DISCUSSION PAPER. Combining Rebates with Carbon Taxes. Optimal Strategies for Coping with Emissions Leakage and Tax Interactions

Transcription

1 DISCUSSION PAPER May 2009 RFF DP Combnng Rebates wth Carbon Taxes Optmal Strateges for Copng wth Emssons Leakage and Tax Interactons Carolyn Fscher and Alan K. Fox 1616 P St. NW Washngton, DC

2 Combnng Rebates wth Carbon Taxes: Optmal Strateges for Copng wth Emssons Leakage and Tax Interactons Carolyn Fscher and Alan K. Fox Abstract Emssons regulatons lke carbon prcng rase the prce of covered sector goods and thus can nteract wth and exacerbate other preexstng dstortons n the economy. One such dstorton s labor taxes. Another s emssons leakage due to the lack of comparable emssons prcng abroad or among other emttng sectors at home. A potental response s to combne the emssons tax wth a rebate to producton to mtgate the prce ncreases. We use an optmal tax framework to solve for the optmal emssons tax and output rebate, gven these dstortons. We then employ a multsector computable general equlbrum model based on the GTAP framework to smulate the effects of a $50 per-ton carbon tax on the major emssons-ntensve sectors n the U.S. economy and estmate optmal rebates by sector. Key Words: carbon tax, tax nteracton, carbon leakage JEL Classfcaton Numbers: Q2, Q43, H2, D58, D Resources for the Future. All rghts reserved. No porton of ths paper may be reproduced wthout permsson of the authors. Dscusson papers are research materals crculated by ther authors for purposes of nformaton and dscusson. They have not necessarly undergone formal peer revew.

3 Contents Introducton... 1 Theoretcal Background... 4 Specfcatons... 5 Decentralzed Markets... 5 Planner Problem... 7 Numercal Model Model Descrpton Polcy Scenaros Results Concluson References... 19

4 Fscher and Fox Combnng Rebates wth Carbon Taxes: Optmal Strateges for Copng wth Emssons Leakage and Tax Interactons Carolyn Fscher and Alan K. Fox Introducton Clmate change s a global problem, but most of the current efforts to ren n greenhouse gas (GHG) emssons through carbon prcng are unlateral, natonal, and even subnatonal n orgn. The European Unon has led wth the mplementaton of ts Emssons Tradng System (ETS). In the Unted States and Canada, n the absence of federal programs, the Northeast s Regonal Greenhouse Gas Intatve (RGGI), Calforna s Global Warmng Solutons Act (AB32), Brtsh Columba s carbon tax, and the Western Clmate Intatve have emerged. However, a serous concern wth these ndvdual ntatves s that ncreases n domestc prces for energy, electrcty, and thereby products may lead to a shft n producton (and even generaton) to unregulated regons. Not only s lost compettveness a concern for polcymakers, but so s the resultng carbon leakage, whch threatens to partally undo the efforts to reduce emssons. These concerns extend to the federal level n the Unted States and for other developed countres that are postonng themselves to take on sgnfcant carbon prcng polces. The Kyoto Protocol, the nternatonal agreement to reduce emssons of the greenhouse gases that cause global clmate change, also recognzes a prncple of common but dfferentated responsbltes, mplyng that comparable carbon constrants wll not be expected from developng countres for some tme. Leakage to unregulated enttes s not strctly an extraterrtoral phenomenon. If the envronmental polcy s narrowly appled to a subset of emttng sectors, producton and emssons (and thereby leakage) can shft to uncovered sectors wthn that same regon. The reasons for ncomplete coverage of a carbon prcng polcy may be techncal or admnstratve or arse from other concerns; however, the strategy s common. For example, the EU ETS apples Carolyn Fscher (fscher@rff.org) s a Senor Fellow at Resources for the Future, Washngton, DC. Alan Fox (Alan.Fox@ustc.gov) s wth the Offce of Economcs of the Unted States Internatonal Trade Commsson (ITC), Washngton, DC. The opnons and deas expressed heren are solely those of the authors and do not represent the vews of RFF or the U.S. ITC or any of ts ndvdual commssoners. Ths lne of research has beneftted from the fnancal support of EPA-STAR and of the Mstra Foundaton ENTWINED program, whch are gratefully acknowledged, as s the hosptalty and support of the Centre for Advanced Study (CAS) at the Norwegan Academy of Scence and Letters n Oslo. 1

5 Fscher and Fox only to a certan set of energy-ntensve sectors representng 46 percent of European emssons, 1 and RGGI apples only to electrcty generaton. Bernard et al. (2007) show that such neffcent reallocatons can be prevented by taxng the unregulated sectors accordng to the emssons emboded n ther output. However, when unregulated sectors resde abroad, such a tax may be prohbted by World Trade Organzaton rules and, n any case, would only apply to mports, not to all the unregulated producton. Among unregulated domestc sectors, a tax on emboded emssons would be as dffcult to mplement as a full downstream emssons tradng program. Barrng such a tax, Bernard et al. show that the next best polcy s to subsdze the output of the regulated sectors. The optmal subsdy then reflects the value of the emssons crowded out by addtonal output n that regulated sector. In practce, such a subsdy could be mplemented n several ways. For example, all or part of the emssons tax revenues could be rebated to partcpants accordng to ther producton, as n the case of the Swedsh NO x tax (see Isaksson and Sterner 2006). Wthn an emssons capand-trade program, the subsdy can be mplemented by updatng the allocaton of permts to frms wthn the affected sectors based on ther output (whch we wll refer to as output-based allocaton, or OBA). The value of addtonal permts represents an ncentve to produce more, offsettng part of the prce ncrease nduced by the emssons regulaton tself. In both the tax and the cap cases, when appled n a multsector settng, the rebate or allocaton can be based on a performance metrc (such as the emssons ntensty correspondng to an average or best avalable technology) multpled by an ndcator of output. A smlar way to combne emssons prcng wth a rebate s wth tradable performance standards; n ths case, each sector must meet an average emssons requrement. The effect s n theory dentcal: each frm must surrender permts accordng to ts emssons and receves an allocaton accordng to ts output. However, n practce t s dffcult to set performance standards such that margnal cost equalzaton s met, unless the permts are also tradable across sectors, and t s more dffcult to ensure that a partcular emssons target s met. These mechansms have been of nterest to researchers concerned about preexstng tax dstortons. Labor taxes dstort the consumpton-lesure tradeoff, and envronmental regulaton that further rases consumer prces exacerbates those costs (a collecton of the lterature s avalable n Goulder 2002). Snce the output subsdy emboded n performance standards lmts 1 (accessed 03/09/09). 2

6 Fscher and Fox those prce ncreases, these mechansms can outperform a system of grandfathered emssons permts (Goulder et al. 1999; Parry and Wllams 1999; Fullerton and Metcalf 2001). In an nternatonal context, trade dstortons can be as mportant as tax dstortons: although all domestc sectors are covered by the emssons tradng program, trade partners are not covered, whch allows carbon to leak as producton shfts to unregulated producers. Fscher and Fox (2007) consder the effects of dfferent allocaton mechansms, ncludng outputbased allocaton, on the effcency and dstrbutonal effects of a comprehensve carbon emssons tradng program n the Unted States, when both tax and trade dstortons are taken nto account. They fnd that the rebates mplct n OBA mtgate tax nteractons, whch can lead to hgher welfare than grandfatherng. However, the rule for determnng the sector allotments, whch determnes the effectve rebate, matters. OBA wth sectoral dstrbutons based on value added generates mplct rebates smlar to a broad-based tax reducton, performng nearly lke auctonng wth revenue recyclng, whch generates the hghest welfare. OBA based on hstorcal emssons supports the output of more pollutng ndustres, whch more effectvely counteracts carbon leakage, but s more costly n welfare terms. However, from a welfare perspectve, when the domestc coverage of the cap-and-trade program s complete, OBA does not domnate full auctonng wth revenue recyclng. We extend ths research by focusng smultaneously on the three major polcy and market mperfectons that can justfy support for output rebates n combnaton wth an emssons prcng program: (1) lack of comparable polces abroad, (2) ncomplete regulatory coverage at home, and (3) tax nteracton. We frst extend the theory of Bernard et al. (2007) to ncorporate tax nteractons, revealng that wth Pgouvan taxaton of the emssons externalty, the optmal rebate to regulated goods reflects not only the margnal beneft of avoded leakage but also a component to adjust for labor tax nteractons. Subsequently, we extend the numercal analyss of Fscher and Fox (2007) to consder the case of more lmted domestc coverage n a carbon prcng scheme for the Unted States, along the lnes of recent proposals for a domestc cap-andtrade program for CO 2 and of trends n sectoral coverage that have emerged n the EU ETS. In keepng wth the theoretcal secton, we model a carbon tax polcy and solve for optmal rebates for each of the covered sectors. 2 The analyss s carred out n a modfed GTAP framework augmented wth a labor-lesure choce to acknowledge the tradeoff between usng 2 Fscher and Fox (2009) explore a broader range of allocaton optons n the context of a cap-and-trade program. 3

7 Fscher and Fox carbon tax revenues to offset labor taxes versus grantng rebates. The results help dentfy lkely sectors and mechansms for mnmzng the costs of broader emssons reductons when consderng the nteracton of one country s emssons polces wth unregulated emssons and labor market dstortons n other countres. However, they also llustrate the potental mportance of other economc dstortons that are present n nternatonal trade but not consdered n theoretcal models. Theoretcal Background To understand the optmal tax problem, we expand upon the analyss of Bernard et al. (2007) to add a labor tax and revenue constrant. We wll not dstngush between global and domestc leakage n ths secton. Consder a smple economy wth two sectors (one wth emssons regulaton and one wthout), two goods, and two factors of producton (labor and emssons). Labor s consdered the thrd good and, wthout loss of generalty, wll be taken as the numérare (w = 1). The followng lst summarzes our notaton: Quanttes Q 1 = Producton n the regulated sector Q 2 = Producton n the unregulated sector L 1 = Labor demand n the regulated sector L 2 = Labor demand n the unregulated sector L = Labor supply l = Lesure = 1 L C 1 = Demand for good produced n the regulated sector C 2 = Demand for good produced n the unregulated sector E 1 = Emssons of pollutant n the regulated sector E 2 = Emssons of pollutant n the unregulated sector Prces p 1 = Consumer prces n the regulated sector p 2 = Consumer prces n the unregulated sector q 1 = Producer prces n the regulated sector q 2 = Producer prces n the unregulated sector w = Labor wage (numérare) τ 1 = Tax on emssons n the regulated sector t 1 = Tax on regulated commodty = Tax on labor t L 4

8 Fscher and Fox Specfcatons The household sector conssts of a representatve consumer, allowng us to avod consderaton of nternatonal equty n the analyss. Utlty s a functon of consumpton of both goods and lesure: U = U C, C,1 ). ( 1 2 L Households also suffer dsutlty as a functon of total emssons: D = D( E 1 + E2 ). The welfare functon, then, s the dfference between consumpton utlty and emssons dsutlty, whch we have assumed to be separable: W = U D = U( C, C,1 L) D( E + E ) Producton n each sector ( = 1, 2) s a functon of labor and emssons: Q = f L, E ). ( Equvalently, labor n each sector can be specfed as a functon of output and emssons: L = L Q, E ). ( Fnally, a government revenue constrant requres that tl+ tq+ τ E+ τ E = G L We assume that the good produced by the unregulated sector s the untaxed commodty. Decentralzed Markets It s well understood that f the socal planner could choose quanttes of output and emssons drectly, the margnal damage of all emssons would be equalzed wth the cost of reducng them and the margnal value of a good s consumpton would equal ts socal margnal cost, nclusve of the externalty. In the decentralzed problem, the planner uses taxes to nfluence market prces and acheve ths outcome. The maxmzaton problems of the consumer and producers form the constrants for the planner, along wth any regulatory constrant. Consumer Problem Takng polluton externaltes as gven, the representatve household maxmzes utlty wth respect to consumpton and lesure, U ( C1, C2,1 L) 5

9 Fscher and Fox subject to a budget constrant: ( λ ) p1 C1 + p2c2 (1 tl ) L = 0. From the consumer problem, we obtan U U U ( C ) = pλ; ( C ) = p λ; ( L) = λ(1 tl). C C l Producer Problems The representatve frm n each sector chooses output and emssons to maxmze profts: π = q Q L ( Q, E ) τ E, from whch we obtan L L ( Q) q = ; ( E) = τ. Q E The frst expresson mples that the output prce equals the margnal cost (or, wth some rearrangng, that the value of the margnal product of labor equals the wage rate). The second means that the labor cost savngs from usng more emssons just equal the tax. Government Revenue The government revenue constrant mples that any shortfall from the consumpton and emssons taxes must be made up by a tax on labor: Market Equlbrum G = tc + τ E + t L L In equlbrum, we have C = Q; C = Q ; L( Q, E ) + L ( Q, E ) + G= L and q = p t, and the revenue constrant s met. Wth well-behaved utlty and producton functons, consumpton of each good s decreasng n ts own costs, whch nclude output and emssons taxes, so dc / dt < 0 (and dc / dτ < 0). Let us defne the goods as substtutes f dc / dt > 0 and complements f dc / dt < 0. These cross-prce effects depend not only on the sgns of the cross-partals n the j j 6

10 Fscher and Fox utlty functon but also on the general equlbrum. Overall labor supply s ncreasng n the margnal utlty of consumpton. Planner Problem The queston at hand s what happens when the polcymaker can nether regulate polluton n Sector 2 nor tax the output of that sector. However, the regulated sector can be taxed or subsdzed as well as regulated. The socal planner maxmzes welfare, W, wth respect to t 1, τ 1, subject to the aforementoned constrants. The essence of the optmal tax problem can be seen by totally dfferentatng welfare: U U U D dw = dc + dc dl de + de C C l E ( ) Usng the frst-order condtons from the consumer problem, ths equaton smplfes to dw D = p dc + p dc (1 t ) dl de + de λ λ ( ) L 1 2 Furthermore, dl = dl1+ dl2and totally dfferentatng the producton functon, we get. Usng the producer frst-order condtons and market equlbrum condtons, ths mples dl = q dc τ de. Substtutng, we get the margnal welfare mpacts of the dfferent polcy levers (relatve to the margnal utlty of consumpton): dw D D = tdc 1 1 τ1 de1 de2 + tldl λ λ λ We can further smplfy by notng that output taxes change emssons by crowdng out (or n) output. Thus, de / dt = m dc / dt for all {,j}, where m s the margnal emssons rate. j j Smlarly, the mpact on emssons n one sector of a change n the emssons tax n the other sector depends on crowdng out: de / dτ j = mdc / dτ j, for j. Note that snce the emssons tax also affects the emssons ntensty, de1/ dτ1 = m1dc1/ dτ1+ C1dm1/ dτ1. Thus, 1 dw dc D t m dc D = τ m dc + t dl λ dt dt λ dt λ dt dt L dw dc D de D dc dl = t τ m + t λ dτ dτ λ dτ λ dτ dτ L

11 Fscher and Fox Solvng for the optmal tax by settng the welfare change equal to zero, and usng the Chan Rule, we get D dc 2 D t dl 1 = m2 + τ1 m1 tl λ dc λ dc { leakage mpact ( ) unnternalzed per-unt damages tax nteracton D D dc 2 dc1 dl τ1 = + m2 t1 tl { λ λ dc1 de1 de { margnal damage unnternalzed leakage ( ) tax nteracton (1) (2) Solvng for the optmal tax combnaton, we get D τ1 = λ and D dc dl 2 t1 = m2 tl λ dc1 dc1 The frst equaton states that the optmal emssons tax equals the margnal damages. The second states that the optmal rebate has two components. Frst, t nternalzed the margnal damages of the emssons generated by the substtuton of consumpton away from the regulated good toward the unregulated good. Second, the rebate s needed to counteract the tax nteracton problem. Thus, a subsdy to Sector 1 to prevent emssons leakage and tax nteracton can be preferred to full recyclng of the envronmental tax revenues. Let us put these results n the context of Bovenberg and de Mooj (1994), Fullerton (1997), Wllams (2000), and others who assume that emssons have a lnear relatonshp wth output. In ther case, E1 = C1 (or m 1 = 1), so the emssons tax would be a commodty tax. Suppose that Sector 2 s a clean good, so de 2 = 0 and m 2 = 0. Then for the drty good we have D τ = t = t λ 1 ; 1 L Ths combnaton s equvalent to ther result of an optmal second-best emssons (commodty) tax beng lower than the Pgouvan level by the tax nteracton effect. Smlarly, n ths case, n whch techncal abatement opportuntes are avalable, f a rebate s not a polcy dl dc 1 8

12 Fscher and Fox D dl opton, we would have τ1 = tl. When a rebate s an opton, though, t s a more λ de 1 effectve means of addressng the tax nteracton effects than dstortng the margnal abatement ncentves. Snce the labor supply dsncentve derves from the product prce ncrease, a prce rebate addresses the tax nteracton problem drectly. Another ssue for output-based rebatng s the accuracy of 100 percent rebatng, whch s the case of dstrbutng all the permts grats accordng to output. The optmal rebate rate s one f the subsdy expendtures equal the emssons tax revenues: D C m dc + t dl D = E L 1 λ dc1 dc1 λ that s, f dc2 D D dl m2 = m1 tl dc1 λ { λ dc emssons leakage costs damages labor tax dstorton In the absence of a labor tax, ths holds f the emssons tradeoff (leakage factor) s oneto-one, such as for two perfect substtutes wth dentcal emssons rates. Wth preexstng labor taxes, ths leakage factor must be hgher than one to justfy 100 percent rebatng because of the opportunty cost of publc revenues. How much hgher than one depends on the relatve sze of the labor tax dstorton to the margnal emssons damages. More generally, wth two drty goods, t s nterestng to see how addtonal polcy constrants affect the second-best (or nth-best) taxes. One constrant could be on the earmarkng of the allocaton. If the resultng rebate s too low (hgh), then Equaton (2) reveals that the optmal emssons prce wll be lower (hgher) than f the leakage and tax dstortons were exactly nternalzed. Meanwhle, f the emssons prce s too low and s restrcted from fully nternalzng the margnal damages, Equaton (1) shows that the optmal rebate s lower, reflectng the need for taxng the emssons emboded n addtonal output that dd not get taxed drectly. To understand the relatve magntude of these knds of polcy constrants on welfare and other measures of the burdens of regulaton, we next turn to a smulaton model. 9

13 Fscher and Fox Numercal Model Model Descrpton We employ a modfed verson of the model used n Fscher and Fox (2007), n whch greater descrptve detal can be found. Ths computable general equlbrum (CGE) model from the Global Trade Analyss Project (GTAP) offers a rchness n calculatng trade mpacts and allows us to evaluate the dstrbutonal and effcency effects of emssons permt allocaton mechansms, spannng a more dverse and dsaggregated set of energy-usng sectors than n most clmate models. Although t does not allow for dynamc responses or for technologcal change, t does allow for captal reallocaton. 3 Our results should therefore be consdered llustratve of short- to medum-term effects (say, three to fve years, a relatvely short perspectve for clmate polcy). The model and smulatons n ths paper are based on Verson 5.4 of the GTAPnGAMS package developed by Thomas Rutherford and documented for Verson 4 of the dataset and model n Rutherford and Paltsev (2000). The GTAP-EG dataset used s a GAMS dataset mergng the GTAP economc data wth nformaton on energy flows. We adapt the framework to employ the Verson 6.0 Release Canddate, whch updates the analyss to 2001, the base year of the more recent GTAP database. 4 The model s a multsector, multregon general equlbrum model of the world economy as of Energy requrements and ther correspondng carbon emssons are ncorporated nto ths framework. The producton functon ncorporates most ntermedate nputs n fxed proporton, although energy nputs are bult nto a separate energy nest. For the chemcals sector, whch ncludes petrochemcals, we dvded ts energy use nto feedstock requrements, whch are treated as ntermedate nputs, and the remander, whch s treated as energy, usng the feedstock use ratos for ol and gas n Lee (2002). Energy producton s a constant elastcty of substtuton (CES) functon nested to three levels. At the lowest level, ol and gas are relatvely substtutable for one another (elastcty = 2) wthn the lqud nest, but lqud energy s less substtutable aganst coal n the nonelectrc nest. Lastly, nonelectrc has low substtutablty (0.1) aganst 3 In the default, captal reallocaton occurs wthn, not across, regons. Paltsev (2001) conducts senstvty analyss wth respect to ths assumpton and fnds that the carbon leakage rate does not change sgnfcantly wth greater nternatonal captal moblty. 4 A more complete dscusson of the energy data used can be found n Complanvlle and van der Mensbrugghe (1998). 10

14 Fscher and Fox electrcty n the energy nest. Energy tself has low substtutablty (0.5) for the labor-captal composte from the value-added nest. Wthn the value-added nest, labor, prvate captal and publc captal have untary elastcty. Foregn and domestc varetes are substtutable for one another through a standard Armngton structure, wth the elastcty of substtuton between the domestc varety and foregn composte set to half the elastcty of substtuton among foregn varetes. The latter elastctes are largely derved from econometrcally based estmates, as n Hertel et al. (2004). Consumpton s a composte of goods, servces, and n our modfcaton, lesure. The energy goods ol, gas, and coal enter nto fnal demand n fxed proportons n the energy nest, and are untary elastc wth electrcty. Ths composte s then substtutable at 0.5 wth other fnal demand goods and servces. Goods and servces (ncludng energy) are then substtutable aganst lesure; the dervaton s gven n Fscher and Fox (2007) and Fox (2002). Government demand s represented by a smlar demand structure and prvate consumpton, wth the excepton of the labor-lesure component. Government demand s held fxed through all the experments, although the fundng mechansm (adjustment of a lump-sum tax or the tax on labor) vares as noted below. Three features added to the GTAP-EG structure allow us to model the mpact of the polcy scenaros. Frst, we add a carbon prce that s appled to the covered sectors. Second, the approprate structure for smulatng a producton rebate must be ncorporated nto the model. Thrd, the household s gven a labor-lesure choce so that labor taxes are dstortng, allowng us to conduct smulatons recyclng revenue from polluton permts to offset the dstortng tax nstrument. Snce we have no data on labor taxes wthn the GTAP-EG database, we assume a labor tax rate of 40 percent wthn Annex B (developng) countres and a 20 percent tax rate wthn all other countres and the Rest of World. To ncorporate the carbon emssons payment requrement, we ntroduce a carbon permt as a Leontef technology n an addtonal composte fossl fuel nest to producton n the covered sectors. The composte of permt and energy nput s then ncluded n the producton block for the output good. In ths manner, one permt s demanded for each unt of carbon that enters nto producton, and we can track polluton permts through the model. Fnally, we ncorporate a rebate n the form of a (negatve) endogenous tax nto the sector s producton functon. 11

15 Fscher and Fox Polcy Scenaros We consder a scenaro n whch the Unted States adopts a carbon tax at $50 per ton C (roughly $14 per ton CO 2, whch s well n the range of many proposed prces). In prncple, an emssons prce can be mplemented by ether a tax or a cap-and-trade program; however, n the latter program, dfferent levels of coverage or amounts of rebates wll affect the equlbrum allowance prce. Many of those prce changes can be mtgated by several of the proposed cost contanment mechansms, lke a safety valve (prce celng), nternatonal lnkng, or bankng and borrowng. For our purposes, fxng the prce allows us to value all emssons consstently across scenaros as we search for the optmal rebate. (In related work, Fscher and Fox 2009 look at the effects on effcency and leakage of dfferent allowance allocaton optons, dependng on dfferent degrees of program coverage.) We observe that sector coverage ranges wdely across programs and proposals. Among the major clmate-change blls put forth n the 110th Congress, coverage of proposed emssons tradng systems has ranged from the electrcty sector alone (Fensten-Carper 5 ) to major sources or sectors (Leberman-McCan 6 and Kerry-Snowe 7 ) to sectors chosen at the dscreton of the Envronmental Protecton Agency (Sanders-Boxer 8 ) to a comprehensve upstream program (Bngaman-Specter 9 ). Among the carbon tax blls, however, coverage has been fully upstream. 10 The European ETS covers only major sources and sectors. We consder a central scenaro n whch the carbon tax s appled downstream to the sx major energy-ntensve sectors: electrcty, petroleum and coal products (refned), ron and steel, chemcals, nonmetallc mnerals (whch nclude cement, glass, and ceramcs), and paper, pulp, and prnt. Ths scenaro follows the mddle-of-the-range proposals that target major pont-source emtters. Gven our model, we must assume that the covered sectors are covered n ther entrety, whch tends to overstate somewhat the magntude of covered polluton. These sectors represent 55 percent of U.S. CO 2 emssons, accordng to our data. In a second scenaro, we nclude transportaton fuels, whch ncreases coverage to 75 percent. 5 S. 317: Electrc Utlty Cap and Trade Act of 2007 and S. 1177: Clean Ar Plannng Act of S. 280: Clmate Stewardshp and Innovaton Act of S. 485: Global Warmng Reducton Act of S. 309: Global Warmng Polluton Reducton Act 9 S.1766: Low Carbon Economy Act of H.R. 2069: Save Our Clmate Act of 2007, and H.R. 3416: Amerca s Energy Securty Trust Fund Act of

16 Fscher and Fox Results Solvng for Optmal Rebates Followng the theoretcal model, our goal s to solve for the optmal rebates to accompany the carbon tax; that s, for the covered sectors, we want to fnd the set of rebates that maxmze global welfare, ncludng the value of emssons reductons. In ths complex CGE model, however, there are many preexstng dstortons besdes just the emssons externalty, ncomplete coverage, and labor taxes. In fact, n the global context of tarffs, terms of trade effects, wealth redstrbuton ncentves, and so on, the optmal polcy seems to nvolve mposng addtonal taxes on U.S. producton. We therefore frst solve for these global welfare-maxmzng taxes on U.S. energy-ntensve sectors n the absence of any clmate polcy, and then see how they change when the carbon tax s mposed. These changes are our measure of the optmal rebates assocated wth the carbon tax, and they reflect the nteracton between the carbon tax and the leakage and tax dstortons. Table 1 reports the magntude of these optmal rebates as a percentage of the value of the emssons n each sector after the carbon tax. It reveals that full rebatng for electrcty and refned petroleum products s not desred. Optmal rebates for electrcty are roughly one-thrd of emssons payments somewhat less when transportaton s excluded and somewhat more when t s ncluded. The coverage of transportaton has an mportant nfluence of the desrablty of rebatng for refnng: when transportaton s excluded, one wants to tax refned petroleum products n addton to levyng the carbon tax on the refnng process, to encourage conservaton among those uncovered sectors. When transportaton s ncluded, however, a small (28 percent) rebate s optmal, and the transportaton sector also gets a 60 percent rebate. For energy-ntensve manufacturng, on the other hand, regardless of the carbon tax coverage of transportaton, optmal rebates are well over 100 percent; n fact they all exceed 200 percent, and for the ron and steel sector, 500 percent. Table 1. Optmal Rebate as a Percentage of Emssons Payments Sector Wthout transportaton Wth transportaton Electrcty Refned petroleum Chemcals Nonmetallc mnerals Pup, paper, and prnt Iron and steel Transportaton 60 13

17 Fscher and Fox Those hgh rates reflect n part the fact that n the model, foregn producers are on average more carbon ntensve than U.S. producers n these sector aggregates, emttng more than twce as much for nonmetallc mnerals and almost three tmes as much for ron and steel. However, these rebates represent the (second-best) nternalzaton of not only carbon leakage to unregulated sectors both foregn and domestc but also nteractons between the carbon tax and all the other preexstng dstortons throughout the system. We fnd smlar results when the objectve functon s restrcted to U.S. welfare (equvalent varaton plus the value of global emssons changes). The U.S. focus ncludes nteractons wth the terms of trade, whch were counterbalanced by foregn changes when maxmzng global welfare was the objectve. Because of these effects, the optmal taxes n the absence of a carbon tax are much lower and are negatve for most energy-ntensve manufacturng, as one mght expect; however, they respond smlarly to the mposton of a carbon tax. When transportaton s not covered, optmal rebates for energy-ntensve manufacturng are all roughly 200 percent, except for steel, whch s a bt over 300 percent. The rebates for electrcty and refned ol products are 36 percent and 47 percent, respectvely. Applyng Optmal Rebates Although the numercal results valdate the theory, some cauton should be taken n applyng them n practce that s, n a world wth other preexstng dstortons. As we noted, n the model, global welfare mproves when addtonal producton taxes are mposed on covered U.S. sectors, whether or not a carbon tax s n place. Therefore, when startng from a baselne wthout these taxes, offerng rebates reduces global welfare. However, t does ncrease U.S. welfare. Furthermore, t s nterestng to observe the general equlbrum effects of the rebate polcy on the pattern of complance and leakage. Table 2 compares the macroeconomc results from the carbon tax alone (wth revenue recyclng) wth the carbon tax plus the above rebates, n the scenaro excludng transportaton. The rebates nearly elmnate the welfare cost for the Unted States, wth very slght declnes n producton, employment, and the real wage but gans n lesure. Although the rebates do mtgate carbon leakage somewhat, domestc emssons also expand; f one values those global emssons changes at the $50 per-ton C tax rate, the U.S. welfare gan s fully offset by these losses. 14

18 Fscher and Fox Table 2. Percentage Change n Summary Indcators Indcator for Unted States Carbon tax alone Wth rebates Welfare (equvalent varaton) Producton Employment Real wage Total domestc emssons Table 3 compares the effects of the two scenaros on dfferent measures of leakage. It reveals that by conventonal measures of leakage that s, the change n foregn emssons as a share of domestc reductons the rebates have lttle effect, ether overall or for the covered sectors as a whole. The reason s that whle foregn emssons expand less wth the rebates 2 percent less overall and 39 percent among covered sectors domestc sectors also reduce less. However, the rebates do change the dstrbuton of leakage among the covered sectors. The energy-ntensve manufacturng sectors exhbt the hghest rates of leakage, n large part because 78 percent of the lost producton n the Unted States due to the carbon tax s made up by producton ncreases abroad. Wth the rebates, that shft n producton drops dramatcally, leakage rates drop by more than a thrd, and the expanson of foregn emssons drop by half, relatve to the carbon tax alone. Table 3. Emssons Leakage Measures Indcator for Unted States Carbon tax alone Wth rebates Foregn leakage relatve to total domestc reductons 14.2% 14.2% Foregn covered sector leakage 12.8% 13.0% Foregn energy-ntensve manufacturng leakage 27.4% 17.1% Foregn and domestc leakage relatve to covered sector 9.4% 4.9% Meanwhle, domestc leakage of emssons to uncovered sectors s substantally reduced. To be more precse, domestc leakage may be somewhat msleadng; domestc uncovered sectors actually reduce ther emssons along wth the covered sectors under the carbon tax, but by far less than f they were themselves covered, and the rebates help capture some of those addtonal reductons. Thus, the rebates have many of the effects one would expect. The (generous) subsdes to energy-ntensve manufacturng nearly elmnate the deteroraton n net exports and reduce the foregn leakage rates for those sectors. The tax on refned ol products nduces addtonal 15

19 Fscher and Fox reductons from the uncovered domestc sectors, although t exacerbates foregn leakage by encouragng more mports of refned products and also by drvng down global ol prces, whch expands consumpton abroad. The partal rebate to electrcty helps downstream sectors, partcularly electrcty-ntensve manufacturng, but the ncrease n emssons more than offsets the decrease n the uncovered sectors. On the other hand, t also helps reduce foregn leakage, n part through downstream demand changes and n part by mtgatng the fall n global prces for natural gas and coal. Overall, the net result s no sgnfcant change n global emssons, relatve to the carbon tax alone. However, the redstrbuton of the emssons reductons to uncovered domestc and foregn sectors lowers the cost of the clmate polcy to the Unted States. Of course, those other sectors bear the cost of the redstrbuton, whch s not globally effcent n ths second-best world; the loss n welfare to other countres s roughly double the welfare gan to the Unted States. Fgure 1 llustrates the nternatonal dstrbuton of the effects of clmate polcy n the Unted States. The rebates (and more pontedly the tax on refned ol products) tend to exacerbate the losses by crude ol producng countres. Asan countres see ther gans from the U.S. carbon tax erode to losses wth the rebates to energy-ntensve manufacturng. However, several trade partners, ncludng Inda, Brazl, and Eastern Europe, do beneft from the rebates. Fgure 1. Percentage Change n Welfare (Equvalent Varaton) by Regon.04%.02% 0% -.02% -.04% -.06% -.08% -.10% -.12% Unted States Canada Europe Japan Other OECD Former Sovet Unon Central European Assocates Chna Inda Brazl Other Asa Mexco, OPEC Rest of World Grand Total C Tax Rebate -.14% 16

20 Fscher and Fox Concluson As natonal and subnatonal clmate polces evolve, concerns naturally arse about the potental for adverse effects on energy-ntensve, trade-senstve ndustres, and for the eroson of the envronmental benefts f an ncrease n domestc producton costs causes more producton actvtes to shft to natons or regons wth weaker clmate mtgaton polces. We fnd some support for these concerns, but we also note that equal or even greater attenton should be pad to the actvtes of domestc sectors left uncovered by the clmate polcy. The revenues from emssons prcng polces are valuable, snce they can offset the taxaton of productve factors lke labor and mprove the effcency of the economy as a whole. Therefore, t s mportant to understand under what crcumstances those valuable revenues should be partly used to mtgate some of the effects of carbon regulaton on covered sectors. By mtgatng prce ncreases of covered sector products, rebates reduce both the nteracton wth preexstng taxes and the loss of compettveness that can lead to leakage. Thus, the optmal rebate s hgher for goods that are stronger substtutes for hgher-emttng unregulated goods and that are stronger complements wth employment. The numercal results llustrate these ponts. Rebates are hgher for the trade-senstve, energy-ntensve manufacturng sectors, such as ron and steel. Refned ol products are hghly complementary wth emssons of uncovered sectors, so they should be taxed n addton to the carbon tax when the transportaton sector s not covered. Electrcty generaton s the largest emtter and has mportant countervalng effects. On the one hand, t s complementary wth downstream emttng sectors, but that ncludes already-covered sectors and many sectors that are trade senstve and subject to foregn leakage. Thus, a modest rebate can help redstrbute emssons reductons among these sectors n a more cost-effectve way at least for the mplementng country but at a cost of forgong sgnfcant reductons. Of course, even f they could calculate optmal rebates, countres are not lkely to be able to mplement rebates that exceed the value of actual emssons labltes, whch can run afoul of trade rules regardng subsdes. Furthermore, from a global perspectve, ncomplete coverage and labor taxaton are not the only market falures. Further research could employ a wder varety of clmate models to nvestgate these ssues of polcy nteractons and better understand the senstvty of leakage and rebate effectveness to numercal model structure and parameters. Addtonal scenaros also mert nvestgaton, ncludng the response of cap-and- 17

21 Fscher and Fox trade programs (where the prce s endogenous), the role of expanded coverage of carbon prcng among tradng partners, and the effects of all these strateges on the least developed countres. 18

22 Fscher and Fox References Bernard, A., C. Fscher, and A. Fox Is There a Ratonale for Output-Based Rebatng of Envronmental Leves? Resource and Energy Economcs 29(2): Bovenberg, A.L., and R.A. de Mooj Envronmental Leves and Dstortonary Taxaton. The Amercan Economc Revew 84(4): Complanvlle, C., and D. van der Mensbrugghe Constructon of an Energy Database for GTAP V4: Concordance wth IEA Energy Statstcs. Manuscrpt. Pars: OECD Development Centre. Aprl. Fscher, C., and A.K. Fox Output-Based Allocaton of Emssons Permts for Mtgatng Tax and Trade Interactons. Land Economcs. Forthcomng n November. Fscher and Fox When Revenue Recyclng Isn t Enough: Permt Allocaton Strateges to Mnmze Intra- and Internatonal Emssons Leakage. Workng Paper. Washngton, DC: Resources for the Future. Fox, A.K Incorporatng Labor-Lesure Choce nto a Statc General Equlbrum Model. Workng paper. Washngton, DC: U.S. Internatonal Trade Commsson. Fullerton, D Envronmental Leves and Dstortonary Taxaton: Comment. The Amercan Economc Revew 87(1): Fullerton, D., and G. Metcalf Envronmental Controls, Scarcty Rents, and Pre-Exstng Dstortons. Journal of Publc Economcs 80: Goulder, L.H., ed Envronmental Polcy Makng n Economes wth Pror Tax Dstortons. Cheltenham, UK: Edward Elgar. Goulder, L.H., I. Parry, R. Wllams III, and D. Burtraw The Cost-Effectveness of Alternatve Instruments for Envronmental Protecton n a Second-Best Settng. Journal of Publc Economcs 72(3): Hertel, Thomas & Hummels, Davd & Ivanc, Maros & Keeney, Roman, How confdent can we be of CGE-based assessments of Free Trade Agreements? Economc Modellng, 24 (4): Isaksson, L., and S. Sterner Refunded Emsson Payments Theory, Dstrbuton of Costs, and Swedsh Experence of NOx Abatement. Ecologcal Economcs 57(1):

23 Fscher and Fox Lee, H.L An Emssons Data Base for Integrated Assessment of Clmate Change Polcy Usng GTAP. GTAP Workng Paper (Draft), Revsed Verson, October. Paltsev, S.V The Kyoto Protocol: Regonal and Sectoral Contrbutons to the Carbon Leakage. Energy Journal 22(4): Parry, I.W.H., and R.C. Wllams Second-Best Evaluaton of Eght Polcy Instruments to Reduce Carbon Emssons. Resource and Energy Economcs 21: Rutherford, T.F., and S.V. Paltsev GTAPnGAMS and GTAP-EG: Global Datasets for Economc Research and Illustratve Models. Workng paper. Boulder, CO: Department of Economcs, Unversty of Colorado. September. Wllams, R.C Health Effects n a Model of Second-Best Envronmental Taxaton or Reconsderng Reconsderng the Tax-Interacton Effect. NBER Workng Paper Cambrdge, MA: Natonal Bureau of Economc Research. 20