an Armngton Extended Clmate Model Based on the fndngs from Stern (2006), we assume the BAU path of emssons wll lead to about 3-degree temperature ncreases around the year 2030, and near 5 degrees by around 2050, although as we note earler these exceed IPCC ranges. For smplcty, we assume that zero growth n the global economy wll lead to no temperature change,.e., c = 0. Wth growth rates and emsson ntenstes for each country for the BAU growth scenaros, we can then calbrate the parameters a and b. We have data for the year 2006 and proectons of emssons and output data for 2030 and 2050. We choose 2006 as the base year, and assume that 25 years later that s by 2030 the global average temperature wll ncrease by 3 degrees and 5 degrees by 2050. We assume that the BAU path mples output growth for each country comparable to that of 2000-2006, whle emsson ntenstes are unchanged from 2006. Table 2 reports the calbrated values of a and b. 8 We are also able to ncorporate nto the model calbraton procedures autonomous (exogenous) mprovements n energy effcency (ntensty). H Table 2: Calbrated Model Parameters n preferences a, b n temperature change functon BAU Damage cost assumed assumng 50-year tme horzon 10% 0.152 10 20% 0.322 50% 1.000 10% 0.366 2030 T 2050 T 3 5 2030 T T 2050 1.5 3 13 20 30 20% 0.776 25% 1.000 10% 0.578 16.7% 1.000 a= 0.044 b=0.287 a= 0.005 b=0.389 Emssons Reductons Table 3 reports the percentage emssons reductons over the commtment perod 2012-2020 mpled by dfferent allocaton formulas, as well as proected 2020 emssons. Gven Chna s hgh growth, Chna accounts for over 50% of global emssons by 2020, and so how dfferent emssons reductons affect Chna s crtcal. Large dfferences occur usng cumulatve rather than annual emssons, wth only small dfferences wth a consumpton base for emssons. The choce of base data of 1990 over 2012 also makes a large dfference. In the case of cumulatve emssons based reductons, we use an upper bound on emsson reductons of 80%. Table 3: Percentage Emsson Reductons over Perod 2012-2020 Impled by Dfferent Country Allocatons of a Global 30% Emssons Reducton 8 Gven the Stern estmates, b < 1 whch mples dmnshng not ncreasng margnal mpacts of growng consumpton on temperature change. Workng Paper 2015-06

Developng Countres and The UNFCCC Process: Some Smulatons From an Armngton Extended Clmate Model Chna Inda Russa Brazl US EU Japan Row 30% proportonal reducton n emssons for each country by 2020; Usng base data of proected 30% 30% 30% 30% 30% 30% 30% 30% 2012 emssons 30% proportonal reducton n emssons for each country by 2020; Usng base data of 1990 45.6% 40.3% 25.5% 19.7% 14.9% 12.7% 10.2% 17.4% emssons 30% proportonal reducton n emssons globally allocated usng cumulatve emssons 1900-2012 15% 17% 33% 26% 55% 80% 49% 21% (80% upper bound) 30% proportonal reducton n emssons for each country by 2020 usng proected 2012 base data and usng consumpton rather than producton 28% 32% 26% 29% 32% 31% 30% 28% 30% reducton globally by 2020 usng proected 2012 1% 29% 29% 29% 29% 33.0% 33.0% 33.0% 29% base data but wth 3% 27% 27% 27% 27% 39.1% 39.1% 39.1% 27% developng country targets 1%, 3%, 5% lower for non OECD 5% 25% 25% 25% 25% 45.2% 45.2% 45.2% 25% IV. Model Results We use the calbrated model to evaluate the mpacts of alternatve ont emssons reductons and possble accompanyng actvty on border adustment and transfers. We frst consder cases nvolvng all countres under the dfferent allocatons of global reductons set out n Table 3. We then later consder accompanyng mechansms, ncludng border tax adustments and/or fnancal transfers beng used n whch we assume there s partcpaton of the OECD countres but wth partcpaton of developng countres lnked to possble trade sanctons. Ths enables us to assess how large these have to be to nduce partcpaton. 14 We frst report results from usng the modelng framework set out above to make calculatons of the welfare mpacts of emssons reductons n Hcksan money metrc form (n $bllon over the commtment perod) by country. These are reported n Table 4 for a 30% equ-proportonal reducton by all countres by 2020 and a 30% reducton by 2030. In these results, gven the damage cost estmate of 5% by 2050 used (from Stern (2006)) all countres lose from partcpaton n clmate arrangements for all three perods out to 2020, 2030 and 2050. Ths ndcates nfeasblty n concludng a negotaton on ths bass, even wth sde payments. For the reductons out to 2020 the largest losses occur for the US and ROW, followed by the EU and Japan. For 50% reductons by 2030, losses ncrease for Chna due to ther hgher growth rate, but fall for the US and the EU due to restraned growng emssons n Chna and Inda. Table 5 then reports welfare mpacts by country for smlar global 30% proportonal reductons n emssons by 2020, but wth changed model assumptons. We frst change the assumed BAU Workng Paper 2015-06

Developng Countres and The UNFCCC Process: Some Smulatons From an Armngton Extended Clmate Model damage cost estmates used n model calbraton. If we lower damage cost estmates to 5% of GDP from 10% of GDP, country losses ncrease as consumpton losses reman, but benefts of slowed global warmng fall. If we ncrease clmate change damage estmates used n calbraton to 20% of GDP, gans accrue to all countres as the benefts of slowed global warmng ncrease. In ths case, an nternatonal negotaton can seemngly conclude. If we lower assumed temperature change, the benefts of slowed global warmng fall. If we dscount GDP growth at 1% and 0.5% for non OECD and OECD respectvely, losses fall as the sze of economes over the commtment perod shrnks. Usng PPP measures for GDP ncreases losses n Chna and Inda as ther economes are proportonally larger. Table 4: Welfare Impacts by Country of Equproportonal Reductons n Emssons for each Country Usng Central Case Model Specfcaton ($ bll, Money Metrc Hcksan measures) Chna Inda Russa Brazl US EU Japan Row 30% reducton by 2020-152.928-43.938-56.150-40.408-405.843-299.403-128.932-515.100 50% reducton by 2030-272.330-51.391-107.040-46.734-32.769-145.795-120.405-648.884 Table 5: Welfare Impacts by Country of 30% Global Proportonal Cuts n Emssons for Each Country by 2030 Usng 2012 Base Data Under Varyng Model Specfcatons ($bll; Money Metrc Hcksan Measures) 15 Chna Inda Russa Brazl US EU Japan Row A. Central case model specfcaton n Table 4-152.928-43.938-56.150-40.408-405.843-299.403-128.932-515.100 B. Varaton model specfcaton Change assumed BAU damage cost estmated n model calbraton of temperature change functon out to 2050 5% -231.212-71.160-79.997-61.413-662.196-480.823-200.078-778.808 20% 8.692 12.252-6.906 2.955 123.307 75.084 17.940 29.330 Change assumed temperature 2030 change to T 1.5 2050, T 3-226.263-69.405-78.525-60.083-645.634-469.148-195.543-762.137 Wth dscountng of GDP at 1% for Non OECD and 0.5% for OECD -141.913-41.051-50.599-37.424-398.992-293.014-125.524-500.368 Workng Paper 2015-06

Developng Countres and The UNFCCC Process: Some Smulatons From an Armngton Extended Clmate Model Wth use of PPP measures of GDP n 2006-260.614-89.514-71.489-46.078-323.776-205.939-100.658-569.233 In Table 6 we report the mpacts of alternatve country allocatons of emssons reductons whch keep the same global total of a 30% global emssons reducton by 2020. These results ndcate sharp changes by country n mpacts as dfferent allocatons are used. Changes n the base date to 1990 nearly double the losses of both Chna and Inda as the hgh growth economes, whle losses of slower growng US and EU fall sharply. Usng hstorcal emssons over the perod 1900-2012 makes a large dfference to both Inda and Chna whose losses nearly dsappear. Losses to the US and the EU both nearly double. Changng to a consumpton bass from producton makes relatvely lttle dfference to country mpacts. The use of dfferental cuts for OECD and non OECD s progressvely more advantageous to Chna, Inda, Brazl and Russa and dsadvantageous to the US and the EU. In Table 7 we report the senstvty of model results on welfare mpacts by country for a 30% proportonal cut n emssons. We vary alternatve sets of key model parameter values. Varyng trade elastctes for all countres together has lttle mpact on model results. Varyng damage costs, as above, has larger mpacts and wth a 20% damage cost estmate losses become gans. If damage costs are hgher n non OECD countres, they beneft more from emsson reductons. Varyng the temperature change upper bound has lttle mpact, as does varyng temperature change dfferentally between Chna and Inda, and other countres n the model. Table 6: Welfare Impacts by Country of Alternatve Globally Equvalent Emssons Reductons by 2020 Relatve to a 30% Proportonal Cut by Country Usng 2012 Proectons as Base Data ($ bll, Money Metrc Hcksan Measures) Chna Inda Russa Brazl US EU Japan Row 16 Central case model specfcaton -152.928-43.938-56.150-40.408-405.843-299.403-128.932-515.100 Change base date to 1990-276.45-95.63-50.016-22.851-219.295-172.099-79.867-277.864 30% global cut allocated usng 1900-2012 emssons by country Change to consumpton bass from producton embedment -7.15-1.68-66.45-30.20-354.02-278.52-190.56-315.50-146.397-46.733-50.413-39.190-441.1-312.773-128.134-493.710 Use of dfferental cuts OECD/ non OECD (non OECD preference) 1% -143.58-40.64-53.24-37.86-497.30-364.78-155.17-483.06 2% -124.89-34.04-47.43-32.75-680.21-495.52-207.63-418.97 3% -106.20-27.44-41.62-27.64-863.13-626.27-260.10-354.90 Workng Paper 2015-06

Developng Countres and The UNFCCC Process: Some Smulatons From an Armngton Extended Clmate Model Table 7: Senstvty of Country Welfare Impacts for 30% Proportonal Reducton by 2020 to Key Parameter Values ($ bll, Money metrc Hcksan measures) Chna Inda Russa Brazl US EU Japan Row Changng trade elastctes, m Varyng damage cost Varyng temperature change upper bound (H) Varyng Temperature change across country ( ) T 0.5 0.5-155.210-44.786-58.444-40.836-422.570-301.116-129.300-518.138 m 0.5 0.9 m (BAU) -152.928-43.938-56.150-40.408-405.843-299.403-128.932-515.100 1.2-102.342-30.075-36.292-27.841-270.675-189.613-86.023-318.845 0.9 m 5% -231.212-71.160-79.997-61.413-662.196-480.823-200.078-778.808 10% (BAU) 5% for OECD country, and 10% for Non- OECD country -152.928-43.938-56.150-40.408-405.843-299.403-128.932-515.100-73.457-16.307-31.937-19.086-405.843-299.403-128.932-247.395 20% 8.692 12.252-6.906 2.955 123.307 75.084 17.940 29.330 10 (BAU) -152.928-43.938-56.150-40.408-405.843-299.403-128.932-515.100 20-142.040-40.173-52.811-37.489-370.405-274.297-119.060-478.425 30-139.488-39.293-52.027-36.804-362.119-268.424-116.748-469.827 If Chna and Inda T 0.5 If Chna and Inda T 1-141.780-40.131-57.194-41.306-416.584-307.038-131.954-526.391-129.493-35.931-58.229-42.196-427.232-314.600-134.948-537.580 17 Workng Paper 2015-06

Developng Countres and The UNFCCC Process: Some Smulatons From an Armngton Extended Clmate Model Table 8 reports the mpacts of 30% proportonal cuts by 2020 and 50% proportonal cuts by 2030 on country GDP and country mports. The mpacts on country GDP also reflect the cost of emssons reductons. Under a 30% equ-proportonal cut the percent reductons n consumpton are smlar. Changes n country mports mrror these falls snce n the model countres only trade a sngle good, and so relatve prce effects of carbon prcng on energy ntensve and non ntensve goods are excluded. Table 9 reports results for the welfare mpacts over the perod 2012 to 2020 f alternatve accompanyng funds of varyng szes accompany the emssons reductons. These funds are assumed to be transferred over the perod 2012 to 2020. Wth transfers of approxmately $150 bllon per year (totalng $1.2 trllon over the eght-year commtment perod) losses for Inda dsappear and for Chna, Russa and Brazl losses become neglgble. Losses to the US, EU and Japan who fnance the transfers double. Even larger redstrbutons occur when transfers of $200 bllon per year occur over the same perod. These results thus hghlght the crtcal role that can be played by transfers of resources n facltatng developng country partcpaton n the post Kyoto process (Sprngmann, 2012). Table 10 reports country welfare mpacts for 30% equ-proportonal emssons reductons n the OECD beng accompaned by alternatve trade related mechansms nvolvng tarffs and export rebates by the OECD and non OECD countres. The frst row reports welfare mpacts for a case where emssons reductons are lmted to the US, EU, Japan and the ROW. In these cases Chna, Inda, Russa and Brazl all beneft from slowed clmate change. These gans then fall as varous measures of ncreasng severty are appled aganst them. 20% and 30% tarffs nduce Chna to partcpate by nflctng net losses; 10% and 30% tarffs play the same role for Brazl. Impacts of border adustments are less pronounced due to the export subsdy rebates nvolved. 18 Table 8: Impacts on Country GDP and Trade of 30% and 50% Equproportonal Cuts by 2020 and 2030 A: 30% proportonal cut by 2020 Chna Inda Russa Brazl US EU Japan Row % change n country GDP -0.79% -0.72% -0.88% -0.79% -0.71% -0.73% -0.76% -0.79% % change n country mports -0.82% -0.59% -1.06% -0.90% -0.49% -0.67% -0.80% -0.84% B: 50% proportonal cut by 2030 Chna Inda Russa Brazl US EU Japan Row % change n country GDP -1.31% -1.13% -1.46% -1.36% -1.05% -1.18% -1.28% -1.32% % change n country mports -1.32% -1.18% -1.41% -1.30% -1.14% -1.19% -1.27% -1.32% Table 9: Welfare Impacts Over the Perod 2012 to 2020 of Alternatve Accompanyng Workng Paper 2015-06

Developng Countres and The UNFCCC Process: Some Smulatons From an Armngton Extended Clmate Model Transfers of Funds to Accompany Equal Country Proportonal Emssons Reductons of 30% by all Countres ($bll, Money Metrc Hcksan measures) Welfare mpact n $ bll of varous accompanyng mechansms to 30% proportonal emssons reducton by 2020 by country 1.Central case wth no accompanyng mechansms 2.Dstrbutng $ 1.2 trllons of accompanyng funds to Non OECD proportonal to GDP pad for by OECD proportonal to GDP 3.Dstrbutng $ 1.6 trllons of accompanyng funds to Non OECD proportonal to GDP pad for by OECD Chna Inda Russa Brazl US EU Japan Row -152.928-43.938-56.150-40.408-405.843-299.403-128.932-515.100-3.381 6.720-8.755-0.103-962.618-700.992-444.474-393.960 71.690 32.154 15.012 20.100-1239.748-901.038-601.163-333.343 19 Table 10: Welfare Impacts of Alternatve Trade Related Mechansms to Accompany Proportonal Emssons Reducton of 30% by 2020 Only by OECD Wth Non Partcpaton by Welfare mpact n $ trl 1.Central case model specfcaton wth partcpaton only by US, EU, Japan, Row n Non OECD ($ bll, Money Metrc Hcksan measures) Chna Inda Russa Brazl US EU Japan Row 77.774 29.827 19.980 21.093-624.785-449.682-184.689-710.842 Workng Paper 2015-06

Developng Countres and The UNFCCC Process: Some Smulatons From an Armngton Extended Clmate Model 30% reducton by 2020 2.Non partcpaton by non OECD plus 10% border adustment n OECD 3.Non partcpaton by non OECD plus 20% border adustment n OECD 4.Non partcpaton by non OECD plus 50% border adustment n OECD 5. Non partcpaton by non OECD plus 10% tarff n OECD 6. Non partcpaton by non OECD plus 20% tarff n OECD 7. Non partcpaton by non OECD plus 210 % tarff n OECD 29.335 32.543-18.567 13.598-816.014-511.382-187.734-599.481-32.026 32.615-60.501 4.247-1020.428-601.217-200.955-539.014-172.57 25.638-118.109-23.926-1855.346-1040.474-305.078-144.382 31.962 38.378-25.296 10.927-853.046-530.503-181.341-604.664-11.869 46.483-68.908 1.168-1086.031-635.944-189.945-549.113-164.73 64.502-142.869-30.323-3134.875-2296.720-658.162-142.076 V. Concludng Remarks 20 Ths paper presents numercal smulaton results for a mult-country Armngton clmate extended general equlbrum model whch captures the beneft sde of clmate change n preferences and allows for the analyss of country welfare mpacts of global carbon emsson reducton arrangements. The model embodes an assumpton of unform percentage clmate change and damages by regon, whch we relax n later senstvty analyss. These results are taken as possbltes for the post Kyoto/post 2012 perod after the Bal roadmap UNFCCC negotaton process whch concluded n Durban n 2011. The man focus s mpacts on large rapdly growng developng countres (Inda, Chna, Brazl, etc.) In the model goods consumpton and temperature change both enter utlty change functons and countres ontly beneft from the emssons reductons of others. Trade effects enter through the heterogenety of country goods, and consumpton reducng emssons reductons have terms of trade effects. The model s calbrated to alternatve Busness as Usual (BAU) scenaros out to 2020, 2030 and also 2050. Counterfactual exercses are then conducted around these varous BAU scenaros. Results show all countres as losng f the damage estmates used n calbraton from clmate change are less than 10%. Ths suggests a possble nablty to conclude a ontly agreed negotaton, unless sanctons of some form are used or damage costs are consderably hgher. Large changes n country mpacts as alternatve reducton arrangements are consdered. Workng Paper 2015-06