TjallingC. Koopmans Research Institute - PDF document

TjallingC. Koopmans Research Institute Utrecht School of Economics Tjalling C. Koopmans Research Institute Discussion Paper Series 04-09 Why Does Centralisation Fail to Internalise Policy Externalities? Robert Dur 1 Hein Roelfsema 2 1 Tinbergen Institute Earasmus University Rotterdam 2 Utrecht School of Economics Utrecht University Jan 2004 Abstract We thank an anonymous referee of , Amihai Glazer, Otto Swank, Huseyin Yildrim, seminar participants in Amsterdam and Rotterdam, and participants of the 2002 Annual Meeting of the Public Choice Society in San Diego, the 2002 Annual Meeting of the European Public Choice Society in Belgirate, and the XIV Villa Mondragone International Economic Seminar in Rome. Dur gratefully acknowledges financial support from NWO, KNAW, and VSNU through a Vernieuwingsimpuls grant. Roelfsema likes to express his appreciation for the hospitality at the Department of Economics of Erasmus University Rotterdam. Keywords: Centralised decision making, strategic delegation, financing rules. Acknowledgements We wish to thank the seminar participants at Nuffield College (University of Oxford), the Tinbergen Institute (University of Amsterdam), Amsterdam Institute for Advanced Labour Studies and the 2003 Meeting of the European Society of Population Economics for valuable comments, and Nigel Stuttard at National Statistics for providing the NES data. Financial support from the Leverhulme Trust project “The labour market consequences of technical and structural change” is gratefully direct all correspondence to the first author. Tel +31 (0) 302537955 E-mail: h.roelfsema@econ.uu.nl : http://www.koopmansinstitute.uu.nl dur@few.eur.nl Hein Roelfsema Utrecht University Utrecht School of Economics Vredenburg 138 3511 BG Utrecht Robert Dur Department of Economics H 7-21 Erasmus University Pobox 1738 3000 DR Rotterdam The Netherlands Fax: +31-10-408 9147 Email: ontwerp voorblad: WRIK Utrecht How to reach the authors Please direct all correspondence to the first author. of research results, and to enhance discussion with colleagues. Please send any comments and suggestions on the Koopmans institute, or this series to P.vanDriel@econ.uu.nl It was founded in 2003, and named after Professor Tjalling C. Koopmans, Dutch-born Nobel Prize laureate in economics of 1975. es the Koopmans Institute publishes results of ongoing re The Netherlands telephone +31 30 253 9800 fax +31 30 253 7373 website www.koopmansinstitute.uu.nl The Tjalling C. Koopmans Institute is the research institute Tjalling C. Koopmans Research Institute Utrecht School of Economics Utrecht University Vredenburg 138 3511 BG Utrecht delegatebargainingtoagentswhodonotcareforpublicgoodsatall,soasto‘force’alargeincreaseinpublicgoodsprovisionintheotherregion.Sincebothpolicymakersdo,nopublicgoodsaresupplied,asunderdecentraliseddecisionmaking,see(A6). forbyhigherforeignpublicgoodssupply.Iflocalpublicgoodsarenearperfectsubstitutes(approaches),areductioninisalmostcompletelycompensatedforbyanincreasein,resultingindelegationtoanextremelyconservativepolicymaker.Clearly,underprovisionofpublicgoodsismoreseverethaninthecasedescribedinthemaintextas,inadditiontotheexternalityproblem,votersappointconservativepolicymakerswhenlocalpublicgoodsarestrategicsubstitutes.CentralisedDecisionMaking.Publicgoodsprovisionresultingfromthebargainbetweenthedelegatesfromthetwocountriesisdescribedby:Giventhepreferencesoftheforeigndelegate,optimalpreferencesofthedomesticdelegatearegivenby: =· 0(+ (+)¸ ³ 2+´ Using(A7)tondthevaluesof ,andimposingsymmetryinequilibrium,weobtain: +(1+Asinthemaintext,wendthatthepolicymakerdelegatesbargain-ingtoapublicgoodloverwhen,whileshedelegatesbargainingtoaconservativewhen.Ifallofthecostareshared(forall.Theintuitionisclear.Comparedtothebargainingoutcomewithsinceredelegation,eachpolicymakerdesireshigherdomesticpublicgoodssupplyandlowerforeignpublicgoodssup-ply.Giventhepreferencesoftheotherdelegate,delegatingtoanagentwithincreasesuptothepolicymaker’soptimallevel(adelegatewithwouldoversupplyevenifallcostsareshared)andreducesthelevelofHence,bydelegatingtoan‘extremelover’ofpublicgoods,thepolicymakerkillstwobirdswithonestone.Thisisincontrasttothecaseinthemaintext,wheredelegatingtoapublicgoodloverincreasesforeignpublicgoodsupplyaswell.Ifthecostofpublicgoodsprovisionarenon-shareable(),forallthereisconservativedelegation,asinthemaintext.However,incontrasttotheresultsinthemaintext,optimalconservativenessincreasesmonotonicallyin.Theintuitionisthatwhenpublicgoodsarestrategicsubstitutes,sendingamoreconservativedelegateincreasesthelevelof.Inthelimitingcasewhereapproaches,policymakers 11AppendixThisAppendixstudiesthecasewherelocalpublicgoodsarestrategicsubstitutes.Usingthesamenotationasinthemaintext,theutilityfunctionisgivenby:TheSocialOptimum.Sociallyoptimalpublicgoodsprovisionisde-scribedby:DecentralisedDecisionMaking.Whenlocalpublicgoodsarestrate-gicsubstitutes,votershaveanincentivetodelegatepolicymakingtoanagentwithpreferencesdierentfromtheirown.Given,policymaker’soptimalpublicgoodsupplyisdescribedby:wheredenotesthepreferencesofthepolicymakerinregion.Giventhepreferencesofthepolicymakerinregion,themedianvoter’soptimalpreferencesofthepolicymakerinregionaredescribedby: =· 0(+ (+)¸ where,using(A3): =0(+) 2)+)0 = Delegatingpolicymakingtoanagentwithstrongerpreferencesforpublicgoodsincreasesdomesticpublicgoodssupply,butreducespublicgoodssupplyintheforeignregion.Substituting(A5)in(A4),andusing(A3)tosimplify,resultsin:=(1Forany,themedianvoterdelegatespolicymakingtosomeonewhocareslessforpublicgoodsthanshedoes.Thereasonisclear.Giventhepreferencesoftheforeignpolicymaker,delegatingtoamoreconservativeagentimplieslessdomesticpublicgoods,butthisispartlycompensated andEmploymentinaUnitedEurope,Cambridge,MA:MITPress.Besley,T.andS.Coate(2003).CentralizedversusDecentralizedProvisionofLocalPublicGoods:APoliticalEconomyApproach,nalofPublicEconomics87(12):2611-2637.Bolton,P.andG.Roland(1997).TheBreakupofNations:APolit-icalEconomyAnalysis,QuarterlyJournalofEconomics112(4):1057-Brueckner,M.(2000).UnanimityversusConsensusBargaining:Strate-gicDelegationandSocialOptimality,Mimeo,ZEI,Bonn.Chari,V.V.,L.E.JonesandR.Marimon(1997).TheEconomicsofSplit-TicketVotinginRepresentativeDemocracies,AmericanEconomicReview87(5):957-976.Cheikbossian,G.(2000).Federalism,DistributivePoliticsandRep-resentativeDemocracy,EconomicsofGovernance2(1):105-122.DelRossi,A.F.andR.P.Inman(1999).ChangingthePriceofPork:TheImpactofLocalCostSharingonLegislators’DemandsforDistributivePublicGoods,JournalofPublicEconomics71(2):247-273.Ellingsen,T.(1998).Externalitiesvs.Internalities:AModelofPoliticalIntegration,JournalofPublicEconomics68(2):251-268.Goyal,S.andK.Staal(2002).ThePoliticalEconomyofRegionalism,EuropeanEconomicReview,forthcoming.Gradstein,M.(2000).ThePoliticalEconomyofSustainableFeder-ations,CESifo,WorkingPaperSeriesNo.315.Mazza,I.andF.vanWinden(2002).DoesCentralizationIncreasetheSizeofGovernment?TheEectsofSeparationofPowersandLob-bying,InternationalTaxandPublicFinance9(4):379-389Oates,W.E.(1972).FiscalFederalism.NewYork,HarcourtBraceJovanovich.Persson,T.andG.Tabellini(1994).DoesCentralizationIncreasetheSizeofGovernment?,EuropeanEconomicReview38(3-4):765-73.Persson,T.andG.Tabellini(2000).PoliticalEconomics:ExplainingEconomicPolicy,Cambridge,MA,MITPress.,B.(1998).DelegationofBargainingandPower,Stock-holmSchoolofEconomics,WorkingPaperSeriesinEconomicsandFi-nanceNo.248. Asecondextensionistoallowforinterregionalheterogeneityinpref-erencesand/ordierencesinlocalcostsofpublicgoods.Dierencesinnon-shareablecostsacrossregionsmayaddtoourunderstandingofwhyitissodiculttoagreeoncommonpolicies.Forinstance,thereluc-tanceoftheUStosigntheKyotoprotocolmayhavemoretodowiththerelativelyhighcostsforgrowingUSindustriestocomplywithre-strictiveglobalemissionstandards,thanwithweakpreferencesoftheUSelectorateforenvironmentalprotection.Anotherinterestingextensionofthemodelwouldbetoallowformorethantwocountries.Thisseemsespeciallyrelevantinthelightoftheen-largementprocessintheEU.Memberstatesthatbenetheavilyfrompoliciesthatarecommonlynancedfearthatenlargementmayerodetheirprivilegedposition,andreducethepossibilityofdeepeningcooper-ation.Further,countriesthathavestrongpreferencesforenvironmentalprotectionfearthatenlargementmaylowerthecommonstandardsinthisdomain.Therefore,enlargementmaychangetheattitudesoflocalpolicymakerstowardspolicymakinginBrusselsandmayinuencetheirdelegationdecision.Ouranalysishasshedlightonproblemsthatarisewith‘cooperativepolicycoordination’,aphenomenonthathasbecomemoreimportantovertime.Increasedinterdependencecreatesincentivesforpolicycoor-dinationataregionalandgloballevel,forinstanceinthecaseofen-vironmentaldegradationorcopingwithlargenumbersofrefugees.Onmostoftheseissues,policymakerscooperateintheinternationalarenathroughpolicycoordination,notthroughcostsharing.Ourmodelseemstobewellplacedtoexplaintheinecienciesthatmayariseinpoliticaldecisionmakingwhencountriesdecidetocooperate,butdonotshare.10ReferencesAlesinaA.,I.AngeloniandL.Schuknecht(2001).WhatDoestheEuro-peanUnionDo?,Cambridge,MA,NBERWorkingPaper8647.AlesinaA.,I.AngeloniandF.Etro(2001a).InstitutionalRulesforFederations,Cambridge,MA,NBERWorkingPaper8646.AlesinaA.,I.AngeloniandF.Etro(2001b).ThePoliticalEconomyofInternationalUnions,Cambridge,MA,NBERWorkingPaper8645.Alesina,A.andE.Spolaore(1997).OntheNumberandSizeofNations,QuarterlyJournalofEconomics112(4):1027-1056.Alesina,A.andR.Wacziarg(1999).IsEuropeGoingtooFar?,Carnegie-RochesterConferenceSeriesOnPublicPolicy51(1):1-42.Bertola,G.,J.F.Jimeno,R.MarimonandC.Pissarides(2001).EUWelfareSystemsandLaborMarkets:DiverseinthePast,IntegratedintheFuture?,inG.Bertola,T.BoeriandG.Nicoletti(eds.).Welfare derspendingemergeswhennancingdiersfromtheoptimalnancingrule.Thepolicymakerswillthereforedecidetoimplementtheoptimumnancingrule.Sharingonlyapartofthecostofpoliciesdecidedonatcentralleveliswidelyobservedinpractice.Forinstance,EUgrantsforlocalprojectsoutoftheRegionalDevelopmentFundhavetobematchedbyequalgrantsfromnationalgovernments,theso-calledco-nancingsystem.AsDelRossiandInman(1999)showinanempiricalstudyonUSlegisla-tors’demandforlocalpublicgoods,co-nancingsignicantlyreducesthelegislators’demandforcentrallynancedprojects.Bonusestocom-pensateforlocalindirectcostsarelessfrequentlyobserved.Onereasonmaybethatnon-shareablecostsarediculttoquantifyexanteandareprobablyhardtoverifyexpost.Hence,itmaybepoliticallydiculttoagreeonthembeforedecisionmakingontheamountsofpublicgoods9ConcludingRemarksThispaperprovidedanexplanationforwhyinsomepolicydomainscooperativecentraliseddecisionmakingonlocalpublicgoodsleadstooverspending,whereasinotherareaspublicspendingonlocalpublicgoodsistoolow.Wearguedthatifcostsoflocalpublicgoodsaresharedamongparticipatingregionsthroughacommonbudget,thedelegationofpublicgoodsloversleadstooversupply.Ifasucientlylargepartofthecostsarenon-shareable,thedelegationofconservativesresultsinunderspending.Lastly,wederivedcostsharingruleswhicheliminatetheincentivestodelegatebargainingatthecentralleveltoagentswithpreferencesdierentfromthedomesticpolicymaker’spreferences.Ouranalysiscanbeextendedinseveralimportantways.Oneistoincorporatechecksandbalances.Charietal.(1997)examinetheroleofapresidentasacheckonoverspending.Theyallowfor‘splitticket’votingonthepreferencesofthedelegatesandthepresident.IntheconstitutionaldebateintheEuropeanUnion,itisstillanopenissuewhetherinstitutionalreformshouldmovetowardsanelectedheadoftheEuropeanCommission,whomayserveasacheckonthemembersoftheCommissiondelegatedbythemembercountries. Introducingseparationofpowersinthebudgetaryprocessmayalsocontributetoecientprovisionofpublicgoods.Charietal.(1997)buildamodelwherevotersineachstatedelegatespending-proneagentstoCongress,butappointaconservativePresidentsoastocurtailexcessivespending.InamodelthatmirrorsdecisionmakingintheEU,MazzaandVanWinden(2002)showthatseparationofpowers,wherethebudgetissetbeforethepolicyselectionstage,reducestheincentivestolobbyforlocalpublicgoodsandthereforethesizeofspendingatacentrallevel. ofspillovereects.Commonnancingofallcost()isonlyoptimalincaseofglobalpublicgoods(Intheotherextremecasewhereallofthecostofpublicgoodsarebornedomestically(=0),(22)equals: (24)Hence,for,asubsidyisneededtoinducepolicymakerstore-frainfromdelegatingbargainingtoaconservativeagent.Thesubsidyincreasesinthesizeofspillovers.Thismaycomeasasurprisesince,startingfrom,optimalconservativenessofthedelegaterstin-creasesinand,from ,decreasesin(seeSection7).Thereasonisthattheeectivenessofthesubsidyinchangingthepol-icymakers’delegationdecisionreducesin.Whenissmall,sendingalessconservativeagenthasarelativelylargeeectonthesupplyofthedomesticpublicgoodandarelativelysmalleectonthesupplyoftheforeignpublicgood.Thenetadditionalreceiptsfromthesubsidyschemearethereforelargewhenislow.Whenapproachesunity,sendingalessconservativeagentincreasesdomesticandforeignpublicgoodsupplybyalmostthesameamount,renderingthesubsidyschemeclosetobudgetaryneutralforeachregion.Hence,alargesubsidyperunitofpublicgoodisneededtoosetasmalldistortioninthepolicymaker’sdelegationdecision.Whenlocalpublicgoodscloselyresembleglobalpublicgoods,theoptimalsubsidyapproachesthetotalcostofpublicgoodsupply().Hence,asforthecase,weconcludethatsharingallofthecostofpublicgoodsthroughacentralbudgetisonlyoptimalincaseofglobalpublicgoods.Using(22),wecanderivewhichpartofthetotalcostofpublicgoodsmustbesharedtoguaranteesociallyoptimalpublicgoodsupply: (+)=2 (25)Hence,sharingnoneofthecostisoptimalonlyintheabsenceofspillovers,whilecommonnancingofallcostisonlyoptimalincaseofglobalpublicgoods.Theoptimumnancingruledescribedby(25)maynotonlybeso-ciallyoptimal,butalsopoliticallyfeasible.Whendecidingonthenancingrule,localpolicymakersrecognisethatequalamountsofpublicgoodsresultforeachregion.Theyalsorecognisethatover-orun- Thiswillnotbethecasewhenthemedianvotersintheregionsdierintheirpreferencesforpublicgoods.Then,policymakerswilltrytomanipulatedecisionmakingonthenancingruleinordertobringthecentraldecisionclosertotheirpreferences. Substituting(9)and(10)into(20),andimposingsymmetryinequilib-riumgives:2(1+ (1+(1+(21)Clearly,thelargeris,thelargeristheoptimalvalueof.Asubsidycreatesapositivebudgetexternalityfromdomesticpublicgoodssupply.Thisinducespolicymakerstodelegatebargainingtolessconservativeagentssoastoincreasedomesticpublicgoodssupply.Theonlyexcep-tioniswhen.Then,thelevelsofdomesticandforeignpublicgoodsareequal,irrespectiveofthepreferencesofthedelegates(seeequation(9)).Hence,inthatcase,thetax/subsidyschemedoesnotaectthebudgetsofthelocalgovernmentsand,hence,thedelegationdecision.Denoteastheoptimalsubsidy,thatis,thesubsidythatresultsinsinceredelegationbybothpolicymakers().Theoptimalsubsidy(tax)is: (22)foranyTheoptimalsubsidyincreasesinandanddecreasesToevaluatethepropertiesofconsiderthesituationwhereallcostsofpublicgoodsarenancedthroughacommonbudget(=0)Inthiscase,(22)reducesto: (1+(23)Hence,intheabsenceofindirectcostandwith,acentraltax()isneededtoeliminatethestrategicdelegationofpublicgoodlovers.Theoptimaltaxdecreasesinthelevelofspillovers.Whenspilloversareabsent,theoptimaltax()impliesthatnoneofthecostofpublicgoodsareactuallyshared.Centraliseddecisionmakingwithanoptimalsubsidythenresultsinthesamepublicgoodssupplyasunderdecentraliseddecisionmaking,whichissociallyoptimalintheabsence Recallthatthebargainingoutcomeisindependentofthetax/subsidyscheme.Hence,wecanuse(9)and(10)tosimplify(20).Obviously,theredoesnotexistanoptimallevelofbecausethetax/subsidyschemedoesnotaectthedelegationdecisioninthatcase.Thisisofnoconcernbecausedelegationisalwayssincereif,seeequation(12)intheprevioussection. ofpublicgoods,anadditionalcentraltaxschemerestrainsthepolicymakersfromdelegatingpublicgoodlovers.Considertheintroductionofacentralsubsidy,or–incase–tax,on.Weassumethattherevenuesofataxarefullyrefundedtotheregionsandthateachregionreceiveshalfofthisfund.Incaseofasubsidy,eachoftheregionspayshalfofthecostofthescheme.Theobjectivefunctionofthedelegateofregion 2(+)+ (18)wherethelasttermisthedierencebetweenregion’ssubsidyrevenues)andtheregion’scontributiontothesubsidyfund( Asintheprevioussections,thebargainingoutcomeisgivenbythemaximumofthesumoftheutilitiesofthedelegates.Asisclearfrom(18),thissumisindependentofthelevelofthesubsidy.Hence,thesubsidyschemedoesnotaecttheoutcomeofthebargainingbythedelegates,theirpreferences.Thesubsidyschemedoes,however,ectthepolicymakers’delegationdecisions.Theobjectivefunctionofthepolicymakerinregion 2(+)+ (19)Clearly,fromtheperspectiveofthepolicymakerinregion,asubsidy(tax)introducesanadditionalbenet(cost)ofunitsofandanad-ditionalcost(benet)ofunitsof.Equilibriumisdescribedbythefollowingrst-orderconditions: 1+(2) 1¸· 1+ 1¸ 2 (20) 1 1¸ 22·0(2) 2+(1) 2¸· 2+ 2¸ 2 2+· 2 2¸ 2 Similarly,onecouldadjusttheshareofthetaxcostthatisnancedthroughthecentralbudgetinordertoeliminatestrategicdelegation.Becausewewanttoallowforthecase,weintroduceanadditionaltax/subsidyscheme.Attheendofthissection,wederivewhichpartofthetotalcostofpublicgoodsmustbesharedsoastoguarenteesociallyoptimalpublicgoodssupply,usingourresultsfortheoptimaltax/subsidyscheme. delegationmitigatesthiseect.However,foralltheformereoutweighsthelatter,therebyincreasingthelevelofpublicgoodsupply.Centralisationimprovessocialwelfareineachregioniftheincreaseinbenetsfromhigherlevelsofpublicgoodsarelargerthantheincreaseincosts.Recognisethatgivensymmetrythewelfareofeachindividualmedianvoterisatamaximumatthesocialoptimum.Althougheachpolicymakerhasanincentivetodelegatestrategicallytoalterthedis-tributionofpublicgoodsinfavourofherregion,theydonotachievethisgoalbecausebothdelegatestrategically.Inequilibrium,welfareinbothregionsislowerthaninthesocialoptimum.Toevaluatetheeectsofcentralisationonsocialwelfareineachdistrict,wesubstitutethelevelsofpublicgoodsarisingundercentralisedanddecentraliseddecisionmak-ing,respectively,into(2).Itfollowsthatcentralisationincreasessocialwelfareif:(1++b(gci)−b(gdi)]−(p+c)(gci−gdi)>0(17)wherethesuperscriptdenotescentraliseddecisionsanddenotesde-centraliseddecisions.Again,therearetwocases.Ifregionsdelegatebar-gainingtopublicgoodslovers,thesupplyofpublicgoodswillbehigherthanthesocialoptimum.AsisshownbyBesleyandCoate(2003),forhighlevelsofcentralisationlikelyimproveswelfare.Thereasonisthatthebenetsfrominternalisingtheexternalitiesarelarge,whereasthedistortionfromstrategicdelegationissmall.TheoppositeistrueforlowlevelsofTherefore,ifregionsdelegatebargainingtopublicgoodslovers,centralisationisonlywelfareimprovinginpolicydomainsthathavelargeexternalities.Ifregionsdelegatebargainingtoconservatives,undercentraliseddecisionmakingthesupplyofpublicgoodsislowerthaninthesocialoptimum,buthighercomparedtodecentralisedde-cisionmaking.Therefore,eventhoughregionsdelegatebargainingtoconservatives,centralisationimprovesthewelfareofeachregion.8OptimalFinancingRulesTheprevioussectionshowedthatwhenalargepartofthetotalcostofpublicgoodsaresharedthroughacommonbudget,overprovisionofpub-licgoodsresults(exceptforthecase);underprovisionoccurswhenalargepartofthecostarebornelocally(exceptforandInthissection,wederivenancingrulesthateliminatetheincentivesforstrategicdelegation.Whenunderprovisionofpublicgoodspersistsun-dercentraliseddecisionmaking,an(additional)centralsubsidyschemewhichintroduces(ormagnies)apositivebudgetexternalityremediesthestrategicdelegationofconservatives.Inthecaseofoverprovision byequalamountswhen;see(10).Inthegeneralcasewherebothand,describedby(12),itdependsonthemagnitudeofthecostparametersandthespilloverectwhetherpolicymakershaveanincentivetodelegatebargainingtopublicgoodloversortoconservatives.Thelargerareindirectcostrelativetodirectcost,themoreconservativearethepreferencesofthedelegates.Theeectofthesizeofspilloversonthedelegationdecisiondependsontherelativeimportanceofdirectandindirectcost.Thisisduetotheambiguouseectofonoptimaldelegate’spreferencesinthecaseislarge,seethediscussionabove.Forvariouscombinationsofparametervalues,theterminbracketsin(12)isone.Hence,policymakersdelegatebargainingtoagentswiththesamepolicypreferencesastheirownandthesocialoptimumisattained.Thisisthecaseif: =1 (15)Ifthelefthandsideof(15)issmallerthantherighthandside,policymakersappointpublicgoodlovers,resultinginoverspending.Ifthelefthandsideof(15)islargerthantherighthandside,conservativesareappointed,resultinginunderprovisionofpublicgoods.Bycomparingequation(4)to(9)withgivenby(12),wederivetheeectofcentralisationofdecisionmakingontheamountsofpublicgoods.Thelevelofishigherundercentraliseddecisionmaking2(1+ (1+)+(1(16)whichholdsunlessbotharezero.Ifbothareequaltozero,centralisationofdecisionmakingdoesnotaectpublicgoodssupplybecausei)therearenoexternalitiestointernaliseandii)thereisnocommonpoolproblem.Dependingonthevaluesofc,ptypesofstrategicdelegationmayoccur.Inthecaseofthedelegationofpublicgoodlovers,boththeinternalisationofexternalitiesaswellasthestrategicdelegationeectpushupthelevelofpublicgoodsascomparedtothedecentralequilibrium.Inthecaseofthedelegationofconservatives,theeectsworkinoppositedirections.Theinternalisationofexternalitiespushesupthelevelofpublicgoods,whereasthestrategic Whenlocalpublicgoodsarestrategicsubstitutes,optimalconservativenessin-creasesmonotonicallyin,assendingamoreconservativedelegateincreasespublicgoodssupply;seetheAppendix. thatcasebecausesendinganagentwithstrongerpreferencesforpublicgoodsdoesnotraiseforeignpublicgoodsupply.Thesocialoptimumisimmunetostrategicdelegationonlywhenpublicgoodsareglobalpublicgoods().Aswealreadyarguedabove,whenbenetsandcostofpublicgoodsareperfectlyinline.Hence,thereisnoincentivetomisrepresentpolicypreferencesatthecentrallevel.Intheotherextremecase,,noneofthecostofpublicgoodsaresharedthroughacommonbudget.Equation(12)thenreduces(1+ (1+(14)Hence,sinceredelegation()isoptimalonlyifandWhen<κ<,thepolicymakerdelegatesbargainingtoa‘conser-vative’agent,i.e.someonewhocareslessforpublicgoodsthanshedoes.Startingfrom,optimal‘conservativeness’rstincreasesin 0 )andthendecreasesin for <κ<).Aswearguedabove,policymakershaveanincentivetomovetheoutcomeofcentraliseddecisionmakingawayfromthesocialoptimum.Ifallofthecostofdomesticpublicgoodsarebornedomes-tically,policymakershaveanincentivetodistortthecentraldecisiontowardslowerdomesticpublicgoodsupplyandtowardshigherforeignpublicgoodsupply.Intuitively,theexternalityprobleminherenttode-centraliseddecisionmakingpersistsundercentraliseddecisionmaking.Whileregionsfullybearthecostofdomesticpublicgoodsupply,theyonlyreapapartofthesocialbenets.Thenegativegapbetweenlocalbenetsandlocalcostofdomesticpublicgoodsincreasesinthesizeofthespillovere.Thisisthereasonwhyoptimalconservativenessincreasesinforlowvaluesof.Whenbecomessucientlylarge,anothereect,workingintheoppositedirection,startsdominatingandoptimalconservativenessdecreasesin.Thisistheeectofconserva-tivenessonforeignpublicgoodsupply.Bydelegatingbargainingatthecentralleveltoamoreconservativeagent,bothdomesticandforeignpublicgoodsupplydecrease(see(10)).Thedecreaseinforeignpublicgoodsupplyisacosttothepolicymakerbecauseshefreeridesonfor-eignpublicgoodsprovision.Thiscostislarger,thelargeristhespillover.Intheextremecaseof,policymakersdelegatebargain-ingtoagentswiththesamepreferencesastheirown,eventhougheachpolicymakerhasanincentivetopushforlowerdomesticpublicgoodsupplyandforhigherforeignpublicgoodsupply.Thereasonisthatsendingamoreconservativeagentreducesdomesticandforeignsupply policymaker,andviceversa.Equilibriumisdescribedbythefollowingrst-orderconditions: 10(1 1(2)¸· 1+ 1¸ 2 (11) 20(2 2(1)¸· 2+ 2¸ 2 Substituting(9)and(10)into(11),andimposingsymmetryinequilib-riumyields:2(1+ (1++(1+(12)Clearly,itisgenerallynotinthepolicymaker’sinteresttodelegatebargainingtoanagentwiththesamepolicypreferencesasherown).ThedelegationdecisiondependscruciallyonthelevelofdirectandindirectcostandthesizeofthespillovereAsabenchmark,considertheBesleyandCoate(2003)caseinwhichallcostaresharedamongthedistrictsthroughacommonbudget().Equation(12)thenreducesto:2(1+ (1+(13)TheterminlargebracketsisalwaysgreaterthanoneunlessHence,thepolicymakerhasanincentivetodelegatebargainingtoa‘publicgoodlover’.Thereasonisacommonpoolproblem.Sinceallcostofpublicgoodsarenancedthroughacommonbudget,whileben-tsare—for—atleasttosomeextentregion-specic,policymakershaveanincentivetopushforhigherdomesticpublicgoodssup-plyandforlowerforeignpublicgoodssupply.Thedelegationofapublicgoodloverhastwoeects.First,itresultsinanincreaseinthedomesticpublicgoodsprovision.Thisraisestheutilityofthelocalpolicymaker.Second,itresultsinanincreaseofforeignpublicgoodsprovision.Thislowerstheutilityofthelocalpolicymaker.However,for,domes-ticpublicgoodsprovisionincreasesbymorethanforeignpublicgoodsprovision(see(10)).Startingfromtheequilibriumwithsinceredelega-tion,bothpolicymakershaveanincentivetosendanagentwhocaresmoreforpublicgoodsthantheydothemselves.InthesymmetricNash-equilibrium,bothpolicymakerssendapublicgoodlover.Asaresult,thereisoverprovisionofpublicgoods.Thisfollowsfromcomparing(3)with(9),withgivenby(13).Overprovisionislargestwhen.Then,.Thecommonpoolproblemismostseverein ouslyandindependentlyfromeachother.Agentsareselectedfromtheregions’populations.Recallthatindividualsineachregiondieronlyintheirrelativepreferenceforpublicgoods,givenbytheparameterWeassumethatcitizens’preferencesaresucientlyvariedsothataninteriorsolutiontothepolicymaker’sselectionproblemisensured.Asintheprevioussection,thebargainingoutcomeisgivenbythemaximumofthesumofutilitiesoftheagentsatthebargainingtable:e:b(g1)+κb(g2)]++b(g2)+κb(g1)]−(p+c)(g1+g2)(8)whereisthepreferenceparameteroftheagentappointedbyregion’spolicymaker.Maximising(8)toresultsin:(9))λd2+λd1κ]b0(g2)−p−c=0Thecomparativestaticsareobtainedbyapplyingtheimplicitfunctiontheoremto(9): 1=0(1) [1+2]1), 1=(2) λd2+λd1κ]b00(g2),(10) 2=0(2) λd2+λd1κ]b00(g2),and 2=(1) λd2+λd1κ]b00(g1)whichareallpositive.Hence,delegatingbargainingtoanagentwithstrongerpreferencesforpublicgoodsresultsinanincreaseinboththedomesticandtheforeignpublicgood.Theincreaseindomesticpublicgoodsprovisionislargerthantheincreaseinforeignpublicgoods,unlesspublicgoodsareglobalpublicgoods().Itisalsoclearthattheectofpolicymaker’sdelegationdecisiononthebargainingoutcomedependsontheotherpolicymaker’sdelegationdecision.Eachpolicymakerselectsadelegatesoastomaximisegivenby(6).Sinceselectiontakesplacesimultaneouslyandindependently,eachpolicymakertakesasgiventhepreferencesofthedelegatefromtheotherregion.IntheNash-equilibrium,region1’spolicymaker’sselectiondecisionisoptimalgiventheselectiondecisionofregion2’s Clearly,theinecienciesthatarisefromstrategicdelegationmaybeavoidedbycoordinatingthedelegationdecision.Ifbothpolicymakerscommittosinceredelegation,thecentralbargainwillproducethesocialwelfaremaximisingleveloflocalpublicgoods.Inpractice,however,itseemsdiculttocommittosuchanagreement,asexantethepreferencesofthedomesticdelegate(and,possibly,eventhoseofthedomesticpolicymaker)arediculttoassessforforeignpolicymakers. Theoptimallevelsofaredescribedby: (7) Comparing(7)with(3),itisclearthateventhoughcentralisationin-creasesbothpolicymakers’welfare,individuallyoptimalprovisionofpublicgoodsgenerallydivergesfromtheamountsarisingundercen-traliseddecisionmakingwithsinceredelegation.Thisconictofinterestgivespolicymakersanincentivetodistortthecentraldecision.Letusconsidertwospecialcases.First,ifindirectcostsarezero,,allofthecostsofpublicgoodsnancedthroughacommonbudget.Then,unless,thesupplyofdomesticpublicgoodsundercentraliseddecisionmakingistoolowfromtheperspectiveofthepolicymaker,whilethesupplyofforeignpublicgoodsistoohigh.Thisisthecommon-poolproblem:commonnancingdrivesawedgebetweenthebenetsandcostoflocalpublicgoods.Whilethebenetofanincreaseinpublicgoodsprovisioninoneoftheregionsislargelyregion-specic,thecostisspreadoverthetworegions.Commonnancingthereforegivesanincentivetobothpolicymakerstopushforahighersupplyofdomesticpublicgoodsandlowersupplyofforeignpublicgoods.Theonlyexceptioniswhenpublicgoodsare‘global’publicgoods,.Then,benetsandcostofpublicgoodsareperfectlyinlinegiventhatallcostareshared.Second,iftaxcostsarezero,,allofthecostarebornelocally.Then,unless,centraliseddecisionmakingresultsintoohighalevelofdomesticpublicgoodsandtoolowalevelofforeignpublicgoodsfromtheperspectiveofeachpolicymaker.Clearly,sinceallcostarebornelocally,thepolicymakerwantstofreerideonaninniteamountofforeignpublicgoods.Thepolicymakerwantstoprovideonlyamoderateamountofdomesticpublicgoods,viz.thesamelevelthatarisesunderdecentraliseddecisionmaking(compare(7)with(4)).Indirectcostthusgiveincentivestopushforlowerdomesticpublicgoodsupplyandforhigherforeignpublicgoodsupply.Giventhat,theonlycaseinwhichapolicymakerdoesnothaveanincentivetomovepublicgoodsprovisionawayfromthesocialoptimumiswhenspilloversareabsent,,i.e.whenthesupplyofpublicgoodsundercentraliseddecisionmakingcoincideswiththatunderdecentraliseddecisionmaking.Tobringthecentraldecisiononpublicgoodsprovisionclosertothepolicymaker’sindividualoptimum,thepolicymakermaydelegatebar-gainingatthecentralleveltoanagentwithpreferencesdierentfromherown.Weassumethatpolicymakersselecttheagentssimultane- Coate(2003),weassumethatthebargainingoutcomeisgivenbythemaximumofthesumofutilitiesoftheagentsatthebargainingtable: (5) (1++b(g1)+b(g2)]−(p+c)(g1+g2)Noticethatisidenticaltothesocialwelfarefunctiongivenby(2).Hence,centraliseddecisionmakingwithsinceredelegationproducesthesociallyoptimallevelsofpublicgoodsdescribedby(3).Centralisationofpoliticaldecisionmakingthuscompletelyresolvestheexternalityprob-lem,providedthatdelegationissincere.Recallthatthesocialoptimummaximisesthesumoftheutilitiesofthemedianvotersinthetwore-gions.Sinceregionsareidentical,itfollowsthatbothmedianvotersarebetteroundercentraliseddecisionmakingwithsinceredelegationcomparedtodecentraliseddecisionmaking.7CentralisedDecisionMakingwithStrategicDel-egationInthissection,werelaxtheassumptionthatdelegationissincere.Weshowthatpolicymakershaveanincentivetomisrepresenttheirpolicypreferencesatthecentrallevel.Asaresult,undercentralisedpoliticaldecisionmakingunderspendingmaypersistoroverspendingmayarise.Toclarifythepolicymakers’motivesformisrepresentingtheirpolicypreferences,werstderivetheamountsofpublicgoodsthatwouldbesetifoneofthepolicymakershadcompletecontrolovercentralpolicy.Giventhatthedirectcostofpublicgoodsnancedthroughacommonbudgetwhileindirectcostarebornelocally,theobjectivefunctionofthepolicymakerfromregionis: (6) Alternatively,wecouldassumethatthebargainingoutcomeisdescribedbytheNashbargainingfunction.Thiswouldgivepolicymakersadditionalincentivestomisrepresenttheirpolicypreferences,particularlywhenthepreferencesofthedel-egatesaectthepolicymakers’outsideoptions,asinSegendor(1998),orwhendelegatesalsobargainaboutside-payments,seeBrueckner(2000).ThiscaseisclosetoBesleyandCoate(2003)’sanalysisofcentraliseddecisionmakingwithanoncooperativelegislature. 5DecentralisedDecisionMakingUnderdecentraliseddecisionmaking,eachregiondecidesindependentlyontheprovisionofpublicgoods.Publicgoodsarenancedlocally.Hence,thepercapitataxequals.Weassumethatineachregionthepolicymaker’spreferencescoincidewiththepreferencesofthemedianvoter.Hence,thepolicymakerchoosestomaximise(1)wherethemedianinregion.Thepolicymakerinregiontakesasgivenwhendecidingon.Optimalpublicgoodsprovisionineachregionunderdecentraliseddecisionmakingisdescribedbythecondition:(4)Asinthesocialoptimum,publicgoodsprovisionunderdecentraliseddecisionmakingincreasesintheintensityofthemedianvoter’sprefer-encesforpublicgoodsanddecreasesinthecosts.Incontrasttothesocialoptimum,theamountofpublicgoodsisindependentofthesizeofthespillovere.Comparing(3)with(4),itfollowsthatfordecentraliseddecisionmakingresultsinunderprovisionofpublicgoods.Theintuitionisclear.Sincethepolicymakersdonottakeintoaccountthepositivespillovereectofpublicgoodsontheutilityofthecitizensintheforeignregion,thesupplyofpublicgoodsistoolow.6CentralisedDecisionMakingwithSincereDele-gationToresolvetheproblemofunderprovisionofpublicgoods,thetworegionsmaydecidetoinstallacentralgovernmentorasupra-nationaldecisionmakingbodytodecideonlocalpublicgoodsprovision.Weassumethatcentraliseddecisionmakingisorganisedasfollows.Eachregionappointsanagenttothecentraldecisionmakingbody.Theagentsfromthetworegionsbargainoverthelevelsofand.Thetaxcostsofpublicgoods,,arenancedthroughacommoncentralbudget.Hence,thepercapitataxineachregionis .Theindirectand,arebornelocally.Assumeforthemomentthatpolicymakersdelegatebargainingatthecentralleveltoagentswiththesamepreferencesastheirown.Werefertothiscaseas‘sinceredelegation’,aspolicymakersmakenoetomisrepresentlocalpreferencesforpublicgoods.FollowingBesleyand Iflocalpublicgoodsarestrategicsubstitutes,votershaveanincentiveforstrate-gicdelegation,seetheAppendix.Asaresult,underprovisionisevenmoreseverethaninthecasedescribedinthemaintext. strongerherpreferenceforpublicgoods.Theparametermeasuresspillovereects.If,spillovereectsareabsent:individ-ualsinregiondonotcareforpublicgoodsprovisioninregion.Thelargeris,thelargeristhespillovereect.If,individualscareequallyforthepublicgoodsproducedintheirownregionastheydoforthepublicgoodsproducedintheotherregion.Then,thepublicgoodsmaybecalled‘global’publicgoods.Theassumptionofseparabilityoflocalpublicgoodsintheutilityfunction(1)maybeconsideredrestrictive.IntheAppendix,weexam-ineanalternativespecicationwherelocalpublicgoodsarestrategicsubstitutes.4TheSocialOptimumBeforeweconsiderpoliticaldecisionmakingonpublicgoods,wederivethesociallyoptimalamountsofpublicgoods.Thesocialopti-mumservesasabenchmarkagainstwhichtoevaluatetheoutcomesofpoliticaldecisionmakingunderdierentinstitutionalstructures.Wede-nethesocialoptimumastheoutcomewhichmaximisestheunweightedsumofutilitiesofallindividualsinbothregions.Sinceindividualsaresymmetricallydistributedovertheinterval ,andwithpopulationsizenormalisedtoone,socialwelfareisequaltothesumoftheutilitiesofthemedianvotersinbothregions.Hence,thesociallyoptimalamountsofpublicgoodsarefoundbymaximising:g:b(g1)+κb(g2)]−(p+c)g1(2))b(g2)+κb(g1)]−(p+c)g2=λm(1++b(g1)+b(g2)]−(p+c)(g1+g2)whereandarethemedianvoter’svaluesofinregion1and2,respectively,whicharethesamesinceregionsareidentical.Sociallyoptimalpublicgoodsprovisionisdescribedbythefollowingconditions:(1+(3)(1+Itisclearfrom(3)thatthesociallyoptimalamountsofandincreaseintheintensityofthemedianvoters’preferencesforpublicgoodsincreaseinspillovers,anddecreaseinthecostofpublicgoods( thedelegatefromtheforeignregion,andhencechangesthebargainingoutcomeinfavourofthedomesticpolicymaker.Brueckner(2000)allowsforbargainingoverside-paymentsinadditiontopolicies.Heshowsthattoattractside-payments,localpolicymakersdelegatepolicymakingauthoritytoagentswhofavourthestatusquooflowspendingonpublicgoods.IncontrasttothesestudiesandtothestudyofBesleyandCoate,ourpaperidentiesthecharacteristicsofpolicydomainsinwhichunderspendingoroverspendingislikelytooccur.3TheModelThemodelrevolvesaroundpoliticaldecisionmakingonpublicgoodsprovisionintworegions.RegionsareidenticalandlabelledPublicgoodsprovisioninoneregionhaspositivespillovereectsontheutilityofindividualsintheotherregion.Theproductionofoneunitofpublicgoodsinaregionentailsapercapitataxcostof.Forconvenience,taxesarenon-distortionary.Inaddition,eachunitofpublicgoodsproducedinaregionhasindirectpercapitautilitycostforallcitizensintheregion.Forsimplicity,weassumethatindirectcostarelinearinpublicgoodsproduction.Thedierencebetweenthetaxcostandtheindirectcostisthattaxcostcanbesharedbetweentheregionsthroughacommoncentralbudgetwhileindirectcostcannot.Individualsineachregiondierintheirpreferencesforpublicgoodsrelativetoprivategoods.Theutilityfunctionofindividualinregionisgivenby::b(gi)+κb(g−i)]+(1)whereistheamountofpublicgoodsprovidedinregionisgrosspercapitaincome,andisthepercapitataxinregionThus,percapitaconsumptionofprivategoodsinregion.Whenpublicgoodsnancedlocally,thepercapitataxequals.Whentaxcostaresharedamongtheregionsthroughacommonbudget, InSection7,weexaminenancingthroughacommonbudgetwithmoresophisticatedsharingrules.Astaxesareassumedtobenondistortionary,wewillhenceforthomitthegrosspercapitaincome.Thefunctionisconcaveandincreasing.Theparameteraccountsfordiinpreferencesforpublicgoodsrelativetoprivateconsumptionamongindividualsinaregion.Individualsineachregionaresymmetricallydistributedovertheinterval .Thelargerisanindividual’s,the OursetupdiersslightlyfromthatofBesleyandCoate(2003)inthetreatmentofspillovers.Theutilityfrompublicgoodsintheirmodeliss(−κ)ln,implyingthatthesizeofspilloversectsthetrade-betweenthedomesticpublicgoodandprivateconsumption.Thisisnotthecaseinourmodel. entailsatrade-obetweenthebenetsofcentralisationofpolicymak-ing(e.g.economiesofscaleandinternalisationofexternalities)andthecostsofpolicyuniformity(aneglectofthediversityinpreferencesforpublicgoods).InOates’analysis,policiesaresetbyasocialplanner.Morerecently,attentionhasshiftedtothepoliticalprocessesthatgov-ernpolicychoicesandtheincentivestocentralisepolicymaking(AlesinaandSpolaore,1997,BoltonandRoland,1997,Ellingson,1998,Alesina,Angeloni,andEtro,2001a,b,GoyalandStaal,2002).Whenpoliciesaresetaccordingtothewishesofamajorityofvotersinthefederation,re-gionswhichhaveminoritypreferencesmaybeworseoundercentraliseddecisionmaking.Thismaygiveregionsanincentivenottojoinafeder-ationevenwhenpotentialeciencygainsarelarge.Recently,Gradstein(2000)arguesthatacommitmenttoanegalitarianbargainingrulemaybeneededtoextractthefullbenetsofcentralisationandtoguaranteethepoliticalsustainabilityofcentraliseddecisionmaking.Political-economicstudiesintheOates’traditionarewellsuitedtoexplainthecostofcentralisationinpolicydomainswherepublicgoodscannotbedierentiatedaccordingtothepreferencesoflocalities.How-ever,inmanycasesitispossibletodecidecentrallyongeographicallyerentiatedlevelsofpublicgoodsinlinewiththediverseregionalpreferencesandcultures.Thisopensuptoredistributiongamesamongregionstogatherinalargershareofcentralspending.InBesleyandCoate(2003),regionsseektoattractalargershareofcentralspendingbydelegatingbargainingtopublicgoodlovers.InPerssonandTabellini(1994)localpolicymakersusecontributionstopersuadethecentralleg-islatortoallocatepublicspendingtowardstheirregion.Asallregionsnon-cooperativelymakethesecontributions,inequilibriumthecentrallegislatorsuppliestoomanylocalpublicgoods.Cheikbossian(2000)pointsoutthatsupplydecisionsonpublicgoodsareoftentakenatthelocallevel.Ifthesegoodsarenancedthroughacentralbudget,votersineachregionhaveanincentivetoappointapublicgoodloverastheirlocalpolicymaker.Inallthesestudies,centralisationofpoliticaldecisionmakingresultsinoverprovisionofpublicgoods.Themaincontributionofourpaperistoexamineunderwhatconditionsunderprovisionofpublicgoodsmaypersistundercentralisedcooperativepoliticaldecisionmaking.Twootherrecentpapershavedealtwiththisissue(Segendor,1998,andBrueckner,2000).Asinourpaper,underspendingemergesbecauselocalpolicymakersdelegatebargainingtoconservatives,butfordierentreasons.Segendor(1998)assumesthatthepreferencesofthedelegatesectpolicyoutcomeseveninthecaseofabreakdowninnegotiations.Then,delegatingbargainingtoaconservativeagentservesasathreatto localgovernments.Theyshowthatoverprovisionofpublicgoodsmayresultfromstrategicdelegationbyjurisdictions.Intheirmodel,thera-tionaleforcentraliseddecisionmakingisthatlocalpublicgoodshavepositivespilloverseectsonwelfareinotherregions.Decentralisedde-cisionmakingthereforeresultsinunderprovisionofpublicgoods.Cen-traliseddecisionmakingcompletelyresolvestheexternalityproblempro-videdthatlocalpolicymakersdelegatebargainingtoagentswhohavethesamepreferencesforpublicgoodsasthemselves.However,whenthecostofpublicgoodsaresharedthroughacommonbudget,policymak-ershaveanincentivetodelegatebargainingto‘publicgoodlovers’.Thedelegationofapersonwithstrongpreferencesforpublicgoodsservestoincreasespendingonlocalpublicgoodsattheexpenseofthecom-monbudget.Sinceinequilibriumalldistrictssendpublicgoodlovers,strategicdelegationresultsinoverprovisionofpublicgoods.Inthispaper,weextendtheanalysisbyBesleyandCoate(2003)toexplainwhyunderprovisionofpublicgoodsmaypersistundercooper-ativecentraliseddecisionmaking.Becauseoursetupisveryclosetotheirs,thetwopapersmayserveascomplementstoexplainwhycoop-erationatacentralisedlevelsometimesresultsinoverspending,whereasinothertimesitresultsinunderspending.Weextendthemodelbyal-lowingforcostwhichcannotbesharedamongdistricts.Examplesofpolicydomainswherenon-shareablecostareimportantincludeenviron-mentalpolicy(whichimposeindirectcostsonlocalindustries),asylumpolicies(whereshorttermadjustmentcostsmayfallonlocalcommuni-ties)andmultilateraleortsforpeacekeeping(thatmayresultinlocalcasualties).Weshowthatifasucientlylargepartofthecostofpublicgoodscannotbesharedamongregions,underprovisionofpublicgoodspersistsundercentraliseddecisionmakingbecauselocalpolicymakersdelegatebargainingto‘conservatives’.Underprovisionofpublicgoodsisstrongestwhenspillovereectsaremoderate.Weshowthatbothintheabsenceofspillovereectsandinthecaseofglobalpublicgoods,centraliseddecisionmakingproducesthesociallyoptimalamountsofpublicgoods.Finally,wederivenancingrulesthateliminatestrate-gicdelegationbylocalpolicymakersandthuspromoteecientpublicgoodsprovision.2RelatedLiteratureAconsiderablebodyofliteratureonfederalismshowswhycentralisa-tionmayproducesuboptimalpolicies.ManyofthecontributionsareinthetraditionofOates’(1972)decentralisationtheorem.AnimportantassumptioninOates’analysisisthatcentralisationimpliespolicyunifor-mity.Then,asOatesshows,theoptimalpoliticaldesignofjurisdictions 1IntroductionCentralisationofpoliticaldecisionmakingoftenfailstoproducethedesiredresults.Forinstance,itisfrequentlyarguedthatdecisionmakingwithintheEuropeanUnionresultsinoverspendingandoverregulationinsomepolicyareas,whiletoolowspendingandtoolittleregulationpersist Utrecht School of Economics Tjalling C. Koopmans Research Institute Discussion Paper Series 04-09 Why Does Centralisation Fail to Internalise Policy Externalities? Robert Dur 1 Hein Roelfsema 2 1 Tinbergen Institute Earasmus University Rotterdam 2 Utrecht School of Economics Utrecht University January 2004 Abstract We thank an anonymous referee of , Amihai Glazer, Otto Swank, Huseyin Yildrim, seminar participants in Amsterdam and Rotterdam, and participants of the 2002 Annual Meeting of the Public Choice Society in San Diego, the 2002 Annual Meeting of the European Public Choice Society in Belgirate, and the XIV Villa Mondragone International Economic Seminar in Rome. Dur gratefully acknowledges financial support from NWO, KNAW, and VSNU through a Vernieuwingsimpuls grant. Roelfsema likes to express his appreciation for the hospitality at the Department of Economics of Erasmus University Rotterdam. Keywords: Centralised decision making, strategic delegation, financing rules. Acknowledgements We thank an anonymous referee for Public Choice, Amihai Glazer, Otto Swank, Huseyin Yildrim, seminar participants in Amsterdam and Rotterdam, and participants of the 2002 Annual Meeting of the Public Choice Society in San Diego, the 2002 Annual Meeting of the European Public Choice Society in Belgirate, and the XIV Villa Mondragone International Economic Seminar in Rome. Dur gratefully acknowledges Financial support from NWO, KNAW, and VSNU through a Vernieuwingsimpuls grant. Roelfsema likes to express his appreciation for the hospitality at the Department of Economics of Erasmus University Rotterdam. Tel +31 (0) 302537955 E-mail: h.roelfsema@econ.uu.nl : http://www.koopmansinstitute.uu.nl dur@few.eur.nl Hein Roelfsema Utrecht University Utrecht School of Economics Vredenburg 138 3511 BG Utrecht Robert Dur Department of Economics H 7-21 Erasmus University Pobox 1738 3000 DR Rotterdam The Netherlands Fax: +31-10-408 9147 Email: ontwerp voorblad: WRIK Utrecht How to reach the authors Please direct all correspondence to the first author. of research results, and to enhance discussion with colleagues. Please send any comments and suggestions on the Koopmans institute, or this series to P.vanDriel@econ.uu.nl It was founded in 2003, and named after Professor Tjalling C. Koopmans, Dutch-born Nobel Prize laureate in economics of 1975. es the Koopmans Institute publishes results of ongoing re The Netherlands telephone +31 30 253 9800 fax +31 30 253 7373 website www.koopmansinstitute.uu.nl The Tjalling C. Koopmans Institute is the research institute Tjalling C. Koopmans Research Institute Utrecht School of Economics Utrecht University Vredenburg 138 3511 BG Utrecht Utrecht School of Economics Tjalling C. Koopmans Research Institute Discussion Paper Series 04-09 Why Does Centralisation Fail to Internalise Policy Externalities? Robert Dur 1 Hein Roelfsema 2 1 Tinbergen Institute Earasmus University Rotterdam 2 Utrecht School of Economics Utrecht University January 2004 Abstract We provide an explanation for why centralisation of political decision making results in overspending in some policy domains, whereas too low spending persists in others. We study a model in which delegates from jurisdictions bargain over local public goods provision. If all of the costs of public goods are shared through a common budget, policy makers delegate bargaining to `public good lovers', resulting in overprovision of public goods. If a sufficiently large part of the costs can no be shared, underprovision persists because policy makers delegate bargaining to `conservatives'. We derive financing rules that eliminate the incentives for strategic delegation. Keywords: Centralised decision making, strategic delegation, financing rules. Acknowledgements We thank an anonymous referee for Public Choice, Amihai Glazer, Otto Swank, Huseyin Yildrim, seminar participants in Amsterdam and Rotterdam, and participants of the 2002 Annual Meeting of the Public Choice Society in San Diego, the 2002 Annual Meeting of the European Public Choice Society in Belgirate, and the XIV Villa Mondragone International Economic Seminar in Rome. Dur gratefully acknowledges Financial support from NWO, KNAW, and VSNU through a Vernieuwingsimpuls grant. Roelfsema likes to express his appreciation for the hospitality at the Department of Economics of Erasmus University Rotterdam. Tel +31 (0) 302537955 E-mail: h.roelfsema@econ.uu.nl : http://www.koopmansinstitute.uu.nl dur@few.eur.nl Hein Roelfsema Utrecht University Utrecht School of Economics Vredenburg 138 3511 BG Utrecht Robert Dur Department of Economics H 7-21 Erasmus University Pobox 1738 3000 DR Rotterdam The Netherlands Fax: +31-10-408 9147 Email: ontwerp voorblad: WRIK Utrecht How to reach the authors Please direct all correspondence to the first author. of research results, and to enhance discussion with colleagues. Please send any comments and suggestions on the Koopmans institute, or this series to P.vanDriel@econ.uu.nl It was founded in 2003, and named after Professor Tjalling C. Koopmans, Dutch-born Nobel Prize laureate in economics of 1975. es the Koopmans Institute publishes results of ongoing re The Netherlands telephone +31 30 253 9800 fax +31 30 253 7373 website www.koopmansinstitute.uu.nl The Tjalling C. Koopmans Institute is the research institute Tjalling C. Koopmans Research Institute Utrecht School of Economics Utrecht University Vredenburg 138 3511 BG Utrecht