• Non ci sono risultati.

Do technology externalities justify restrictions on emission permit trading?

N/A
N/A
Protected

Academic year: 2021

Condividi "Do technology externalities justify restrictions on emission permit trading?"

Copied!
23
0
0

Testo completo

(1)

ContentslistsavailableatSciVerseScienceDirect

Resource

and

Energy

Economics

j ou r n a l h o m e p a g e :w w w . e l s e v i e r . c o m / l o c a t e / r e e

Do

technology

externalities

justify

restrictions

on

emission

permit

trading?

Enrica

De

Cian

,

Massimo

Tavoni

FondazioneEniEnricoMatteiandCentroEuro-MediterraneoperiCambiamentiClimatici(CMCC),Italy

a

r

t

i

c

l

e

i

n

f

o

Articlehistory:

Received27February2010

Receivedinrevisedform29May2012

Accepted31May2012

Availableonline15June2012

JELclassification: Q54 Q55 Q43 H23 Keywords: Energy-economymodelling Emissiontrading Technologyspillovers

a

b

s

t

r

a

c

t

Internationalemissiontrading isanimportantflexibility mech-anism,butitsusehasbeenoftenrestrictedonthegroundthat accesstointernationalcarboncreditscanunderminethedomestic abatement effortreducingtheincentivetoinnovateand, even-tually,loweringthepaceofclimatepolicy-inducedtechnological change.Thispaperexaminestheeconomicsthatisbehindthese concernsbystudyinghowacaptothetradeofcarbonoffsets influ-encesinnovation,technologicalchange,andwelfare.Byusinga standardgameofabatementandR&D,weinvestigatethemain mechanismsthatshapetheserelationships.Wealsousea numer-icalintegratedassessmentmodelthatfeaturesenvironmentaland technologyexternalitiestoquantifyhowlimitstothevolume,the timing,andtheregionalallocationofcarbonoffsetsaffectclimate policycostsandtheincentivetoinvestininnovationand low-carbontechnologies.

Resultsindicate that,formoderatecapsontheamount trad-ableemissionspermitsandsufficientlyhightechnologyspillovers, globalinnovationandtechnicalchangewouldincreaseandthatthis additionalinnovativeeffortcouldleadtoeconomicefficiencygains. Thenumericalanalysisconfirmsthatwhenconstraintsareclose

ThispaperispartoftheresearchcarriedoutbytheSustainableDevelopmentProgrammeofFondazioneEnricoMattei.

FinancialcontributionfromthePLANETSprojectfundedbytheEuropeanCommissionunderthe7thframeworkprogrammeis

gratefullyacknowledged.Theusualdisclaimerapplies.

WewouldliketoespeciallythankProf.SjakSmuldersforthemanycommentsandsuggestionswhichhavehelpedto

consid-erablyimprovethismanuscript.

∗ Correspondingauthorat:FondazioneEniEnricoMattei,IsoladiSanGiorgioMaggiore,30124Venice,Italy.Tel.:+39041

2700450;fax:+390412700413.

E-mailaddress:[email protected](E.DeCian).

0928-7655/$–seefrontmatter©2012ElsevierB.V.Allrightsreserved.

(2)

to15%ofdomesticabatement,efficiencylossesaresmallbecause theyarepartlycompensatedbymoretechnologicalspilloversand lowerenergyprices.Underabroadrangeofparameters,restrictions arecostlyfortheconstrainedcountries,butalwaysbeneficialfor unconstrainedones.

©2012ElsevierB.V.Allrightsreserved.

1. Introduction

SincetheKyotoProtocol,theinternationaltradeofcarbonallowancesandcreditshasbeena fun-damentalelementtoensureflexibilityandcostefficiencyofclimatechangepolicies.International emissiontradinghasthepotentialtoreducecompliancecostsbecauseitallowsexploitinglowcost emissionreductionoptionsnomatterwheretheyarelocated.Regionalandsub-regionalcarbon mar-ketinitiatives,eitherregulatedorvoluntary,haveemerged,especiallyindevelopedcountries.1The exchangeofcarbonallowancesandcredits,especiallyattheinternationallevel,hasalsoraisedsome concerns.Ifnotwell-designed,cap-and-tradeschemesmightunderminetheenvironmental effective-nessofclimatepolicies,reduceinnovationincentives,andultimatelyhinderthedeploymentofclean technologies.Asamatteroffact,thelackofagreementontheimplementationproceduresoftheKyoto flexiblemechanismswasthecauseofthefailureofthenegotiationsintheHaguein2000.In2004, theEUlinkingdirectiveestablishedthepossibilitytousetheKyotocreditsintheEU-ETSwithoutany limit.DuringthesecondPhaseofemissiontrading,theEUCommissionsintroducedquantitativeand qualitativerestrictionsontheuseoftheseoffsets2(deSépibus,2008).Thisrevisionwasmeantto avoidapricecollapse,asobservedduringthefirstPhase.Anotherargumentwasthatemissiontrading couldreducetheincentivetoinnovate(Hourcadeetal.,1999).

Whenfacingthechoicebetweeninnovateorpurchasecarboncredits,apollutingcountrywould selectthecheapestoption(Driesen,2003).Ifpermitsarecheapercomparedtothecostofinvestingin mitigationoptionsatHome,thepossibilityoftradingcreatesanincentivetoshiftabatementabroad, reducingtotalcompliancecosts,butalsoloweringtheincentivetocarryoutinnovation.Considering thesearguments,mostcap-and-tradeschemesincludeceilingsontheuseofinternationalcarbon offsets.Iflimitstopurchasecreditsarenotlarge,economiclossesmightbemodestandtherefore itmaybeworthwhileoneconomicgroundstousethem(KarpandZhao,2009).Theselimitscould avoidhugefinancialtransfers,maketheagreementmoreappealingforindustrialisedcountries,and stimulateinnovation.

Theargumentinfavourofceilingsmightbereinforcedinasecond-bestworld.Unrestricted emis-siontradingalwaysgeneratesefficiencygainsinafirst-bestworldwhentheonlydistortionisglobal pollution(seeforexampleWeyantandHill,1999;Bohm,1999;Chanderetal.,2002;Richelsetal., 2007).However,inasecond-bestworldaddressingonlyoneofthevariousmarketdistortionswill notnecessarilyimprovewelfare(LipseyandLancaster,1956).Environmentaleconomicssuggeststhat multiplepolicyinstrumentsshouldbeemployed(Jaffeetal.,2003,2005;BennearandStavins,2007). Inpractice,however,designingseveralinstrumentscanbecomplicated,anddespitethevariousforms ofregulationusedtoday,amarket-basedsolutionaimedatpricingCO2iswidelyregarded(atleast

amongeconomists)asthemostefficientsolutionforthecaseofglobalwarming.

Theeconomicefficiencyofemissiontradinginthepresenceofvariousexternalitieshasbeen ana-lysedmostlyusingsimpleanalyticalmodels.However,hardlyanynumericalevaluationhasconsidered second-bestinteractionswithtechnologyexternalities.Mostsecond-bestquantitativeassessments haveexploretheroleofpre-existingdistortionarytaxes(Babikeretal.,2004;Paltsevetal.,2007; McKibbinetal.,1999).OnlyBuonannoetal.(2000)assesstheprosandconsofintroducingceilingsto emissiontradinginamodelwithendogenoustechnicalchange(ETC-RICE).Theyfindlittlesupportfor

1Forareviewofcap-and-tradeschemesaroundtheworldseeCapoorandAmbrosi(2009).

2Inthepaperwedonotdistinguishbetweentradeincarboncreditsandallowances,whichforthepurposeofouranalysis

(3)

quantitativerestrictions,whichhavenegativeimpactsonbothefficiencyandequity.Although quan-titativerestrictionsfostertechnologicalinnovation,itslong-runpositiveeffectsoneconomicgrowth arelowerthantheadditionaleconomiccosts.Theirmodeldoesnotincludeinternationalspilloversof knowledgeandexperienceandthereforeitunderestimatesthebenefitsoftechnologicalinnovation.

Adifferentissue,whichhowevercanultimatelyhavethesameeffectofreducingthevolumeof internationalcarbontrade,relatestothedifficultiesofputtinginplaceawell-functioninginternational carbonmarket.TheCleanDevelopmentMechanism(CDM)oftheKyotoProtocol,thelargestexisting internationalcarbonmarket,hasrevealedsevereproblemsconcerningmonitoring,creditverification, andqualitycertificationofprojects(WaraandVictor,2008).Aslongastheseproblemsarenotsolved, institutionalandadministrativeissueswilldelaytheestablishmentofaninternationalmarket.

Thegeneralquestionweaddressinthispaperiswhetherrestrictingthetradeofinternational carbonoffsetsisasecond-bestpolicyinthepresenceofenvironmental,technology,andeconomic externalities.Marketfailuresexistnotonlyintheprovisionoftheinternationalenvironmentalgood, suchasemissionreduction,but alsoinmarketsfor innovationandtechnologydiffusion.Climate policycaninducesecond-ordereffectsoninnovationandtechnicalchange,butitisdesignedtotackle theenvironmentalexternality,notthetechnologyone.Forthisreason,limitinginternationalemission tradingwouldleadtowelfarelossesinafirst-bestworld,butthismightnotbethecaseinasecond-best world.

Thepapermakesseveralcontributionstotheliterature.First,weuseastandardgameof abate-mentandR&Dwithtwoexternalities(environmentaldamageandinnovation)andtworegionsto evaluatetheimplicationsoflimitingcarbontradingoninnovationandwelfare.Thisprovidesnovel insightsontheinterplaybetweeninnovationandclimatepolicies.Second,weextendtheanalysis usingarichnumericalintegratedassessmentmodelthatfeaturesenvironmental,technology,aswell ascarbonleakagethroughinternationalenergymarkets.Thisallowstogenerateoneoftheeveryfew quantitativeestimatesoftheeconomicimplicationsoftraderestrictionsonemissionpermits.

Ourresultsindicatethatsomerestrictionsoninternationaltradingofemissionallowancescanbe justifiedwhentheonlyinstrumentavailabletopolicymakersisthatofacarbonmarket.Theanalysis showsthat,underplausibleassumptionsabouttechnicalchange,permittradingrestrictionleadsto moreglobalinnovationandthat,whenthelimitationismoderateandspilloverssufficientlyhigh,the efficiencylossescouldbesmallorevennegative.Specifically,whendealingwithaclimate stabili-sationpolicywheredevelopedcountriesinitiallytakeonthelargestobligations,limitingaccessto internationaloffsetswouldmakeabatementmorecostlygloballyandfortheOECDasawhole,but theefficiencylossesaresmallformoderatetradingrestrictions.Constrainingoffsetstoatmost15%of regionalabatement,approximatelyinlinewithwhatwasproposedintheEUandUS,wouldnotaffect welfareintheUS,thoughitwouldresultinalossforEurope.However,financialflowstodeveloping countrieswouldbesignificantlyreduced.Despitetherevenuelossonthecarbonmarket,non-OECD regionswouldbecompensatedbylowerenergypricesandhighertechnologicalspillovers,thelatter duetomoreinnovationbeingcarriedoutinconstrainedregions(OECD).Theseresultsareshownto berobusttoabroadsetofparametersregardingtheprocessoftechnologicalchangeandknowledge spillovers.

Theremainderofthepaperisorganisedasfollows.Section2discussestheconnectionsbetween R&D,spillovers,and permittraderestrictionsusingastandard modelwithmultipleexternalities. Section3introducesanumericalintegratedassessmentmodelandextendstheanalysistoamore generalset-up.Section4illustratesthemacroeconomicimplicationsoftheclimatepolicyunder dif-ferentcarbonmarketdesigns.Giventhecriticalroleoftechnologyexternalities,Section4explores thesensitivityofmajorresultstovariationsinkeyparameters.Section5concludesandaddssome policyconsiderations.

2. Restrictedpermittradeandtechnologicalchangeinastandardanalyticalframework

2.1. Contributionofexistingliterature

Anumberofstudieshaveinvestigatedhowmarketfailuresassociatedwithinnovationandthe dif-fusionoftechnologiesaffecttheoptimalenvironmentalpolicyandtheequivalencebetweendifferent

(4)

market-basedinstrumentssuchastaxandemissiontrading.Awell-knownresultisthat,when tech-nologyexternalitiesarenotinternalised,theoptimaltaxorthepermitpriceshouldbehigherthanthe Pigovianlevel.GolombekandHoel(2006)analysethewelfareimplicationsofcarbonpermittradeina second-bestworldwheretechnologyexternalitiesarenotinternalised.Theyfindthatthesecond-best optimumischaracterisedbymarginalcostsofabatementexceedingthePigoviantaxinallcountries. Atighteremissionrequirementisthusawayofcompensatingforthelackoftechnologypolicy.Ina follow-uppaper,GolombekandHoel(2008)findthatthepriceofcarbonshoulddifferacross hetero-geneouscountriesinthesecond-bestagreementandtheyconcludethatwhetheremissiontrading iswelfareenhancingornot,isanempiricalquestion.Gerlaghetal.(2009),usingamodelwithout spilloversbutwithfinitepatentlifetime,showthatthedifferencewiththePigoviantaxshouldmimic theoptimalresearchsubsidyortax.Infact,whenthetechnologyexternalityisnotregulatedbyan explicittargetedpolicy,theequilibriumlevelofinnovationtendstobelowerthanfirst-best(Golombek andHoel,2004,2005).AsrecentlypointedoutbyHealandTarui(2010),thisconventionalwisdom hingesontheassumedrelationshipbetweeninnovationandcoststructure.Thecrucialassumption isthattechnologyimprovementsreducemarginalabatementcosts.Bakeretal.(2008)discusshow abatementcostcurvescanchangeinmanydifferentwaysasaresultoftechnicalchange.Technical changecanactuallyleadtotheadoptionoftechnologieswithhighermarginalabatementcosts com-paredtotheoldones.Asexampleofabatementoptionswithadifferentcoststructure,HealandTarui (2010)discusstheapplicationofcarboncaptureandstorage(CCS)technologytocoalpower genera-tion.Contrarytorenewablesources,whichhavehighfixedcosts,butnearlyzerovariableones,CCSis characterisedbyhighvariablescoststhatincreasewiththeCO2price.HealandTaruishowthatwhen

technicalchangeleadstotheadoptionofatechnologywiththiscoststructure,theequilibriumR&D investmentsmightbelowerthanfirst-best,dependingalsoontheextentofinternationalspillovers.

Theexistingliteraturehasfocussedontheimplicationsofahighersecond-bestcarbonpriceon innovationandtechnologicalchange,butithasnotexploredthelinkbetweenrestrictionsonpermit trade,innovation,and welfare,which isthefocusofthispaper.Weaskthequestionwhether,in thepresenceofmultipleexternalities,restrictingthetradeofcarbonpermitscanbeasecond-best policyrecommendation,andifso,underwhichconditions.Theremainingofthissectionintroducesa standardsimultaneousgameofabatementandR&Dinvestments,whilethenextsectiongeneralises theanalysisbyusinganumericalgeneralequilibriumcalibratedmodel.

2.2. Amodelofinnovationandabatement

Consideratwo-regionmodelwhereeachcountrychoosesinvestmentsininnovation,x,and abate-ment,a.Bothregionshavethesameabatementcost,whichisanincreasingandstrictlyconvexfunction inabatement,Ca(X,a)>0,Caa(X,a)>0,butdecreasingintheleveloftotalknowledgestockX,CX(X,

a)<0,CXX(X,a)>0.InnovationinvestmentcostsaregivenbyG(x),whereGx>0,Gxx≥0.Thelevelof

knowledgestockintheHomeregiondependsonthesumofitsowninvestments(x)andthoseof theForeignregion(¯x): X=x+ ¯x,where∈[0,1]representstheextentofinternationaltechnology spillovers.Thatis,onlyafractionoftheR&Dinvestmentscarriedintheotherregionisappropriated domestically.Forcomparabilitywiththesecondpartofthepaper,weassumethatinnovationlowers marginalabatementcosts,CaX=CXa<0.Thismechanismisalsomaintainedinthenumericalmodel

usedinSection3,wheretechnicalchangereducesthecostsofcapital-intensivetechnologieswith verylowvariablecosts.3

Thefirst-bestParetolevelofinnovationinvestmentsandabatement{xp,ap}minimisesthetotal

socialcosts,includingenvironmentaldamageD,whichisafunctionoftotalemissions(orequivalently abatement,ifweassumeexogenousbaselineemissions),C(X,a)+G(x)+C( ¯X, ¯a)+G(¯x)+D(a+ ¯a).The firstorderconditions(FOCs)fortheHomeregionare:

x:Cx(Xp,ap)+Cx( ¯Xp,ap)+Gx(xP)=0 (1)

a:Ca(Xp,ap)=Da(ap+ ¯ap) (2)

(5)

Letusnowconsideracost-effectivepolicythatinternalisestheenvironmentalexternalitythrough aninternationalemissiontradingscheme,andinwhichregionsbehavenon-cooperativelyon inno-vation.Thesecond-bestleveloftechnologyinvestmentsandabatementinthecaseofnoconstraints (freetrade)ontheuseofcarbontrading,{xFT,aFT},isobtainedbyeachregionminimisingtheircosts,

includingthecostsofbuying(orprofitsofselling)CO2permitsatthepermitpricep,andtakingthe

otherregions’innovationexpendituresasgiven:p(a0− a)+C(X,a)+G(x).Theinitialendowmentof

abatementisequaltoa0intheHomeregionandto1− a0intheForeignone,wheretotalabatement

normalisedto1.ThisprocessleadstotheFOCs:

x:CX(XFT,aFT)+Gx(xFT)=0 (3)

a:Ca( ¯XFT,aFT)=pFT (4)

IftotalabatementissetsuchthatthemarginalabatementcostspequalsmarginaldamageD,then wehavethatxFT<xpif>0,sinceC

X(X,a)isincreasinginx.Thatis,whensettingthepermitpriceatthe

Pigouvianlevel,theinnovationlevelinthequotaschemewhereR&Dexpendituresaredeterminedina non-cooperativemannerislessthanoptimal.Toinducethefirst-bestlevelofinnovation,theemission quotaandtheresultingpriceneedtobemademorestringent.Thisisthestandardresulthighlighted inSection2.1.

Sincethefocusofthispaperistherelationshipbetweeninternationalpermittraderestrictionsand theincentivetoinvestininnovationwhenthereareinternationalspillovers,weconsiderthesame internationalquotaschemeasabove,butnowaddaconstraintonthenumberofemissionpermitsthat canbetraded.Wewanttounderstandwhethersuchrestrictionscanfostermoreglobalinnovation, andifsowhethertheyallowreducingthewelfaregapbetweenthefirstbestandthenon-cooperative solutions.Eachregionminimisestotalcostssubjecttotheadditionalconstraintthataminimumlevel ofabatement˛mustbeaccomplisheddomestically,aLIM˛.SupposetheHomeregionwillhaveto

abateatleast(aLIM=˛).TheFOCsthenbecome:

x:CX(XLIM,aLIM)+Gx(xLIM)=0 (5)

a:Ca(XLIM,aLIM)−=pLIM (6)

(aLIM−˛)=0 (7)

whereisthemultiplierassociatedwiththepermittradelimitintheHomeregion.Whenthe con-straintisbinding(aLIM=˛),themarginalabatementcostsacrossregionsarenolongerequal.Marginal

costsintheHomeregion,Ca(XLIM,˛)=pLIM+,becomelargerthanthoseintheForeignregion,where

theyequalthepermitprice,Ca( ¯XLIM, ¯aLIM)=pLIM.Atraderestrictionhasanambiguouseffectonthe

internationalpriceofcarbon.Ontheonehand,fewerpermitswouldbetraded,leadingtoalower inter-nationalprice.Ontheotherhand,alowerlevelofinnovationintheForeigncountrycouldincrease marginalabatementcostsandhencetheinternationalpriceofcarbon(seeLemma1).

Thebuyer(constrained)Homeregionwillincurintheextracostofraisingthedomesticeffortabove theefficientlevel,dependingonthestringencyofthetradeconstraint˛.Replacinga=˛and ¯a =1˛ intheFOCsforinnovationandbytotallydifferentiatingthem,theeffectofmakingthepermittrading capmorestringentoninnovationisimmediatelyobtained.

Lemma1. Inatradableemissionquotasystemwhereregionsbehavenon-cooperativelywithrespectto theinnovationexternality,restrictingemissionpermitstradingincreasesR&DinvestmentsintheHome restrictedregiondx/d˛>0ifinnovationreducesmarginalabatementcosts,CaX<0.InnovationintheForeign

unrestrictedregionisnon-increasinginthetraderestriction,d¯x/d˛≤0.

Lemma1indicatesthat,aslongasinnovationlowersmarginalabatementcosts,CaX<0,arestriction

onpermittradeincreasesR&Dinvestmentsintheconstrainedregion.Becausethetotalamountof abatementisfixedto1,theotherregionwillreduceabatementandthereforeinnovationwilldecrease. Thepreciseeffectalsodependsontheextentofspillovers.Foragivenpermittraderestriction,the largerthemagnitudeofspillovers,thehighertheincreaseininnovationintheHomeregionandthe reductionintheForeignone.

(6)

GlobalinnovationwillincreaseiftheadditionalinvestmentintheconstrainedHomeregionis higherthanthereductionintheunconstrainedForeignone,(dx/d˛)+(d¯x/d˛)≥0,orequivalentlyif theinnovationintheconstrainedregionisconvexinthelevelofpermittraderestriction.Thefollowing propositionclarifiesthiseffect.

Proposition1.

a. Inatradableemissionquotasystemwhereregionsbehavenon-cooperativelywithrespecttothe inno-vationexternality,restrictingemissionpermitstradingincreasesglobalR&Dexpenditures,(dx/d˛)+ (d¯x/d˛)≥0,if: −C˛X( ¯XLIM,1−˛) Gxx(¯xLIM)+(1−)Cxx( ¯XLIM,1−˛) < −C˛X(XLIM,˛) Gxx(xLIM)+(1−)Cxx(XLIM,˛) (i) b.TotalR&DalwaysincreasesinthecornersolutioncaseofzeroR&Dinvestmentsintheunrestricted

Foreignregion(¯x =0).

Limitingtradeaffectsglobalwelfarethroughtwochannels.Bynotequalisingmarginalabatement costsacrosscountries,itgeneratesanefficiencyloss.BymodifyingtheincentivestoinvestinR&D,itcan generateinnovationbenefits.Corollary1summarisesthetrade-offbetweenthesetwomechanisms.

Corollary1. Foradeviationfromthefreetradeabatementallocationduetothetradingrestriction,total costsdecreaseif:

C˛(X,˛)−C˛( ¯X,1−˛) CX(X,˛) + CX( ¯X,1−˛) CX(X,˛) dx d˛+ d¯x d˛>0 (ii)

Conditions(i)and(ii)dependonthespecificationoftheabatementandinvestmentcostfunctions. Therelationbetweentechnicalchangeandthemarginalcostofemissionreductionisparticularly importantincondition(i).Ifthemarginalbenefitofinnovationincreasesinabatement, thenthe numeratorof(i),−CX˛,wouldincreaseinthetraderestriction.4ThisisthecaseinwhichR&Dpivots

downthemarginalabatementcostcurve,thusreducingmarginalabatementcostsespeciallyforhigh levelofabatement.Forexample,thinkofaR&Dprogrammethatleadstotheadoptionofa break-throughtechnologywithverylowvariablecosts.Eveninthiscase,though,condition(i)wouldbemet onlyifCXXandGxxdidnotchangeintheleveloftraderestriction.Forexample,aconstantGxxwould

implyGtobeeitherlinearorquadratic,whichispossiblyareasonableassumption.Butthismight notapplytoCXXRecallinfactthat,whileinnovationcostsGdependonlyoninnovation,Cdepends

oninnovationandabatement.Abatementandinnovationbothincreasewiththetraderestriction, buttheireffectonCXXhasanoppositesign.WhethertotalR&Dincreasesinthetraderestrictionwill

dependontheassumedfunctionalformsoftheabatementandtheR&Dcostfunctions.

Condition(ii)statestherequirementforaglobalwelfaregain.Welfaregainscanariseifthe effi-ciencyloss(firstterm)iscompensatedbytheinnovationgain(secondterm).Letusassumethatthe traderestrictionisamarginalonefromthefreetradeallocation(e.g.,˛ ∼= 0.5).Thenthefirsttermin(ii) iszero.Thesecondtermisrelatedtothechangeinglobalinnovation((dx/d˛)+(d¯x/d˛))butwithan additionalweight,CX( ¯X,1−˛)/CX(X,˛),whichmeasurestherelativereturntoinnovationbetween

ForeignandHome.Inthecaseoffullspillovers(=1),thisweightisequaltooneandthencondition (ii)collapsestocondition(i).Thatis,foramarginaldeviationfromfreetradeandfullspillovers,when totalinnovationincreases,costsdecrease.Costwouldalsodecreaseincaseofacornersolutionof zeroforeigninvestment(¯x=0),orifspilloversorreturntoinnovationweresufficientlyasymmetric andlargerforForeignthanforHome(<¯ or CX<CX).Thiscouldindeedholdifwewereto

iden-tifyForeignwithdevelopingcountrieswhicharedistantfromthetechnologyfrontier,somethingwe representintheintegratedassessmentmodelusedinSection3.

For<1andsymmetricregions,however,CX( ¯X,1−˛)/CX(X,˛)islessthanoneandcondition(i)is

anecessarybutnotsufficientconditionfor(ii),sincetheinnovationbenefitsintheForeignregion(for

4RememberthatC

(7)

2 3 4 5 6 7 8 9 0.95 0.85 0.75 0.65 0.55 0.45 Total innovaon Trade restricon, α

Fig.1.TotalR&Dexpendituresforthecaseofb=3andc=1,forrangingfrom0(upperline)to1(lowerline)withincrements

of0.1.Thehorizontalaxisplotsthelevelofpermittraderestriction˛from0.5(norestriction)to1.Theverticalaxisplotstotal

innovation.ThediamondsshowthetotallevelofinnovationintheParetocase,forthecorrespondingvalueof.

amarginaldeviationfromfulltrade)aresmallerthanatHome,duetotheincompleteappropriability ofinnovation.Moreover,whendeviationsfromfreetradearenotmarginal,theefficiencylossdue tothemisallocationofabatement(firsttermof(ii))wouldnolongerbeinsignificant.Thus,wecan expectpermittraderestrictiontoleadtowelfaregainsonlyforsmallconstraintsandsufficientlyhigh spillovers.WefurtherexplorethisresultinthenextsectionforagivenfunctionformofC(X,a). 2.3. Anumericalexample

Letusassumethatabatementcostisapowerlawfunctionofabatementandinnovation,andthat thecostofinnovationequalstheR&Dexpenditures:

C(X,a)= ˇab

(X+1)c and G(x)=x

withbandc(greaterthan0)controllingthecurvatureofmarginalabatementandinnovation.Thisis aquitegeneralfunctionalformwhichisoftenusedincalibratedmodels.

Solvingfortheoptimallevelofinnovationforbothregions(x, ¯x),itisstraightforwardtoshowthat, foraninteriorsolution,condition(i)issatisfiedwheneverb/(c+1)>1.Itisalsoeasilyshownthat thisinequalityisalwayssatisfiedifweassumeC(X,a)tobestrictlyconvex,thatisifCXXCaa−CXa2 >0.

Thatis,incondition(i),thestrictconvexityofC(X,a)ensuresthattheoptimallevelofinnovationinthe constrainedregionisalsoconvexinthelevelofthetraderestriction.Thus,theincreaseininnovation intherestrictedregionmorethancompensatesthedecreaseintheunrestrictedone,evenwhenthisis greaterthanzero.Ifb/(c+1)=1theresponsesinthetworegionsareequal,andthustotalinnovation isconstantintherestriction,untilthepointwhere ¯xgoestozero.

WepictoriallyshowtheimpactofpermittraderestrictionsontotalinnovationinFig.1(forˇ=100, b=3andc=1)fordifferentvaluesoftheinternationalknowledgespilloverparameter,.Thechart showsthat,asexpectedsinceb/(c+1)>1,totalinnovationincreasesinthetraderestriction.The increaseismilderforinteriorsolutionsthanforcornersolutions,whicharereacheddependingonthe levelofrestrictionandofknowledgespillover.Sinceforthechosensetofparametersacornersolution inwhich ¯x =0isachievedforalllevelsofspillover,forstringentenoughtradecapstotalinnovation eventuallyconvergestothesamevalue,giveninthechartbythelowerline.Convergenceisfasterfor higherspillovers.Themagnitudeofspilloversisalsoveryimportantfordeterminingthedifferencesin theinnovationeffortbetweenthenon-cooperativesolutionsandtheParetoones,showninthechart bythediamonds.Asexpected,theunrestrictednon-cooperativecasesyieldglobalR&Dexpenditures

(8)

9 10 11 12 13 14 15 16 17 18 19 0.95 0.85 0.75 0.65 0.55 0.45 Total costs

Traderestricon, α

Fig.2. Totalcostsforthecaseofb=3andc=1,forrangingfrom0(upperline)to1(lowerline)withincrementsof0.1.The

horizontalaxisplotsthelevelofpermittraderestriction˛from0.5(norestriction)to1.Theverticalaxisplotstotalcosts.The

diamondsshowthetotalcostsintheParetocase,forthecorrespondingvalueof.

thatarealwayslowerthantheParetoones.Thisgapdependsontheextentoftechnologyspillovers. ThedistancebetweenParetoandnon-cooperativeinnovationiszerowhentherearenoexternalities, =0.Fig.1representsthiscasebytheupperlineanddiamond,whichinfactareperfectlyaligned.As spilloversincrease,thegapbetweenthediamondandthecorrespondinglineincreasesanditreaches itsmaximumwhenspilloversarefull,representedinthechartbythelowerline.Traderestrictions,by fosteringmoreR&Dexpenditure,helptoclosetheinnovationgapwiththeParetosolution.Asevident fromthechart,iftradeissufficientlyrestricted,totalinnovationcanincreasebeyondtheglobally optimalParetolevel.

Asasecondrelevantquestion,weaskwhethersuchtraderestrictionscanbewelfareimproving. Asdiscussedintheprevioussection,restrictingthepossibilityoftradingpollutionpermitswould introducetwosourcesofinefficiency.First,itwouldshiftabatementawayfromtheoptimalallocation ensuredbytheequalisationofmarginalabatementcosts.And,itwouldrequireadditionalresources forR&Dexpenditure.However,bypartlyinternalisingtheinnovationexternality,thetradeceiling couldprovidewelfarebenefitsandreducethewelfaregapbetweenthedecentralisedoutcomeand thegloballyoptimalsolution.

Fig.2reportsthetotalcostsforvariousrestrictionsandspillovers(andagainforˇ=100,b=3and c=1).Onecannoticethatforhighlevelofspilloversandmildrestrictionsonpermittrading,thetotal costsarelowerthanfortheunrestrictedcase(˛=0.5).Thatis,thereappeartobecaseswherepermit traderestrictionsarewelfareenhancing,andthesecasesareinlinewithwhatstatedincondition(ii). Asexpected,thecostsremainalwayshigherthanintheParetocase(againrepresentedbydiamonds), sincetraderestrictionisnotthemostefficientwaytointernalisetheinnovationexternality. Nonethe-less,theresultisinterestingsinceitpointstopotentialefficiencygainsinducedbythesecond-best policyoflimitingpermittrading.

Summingup,inthissectionwehaveshownthatforcertainassumptionsregardingtheway tech-nicalchangeaffectsmarginalabatementcost,restrictionsontheamountoftradableemissionpermits canincreasethetotalinnovationeffort.Wehavealsoshownthatthisincreasedinnovativeeffortmight leadtopotentialeconomicefficiencygains,ifpermittraderestrictionsaremoderateandspillovers sufficientlyhigh.Intheend,theoveralleffectofrestrictingtradeontheregionalandtotalcostsof theclimatepolicydependsontheparameterisationofthevariousfunctions,especiallytheimpactsof innovationontheperformanceofabatingtechnology.Albeitdifficulttocharacteriseempirically,an integratedassessmentframeworkisneededtomovebeyondthelimitationsoftheanalysismadein thissection,especiallyforwhatregardstheroleofthedynamics,therepresentationofthetechnology spilloversandtechnologicalchange,issuesofleakageandotherpotentialeffectssuchastheimpacton

(9)

internationalenergymarkets.Totacklealltheseissuesinaunifiedframework,weturntoacalibrated generalequilibriummodel.Wefirstintroducethemodelandthescenarioset-up,andthenprovide complementaryinsightstotheonespresentedinthissection.

3. Afullyintegratednumericalanalysis

3.1. Modelstructureandsetupofthenumericalassessment

ThenumericalanalysisusestheWITCHmodel,5anenergy-economymodelthatfeaturesmultiple externalities.AfulldescriptionofthemodelcanbefoundinBosettietal.(2006,2009a).Herewe brieflydiscusshowtheexternalitiesarerepresentedinthemodel.

WITCHisadynamic,optimalgrowthmodelwithafocusontheenergysectorandonGHGmitigation options.Itconsistsof12aggregatedregions,denotedwithn.Oneofitsdistinguishingfeaturesis thegame-theoreticset-upthatallowsproducingtwodifferentsolutions.Themodelcanberunina cooperativemodewhichjointlymaximisesglobalsocialwelfareandinternalisesenvironmentaland economicexternalities,generatingafirst-bestoptimum.Inaddition,itcanprovidethedecentralised, ornon-cooperativesolution,byoptimisingthewelfareoftheindividualregions,takingasgiveneach otherregion’schoice.Thisisdonethroughaniterativeprocedure,whichiscapableofreproducingthe outcomeofanon-cooperative,simultaneous,openmembershipgamewithfullinformation,andthus achieveasecond-bestNashequilibrium.Withthissetting,itispossibletoexploretheconsequences ofdealingwithdifferentsourcesofmarketfailure.Themodelspecificallyincorporatestheclimate changeexternality,thetechnologyexternalityviainternationalknowledgeandexperiencespillovers, andcarbonleakagethroughinternationalenergymarkets.

Theclimateexternalityisaccountedforbyadamagefunctiondrivenbyglobalaveragetemperature, computedthroughasimplifiedclimatemodule.Thispaper,though,providesacost-effectiveanalysis ofaclimatepolicyaimedatstabilisingtheconcentrationsofgreenhousegasesintheatmosphereat agivenlevel,6essentiallyputtingacaponthecumulativeCO

2-eqemissions.Inthesamespiritofthe

precedingsection,weassumeglobalemissionreductionsareachievedthroughaninternationalcarbon market,whichisoperativefrom2020.FollowingtheUNFCCCdistinctionbetweenindustrialisedand developingcountries(intheprecedingsectionHome andForeignregion,respectively),OECDare allocatedstrictreductionsrequiringa90%cutofGHGsby2050comparedto2005levels.Developing countries(non-OECD)areallowedtoincreaseemissionsby20%which,comparedtothemodelbaseline projections,impliesareductionofalmost50%,7orofabout24GtCO

2,comparedto22GtCO2forthe

OECD.

Thetechnologyexternality,inWITCH, ismodelledviainternationalspilloversofknowledgeand experienceacrosscountries(Bosettietal.,2008).Themodelincludesanumberofdifferentlow-carbon mitigationoptions,tech.Ineachcountry,theirtechnologylevel(xtech,n,t)isdescribedbytheglobal

stockofenergyR&D(R&Dtech,n,t),whichmeasuresknowledge,andbythecumulativeinstalledcapacity

(CCtech,n)whichisaproxyofexperienceorLearning-by-Doing.Ateachpointintime,newideasadd

totheexistingR&Dstock,whichdepreciatesattherateı.Thegenerationofnewideasisdescribed bytheinnovationpossibilityfrontier,whichismodelledasaCobb-Douglascombinationbetween domesticinvestments(I R&Dtech,n,t),domesticknowledgestock(R&Dtech,n,t),andforeignknowledge

stock(SPILLtech,n,t):

R&Dtech,n,t+1=R&Dtech,n,t(1−ı)+IR&D˛tech,n,tR&D ˇ

tech,n,tSPILLεtech,n,t (8)

SPILLtech,n,t=



R&Dtech,n,t j∈OECDR&Dtech,j,t

(



j∈OECDR&Dtech,j,t−R&Dtech,n,t) (9)

5Seewww.witchmodel.orgformodeldescriptionandrelatedpapers.

6Specifically530ppmofCO

2-eq,or3.5W/m2.

7Commitmentstartingdateshavealsobeendifferentiatedbetweendevelopedanddevelopingregions,witha10year-delay

innon-OECDregions.Weassumethatbeyond2050,allregionswillcooperatetoreachthelong-termstabilisationtargetand

(10)

The spillover term (SPILLtech,n,t) depends on the interaction between the countries’

absorp-tive capacity, measured by the ratio of the stock of the country to that of the frontier



R&Dtech,n,t/



j∈ OECDR&Dtech,j,t



,andthedistanceofeachregionfromthetechnologyfrontieritself



j∈ OECDR&Dtech,j,t−R&Dtech,n,t



.Thetechnologyfrontieriscalculatedasthetotalstockof knowl-edgeinOECDcountries.Thisformulationimpliesthatforeignknowledgecanbringsomebenefitsonly ifadomesticR&Dprogrammeisestablished.Knowledgespilloversarenotcompletelyfree,butthey requirededicatedinvestments.Inthenotationofthemodelusedintheprevioussection,thiscase correspondsto<1.8

Incontrast,spilloversofexperiencearecompletelyfreewithinthetimeframeofthemodel,which isoffiveyears,t=5.Worldcumulativeinstalledcapacityintechnologytechattimetiscomputedby addinginvestmentsinallregions(Itech,n,t):

CCtech,t+1=



t



nItech,n,t (10)

Globallearningimpliesthatinvestmentsineachgivenregionequallycontributetheformationof aglobalstockofexperience,whichinturnaffectsinvestmentcostsinallregions,asdescribedinEq. (11).Inthenotationofthemodelusedintheprevioussection,thiscasecorrespondsto=1.Regarding themannertechnologylevel(xtech,n,t)impactsthemarginalcostofabatement(Cax),theWITCHmodel

distinguishesbetweentechnologicalchangefordecarbonisingtheenergysystemandforimproving energyefficiency.

Decarbonisationoccursbecauseahighertechnologylevelleadstolowerinvestmentcostsof low-carbontechnologies.Thisprocessisdescribedbyaso-calledtwofactorlearningcurveofthisform:

ICtech,n,t ICtech,n,0 =



R&Dtech,n,t−2 R&Dtech,n,0

−c

CCtech,t CCtech,0

−b

(11) ThetwoexponentsaretheLearning-by-Doing(LBD)index(−b)andtheLearning-by-Researching (LBR)index(−c).Theydefinethespeedoflearning.Moreprecisely,thelearningratios,definedasone minustheprogressratio,LBRratio=1−2−c;LBDratio=1−2−b,determinetherateatwhich invest-mentcostsdeclineeachtimethecumulativecapacityortheknowledgestockdoubles.Atwotime period(correspondingto10years)lagisassumedforR&D,tocapturetheinertiaofbringingresearch tothemarket.Sincethetechnologiesthataresubjecttolearningareassumedtobeofrenewabletype, theinvestmentcostscoincidewiththemarginalcosts.Inthisformulationtechnicalchangereducesthe marginalcostofabatement,asinthemodelpresentedinSection2,wheretheroleofthisassumption wasdiscussed.

Energyefficiencyismodelledthroughimprovementsintheproductivityoftheenergyinput(ENn,t)

intheproductionofthefinalgoodsector.Theenergyknowledgestock(R&DEE,n,t)augmentsthe

quan-tityoffinalenergyservices(ESn,t)thatcanbeprovidedofeachgivenunitofphysicalenergy(ENn,t)

accordingtoaconstantelasticityofsubstitutionproductionfunction: ESn,t=



a1R&DEE,n,t+a2ENn,t

1/

(12) Finally,thelastchannelofinteractionacrossregionsisthatoftheinternationalenergyandcarbon market.InternationalpricesoffossilfuelsandCO2aredeterminedbytheequilibriumofdemandand

supplyatthegloballevel.Thus,differentdomesticabatementgoalsasaresultofconstraintsonthe amountofcarbontradingcanhaveanimpactonthepriceandtradeofsuchgoods.

Thestructureofthisnumericalmodelallowsustoconsiderthemostsalientfeaturesofthe cli-matemitigationpolicy,andtodosoinasecond-bestsetting.Althoughthismodelprovidesamuch richerframeworkthananalyticalonesandrepresentsastep-upoverstandardintegratedassessment modellingthatnormallyfeaturesonlytheclimateexternality,itmustnonethelessberecognisedthat itdoesnotthoroughlyrepresentasecond-bestworld.Forexample,nointernationaltradeofcapital

8Withtheparameterwegenerallydenotestheextentofspillovers,followingthenotationusedinSection2.InWITCHthis

(11)

Table1

Globalandregionalmacroeconomiccostsofclimatestabilisationpolicyunderthe“fulltradehypothesis.”Consumptionlosses

discountedat3%discountrate,2005–2100.

Consumption OECD Non-OECD World

2005US$Trillion 26.19 1.88 28.07 PercentageofBaUconsumption 1.68% 0.18% 1.07% Lossesaregrossofclimatechangedamages.Negativenumbersindicategains.

Table2

Internationalcarbonmarketunderfulltrade. Mktsize (GtCO2-eq) GtCO2-eq abated Internationally tradedabatement Carbonprice (2005 US$/tCO2) Mktvalue (2005US$ Billion) Mktvalue (%ofGWP) 2020 2.8 7 41% 4 11 0.02% 2025 3.0 14 22% 26 76 0.10% 2030 3.3 20 16% 58 193 0.22% 2050 7.1 46 16% 319 2260 1.74%

isassumedandpre-existingdistortionarytaxesandsubsidiesarenotaccountedforinthisexercise.

Ourmodelcapturesmarketfailuresrelatedtointernationalspilloversonlyintheenergysector,as

nogeneralpurposeR&Disassumed.Nolearningisconsideredforknown,yetpotentiallyimprovable

technologies,suchasnuclearpowerandcarboncaptureandstorage(CCS).Thus,thisexerciseprovides

anaccountofonlysomeofthemostrelevantsourcesofglobalinteraction.

3.2. Climatepolicyunderfullinternationalcarbontrading

Webeginbyevaluatingtheimplicationsofacost-effectivepolicyimplementedthroughan

inter-nationalquotaschemeinthecaseofnoconstraintsontheuseofcarbontrading(fulltrade).9 As

showninTable1,giventheregionalabatementcommitments,OECDcountriesfacethehighestcosts non-OECDcountriesslightlygainuntilmidcentury,mostlyduetopermitrevenues,butloseinthe long-term.

The1%globalfigurelieswithintherangereportedintheliterature(seeforexample,IPCC,2007; Ludereretal.,2009).Lowcostsareduetothescenariodesign,theassumptionofimmediate partici-pationandoffullflexibilityamonggreenhousegasesabatementoptions.Departurefromanyofthese assumptionsincreasescostssubstantially(Bosettietal.,2009b;Clarkeetal.,2009).

Withoutanyrestriction,countriesrelyontheinternationalcarbonmarket,andthissuggeststhatit isanimportantelementofeconomicflexibility.Table2showstheevolutionofthecarbonmarketover time.Theinitiallylowcarbonprice(duetotheassumptionofglobalparticipationandwidebasket ofGHGs)issuchthatthemarketvalueiscontainedattheoutset,butgrowssignificantlyovertime, drivenbytheconvexpathofthecarbonprice.Themarketsizeisalreadysignificantin2020,when almost3BilliontonnesofCO2(equaltoroughlyhalfoftoday’semissionsintheUS)areexchanged

internationally,enoughtorequiretheestablishmentofconsiderableinstitutionaland monitoring capacity.

However,aperfectlyfunctioninginternationalcarbonmarketisnotlikelybeavailablein2020. Recentnegotiationshaveshownthatissuesofmonitoringandverificationarequitecomplextobedealt withattheinternationallevel.Thelackofinstitutionalcapacity,reliablemonitoring,andverification systemsinsomecountriesmightdelayorfragmenttheestablishmentofaglobalmarket.Or,countries thathavealreadyimplementedemissiontradingschemes,suchastheEU,mightlimittheamountof internationalcreditsthatcanbeusedinthedomestictradingscheme.Theconsequencesofthesethree possibleoutcomesaredescribedinthenextsection.

9Fullcooperationisassumedtobesupportedbyafullyfledged,perfectlyfunctioning,internationalcarbonmarket,whichis

(12)

Table3

Regionalmacroeconomiclossesofclimatepolicywhentheuseofinternationaloffsetsisdelayed(“when”dimension).

Con-sumptionlossesdiscountedat3%discountrate,2005–2100.

OECD Non-OECD World

Fulloffsets 1.68% 0.18% 1.07%

From2030 1.81% 0.11% 1.12%

From2035 1.90% 0.08% 1.17%

From2040 2.26% 0.05% 1.37%

From2045 2.70% 0.08% 1.64%

Negativeentriesindicateconsumptiongains.

3.3. Delayed,fragmented,andconstrainedinternationalcarbontrade

Followingtheabovearguments,letusassumethattheestablishmentofaninternationalcarbon

marketisdelayedtoafuturedatebetween2030and2045(wenamethisthe“when”scenario).Table3

reportsconsumptionlosseswhenaccesstointernationaloffsetsisdelayedovertimeby10,15,20,or 25years.Postponingtheestablishmentofaninternationalcarbonmarketincreasesglobal macroe-conomiccostsmoderatelyfordelaysuntil2030–2035,butwaitinguntil2045raisesthepolicybillby about50%.ThisisdrivenbyadditionallossesinOECDregions.Instead,innon-OECDthe(moderate) consumptioncostsarehalved,reducingtheglobalpenalty.Welfareimprovementsintheseregions areasecond-bestresult,drivenbypositivetechnologyspilloversandlowerenergyprices.Theyalso dependontheassumptionthatnon-OECDcountriescommittomilderemissionreductionobjectives comparedtoOECD.

Whentradeisrestricted,moreexpensivemitigationoptionsareadopted,drivingupcompliance costs.10Withlimitedaccesstointernationaloffsets,OECDregionsfurtherdecreasetheirdemandof allfossilfuels(traditionalgas,coal,andoil),oilinparticular.ThereduceduseofoilinOECDcountries lowersitsinternationalprice,benefitingnon-OECDregions.Thiseffectisalsoknownasenergymarket effect,whichisoneofthepossiblesourcesofcarbonleakage.Asecondchannelthatisnotdiscussed inthispaperworksthroughchangesinterms-of-tradeandinternationalcompetitivenessof energy-intensiveindustries.11ThecontractionofoildemandinOECDcountriesisonlypartiallyoffsetby carbonleakageinnon-OECD,whichincreasestheoildemandby11%(onaverage)between2020and 2040.Fig.3,upperpanel,showsthatintheextremecaseofa2045delay,thesetwoeffectsresultina globaldemandcutbackof16%in2040.Theinternationalpriceisalmost4%lowercomparedtothefull tradecase.Itshouldbementionedthatmostoftheoilpricereductionoccurswhenimplementinga stabilisationpolicy,alreadyinthefreetradescenario,wheretheoilpricefallsbyabout30%in2050and 70%in2010comparedtothebaseline.Forthisreason,theadditionalpricereductionduetorestricted emissiontradeisrelativelycontained.

AsecondresponsetothetraderestrictionistoinvestmoreincleanR&Dandcarbonfreepower gen-erationtechnologies.ThisfindinggeneralisestheresultinProposition1.Aslongastechnicalchange reducesmarginalabatementcosts,permittradelimitsstimulateinvestmentsintechnologiessubject totechnologicalchange.Letusconsider,forexample,thedynamicsofinvestmentsinrenewable elec-tricity,whosemarginalcostdecreaseswiththeglobalcumulativecapacity,asdescribedinEq.(11).The globalpathofinvestmentsinwindandsolararedepictedinFig.3,bottompanel.Duringthetransition periodwithoutpermittrading(2020–2040),OECDinvestmentsinwindandsolarareraisedtomatch themorestringentabatementtarget(

x/

˛>0).Investmentsinnon-OECDinitiallyremainatlevels comparableorslightlylowerthanthefulltradecase,buteventuallyincreaseaswell.Thispatternis duetothereductioninglobalinvestmentcostscausedbyadditionalinvestmenteffortintheOECD. Investmentsdropbelowthefulltradecaseonlyin2040,inanticipationoftheopeningtointernational

10Weanalysetheseresultsforthemoststringentcaseinwhichtradestartsin2045.Theothercaseshavesimilar,butsmaller

magnitudeeffects.

11BurniauxandOliveira-Martins(2000)assessedtherelativecontributionofeachofthetwochannelsandconcludedthat

(13)

-5.00% -4.00% -3.00% -2.00% -1.00% 0.00% 1.00% -20% -15% -10% -5% 0% 5% 2040 2035 2030 2025 2020 2015 2010

Percentage change w.r.t. full trade

Percentage change w.r.t. full trade

World consumption International price (Rhs axis)

-4.25% -2.25% -0.25% 1.75% 3.75% 5.75% 7.75% -15 -10 -5 0 5 10 15 20 25 2010 2015 2020 2025 2030 2035 2040

Percentage change w.r.t. full trade

Difference w.r.t. full trade(US$ Billion)

Additional global investments Global investment cost reduction (RHS axis)

Fig.3. Internationaloilmarket(leftpanel)andrenewableelectricity(rightpanel)whendelayingtheuseofinternationaloffsets

to2045.

offsetsin2045,buttheyremainhigherthereafter.Theadditionalinvestmentcostreductionforwind andsolarinducedbythetraderestrictiondecreasesovertimebecauseofthedecreasingmarginal benefitsofinvestments,butnonethelessremainspositivethroughoutthecentury.Suchpath depen-dencieshelplowertheclimatepolicycostsinthelong-term,butrequireanadditionaleffortinthe beginning,whichismostlyfacedbydevelopedcountries.Asimilarresultisobservedforinvestments inenergyR&D,whichgloballyincreasethroughoutthecentury,butbyalargermagnitudeduring thetransitionperiodwithoutemissiontrading.Theglobalincreaseismostlydrivenbyadditional investmentsinOECDcountries.Non-OECDcountriesslightlyreducetheirR&Deffortinthe short-run,butincreaseitafter2035.Sincebothknowledgestockandnewinstalledcapacityinlow-carbon technologiesgenerateinternationalpositivespillovers,OECDadditionalinvestmentsreducemarginal abatementcostsinallotherregionsaswell.Inotherwords,theadditionalcostspaidbyOECDregions leadtoadditionalbenefitsinnon-OECD.

Anothersituationthatcouldariseisafragmentedinternationalcarbonmarket,atleastduring thetransitiontowardsmoreconsolidatedprocedures.AsnotedbyFrankel(2008),itcouldbethat theeligibilitytosellpermitsisrestrictedtocountrieswithgoodinternationalgovernanceratings,or thathavecommittedtoinvesttherevenueingreenprojects,orthathavedemonstratedabilityto abidetocommitments.Weanalysetheimplicationsofdelayingtheparticipationofthreegroupsof developingregionsthatintheglobalcarbonmarketwouldplaytheroleofnetsellers:EnergyExporting Regions(EEX:MiddleEastandNorthAfricaandRussia),DevelopingAsia(DA:ChinaandIndia)and RestofWorld(ROW:LatinAmericaandSub-SaharanAfrica).Inthisscenario,aninternationalmarket

(14)

Table4

Regionalmacroeconomiclossesofclimatestabilisationpolicywhenselectedregionsjoincap-and-tradein2045(“where”

dimension).Consumptionlossesdiscountedat3%discountrate,2005–2100.

OECD Non-OECD World

Fulloffsets 1.68% 0.18% 1.07%

DAjoinin2045 1.82% 0.12% 1.14%

ROWjoinin2045 1.73% 0.16% 1.10% EEXjoinin2045 1.73% 0.17% 1.10% Negativeentriesindicateconsumptiongains.

Table5

Regionalmacroeconomiclossesofclimatestabilisationpolicywhenlimitingtheuseofinternationaloffsetsuntil2045(“how much”dimension).Consumptionlossesdiscountedat3%discountrate,2005–2100.

OECD Non-OECD World

Fulloffsets 1.68% 0.18% 1.07%

20% 1.76% 0.14% 1.11%

15% 1.84% 0.12% 1.15%

10% 1.97% 0.11% 1.22%

0% 2.70% 0.08% 1.64%

Negativeentriesindicateconsumptiongains.

isassumedtoalreadyexistin2020,buttheaboveregionsjoinin2045(wenamedthisthe“where”

scenario).

Excludingeitherregionfromtheinternationalcarbonmarkethaslimitedeffectsonglobal

macro-economic costs (Table 4), essentially becausethe others compensate the absenceof a seller by

increasingsupply.Theonlycasethatleadstoavisibleadditionalglobalpenaltyistheexclusionof DevelopingAsia,whichcoversmorethan50%ofthedemandofpermits.Forexample,thepriceof carbonin2025woulddoubleifDevelopingAsiawerenotincludedinthemarket,whereasitwould increasebyonlyafewpercentagepointsifotherregionsweretodelay.Thisresultstemsfromthefact thatemergingeconomiessuchasChinaandIndiaareassumedtohostthelargestbaseofmitigation opportunities.ItalsoreinforcestheknownargumentthatanyclimatepolicyhastolookEasttobe effective.

Section3.2hasshownhowrapidtheexpansionofthecarbonmarketcouldbe(seeTable2).Itcould actuallymobilisehugefinancialresources.WefindthatthenetoutflowsfromtheOECDregionsto developingcountriescouldincreaseuptoUS$1.7Trillionin2050,morethan2%ofOECDGDPinthat year.Therefore,OECDcountriesfaceatrade-off.Ontheonehand,laissezfairwouldmaximise cost-effectivenessandcontaineconomiccosts.Ontheotherhand,itwouldentailsubstantialfinancialflows, whichmayunderminethewillingnessofindustrialisedcountriestocommit.Iftheclimateexternality weretheonlydistortionintheeconomy,restrictionstocarbontradewouldcreateawedgeinmarginal abatementcostsacrosscountries,reducingeconomicefficiency.However,sincetherearetechnology externalities,restrictingtrademightbeasecond-bestpolicyrecommendation.

Theremainderofthesectionexploresthiscase.Weassumethataninternationalcarbonmarket isalreadyestablishedin2020,butquantitativerestrictionsontheuseofinternationalcarboncredits areenforced(wenamethisthe“howmuch”scenario).Until2045,onlyafractionofabatementfrom 20,15,10,0%ofregionalabatement,canbemetthroughtheinternationalcarbonmarket.The0%case coincideswiththescenarioinwhichtradeispostponedto2045(the“2045”caseconsideredabove). ThestylisedanalysisinSection2alreadypointsoutthatrestrictingtradetendstocauseglobal effi-ciencylosses,unlesstherestrictionismildandspilloversaresufficientlyhigh(seeFig.2andCorollary 1).Table5confirmsthat,whenonlyafractionoftotalabatementineachregioncanbemetwith internationaloffsets,globalconsumptionlossesincrease.Amildrestrictionsuchas20%createscosts thatarecomparabletopostponingtradeto2030(1.12%consumptionloss).Tighterrestrictions(10%) havecostscomparabletopostponingtradetoatleast2035(1.17%consumptionloss).

Thedynamicsofregionalcostsdependsnotonlyontheinteractionsbetweentheeffectsoutlined before(energymarketandtechnologyspillovers),butalsoonthetradingpositionofeachregion.In

(15)

0 20 40 60 80 100 120 140 2030 2025 2020 $/tCO2e Europe USA Global

Fig.4.MarginalcostofcarboninEuropeandUSfora15%limitoninternationaloffsetsuntil2045,andglobalcarbonprice

whenthemarketisnotrestricted(Global).

theshort-term,netsellers(non-OECD)tendtolosebecauseofthecontractionofthemarketand,as aconsequence,ofthecarbonprice.Thenon-OECDregionssupplyisrationedandthereforetheyemit more.However,theyanticipatethatafter2045therewillbealagerdemandofpermitsandtherefore theystillkeepemissionsbelowtheinitialallocation.Throughoutthecentury,limitingtradeto10%of regionalabatementincreasesnon-OECDwelfarebyabout40%becauserevenuelossesonthecarbon marketareoffsetbytheenergymarketeffectsandtechnologyspillovers.

Theeffectonnetbuyers(OECD)dependsonwhetherthelimitisbindingornot.Whenthelimitis binding,theregionlosesbecausemoreexpensiveabatementoptionshavetosubstituteforimported credits.Ifthelimitisnotbinding,theregiongainsbecausecarbonpermitsarecheaper,pLIM<pFT.

Withoutanyceiling,andgiventhisallowancesdistribution,developedcountrieswouldfinditoptimal tobuybetween18and54%oftheirabatementontheinternationalmarket.Thisimpliesthatamild restrictionof20%isbindingformostdevelopedcountries,especiallyintheshort-runandinEurope. OnlytheUnitedStatesisrightatthelimit,andin2030the20%restrictionisnotbinding.12Asa consequence,becauseequivalentoffsetrestrictionsleadtorelativelymoreabatementinEuropeand otherOECDcountries,moreR&DandcleaninvestmentsareinducedoutsidetheUS.A15%limiton tradewouldraisethecostofcarbonsignificantlymoreinEuropethanintheUS,asshowninFig.4. IntheUS,the15%restrictionisbinding,butspilloversandtheenergymarketeffectscompensatethe additionalcosts,leadingtoshort-termwelfaregainsandonlytoa1%long-termadditionalpenalty.As aconsequence,theUSwouldhaveeconomicbenefitsfromimposingaslightlyhighercarbontaxand relyinglessoninternationaloffsets,providedEuropeandtherestofOECDwouldalsobewillingto adoptequallystringent(butwithahigherimpactonmarginalcostsofabatement)measures.Thislast conditionisimportantbecausetheUSwouldprofitfromthereductioninoilpricesandininvestment costsoflow-carbontechnologiesinducedbythetightercommitmentintherestofOECD,too.

Furtheranalysis13indicatesthatthetechnologyspillovereffectissomewhatlargerthantheenergy marketeffect,thoughinbothcasestheefficiencylossesaresmallcomparedtotheoverallpolicy costsshowninTable2byafewpercentagepoints.Thereasonforthisisthatthestringentclimate stabilisationpolicyconsideredinthispaperrequiresadrasticswitchoftheenergysystemfromafossil fuelbasedtoalow-carbononeeveninthecasewithfullaccesstointernationaloffsets.Thislimitsthe additionalroomforreductioninthepricesofenergyandlowcarbonfuelswhentradeisconstrained.

12ItisinterestingtonotethatBuonannoetal.(2000),inthedifferentpolicysettingoftheKyotoProtocolandwithadifferent

modelwithoutinternationaltechnologyspilloversfoundasimilarresultfortheUS.

13Wequantifythecontributionoftheenergymarketeffectandtechnologyspillovers,byconsideringthreevariationsofthe

15%scenarioinwhicheitherorbotheffectsareswitchedoff.Specifically,weranthreeadditionalsimulationsofthe15%trade

restrictioncaseinwhichwefixedthecostsofeitherfossilfuelsorlow-carbontechnologiescharacterisedbyspilloverstothe

(16)

ThetechnologyspilloverschannelisfoundtobehigherinOECDregions.Despitebeingclosertothe technologyfrontier,developedcountrieshavetocomplywithaverystringentdomesticcommitment (especiallywhentradeislimited)thatrequiresfastandmajorinvestmentsinlow-carbonmitigation options,suchaswindandsolar.Thus,thegainsofequalreductionofinvestmentcostsin renew-ables(givenbytheassumptionofgloballearningbydoing)benefitsdevelopedcountriesmorethan developingones.Onthecontrary,theenergymarketeffectislargerfornon-OECDcountries,which areconfrontedwithlessdemandingmitigationobligationsandcanprocrastinatetheconsumptionof fossilfuels.

4. Sensitivityanalysis

Thenumericalanalysisdescribedintheprevioussectionhasshownthat,whenthereare inefficien-ciesinothermarkets,limitingpermittradecanincreaseglobalinnovationandtechnologicalchange, leadingtopotentialefficiencygains.Welfareeffectstendtobenegativeinbuyerregions(OECD),but positiveinnon-OECD(seeforexampleTable5),becauseofloweroilpricesandhigher technologi-calspillovers,thelatterinducedbythemoreinnovationbeingcarriedoutintheOECDregions.The resultspresentedinSection3.2extendthevalidityofcondition(i)and(ii)toamoregenericframework characterisedbymultipleexternalitiesanddifferentabatementoptions.

Inthisremainingsectionwetestwhetherthenumericalresultsareidiosyncraticduetospecific parameterchoicesbyvaryingthemostrelevantcalibratedinputs.Wefocusthesensitivityanalysison theextentofinternationalknowledgespillovers(theparameterεinEq.(8))andtheimpactof inno-vationandexperienceontheperformanceofabatingtechnologies(theLBRandLBDindicesdenoted withtheparameterscandbinEq.(11)).Thesetwoparametersarekeybecausetheyaffecttheextent towhichknowledgeandexperiencereducemarginalabatementcosts,inournotationCax.

Itisimportanttostressthedifferentroleoftheextentofspilloversandthelearningrates.Spillovers ofexperiencearefree,orusingthenotationofSection2,is1.Learningisassumedtobeaglobal processthatreducestechnologycostsanywhereintheworld,nomatterwhereinvestmentsoccur.We assumetherearenobarrierstothediffusionofnewtechnologieswhich,oncecommercialised,canbe adoptedwithinthetimeframeoffiveyears.Incontrast,knowledgespilloversrequireR&Dcapacity atHomeand,usingthenotationofSection2,islessthan1.Inthesensitivityanalysiswevarythe extentofinternationalspilloversofknowledge(thevalueofεinEq.(8)),butnotthatofexperience spillovers.Forwhatconcernstheroleofexperience,weexploredtheimpactofdifferentdegreeof learning.

Theempiricalevidencethatguidestheselectionoftheparametervaluesforinternationalspillovers ofknowledgeisscarce,althoughanincreasingnumberofpaperssupporttheirexistence(starting fromGrossmanandHelpman,1991,tothemorerecentcontributionsofKeller,2004andVerdolini andGaleotti,2009).Inaddition,severalstudiessupportthehypothesisthatdomesticR&Deffortis neededtoabsorbinternationalknowledge(startingfromSchmookler,1966tothemorerecentworks byKeller,2004;Dechezleprêtreetal.,2008;Seresetal.,2009;VerdoliniandGaleotti,2009).Asa centralvalueinthemodel,weassumealowenoughvaluecomparedtothecontributionofdomestic investmentsandpastknowledgestock,ε=0.15.Toaccountfordiminishingreturns,thesumofthe elasticitiesofthethreeinputsinEq.(8)islessthanone.Thesensitivityanalysisonεcoversasufficiently largespectrumofvalues,from0.075(−50%)to0.22(+0.50%).Thelatterisaconsiderablyhighvalue, implyingacontributionlargerthanthat ofdomesticinvestments(0.18)andhalfthatofthepast knowledgestock(0.53).

ThelearningratiosinEq.(11)describetheeffectivenessofinnovationprecedingthebreakthrough (Learning-by-Researching,c)andofexperiencefollowingtheadoptionofthetechnology (Learning-by-Doing,b)atreducingthemarginalcostoftheabatingoption.Learning-by-Doingisglobal(=1) whereasLearning-by-ResearchingismostlydrivenbydomesticR&Dinvestmentsbecausethe con-tributionoftheinternationalpoolofknowledgeis only0.15.Theempiricalevidenceonlearning (by-Doing) rates spansrelativelywide ranges.On average, the costsof energy supply technolo-giesdeclineatratesof16±9%witheachdoublingincumulativeproduction(Weissetal.,2010). The value chosen in WITCHis lower than those typically estimated in single factor experience curvessincetechnologicalprogressresultsinpartfromdedicatedR&Dinvestments.Theliteratureon

(17)

Table6

Keyparametervaluesandsensitivityanalysis.Learningratios(LBRandLBD)andcontributionofspilloverstoknowledge

production(ε). Central[−50%to+50%] Literature LBR(parametercinEq.(11)) 0.13 0.13a [0.07–0.19] [0.07–0.19] [LBRm50–LBRp50] LBD(parameterbinEq.(11)) 0.1/0.14 0.16b [0.05–0.14]/[0.07–0.20] [0.03–0.25] [LBDm50–LBDp50]

InternationalknowledgespilloversSPILL(parameterεinEq.(8)) 0.15

[0.075–0.225] [SPILLm50–SPILLp50]

aVarioussources.ForspecificreferencesseeBosettietal.(2009a).

bWeissetal.(2010).

Learning-by-Researchingpointsatanaveragelearningrateof13%.Table6summarisesthecentral valueschoseninthemodelalongwiththerangeexploredinthesensitivityanalysis,whichcoversthe valuesreportedintheliterature(lastcolumn).

Therobustnessofournumericalresultstothesekeyparametersistestedbysimulatinganumber ofadditionalscenariosinwhichLBR,LBD,andSPILLareperturbedtothevalueindicatedinTable6. Wealsoanalysetheextremecasesinwhichtechnologicalexternalitiesareexcluded.Weconsiderthe fourceilingsontradeexaminedinSection3.2,namely0,10,15,and20%.Itshouldberecalledthat theselimitsareremovedafter2045,whenfulltradeisallowed.

Fig.5showstherelationshipbetweeninvestmentsininnovationandcleanenergytechnologies andpermittraderestrictions.Itconfirmsthepatternobtainedwiththesimplernumericalmodel,in Fig.1.LimitingtradeincreasesR&Dandinvestmentsintechnologiessubjecttotechnologicalprogress, foralltheparameterisationsconsidered.Fortraderestrictionsbetween20and15%,varying knowl-edgespillovershasmildimpactsoninnovationinvestmentsandallmarkersareclusteredaroundthe

0 50 100 150 200 250 300 20% 15% 10% 0% US$ B illio n s -d iff er ence comp ar ed to fu ll tr ade Traderestricon, α LBDp50 LBDm50 LBRp50 LBRm50 SPILLp50 SPILLm50 SPILLm25 SPILLp25 LBDp25 LBDm25 LBRm25 LBRp25 LBDm40 LBDp40 LBRm40 LBRp40 Central w/oLBD and LBR w/oLBD,LBR,SPILL

Fig.5. Technologicalchangeforatraderestrictionof20%,15%,10%and0%until2045,andforrangingparameters.Differencein

cumulativeinvestments(2005–2040)inenergyR&Dandrenewableelectricityrelativetothefreetradecase(Billions2005US$).

(18)

-20 -15 -10 -5 0 -2.50 -1.50 -0.50 0.50 1.50 2.50 Central LBDp50 LBRp50 LBRm50 SPILLp50 LBDp25 LBRm25 LBDm25 LBRp25 SPILLm50 SPILLm25 LBDm40 w/oLBD,LBR, SPILL w/oLBD andLBR LBRm40 LBDp40 LBRp40 LBDm50 SPILLp25 Disc ounted consumpon in OECD. Differenc e relave to the full trade case ( Trillions 2005U S$)

Discountedconsumpon in non-OECD. Difference relavetofull trade (Trillions2005US$)

Fig.6.Efficiencygainstraderestrictionsequalto20%(topmarkers),15%,10%,and0%(bottommarkers)forranging

param-eters.Lowerleftquadrant:worldwithouttechnologyexternalities.Lowerrightquadrant:differentparameterisationsofthe

technologymechanismsinthecalibratemodel.Differenceindiscountedconsumptionrelativetothefreetradecase(Trillions

2005US$).TrianglemarkersdenotesensitivityonLBDrates,diamondsonLBRrates,starsonSPILL.Greenmarkersdenotean

increaseintheparametervaluewhereaspinkadecrease.Thefouryellowdiamondsrepresentthecentralcase.Thebluecircles

arecasesinwhichspilloversareswitchedoff.(Forinterpretationofthereferencestocolourinthisfigurelegend,thereaderis

referredtothewebversionofthearticle.)

centralvaluecase(yellowdiamond).Forthestricterlimitof0%,thechartshowsthatlarger incre-mentsininvestmentsoccurwhentheextentofspilloversisincreased(greenstars).Whentheextent ofspilloversislowercomparedtothecentralcase,technologyinvestmentsstillincrease,butbyaless amount.Thisisshownbythepinkstarslocatedbelowtheyellowdiamonds(centralcase).Limiting tradealwaysstimulatesglobalinvestmentsintechnologicalchange.Identifyingunderwhat condi-tionsthehighestincrementoccursiscomplicatedbythepresenceofseveraltechnologychannelsas wellastheothermechanisms,namelytheinternationalmarketofcarbonallowancesandoffossil fuels.

RegionalandglobalwelfareanalysisisillustratedinFig.6,whichshowstheeconomicefficiency gainsfromlimitedtrade.Moreprecisely,itshowsconsumptionchangesindiscountedUS$Trillion comparedtothefreetraderegime,forvariouscarbontraderestrictions.Positivenumbersaregains, indicatingthatconsumptionwithrestrictedtradeishigherthanwithfreetrade.Thevaluesonthe axesareconsumptionchangesforOECD(y-axis)andnon-OECD(xaxis).Togiveasenseofmagnitude, globalpolicycostsunderfreetradeamounttoaconsumptionlossof28US$Trillion(ofwhich1.88 innon-OECD, seeTable2).Thechartdepictstwodifferenthypotheticalworlds.In thelowerleft quadrant,aworldwithouttechnologyexternalitiesisshown.Inthelowerrightquadrant,different parameterisationsofthethreetechnologymechanismsincludedinthemodelarepresented.Efficiency gainsvarywithrangingtradelimits(vertically,fromthetopmarkerswhichrepresentthe20%limit tothebottommarkers,the0%limit)andoverthespaceofdifferentparameterisations(horizontally, fromthecasesofnoexternalitiesintheleftquadranttocaseswithdifferentparameterisationof externalitiesontheright).

(19)

In theabsenceoftechnologyexternalities, limitingthetradeofemissionpermitsworsensthe welfareinbothregions(leftquadrant).Thechartrepresentsthisbythetwosetsofmarkersnamed w/oLBD,LBR,SPILLandw/oLBDandLBR.Thesetwocasesillustratethehypotheticalsituationin whichtechnologicalchangeisfrozen14andtheonlymannertoreducemarginalabatementcostsisby improvingenergyefficiency.Wheninternationalspilloversareexcluded(w/oLBD,LBR,SPILL),energy efficiencycanonlybeincreasedbydomesticR&Dinvestments.Whenspilloversareincluded(w/oLBD andLBRcase),R&Dinvestmentsineachgivenregionaffectenergyefficiencyinotherregionsaswell,as describedinEq.(9).Thus,thedifferencebetweenthesetwosetsofmarkersgivesanindicationofthe welfareeffectassociatedwiththeunintendeddiffusionofknowledge.Inthepresenceoftechnology externalities,limitingpermittradealwaysbringsefficiencygainstonon-OECDcountries,butinduces efficiencylossestoOECDones(Fig.6,rightquadrant).Thisholdsforabroadenoughrangeofparameter values.

Globalwelfarearealwayshigherunderthefreetraderegime,suggestingthatthetechnology pos-itiveexternalityisnotenoughtooutweightheadditionalcostsofinnovationandtheefficiencyloss duetononequalisedmarginalabatementcosts.However,thenumericalanalysisinSection2.3points atcasesinwhichglobalgainsarepossible,especiallywhentraderestrictionsaremild.Fig.6shows that,whentheceilingis20%,theLBDm40andLBDm50casesarecharacterisedbyglobalwelfaregains. Thisresultisthecombinationofanumberofelements.First,the20%limitisbindingonlyforEurope whereasotherOECDcountries,namelytheUS,gainbecausecarbonpermitsarecheaper.Thissituation alsocreatesanincentivetoanticipateinnovationinChinaandLatinaAmerica,whicharethemajor sellersonthecarbonmarket.Inaddition,lowLBDratesarecompensatedforbyalargerinnovation effort.Whentheefficacyofexperiencetoreducetechnologycostsislow,asinthecasesLBDm#,more R&Dexpenditureisneededtocomplementthe(weaker)effectoflearning.Infact,whenLBDratesare lowtradelimitsincreasebreakthroughR&Devenmore.

Fig.6alsohighlightsthatthemagnitudeofwelfareeffectsvarieswiththenatureofthe exter-nalityconsideredandtheextentofspillovers.Thelargestimprovementsinnon-OECDarisewhen learningthroughexperienceismoreeffectiveatreducingmarginalabatementcosts(thegreen tri-anglemarkersdenotedwithLBDp#).HigherLBDratesimplythatthemarginalbenefitofagiven increaseinglobalinstalledcapacityislarger.HigherLBRrateshaveaverysimilareffect,butthe influ-enceonwelfareissmallerbecausetheextentofknowledgespilloversislessthanthatofexperience spillovers.VaryingthedegreeofR&Dspilloversisassociatedwithlimitedwelfareeffectsbecause ofthepartialappropriabilityofknowledge.Inaddition,thisperturbationchangesonlythedegree ofinternationalknowledgediffusion(ε),butitdoesnotmodifytheimpactonthecostsofabating technologies(Cax).

Allin all,policycostsvary significantlyacrossdifferentspecifications.Ifweconsiderthetwo extremecaseswithhighandlowLBDrates,globalconsumptioncostsunderfreetradealmostdouble, from0.79%to1.56%,anabsolutedifferenceofabout20US$Trillion(9US$Trillioninnon-OECD). Excludinginducedtechnicalchange(w/oLBD,LBR,SPILL)wouldincreaseglobalcoststo3.5%.Despite thesedifferences,theeffectoflesstradeoninnovationandwelfareisrobustacrossdifferent param-eterisations.

5. Someconcludingremarksandpolicyconsiderations

Thispaperexplorestherelationshipbetweentradeofcarbonallowances,theincentiveto inno-vate,technologydynamics,andeconomicefficiencywhenmultipleexternalitiesaresimultaneously considered.Usingastandardmodel,weidentifytheconditionsunderwhichrestrictionsonemission permittradingincreaseglobalinnovationandwelfare.Wethengeneralisetheanalysisusinga numer-icalintegratedassessmentmodelthatfeaturesmultipleexternalitiesandendogenoustechnological change.

Resultsindicatethatmildrestrictionstotheinternationaltradingofemissionallowancescanbe justifiedwhen emissiontradingis theonlyinstrumentavailabletopolicymakersandregionsact

(20)

0.50% 0.75% 1.00% 1.25% 1.50% 1.75% 2.00% 350 370 390 410 430 450 Consumpon lo ss (2005 -2100)

Cumulavetradedemissions

(GtCO2-eq 2005-2100)

Howmuch When Where

FullTrade 0% 2040 2035 2030 10% 15% 20% no China/India noEEX no ROW

Fig.7. Carbonmarketandglobalmacroeconomiccosts.Consumptionlossesdiscountedat3%discountrate.

non-cooperativelyoninnovation.Forcertainassumptionsregardingthewaytechnicalchangeaffects marginalabatementcost,restrictionsontheamountoftradableemissionpermitscanincreaseglobal innovationandtechnicalchange,leadingtoeconomicefficiencygains.Thesecondpartofthepaper considersaglobalstabilisationpolicyinwhichthedeveloped OECDcountriestakeonthelargest obligationsintermsofemissionreductionand,asaconsequence,arethemajorbuyersonthe inter-nationalcarbonmarket.Thenumericalanalysisshowsthatlimitingtheaccesstointernationalcarbon creditswouldmakeabatementmorecostlygloballyandfortheOECDasawhole.However,efficiency lossesaresmallformoderaterestrictionsthatlimittheuseofinternationaloffsetstoatmost15% ofdomesticregionalabatement.Internationalfinancialtransferwouldbesignificantlyreducedand moreinnovativeactivitywouldbeundertaken.Innon-OECDregions,therevenuelossesonthecarbon marketwouldbecompensatedbylowerenergypricesandhighertechnologicalspillovers.Thelatter isduetomoreinnovationbeingcarriedoutinconstrainedOECDregions.

Extensivesensitivityanalysisonthekeyparameterscontrollingtherateoftechnologicalchange andspilloversgeneralisesthevalidityoftheseresults. Forthewholespectrumofparameter val-uesanalysed,limitingpermittradealwaysbringsefficiencygainstonon-OECDcountries,provided therearetechnologyexternalities.Infact,whentheseareexcluded,bothdevelopedanddeveloping countrieswouldbeharmedbythebarrierstoemissiontrading.

Fig.7summarisestherelationshipbetweenthesizeofcarbonmarketandtheglobaleconomic costsoftheclimatepolicy.Whenmovingfromthefulltraderegimetothelimitingcaseofnopermit tradeuntil2045,thecostsoftheclimatepolicyincreasesquitesignificantly,byabout50%. How-ever,therelationbetweentradelimits andpolicycostsis quiteflatfor moderaterestrictions.A limitof15%(85%ofabatementmustbeachieveddomestically)wouldbecomparableto postpon-ingtradebetween2030and2035ortoexcludethelargersellerfromtrade.Thiswouldresultin aslightglobalefficiencyloss.Almostallpenaltiesthatarisefromtraderestrictionsarepaidbythe developedcountries,sincedevelopingcountriesarecompensatedbythepositiveexternalitiesof tech-nologicalchangeandenergymarketeffects.Afterexaminingtheeffectsonmacroeconomiccosts, financialflows,andinnovation,wecanconcludethatamoderatequantitativelimittotheuseof inter-nationaloffsets,suchasthe15%proposedinEuropeandintheUS,mightbeasecond-bestpolicy recommendation.

(21)

AppendixA.

ProofofLemma1. ReplacingthetradelimitinthefirstorderconditionsoftheHomeandForeign regions,a=˛and ¯a=1−˛,weobtain:

Homeregion

x:CX(XLIM,˛)+Gx(xLIM)=0 (a1)

a:Ca(XLIM,˛)−=pLIM (a2)

Foreignregion

x:Cx( ¯XLIM,1−˛)+Gx(¯xLIM)=0 (a3)

a:Ca( ¯XLIM,1−˛)=pLIM (a4)

BytotallydifferentiatingtheFOCsforinnovation–Eqs.(a1)and(a3)–weimmediatelyobtain theeffectoftighteningthetradepermitlimitoninnovation.Tosimplifythenotationwedropthe argumentofthecostfunctionsanddenotetheForeignregionwithabaroverthecostfunctions, CXX=CXX( ¯XLIM,1−˛),Gxx=Gxx(¯xLIM)andCX˛=CX˛( ¯XLIM,1−˛).15Usingthisnotationweobtain:

CXX+Gxx CXX CXX CXX+Gxx

dx d˛ d¯x d˛

=

−CX˛ CX˛

ApplyingCramer’srule,itcanbeshownthatinnovationintheHomeregionincreaseswiththe tradelimit: dx d˛= −CX˛(CXX+Gxx)−CXXCX˛ (CXX+Gxx)(CXX+Gxx)−(CXX)(CXX) >0 (a5)

becauseCX˛<0,CX˛<0,CXX(x,a)>0andGxx≥0,−CX˛(CXX+Gxx)−CXXCX˛≥0.Thedenominator

isalwayspositivebecause≤0,andthus(CXX+Gxx)(CXX+Gxx)>(CXX)(CXX).

Becausethetotalamountofabatementisgiven,theotherregionwillreducehisabatementeffort andthereforeinnovationwilldecreaseaccordingtoasimilarexpression:

d¯x d˛=

(CXX+Gxx)CX˛−(−CX˛)CXX

(CXX+Gxx)(CXX+Gxx)−(CXX)(CXX)

<0 (a6)

becauseCX˛<0,CX˛<0,CXX(x,a)>0andGxx≥0,hence(CXX+Gxx)CX˛+(CX˛)CXX<0.

ProofofProposition1a. Usingconditions(a5)and(a6)totalinnovationisgivenbythefollowing expression: dx d˛+ d¯x d˛= −CX˛(CXX+Gxx)−CXXCX˛ (CXX+Gxx)(CXX+Gxx)−(CXX)(CXX) + (CXX+Gxx)CX˛−(−CX˛)CXX (CXX+Gxx)(CXX+Gxx)−(CXX)(CXX)

Letusdefine=(CXX+Gxx)(CXX+Gxx)−(CXX)(CXX),where>0

dx d˛+ d¯x d˛= 1 ([GxxCX˛−CX˛Gxx]+[CXXCX˛−CX˛CXX]−[CXXCX˛−CX˛CXX]) Itfollowsthattotalinnovationincreases

dx d˛+

d¯x d˛>0

15NotealsothechangeinnotationfromC

X˛toCXatoemphasisethatweconsidermarginalvariationsinthetraderestriction,

Riferimenti

Documenti correlati

Simulations and theoretical analysis based on the imprecise computation model show that the integration of IDTH and IDTH+ can achieve improved network performance in terms

We evaluated the differences at the neural population level between the ICB+ and ICB − conditions using a broad spectrum of neural features, divided into 3 clas- ses: firing

However, we also observed that the prednisone dose at the time of flare was lower in patients flaring with PMR or LVV compared with those flaring with cranial manifestations or

In the last five years, several studies (Table 2) compared survival and functional outcomes of patients affected by advanced stage LC (T3 -T4a) and treated distinctly by

In par- ticular, we consider three abstraction approaches: (i) an approximation of the stochastic discrete dynamics with continuous deterministic ones, applied to all model

Data-set consisting of accelerator pedal position α (input data; first graph), dynamometer torque T D and shaft torque T ST (output data; second graph and third graph,

[r]

35 Moncalieri, Archivio Storico del Comune, Serie Generale, mazzo 2055, nn 15 e 18 Nel primo caso si tratta di un bifoglio membranaceo rovinatissimo e con grosse lacune (mm.