American Political Science Review Vol - PDF document

AmericanPoliticalScienceReviewVol.99,No.1February2005 WhoSurfs,WhoManipulates?TheDeterminantsofOpportunisticElectionTimingandElectorallyMotivatedEconomicInterventionMARKANDREASKAYSERUniversityofRochesterandUniversityofOxfordnthispaper,IdevelopacareerconcernsmodelofgovernmentpolicychoicewithinadynamicoptimalstoppingframeworktopredictthedegreeofsurÞng(opportunistictiming)andmanipulation(politicallymotivatedeconomicintervention)underalternateinstitutionalstructures.Amongotherresults,IÞndthatthelikelihoodofopportunisticelectionsriseswithexogenouseconomicperformance, MarkAndreasKayserisAssistantProfessor,DepartmentofPolit-icalScience,UniversityofRochester,Rochester,NY14627-0146,andPostdoctoralPrizeResearchFellow,NufeldCollege,OxfordUniversity,OxfordOX11NF,UK(mark.kayser@rochester.edu).Iwouldliketoexpressmygratitudetothosewhoprovidedcom-mentsonorsuggestionsforthispaper,especiallyKathleenBawn, OnenotableexceptiontotheempiricalorientationofthisresearchareawasChappellandPeel’s(1979)earlymodelofthepoliticalbusi-nesscyclewithendogenouselections.Thismodel,however,providesnouncertaintyintheelectioncallingdecision. WhoSurfs,WhoManipulates?February2005 Lewis-Beck[1988]1990)ortherationalexpectationsofvoters(e.g.,CukiermanandMeltzer1986,PersonandTabellini1990,Rogoff1990,andRogoffandSibert1988).Researchers,leftwithastrongincen-tiveforgovernmentstomanipulatetheeconomy—voters,afterall,punishgovernmentsforpooreco-nomicperformance—–butscantevidenceofcyclesinmacroeconomicaggregates,turnedtheirattentiontoaner-grainedinstrument:cyclesinscalbalances.Theintuitionissimple:Itiseasierforgovernmentstomanipulatescalcomponentstotargetelectorallyimportantconstituenciesandsignalcompetencetora-tionalvoters(Rogoff1990)thantoengineertimelyimprovementsinoutputorothermacroeconomicag-gregates.Yetagain,however,empiricsraisedanewpuzzle:“Politicalcycles”arerarelyfoundinde-velopedcountries(Alesina,Cohen,andRoubini1992),emerginginsteadprimarilyinthedevelopingworld.Thismaybeattributabletoweakerinstitutionsandstrongergovernments(Schuknecht1996)ortoloweraudiencecostsofmanipulation(Gonzalez1999)inlessdevelopedcountries.Iarguethatitisalsoattributabletothepreponderanceofendogenouselectiontimingindevelopedcountries,specicallytheOECD.Ofthe24OECDmembercountriesin1990,onlyNorway,Switzerland,andtheUnitedStatesfullyprecludeearlyelections.Opportunisticelectiontimingmattersforexplainingtheabsenceofpolitical–economiccycles,regardlessofwhethertheyareinscalbalancesormacroeconomicaggregates.Surprisingly,however,aftertheinitialburstofinterestinthemoderatinginuenceofsurngonmanipulation,theeldstalled.Acombinationofdi-minishingmarginalreturnsfromdemonstratingthesameeffectinadditionalcountriesandtheabsenceoftestabletheoriesofhowinstitutionsaffectsurngandmanipulationdampenedinterest.Iseekheretoredressthisproblembydevelopinganexplicitmodelwithempiricallytestablepredictions.Tobeprecise,Iembedacareerconcernsmodelofgovernmentpol-icychoicewithinadynamicoptimalstoppingframe-worktopredictthedegreeofsurngandmanipula-tionunderalternateinstitutional,governmental,andconstituencyfeatures.Amongotherresults,Indthatthelikelihoodofearlyelectionsriseswiththevarianceofexogenousshockstovoters’welfare,withlonger Indeed,macroeconomicmanipulationnotonlymaybeineffectivebutalsomaybackreasrationalvoters,behavingas“scalconser-vatives,”recognizepoliticallyengineeredelectionyearboomsandpunishincumbentsforexpectedpostelectioneconomicdistortions(Peltzman1992).ClarkandHallerberg(2000)offeranadditionalexplanation:Givencapitalmobility,xedandexibleexchangerateregimesconstrainmonetarypolicyandscalpolicy,respectively.However,theconstraintsonopportunisticelectioncallingvaryconsiderablyamongendogenoustimingcountries.WiththenotableexceptionofarticlesbyHeckelmanandBerument(1998)andReid(1998).AsIexplainlater,votersinthisframeworkcannotdirectlyobservegovernmentcompetence;theyinferitfromtheirwelfare.Consistentwithempiricalliterature,theydonotdistinguishbetweenwelfareshocksthatarebeyondthecontroloftheirelectedrepresentativesandthosethataremorecrediblyattributabletogovernment.constitutionalinterelectionperiods(CIEP),andwithuncertaintyaboutthefuturebutdiminisheswiththevalueofofce-holding;manipulationdecreasesinthevarianceofwelfareshocksbutincreaseswiththeCIEPandwiththevalueofofce-holding;andassurngisincreasinginwelfareshocksbutmanipulationdecreas-ing,surngeffectivelysubstitutesformanipulation.Theseresultsarestronglyprescriptiveforinstitu-tionalengineeringandsuggestpossibleunintendedconsequencesforcertainpoliticalandconstitutionalarrangements.Consider,forexample,howincreasinginternationaleconomicintegrationmighthavenotablydifferenteffectsonpoliticsinendogenoustimingandxed-timingcountries.Asbothmonetaryandscalin-terventionintheeconomybecomeincreasinglycon-strainedbyeconomicandpoliticalintegration,govern-mentsabletotimetheirelectionsstrategicallyshouldenjoyanelectoraladvantagerelativetotheirxed-timingpeers.Thosegovernmentswiththefewestinsti-tutionalandpoliticalbarrierstoopportunisticelectiontiming,thesmallest,mostvolatileeconomies,andthelongestCIEPsshouldenjoythegreatestadvantageintheirreelectioncampaigns.Wheremanipulationisleastconstrainedandeconomicperformancemostvolatile—–primarilyinthelesseconomicallyintegrateddevelopingworld—–itistheelectoratethatmaygainthemostfromopportunisticelectiontiming.Thosestatesthatleastimpedethestrategictimingofelections,pri-marilyparliamentarydemocracieswithlongmaximumterms,shouldexperiencelesspreelectionmanipulationandconcomitantlylessdistortioninthepostelectioneconomy.Thesecondgoalofthispaperistoimprovetheun-derstandingofopportunisticelectiontimingitselfbyreplicatingformajoritygovernmentswhatDiermeierandStevenson’s(2000)LupiaandStrom’s(1995)coali-tionbargainingmodelshavedoneforthestudyofgovernmentterminationundercoalitiongovernments.Bymakingcoalitionbargainingexplicit,theymakedissolutionandelections—–whetheropportunisticorinvoluntary—–aconsequenceofthestrategicinterac-tionofcoalitionmembersratherthanthevaguelyepi-demiologicalprocessthatdominatedearliergovern-mentdurationliterature.Afterall,asGrofmanandRoozendaal(1994)oncecommented,itisthechoicesofactorswithincircumstancescreatedbyexogenousshocksthatbringcabinetsdown.Electiontimingbysingle-partygovernmentsisconsiderablysimplerthandynamiccoalitionbargainingbutmuchcanstillbegainedfromanexplicitconsiderationoftheactors’choiceswithincircumstancescreatedbyexogenousshocks.Single-partygovernments,forexample,maxi-mizeboththedurationofthecurrenttermandtheprobabilityofreelection,not,asthetraditionalgovern-mentdurationliteratureassumes,justtimeinofce.Imodelthistrade-offbetweenextendingthecurrenttermandwinninganotherasadynamicoptimiza-tionproblem.Thisisnottherstmodelofstrategicelectiontiming—–signalinggamesbySmith(1996)and SeeGrofmanandRoozendaal’s(1997)reviewarticleforahistoryofthecoalitionstabilityanddurationliterature. AmericanPoliticalScienceReviewVol.99,No.1 Terrones(1989)ndequilibriaonnoearlyelectionsandearlyelections,respectively,anddynamicpro-grammingmodelsbyBalke(1990)andIto(1990)yieldnumericalpredictionsofelectiontiming—–butthismodel,inadditiontoexplicitconsiderationofmanip-ulation,offerstherstanalyticpredictionsofwhenelectionsshouldbecalledandunderwhichinstitutionstheyshouldbemostprevalent.Theremainderofthispaperdevelops,extends,anddiscussestheimportanceofadynamicstochasticmodelofopportunisticelectiontimingandeconomicmanip-ulation.Ibeginwithaparsimoniouselectiontimingmodelillustratinghowgovernmentsformanite-horizonoptimalstrategyforcallingelectionstomaxi-mizetheirreelectionprobability.Governmentsinthisinitialsetuponlydecidebetweencallingandnotcall-inganelectionineachperiodoftheirterm;theyonlycareaboutreelection;andtheyincreasetheirreelec-tionprobabilitybycoordinatingelectionswithfavor-able,exogenouslydeterminedstatesoftheworld.Asubsequentsectionthenbuildsontheinitialmodelbyconsideringutilityfromofce-holding,notjustreelec-tion.Earlyelectionscostgovernmentstheremainderoftheircurrentterm,atrade-offthatmattersifofce-holdingitselfisdesired.Inextconstrainthestochasticstatevariabletoeconomicshocksandmakethevoters’decisioncalculusexplicit.Othernoneconomiceventscertainlymatterforelectiontiming,butparsimony,aswellastheupcomingexplorationofhowtimingsub-stitutesforeconomicmanipulation,demandsafocusonmacroeconomicmanagement.Votersinthisthirdsectionofthemodelnowinfercompetencefromthegovernment’sprovisionofpublicgoods,agov-ernmentactivitythatisconstrainedbytheexogenousperformanceoftheeconomy.Thenalextensionofthemodel,afterabriefdigressionontheintuitionoftheelectiontimingmodel,allowsgovernmentstocir-cumventthisbudgetconstraintbyincreasingcurrent-periodspendingviaadistortionarytaxonthepostelec-tionfuture.Comparativestaticsthensuggest,amongotherresults,thatsurngdoessubstituteformanip-ulation,thatcountrieswithmorevolatileeconomiessurfmore(andmanipulateless),thatlongermaximum-termlengths(CIEPs)increaseboththelikelihoodofopportunisticelectionsandpoliticalinterventionintheeconomy,andthatthevalueofofceholdingdecreasessurngbutincreasesmanipulation. EarlyelectionsinTerrones’modelsignalhighcompetencebecausealow-competencegovernment,knowingitwouldstandlittlechanceofreelection,shouldsimplymaximizeitstimeinofcebypostpon-ingelectionstothelastpossiblemoment.EarlyelectionsinSmith’smodelsignallowcompetencebecausehighcompetencegovernmentshavegreatercondenceintheirabilitytoperformwellinthefuture.LaterempiricalworkbySmith(2003)conrmsthespirit,ifnottheequilibrium,ofhissignalingmodel.Britishgovernmentscallearlyelectionsbut,becauseearlyelectionssignalgovernmentforeknowl-edgeofimminentdownturns,voterspenalizethemfordoingso.Thepenalty,however,isnotsufcienttoprecludeopportunisticelectioncalling.Thestochasticmodelpresentedherepredictsthatgovern-mentscallelectionsduringexceptionallylargeexpansionsunlikelytorecurinthesameterm.Thisisoften,butnotalways,observationallyequivalenttocallingelectionsbeforedownturns.Asamajorityofparliamentarydemocraciesallowforendogenouselectiontiming,theimplicationsofthisresearchareneitherobscurenortrivial.Aclearerunderstandingofwhyandwhenelectionsarecalledpromisesbroadlyapplicableinsightsintoelectoralpol-iticsandthepoliticalbusinesscycle.THEBASICMODELElectiontimingisquintessentiallyaproblemofop-timizationunderuncertainty.Theincumbentgovern-mentassesseselectoralconditionsineachperiodofitstermanddecideswhethertocallanelectionortoproceedtothenextperiod,notknowingwhatthatperiodholds.Opportunetimingcanincreaseboththeprobabilityofelectoralvictoryandlegislativeefcacythroughgreaterseatsharebutalsoeffectivelystops(i.e.,resets)thegame.Thus,electiontimingisfun-damentallyanite-horizonoptimalstoppingproblem.Thegovernmenthasaxednumberofperiodsinwhichtocallanelectionbeforeoneisimposedandthereforetimestheelectiontocoincidewiththemostelectorallyadvantageouscircumstances.Recognizingelectiontimingasanoptimalstoppingproblemallowsonetomodelagovernment’sdynamicdecisionprocessexplicitly.Dynamicprogrammingtechniques,asexplainedbelow,permitcurrent-periodoptionstoberepeatedlycomparedtotheexpectedvalueoffutureopportunities.Althoughtheunknownvaluesofseveralparameterspreventusfrompredict-ingthemagnitudeofdifferentvariables’effectsontheprobabilityofelectionsinthismodel,itisneverthelesspossibletopredictthesignofthateffect.Imaginefornowthesimplestpossiblearrangement:Votershavenomemoryofpreviousperiods;thegov-ernment’sreelectionprobability,,ineachperiod,...,
,isstrictlyincreasingintherandomstatevariable,,distributed,forsimplicity,uniformlywithdensity;eachdrawofisindependentofothers;thevalueofofce-holdingisconstantovertime;andgovernmentssimplymaximizetheirchanceofreelec-tion,bytimingelectionstocoincidewiththehighestthattheybelievewilloccurintheirterm.Giventhatthegovernmentcannotseefuturestates,howdoesitdeterminethatthecurrentstateisthebestthattheywilllikelyseebeforetheirtermexpires?Thegovernmentknowsthedistributionfromwhichisdrawnand,asinanystationaryMarkovdecisionproblem,playsastrategytomaximizeitspayoff:Callanelectiononlyifthecurrentstateexceedsthebestexpectedfuturestate.Assumingoptimizingbehaviorineachperiodallowsustocalculatethegov-ernment’sexpectedfuturestateviabackwardinduc-tion.Thus,forexample,inthepenultimateperiod, Estimatingdiscretechoicedynamicprogrammingparametersisoftenpossible,however,andhasreceivedconsiderableattentionfromsomeeconometriciansandlaboreconomists.EcksteinandWolpin(1989)offerasurveyofseveralapproachesandnewertech-niques(e.g.,KeaneandWolpin1994)continuetoemerge.Inthecaseofthispaper,onepossibilitymightbetoemploypollingdatainplaceofthecurrentstochasticeconomicshocks. WhoSurfs,WhoManipulates?February2005 1,thegovernment’sexpectedstateinthesubse-quentandnalperiod,),issimplytheexpectedvalueoftherandomvariableµ,µ 2,1+1 i.e.,one.Inthispenultimateperiodthegovernmentwillcallanelectionifthecurrentstateofthecoun-try,,isgreaterthantheexpectedfuturestate,1.Thislogicisthenextendedtothepre-cedingperiod,2.Theexpectedstatecontinuing2to1issimplythevalueofplayinganoptimalstrategyat1,i.e.,theaverageexpectedstatefrom1’stwopossibleoutcomesoverallpos-siblevaluesof:(1)continuingtothenalperiod)[payoff1]and(2)callinganelec-)[payoffExpressedmoresuccinctly,thegovernment’sdeci-sionateachsolvestheprogram 21Š1 2Et(µt+1)}µdU1Š1 2,1+1 whichimpliesthatthevalueofplayinganoptimalstrat-egyat 2E(µt+1)}µdµ+
1+1 or,moreexplicitly, 2E(µt+1)dµ+
1+1 Thustheexpectedstatefromplayinganoptimalstrat-egycanbecalculatedatanyperiodandtheexpectedvalueofcontinuingisjustthevalueofanoptimalstrat-egyinthesubsequentperiod.),theex-pectedutilityofcontinuinginofce,isamonotonicallydecreasingfunctionoftime,convergingto),i.e.,unity,inthepenultimateperiod,itisleastlikelythatcurrent-periodcircumstanceswillexceedtheexpectedfuturecircumstancesatthebeginningofaterm.Asremainingtenurewanes,however,theexpectedfuturestatesteadilydeclines,therebyincreasingtheprobabil-itythatacurrentperiodrandomdrawofwillexceeditandtriggeranelection. Keepinmindthedistinctionbetween)and):Theformerissimplytheexpectedvalueofasingledrawfromthestatevariable)distribution,whereasthelatteristheexpectedutilityofplayinganoptimalstrategyatEquations(1)and(2)dependontheBellmanoptimalityequation.SeeSundaram([1996]1999,274)foranexplanationofhowadeci-sionruleconstitutesanoptimalstrategyinnitehorizondynamicprogramming.SeeKreps(1990,Appendix2.2)foramathematicallysimilaroptimalstoppingexample.Theexpectedutilityofcontinuingdeclineswithtimebecausethenumberoffuturedrawsdecreaseswithtime.Thisprocess,toborrowfromtheextensiveoptimalsearchandstoppingapplicationsinlaborGovernments,however,deriveutilityfrommanytime-dependentactionsinofce,notjustfromreelectionperse.Inthiscontext,electiontoofcebecomesonlyanecessarycondition:Governmentsmaximizetheirreelectionprobabilitiesinordertoextendtheirtimeinofcebutmusttradeoffaprobability-weightednewtermagainsttheremainderoftheircurrentterm.Call-inganelectiontoosooncoststhegovernmenttheutilityfromtheremainderofthecurrentterm;waitingtoolongreducesthechancesofanotherfavorableperiodoccurringbeforemandatorydissolution.Thusaratio-nalincumbentattemptstocallelectionsinthelastbestperiodpossibleWhenutilitycomesfromofce-holding,theincum-bent’sdecisionbecomesEachperiodinofcethegovernmentgainsoneunitofutility—–considerthisanegorent,–andreceivesnewinformationaboutthecurrentstateofthecountry.Therationalofce-seekingincumbentcontinuesinofceuntiltheexpectedutilityofcallinganelectionexceedstheexpectedutilityofcontinuinginofce,knowingthatfutureelectionsmayarriveunderlessfortunatecircumstances.Irepresent,thevalueofcontinu-ing,asThesecondtermintheequation,(,capturestheremainingutilityinthecurrentterminofceandshrinksasthetermprogresses,providingadiminishingincentivetoforgofavorableelectionopportunities.themaximumtermlengthinperiodsandthecurrentperiod.TherstterminEq.(4)representstheotherim-portantelementindeterminingelectiontiming:theexpectedvalueofcallinganelectioninthefuture.Whenincumbentsexpectfavorablecircumstancesforanelectioninthefuture,itiseasiertoforgocurrentopportunities.Moreprecisely,theexpectedvalueofafutureelectionistheproductofthereelectionprobabil-itygeneratedbytheexpectedfuturestatestateE(µt+1)]andtheutilityofanewterminofce,,appropri-atelytimediscounted.Againstthisexpectedvalueofcontinuing,politiciansmustweightheutilityofcallingelections.Initsmostbasicformtheexpectedutilityofelectionsinanyperiodissimplythemaximumlengthofanothertermweightedbythecurrent-periodreelectionprobability, economics,isakintothatfacedbyajobseekerwithaxednumberofsequentialjobofferswithsalariesdrawnfromthesamedistri-bution.Atthenalofferthejobseekerwillhavetoacceptedtheexpectedvalueofasingledrawfromthedistribution,i.e.,itsmean.Thepenultimateofferwillthereforeonlybeacceptedifitexceedstheexpectedvalueofthenaloffer,andsoon,backwardtotherstoffer. AmericanPoliticalScienceReviewVol.99,No.1 VOTERSUntilnow,thegovernment’sreelectionprobablyhassimplybeendenedasafunctionofthestatevari-ablewithoutexplicitconsiderationofhowvotersformtheirpreferences.Inowmakethevoters’preferencemechanismexplicitwithrespecttoeconomicwelfare.Iredenethestatevariableasastochasticshocktotheeconomytogenerateatractablemodelwithbudgetconstraintsthatcanlatertradeoffeconomicmanipu-lationandelectiontiming.Thisisnottoclaimthatnoneconomiceventsdonotaffectelectiontiming—theydo—–butsimplytorestrictthismodeltoeffectsoftheeconomy.Adoptingacareerconcernsframework(Dewatripont,Jewitt,andTirole1999a,b;Holstr1982;PerssonandTabellini2000)positingasymmetricinformation,Icannowmakethevoters’selectionprocess—–andhencethegovernment’sreelectionprobabilitycalculation—–explicit.Votersholdgovern-mentsaccountableforeconomicperformancebut,unabletoobservegovernmentcompetencedirectly,inferitfromtheprovisionofpublicgoods.Moreexplicitly,supposethatgovernmentprovisionofpublicgoods,,isconstrainedby),whererepresentsthetaxrate,,income;,aggregatetwo-periodeconomicshocks;and,ahiddenanddistortingtaxthatshiftsresourcesfromthefuturetothepresent,improvingcurrentperiodwelfareatthecostoftheequivalentamountplusnegativeeconomicdistortions,),inthesubsequentperiod.Knowingthatmorecompetentgovernmentsprovidemorepublicgoods,butimpotenttoinuencegovernmentsaretemptedtocreatetheappearanceofgreatercompetencebyincreasing.Whilevoterscanimmediatelyobserve,and,bothremainatleastpartiallyunobservable(primesindicateinferredvariables).Votersthereforeinferoverallgovernmentcompetenceas).Pastaggregatecompetence,,isbothobservableandrelatedtopresentcompetencebutthecurrentperiod’scompetencecannotbeobserveddirectly.Imagineaggregatecompetence,,asthesumoftherandomandseriallyuncorrelatedeconomicshockvariableoverthepresentandimmediatelyprecedingperiod,,whereisobservedbygovernmentandvotersalikebutknowledgeofisreservedforthegovernment.Giventhattherandomvariableisdistributeduniformlywithmeanoneanddensity,µ 2,1+1 ],voterschoosethegovernmentifitsinferredcurrentcompetenceisgreaterthantheopposition’sexpectedcompetence, Thetaxrate,,remainsxedinthismodelintheinterestofsimplicity.Transformingintoachoicevariablewouldmandateafullwelfarefunctionforvotersandunnecessarycomplexity.Withgovernmentsmayonlychoosebetweensurngandmanipulation.Unityischosensimplytosimplifylateralgebraandhasnosub-stantiveeffectonthemodel’scomparativestatics.Objectionsthatgovernmentcompetenceincreaseswelfarebeyondthegovernment’sresourcesareeasilymetbyreducing)toone-halforless.Notethatvariablesnotdirectlyobservedbytheelectoratearedenotedbyprime.Thus,thegovernment’sreelectionprobabilityis0if1ifsimplytheprobabilitythatitsinferredcurrentcompe-tenceexceedstheopposition’sexpectedcompetence.Moreexplicitly, or,reexpressingandrearranging, Keenobserverswillnoticethepossibilityofboundedrationality.Votersbasetheirassessmentsofgovernmentcompetence—–andhenceelectionpreferences—–oneconomicperformance,yetgovern-mentshavelittledirectinuenceovertheshort-termperformanceofthemacroeconomy.Assumingsomepricestickiness,assomerecentrealbusinesscyclemodelsattempt(cf.McCallum1999),buildsinalagoninationthatthenresemblesthemodelofferedhere:Governmentscanproviderealshort-termstim-ulusbutonlyatthecostoffuturedistortions.Intheabsenceofsuchpricestickiness,governmentswouldbeconstrained—–asinpoliticalbudgetcyclemodels—scalmanipulation,whichobviouslyalsogeneratesanintertemporaltransfer.Thus,Icontinuetoassumethatmanipulationisonlypossibleincombinationwithcostlytransfersfromthefuturebutthat,consistentwithempiricalpoliticalscience,votersneverthelessrewardandpunishgovernmentsonthebasisofeconomicper-formance(Lewis-BeckandStegmaier2000).Voters,therefore,maybefullyrationalorboundedlyrational,dependingonthecosttothemofthegovern-mentinuencingtheeconomicvariablefromwhichtheyinfercompetence.Inthismodelthegovernmentcanonlysignalcompetencebyincreasingpublicgoods,eitherthoughopportunetiming(hencea)orbyincreasing,ahiddenanddistortionarytaxonthefuture.Althoughexogenoustothemodel,ifthevaluetothevotersofthecompetencesignalin-ferredfromexceedsthetime-discountedfuturedis-tortionaryeffectsof,thenthevotersarefullyrationalinmuchthesamewaytheyareinrationalpoliticalbudgetcyclemodels(e.g.,Rogoff1990).Alternatively,ifthevalueofthecompetencesignaldoesnotexceedthatofthediscountedfuturedistortions,votersareboundedlyrational.Empirically,therearenumerousexamplesofvotersholdingpoliticiansaccountableforactsofGodthattheycanatbestmitigatebutcannot SeethedebatesovermonetarypolicyineffectivenessintherealbusinesscycleliteraturelaunchedbyKydlandandPrescott(1980,1982)followingtherationalexpectationscontributionsofLucas(1972)andothers.Notethatthestandardassumptionintherealbusinesscycleliteratureisthatthecyclicalrealeffectsofmonetarypolicyareverysmallornonexistent.Empiricalsupportforthepolicyineffectivenessassumptionisprovided,amongothers,byLittermanandWeiss(1989). WhoSurfs,WhoManipulates?February2005 FIGURE1.ElectionCallingOverTime Thisbehaviormayeveninducegovernmentstoengageincostlyactsofsignalingthatreducefuturewelfare.Theframeworkdevelopedhereissufcientlygeneraltoaccommodateeitherassumptionaboutvoterbehavior.Finally,consistentwiththisdiscussion,notethatre-electionprobabilitiesareincreasinginEq.(7),providinggovernmentswithanincentivetomanipulatetheeconomyinelectionperiodsforelec-toraladvantage.Toprovidemorerealism(andanin-teriorsolution),Iexplorethederivationofoptimal,below;butrstletusconsiderthemodel’sbroaderintuition.THEINTUITIONThusfarIhaveestablishedaframeworkforhowgov-ernmentscalculatereelectionprospectsandhowvotersassessgovernmentcompetence.Inbrief,governments’expectedfutureelection-periodcompetencedimin-ishesastheremainingtime(andhenceelection-callingopportunities)wanes;votersinfercompetenceandse-lectcandidatesbasedontheirownwelfare;andgovern-mentsrepeatedlycomparecurrentandexpectedfutureutilityindecidingwhentocallanelection.Temporarilydisallowingeconomicmanipulationandadaptingthegovernment’selection-callingdecisiontoafullydy-namicsettinginwhichfutureperiodswithinthepresenttermarediscountedrelativetomoreimmediateperi- TherecentpaperbyAchenandBartels(2002)showinghowNewJerseyvoterspunishedlocalpoliticiansforsharkattacks(amongotherexamples)isanamusingillustrationofthis.SusanHansen(1999)showsthatevencompetentstategovernorsareoftenpunishedbyvotersfornationaleconomictrendssuchasunemploymentoverwhichtheyhavelittleinuence.odsnowallowsustostrengthenthemodel’sintuitionbysimulatinganincumbent’speriod-by-periodelection-callingdecision.Aperiod-by-perioddecisionprocessrequiresustocalculatethepresentvalueofeachterm’segorent“revenuestream”atthetimeofeachelectioncallingdecision,i.e.,ateach.Notonlydogovernmentsviewfuturetermsaslessvaluablethanthepresentterm,butwithinagiventerm,adistantperiodislessvalu-ablethanthecurrentperiod.Hence,thegovernment’selection-callingdecisionshouldbeexpressedaspReThisdecisionisbestpresentedgraphically.Figure1simulatesthecalling(solidline)andcontinuing(dottedline)decisionovertime,assuming02,and60.Obviously,whencallingexceedscontinuing,anelectioniscalled.Towardthebeginningofaterm,thetimeremaininginthecurrentterm,,togetherwithhighexpectedfuturepopularity,),ensuresthatthevalueofcontinuinginofceexceedsthevalueofcallingelections.Astimeinofceprogresses,however,)diminish,loweringthevalueofwaitingtotherangewhereanexogenousstochasticeventmaymakeasnapelectionthemoreappealingoption.Fromthegovernment’sperspective,when,circumstancesarebetterthanexpectedandlikelyonlytodeteriorateinthefuture. AmericanPoliticalScienceReviewVol.99,No.1 FIGURE2.ElectionCalling,LongVoterMemories Whatifvoters’affectionsarenotsockle?Althoughextraneoustothemodelathand,itisnotdifculttosimulatevoterswhorecallpreviouseventsandup-datetheirassessmentofgovernmentcompetencemoreslowly.Constructingasalinearcombinationinwhichpastreelectionstandingiscombinedwithneweventsinproportiontothestrengthofvotermemorydoesnotchangethemodel’sfundamentals.Supposethathat,1],capturesthestrengthoftheprevi-ousperiod’sinuenceoncurrentreelectionprospectsrepresentsthecurrentperiodinstantaneousreelectionprobability,i.e.,thegovernment’schanceofreelectionifvotersbasetheirdecisionononlythecurrentperiod’sevents.Athigherlevelsofnewdevelopmentsaffectreelectionprospectslessand,obversely,lowerlevelsofweightpresenteventsmore.Votermemoryreducesthevolatilityofgovernmentreelectionprobabilitiesandconsequentlymaylowerthelikelihoodofearlyelections.Figure2(7,and60)illustratesthis:,thevalueofcallingateach,isvisiblylessvolatilethaninFigure1.Abroaddenitionofcollectivevotermemory,suchasonethatincludeseducationlevelsandtheindepen-denceandqualityofthepressinadditiontoinnatehumanmemory,wouldimplythatsocietieswithhigher(lower)levelsofeducationandabetter(worse)pressmaybemore(less)resistanttoopportunisticelectioncalling.Speculationaside,bothFigure1andFigure2illustratethemodel’sfundamentalintuition:Ahighinitialvalueofcontinuingdescendsintoarangewhereitmaybeexceededbystochasticcompetenceshocksasagovernmentages.OPTIMALMANIPULATIONAsthemodeliscurrentlyconstituted,reelection-mindedgovernmentsshouldsimplyincreaseeconomicmanipulationuntiltheirprobabilityofreelectionap-proachesunity,engenderinganeconomicallydestruc-tiveviciouscycleasexpectationsalsoadjustupward.Asvotersdonotwitnessthedistortionaryeffectsoftilaftertheelection,governmentsshouldconsistentlyinatetheeconomybeyondexpectationstoensurevictory.However,agovernmentthatcaresaboutvoterwel-fare,or,morecynically,itsownreputationandfuturereelectionbids,willavoidthiscornersolutionbytrad-ingoffreelectionprobabilityandvoterwelfare.Pre-electionmanipulationinducespostelectionpenanceintheformof(1)lowergovernmentrevenue,(2)lowerpublicgoodsprovisionas),and(3)negativedistortionaryeffects,),suchthatInshort,governmentalsinsofthepastrevisitthevot-ers’presentasthedistortionaryeffectsofpreviousperiodmanipulationandlowerlevelsofpublicgoodsasthegovernmentrepayselectionperiodobligations.Incumbentsthusmaximizeacombinationofreelectionconditionalutilityandvoterwelfareinchoosingtheoptimallevelofeconomicmanipulation, WhoSurfs,WhoManipulates?February2005 or,equivalently, 2+
RstŠst stŠµtŠ1stŠst y(1ŠT)ŠVt+1(st)+(µt+µt+1)(TyŠst)],(12)which,takingexpectations,1,andretainingonlynecessarysubscripts,yields s=µtŠ1 s+s+µtŠ1s+µtŠ1s Multiplyingthroughbytheinverseof)andsetting(inequilibrium)providestheFOC: (s)=µt+1 Withalittlehelpfromtheimplicitfunctiontheorem,onecannowseethatoptimalmanipulationisdecreas-ingincompetence: µtT+s)2 Governments,concernedwiththeirlegaciesandcon-stituents’welfare,arenotwillingtoinducetoosevereapostelectioneconomicdistortioninreturnformarginalincreaseinreelectionprobability.Whilestrongincen-tivesexisttomanipulatetheeconomyforelectoralgaininelectionperiods,thedegreeofsuchinterventionistemperedbyconcernsabouteconomicrepercussions.COMPARATIVESTATICSSowhatdoesthismeanforelectiontiming?Thecom-parativestaticsofthismodelbearstrongimplicationsforoptimalelectiontiming.Understandingtheeffect,,µ,andhas—–asIshowbelow—implicationsforoptimaltimingand,consequently,in-stitutionaldesign.Wewillalsoseethatopportunistictiming,inturn,hasitsownimplications,especiallyinrelationtomanipulation.Asingleobservation,posedasapropositionbelow,allowsforrelativelysimplecomparativestaticsonelec-tiontiming.Proposition1.LetEbetheexpectedperiodinwhichelectionsarecalledunderoptimalelectiontiming.ThenEisstrictlydecreasinginE Solvingexpression(7)foryieldsthecriticalvalue,istheareainthe,thatis,,+(1/2)Šµcrit]or,moreexplicitly, 2+[sŠs sŠµtŠ1sŠs ].PerssonandTabellini(2000)offeranexcellentexplanationofsimilarprobabilisticvotingtechniques.Theexpectationstermrepresentsexpectedwelfareintheperiodfollow-ingtheelectionperiod.Iassumethatrevenues,,borrowedintheelectionperiodmustberepaid,therebyreducingtheprovisionofpublicservicesinthesucceedingperiod,GivenEq.(8),anyincreasein),thecentralcom-ponentincalling,yieldsagreaterincreaseintheprimarycomponentincontinuing,inallperiodspriorto1.Theexpectedvalueofplayinganop-timalstrategyovermultiplefutureperiodsisgreaterthanthe(expected)valueofthesingledrawinthecurrentperiod,soanyincreaseinsingledrawisampliedin).Thisimpliesthatthevalueofcontinuingincreasesin)atafasterratethanthevalueofelectioncalling,yetascallingisimmediateandcontinuingdeferred,onlythelatteristimediscounted.Aslongasagovernmentdiscountsthefutureataratesufcienttooffsettherateatwhich)surpasses),thenGiventhat1obtains,itisnowpossibletodrawanumberofconclusionsabouttheeffectofseveralsocietalandinstitutionalfeaturesonelectiontimingandmanipulation.Proposition1impliesthatopportunistictimingincreases(i.e.,decreases)in.Governmentscannotinuenceshort-termeco-nomicconditions(shortofdistortionarymanipula-tion)buttheyareabletocallelectionstocorrespondwithopportunecircumstances.Timingelectionstocorrespondtoperiodsofbettereconomicperfor-mance(greater)increasespublicgoodsprovision,theelectorate’sestimateofgovernmentcompe-tence,andtheincumbent’scurrentperiodreelectionprobability.Opportunisticelectionsshouldthere-forebemorelikelyinperiodsofexceptionallystrongeconomicperformance..Themoreagovernmentdiscountsthefuture,the,thevalueofcontinuing,isreducedrelative,calling.Thissuggeststhatminoritygovern-mentsorgovernmentswithnarrowmajoritiesorlowpartydiscipline—–thatis,anygovernmentwithlowexpectationsofstayinginpowerforlong—bemoreinclinedtoopportunistictiming..Longermaximumtermspostponetheexpectedelectionperiod,,butincreaseopportunisticelec-tioncallingbydecreasing.Theprioroccursbecausetheremainingtermsacricedbyearlyelec-tionsincreases;thelatteroccursbecauselongertermsraisethevalueofagiventerm,therebyin-vokingproposition1.Opportunisticelectioncallingshouldthereforebemorecommoninstateswithlongermaximumterms. Theconditionthat1isveryreasonable.Forexample,evenneartheendofatermwhere)isgreatestandtheexponentialdiscountingslightest,given 4,3 andthreeremainingperiodspriortomandatoryelections,anygreaterthan.02sufcestoensurethatanincreasein)willraisetheutilityofcallingmorethancontinuing.Inthepreced-ingperiod,4,thediscountthresholdfallsto.018,andbytherstperiodofa60-periodterm(thinkofmonthlyperiodsinave-yearterm),anyvalueofgreaterthan.002willensurethat1andtherefore0.Thus,anythingthatraises)yieldsmoreopportunisticelectioncallingandearlierelectionsif1andreducesoppor-tunisticelectioncallingandextendsexpectedgovernmenttenureif AmericanPoliticalScienceReviewVol.99,No.1 Butopportunistictimingdecreasesin.Thegreaterthevalueofofce-holding,thelessopportunisticthegovernment.Anincreaseinthevalueofofce-holding(fewerchecksonpower,weakeropposition,etc.)isreducedbyreelectionuncertaintyinthecallingfunctioncommontobutisunmodiedintheremainderofthepresenttermuniqueto.Thehigherthedensityof,thelowerthevari-anceof,thelowertheprobabilityofadrawofsufcientlyabove)fortoexceed.Stateddifferently,thefrequencyofopportunisticelectionsispositivelyassociatedwiththevarianceofeco-nomicperformance.Opportunisticelectiontimingshouldthereforebemorecommonincountrieswithvolatileeconomicperformance,suchasdevelopingcountriesanddevelopedcountrieswithsectorallyconcentratedeconomies.Similarly,Eq.(14)showsthatmanipulation,,in-creasesin.Agreatervalueofofce-holdingincreasestheoptimumlevelofreelectionmotivatedinter-ventionintheeconomy.Strongergovernmentsshouldthereforebeexpectedtomanipulatetheireconomiesmore,notjustbecausetheycan,butbe-causetheyhaveastrongerincentivetodoso..Longermaximumtermlengthsincreasethevalueofofce-holdingandthelevelofelection-motivatedeconomicmanipulation.Allelseequal(especiallyelectiontiming),economicmanipulationshouldbegreaterinstateswithlongermaximumterms.Sincestateswithlongertermlengthsarealsomorelikelytoreducemanipulationbycallingopportunisticelections,empiricaltestingrequirescarefulselec-.Ahigherdensityinthedistributionofexogenouseconomicshocksimpliesmoremanipulation.Imag-ineaninnitedensitysothateverydrawof);thenonlymanipulationwouldremainasameansofincreasingperceivedcompetence.Manip-ulationshouldthereforebegreaterinstateswithstableeconomiesthat,byvaryinglittle,offerfewbenetsfromopportunistictiming.Butmanipulationdecreasesin.Positiveexogenouseconomicshockssubstituteforeconomicmanipulation,loweringtheneedfordistortionarymanipulationbyraisingthereelectionprobability.Electionsheldinperiodsofexceptionaleconomicperformanceshouldthereforebeassoci-atedwithlesseconomicmanipulation.Finally,itisnowpossibletoseethatgreatereco-nomicshocksincreaseopportunistictimingbut(recall-ingEq.[15])decreasemanipulation.Thus,surngandmanipulationareinverselyrelatedin.Opportunistictimingeffectivelymakes,anexogenousrandomvari-able,achoicevariable.Institutionalarrangementsthatenableopportunistictimingenableahigherelectionandconsequentlyreducemanipulation.Withtheexceptionofthreeelectionscalledintherst19monthsinofcebygovernmentshopingtoimproveweakparliamentarypositions,noneofthe16BritishgeneralelectionssincetheSecondWorldWarhavebeencalledbefore40months,two-thirdsofthemax-imumve-yearterm(Butler1995[1989];Keesing’s).Yetonlytwogovernments(underAlecDouglas-Homein1964andJohnMajorin1997)haveruntheiren-tireterm.NewZealandexhibitsastrikinglydifferentpattern:There,itisextremelyrareforaparliamentnottorunitsentirethree-yearcourse,althoughop-portunisticelectionsareclearlyallowed.SinceWorldWarTwo,only3of20NewZealandparliamentshavebeendissolvedearly(byHollandin1951,Muldoonin1984,andClarkin2002).InGreatBritainpoliti-calbusinesscyclesinmacroeconomicaggregatesarelargelyabsent;inNewZealandexceptionaleconomicgrowthpriortoelectionshasbeenhighlyapparent(cf.Alesina,Roubini,andCohen1997).Thecompar-ativestaticsderivedinthispaperbroadlycorrespondtotheseempiricalpatterns.Stateswithlongermaxi-mumterms,suchastheUnitedKingdom,shouldex-periencemorefrequentopportunisticelectioncallingthantheirshorter-termpeers,suchasNewZealand.Itthenfollowsthatcountrieswithmoreopportunisticallytimedelections,suchastheUnitedKingdom,shouldalsohavelessevidentpoliticalbusinesscyclesthanlessopportunisticcountrieslikeNewZealand.Thispaperprovidessometentativeexplanationsfortheseobservationsandsuggeststheexistenceofadditionalregularities.Morespecically,Ihavefoundthatsurngincreasesinexogenouseconomicperformance,infutureelec-toraluncertainty,inthematurityofaparliament,inthemaximumlengthofatermandinthevarianceofeconomicperformancebutdecreasesinthevalueofofce-holding.Manipulationincreasesinthevalueofofce-holdingandinthemaximumlengthofatermbutdecreasesineconomicperformanceandinthevari-anceofshockstovoterwelfare.Importantly,bettereconomicperformanceincreasesopportunistictimingbutdiminisheselection-motivatedeconomicmanipula-tion,implyinganinverserelationshipbetweensurngandmanipulation.Earlierworkhashighlightedtheendogeneitybiasinherentinempiricaltestsforpoliticalbusiness(andbudget)cycleswhenopportunisticelectioncallingispermittedbutnotmodeled(e.g.,HeckelmanandBerument1998).Yetthisistherstpapertodelineatetherelationshipbetweensurngandmanipulation.Thispaperexplainswhy,where,andtowhatmagni-tudeopportunistictimingshouldaffectmanipulation.Wheregovernmentshavetheleastinuenceovertheirdomesticeconomyandeconomicperformanceisvolatile,endogenouselectiontimingmayreducethemanipulationthemost.Thetemperingeffectofsurngonmanipulationmaybegreatestintrade-exposedde-velopingcountrieswithvolatiletermsoftrade.Mexico,forexample,mighthaveavoideditsnotoriouscycleofpreelectioncurrencyovervaluationandpostelection WhoSurfs,WhoManipulates?February2005 devaluationifitpermittedendogenouselectiontim-ing.Canada,wheresomenewspapersandthecurrentoppositionarecallingonthefederalgovernmenttofollowBritishColumbiaandOntario’srecentadoptionofxedelectiontiming,mightbewisetoconsidertheimplicationsforpreelectionmacroeconomicmanage-Theimplicationsforelectiontimingpersearenolessimportantthanthoseforthepoliticalbusinesscy-cle.Themodelsuggeststhatmostsingle-partygovern-mentsshouldbehighlyopportunisticincallingelec-tions.Governmentsabletocoordinateelectionswithanexpandingeconomyshouldenjoygreaterincum-bencyadvantagethantheirxed-timingpeers.More-over,constraintsonpossibletoolsofmonetaryandscalmanipulation—–forexample,theEuropeanEco-nomicandMonetaryUnion,increasingcapitalmobil-ity,theEuropeanUnion’sgrowthandstabilitypact,orsimplyagrowingshareofexportsinGDP—onlyincreasethisdifference.Finally,thispaperprovidesacentralroleforstrategicdecisionmakinginresearchongovernmenttermina-tion.Traditionalempiricalresearchongovernmentter-mination,constrainedbythelimitsofdiscretetimeandhazardmodels,haslargelyignoredstrategicbehavior(e.g.,Warwick1994).Formalworkhasincorporatedstrategicbehaviorbuthasbeeneithernumeric(e.g.,Balke1990)ornotfullydynamic(e.g.,LupiaandStrom1995).Thepresentdynamicstochasticmodel,togetherwithadvancesintheestimationofdynamicstochasticdiscretechoicemodels(e.g.,KeaneandWolpin1994),shouldoffernewopportunitiesforpredictingandesti-matingtheeffectsofvariouspolitical,institutional,andeconomicarrangementsonsurngandmanipulation.Achen,ChristopherH.,andLarryM.Bartels.2002.“BlindRet-rospection:ElectoralResponsestoDrought,Flu,andSharkAt-tacks.”PaperpresentedattheAnnualMeetingoftheAmericanPoliticalScienceAssociation,Boston.Alesina,Alberto,GeraldCohen,andNourielRoubini.1992.“MacroeconomicPolicyandElectionsinOECDDemocra-cies.”InPoliticalEconomy,Growth,andBusinessCycles,ed.,A.Cukierman,Z.Hercowitz,andL.Leiderman.Cambridge,MA:MITPress.Alesina,Alberto,andNourielRoubiniwithGeraldCohen.1997.PoliticalCyclesandtheMacroeconomy.Cambridge,MA:MITPress.Balke,NathanS.1990.“TheRationalTimingofParliamentaryElec-tions.”PublicChoice65(June):201–16.Butler,David.[1989]1995.BritishGeneralElectionsSince1945Oxford:Blackwell.Butler,David,andDennisKavanagh.TheBritishGeneralElectionsof1983.NewYork:St.Martin’sPress.(AlsoseeButlervolumesonpreviousandsubsequentBritishgeneralelections.)Cargill,ThomasF.,andMichaelM.Hutchinson.1991.“PoliticalBusinessCycleswithEndogenousElectionTiming:EvidencefromJapan.”ReviewofEconomicsandStatistics73(November):733–Chappell,D.,andD.A.Peel.1979.“OnthePoliticalTheoryoftheBusinessCycle.”EconomicsLetters2(4):327–32.Chowdhurry,AbdurR.1993.“PoliticalSurngoverEconomicWaves:ParliamentaryElectionTiminginIndia.”AmericanJour-nalofPoliticalScience37(November):1100–18.Clark,WilliamR.,andMarkHallerberg.2000.“MobileCapital,DomesticInstitutions,andElectorallyInducedMonetaryandFiscalPolicy.”AmericanPoliticalScienceReview94(June):Crossman,Richard.1979.TheCrossmanDiaries:SelectionsfromtheDiariesofaCabinetMinister1964Ð1970,ed.AnthonyHoward,London:MagnumBooks.Cukierman,Alex,andAllanMeltzer.1986.“APositiveTheoryofDiscretionaryPolicy,theCostofDemocraticGovernment,andtheBenetsofaConstitution.”EconomicInquiry24(July):367–Dewatripont,M.,I.Jewitt,andJ.Tirole.1999a.“TheEconomicsofCareerConcerns.PartI:ComparingInformationStructures.”ReviewofEconomicStudies66(January):183–98.Dewatripont,M.,I.Jewitt,andJ.Tirole.1999b.“TheEconomicsofCareerConcerns.PartII:ApplicationtoMissionsandAccount-abilityofGovernmentAgencies.”ReviewofEconomicStudies(January):199–217.Diermeier,Daniel,andRandolphT.Stevenson.2000.“CabinetTer-minationsandCriticalEvents.”AmericanPoliticalScienceReview94(September):627–40.Drazen,Allan.2000.PoliticalEconomyinMacroeconomics.ton,NJ:PrincetonUniversityPress.Eckstein,Zvi,andKennethI.Wolpin.1989.“TheSpecicationandEstimationofDynamicStochasticDiscreteChoiceModels:ASur-vey.”JournalofHumanResources24(4):562–98.Gonzalez,MariadelosAngeles.1999.OnElections,DemocracyandMacroeconomicPolicy.Dissertation.PrincetonUniversity.Grofman,Bernard,andPeterVanRoozendaal.1994.“TowardaTheoreticalExplanationofPrematureCabinetTermination.”EuropeanJournalofPoliticalResearch26(2):155–70.Grofman,Bernard,andPeterVanRoozendaal.1997.“ModellingCabinetDurabilityandTermination.”BritishJournalofPolitical27(3):419–51.Hansen,SusanB.1999.“LifeIsNotFair:Governors’JobPerfor-manceRatingsandStateEconomies.”PoliticalResearchQuarterly52(March):167–88.Heckelman,J.C.,andHakanBerument.1998.“PoliticalBusinessCyclesandEndogenousElections.”SouthernEconomicJournal64(April):987–1000.om,B.1982.“ManagerialIncentiveProblems—–ADynamicPerspective.”Reprintedin1999inReviewofEconomicStudies(January):169–82.Inoguchi,Takashi.1979.“PoliticalSurngoverEconomicWaves:ASimpleModeloftheJapanesePoliticalEconomicSystem.”Unpublishedmanuscript.Ito,Takatoshi.1990.“TheTimingofElectionsandPoliticalBusinessCyclesinJapan.”JournalofAsianEconomics1(1):135–56.Ito,Takatoshi,andJinHyukPark.1988.“PoliticalBusinessCy-clesintheParliamentarySystem.”EconomicsLetters27(3):Keane,MichaelP.,andKennethI.Wolpin.1994.“TheSolutionandEstimationofDiscrete-ChoiceDynamic-ProgrammingModelsbySimulationandInterpolation:MonteCarloEvidence.”ReviewofEconomicsandStatistics76(4):648–72.Kreps,DavidM.1990.ACourseinMicroeconomicTheory.ton,NJ:PrincetonUniversityPress.Kydland,FinnE.,andEdwardC.Prescott.1980.“ACompetitiveTheoryofFluctuationsandtheFeasibilityofStabilizationPolicy.”RationalExpectationsandEconomicPolicy,ed.StanleyFisher.Chicago:UniversityofChicagoPress.Kydland,FinnE.,andEdwardC.Prescott.1982.“TimetoBuildandAggregateFluctuations.”50(6):1345–70.Lewis-Beck,Michael.[1988]1990.EconomicsandElections:TheMajorWesternDemocracies.AnnArbor:UniversityofMichiganPress.Lewis-Beck,Michael,S.,andMaryStegmaier.2000.“EconomicDe-terminantsofElectoralOutcomes.”AnnualReviewofPolitical3:183–219.Litterman,RobertB.,andLaurenceWeiss.1989.“Money,RealInter-estRates,andOutput:AReinterpretationofPostwarU.S.Data.”53(1):129–56.Lucas,RobertE.,Jr.1972.“ExpectationsandtheNeutralityofMoney.”JournalofEconomicTheory4(2):103–24.Lupia,Arthur,andKaareStrom.1995.“CoalitionTerminationandtheStrategicTimingofElections.”AmericanPoliticalScience89(3):648–65. AmericanPoliticalScienceReviewVol.99,No.1 McCallum,BennettT.1999.“RecentDevelopmentsinMonetaryPolicyAnalysis:TheRolesofTheoryandEvidence.”NBERWork-ingPaper7088.Nordhaus,WilliamD.1975.“ThePoliticalBusinessCycle.”ofEconomicStudies42(2):169–90.Palmer,HarveyD.,andGuyD.Whitten.2000.“GovernmentCompetence,EconomicPerformanceandEndogenousElectionDates.”ElectoralStudies19(June):413–26.Peltzman,Sam.1992.“VotersasFiscalConservatives.”JournalofEconomics107(May):327–61.Persson,Torsten,andGuidoTabellini.1990.MacroeconomicPolicy,CredibilityandPolitics.London:HarwoodAcademy,78–98.Persson,Torsten,andGuidoTabellini.2000.PoliticalEconomics:ExplainingEconomicPolicy.Cambridge,MA:MITPress.Reid,B.G.1998.“EndogenousElections,ElectoralBudgetCyclesandCanadianProvincialElections.”PublicChoice97(1–2):35–48.Rogoff,Kenneth.1990.“EquilibriumPoliticalBudgetCycles.”Amer-icanEconomicReview80(1):21–36.Rogoff,Kenneth,andAnneSibert.1988.“ElectionsandMacroeco-nomicPolicyCycles.”ReviewofEconomicStudies55(1):1–16.Roper,StevenD.,andChristopherAndrews.2002.“TiminganElec-tion:TheImpactonthePartyinGovernment.”AmericanReviewofPolitics23(Winter):305–18.Schuknecht,Ludger.1996.“PoliticalBusinessCyclesandFiscalDis-ciplineandFiscalPoliciesinDevelopingCountries.”Kyklos49(2):Smith,Alastair.1996.“EndogenousElectionTiminginMajoritarianParliamentarySystems.”EconomicsandPolitics8(2):85–110.Smith,Alastair.2003.“ElectionTiminginMajoritarianParliaments.”BritishJournalofPoliticalScience33(3):397–418.Sundaram,RangarajanK.[1996]1999.AFirstCourseinOptimiza-tionTheory.NewYork:CambridgeUniversityPress.Suzuki,Motoshi.1992.“PoliticalBusinessCyclesinthePublicMind.”AmericanPoliticalScienceReview86(December):989–Terrones,MarcoEmilio.1989.“MacroeconomicPolicyCyclesUnderAlternativeElectoralSystems.”ResearchReport8905.Depart-mentofEconomics,UniversityofWesternOntario.Warwick,Paul.1994.GovernmentSurvivalinParliamentaryDemoc-racies.NewYork:CambridgeUniversityPress.