A decade of dissent explaining the dissent voting

《A decade of dissent explaining the dissent voting》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。

PublicChoiceDOI10.1007/s11127-010-9597-6Adecadeofdissent:explainingthedissentvotingbehaviorofBankofEnglandMPCmembersMarkN.Harris·PaulLevine·ChristopherSpencerReceived:4June2008/Accepted:7January2010©SpringerScience+BusinessMedia,LLC2010AbstractWeexaminethedissentvotingrecordoftheBankofEnglandMonetaryPol-icyCommittee.ContrarytofindingsintheFOMCliterature(forexampleHavrileskyandSchweitzerinThePoliticalEconomyofAmericanMonetaryPolicy,pp.197–210,1990;Chappelletal.inQ.J.Econ.108(1):185–218,1993),theeffectsofmembers’careerback-groundsandthepoliticalchannelofappointmentonvotingbehaviorarenegligible,re-flectingthedistinctinstitutionalconstraintsandincentivesassociatedwithUKmonetarypolicy.Ourfindingsalsosuggestthatliteraturewhichcharacterizesvotingbehaviorasbe-ingpredominantlydeterminedbymembers’internalorexternalstatusisoverlysimplistic.Thisviewissupportedbyeconometricresultsappertainingtotheintroductionofmember-specificfixed-effects,whichaccountforpossibleunobservedheterogeneity.KeywordsBankofEngland·MonetaryPolicyCommittee·Careerbackgroundeffects·Dissentvoting·Appointmentchannels·Unobservedheterogeneity1IntroductionAneglectedaspectinthegrowingliteratureoninterestratesettingbytheBankofEng-landMonetaryPolicyCommittee(MPC),istheeffectofpoliticalinfluenceandconstituentM.N.HarrisDepartmentofEconometricsandBusinessStatistics,MonashUniversity,Clayton,Victoria3800,Australiae-mail:mark.harris@buseco.monash.edu.auP.LevineDepartmentofEconomics,UniversityofSurrey,Guildford,GU27XJ,UKe-mail:p.levine@surrey.ac.ukC.Spencer()DepartmentofEconomics,LoughboroughUniversity,Leicestershire,LE113TU,UKe-mail:c.g.spencer@lboro.ac.uk PublicChoicegroupsonvotingbehavior,arisingeitherthroughtheappointmentsprocedureorasacon-sequenceofmembers’careerbackgrounds.ThisisincontrasttotheliteratureonFed-eralOpenMarketCommittee(FOMC)decisionmaking,wherethepolicychoicesofitsmembers,includingdissents,arenotonlyrigorouslymodeledasafunctionofthesechar-acteristics,butfoundtobeassociatedwithsuchfactorsineconometricestimationandstatisticalanalysis(seeamongstothers,Gildea1990;HavrileskyandSchweitzer1990;HavrileskyandGildea1992;Chappelletal.1993;KeechandMorris1997;Krause1994,1996;ChappellandMcGregor2000,andmorerecently,Adolph2005;Chappelletal.2005;McGregor2007).1Thefactthatsucheffectshavehithertoremaineduninvestigatedissur-prising,particularlywhenviewedagainstthebackdropofmajorchangestotheUKmonetarypolicyframework,whichwereintroducedinMay1997bythenewlyelectedLabourParty.Thesechangesnotonlywitnessedtheestablishmentofthenine-memberMPC,butcoin-cidedwithashifttoso-calledoperationalindependencewherebyinterest-ratesettingwasnolongerunderthedirectcontrolofpoliticians.Asthislegislativeinnovationwasclearlyintroducedwithaviewtoimprovingmacroeconomicoutcomesbyisolatingmonetarypol-icydecisionsfrompoliticalpressure,thequestionofwhethervotesonthepolicyratemaybesubjecttopoliticalorconstituentgroupinfluenceisnotaninsignificantone.2Inaimingtoassesstheeffectsoutlinedabove,thispaperextendstheexistingliteratureonMPCdecisionmakinginanumberofways.Usingorderedprobitanalysis,wetestthehypothesisthatthepoliticalandnon-politicalchannelsthroughwhichMPCmembersareappointedaffectthedecisiontodissent,asdodifferencesinmembers’careerbackgrounds:specifically,theimpactofthesecharacteristicsondissentingisestimatedaspartofawiderestimationstrategywhichalsocontrolsforthecorrespondingimpactofamember’stype(i.e.,internalversusexternalmember),theMPC’sinflationandoutputprojections,anddif-ferentchairmen.Ouranalysisaddstothestrandofpartisanshipliterature,asexemplifiedby(Chappelletal.1993),thatinvestigatestheroleforpoliticalinfluenceinmonetarypoli-cymakingthroughtheappointmentschannel.WealsobuildonrelatedworkbyHavrileskyandSchweitzer(1990),whopredictthatdifferentcareerbackgroundsinfluenceanFOMCmember’spropensitytodissentonthesideofmonetaryeaseortightness.Here,theap-pointmentofmemberswithparticularcareerbackgroundscanbeenvisagedasmeetingtherepresentationaldemandsofconstituentsororganizedinterests,whomayhavespecificpol-icypreferences.Inthecontextofthispaperthesebackgroundsarereadilyquantifiable,andcorrespondtoyearsspentworkinginthefollowingareasororganizations:academia;theBankofEngland;bankingandfinance;government;industry;andnon-governmentalorga-nizations.Unobservedheterogeneityintheformofmember-specificfixed-effectsisalsointroducedintotheestimationstrategy.ThisinnovationismotivatedbytheincreasinglyacceptedviewthattheBankofEnglandMPCis‘individualistic’(Blinder2007)—thatis,comparedto1ThereasonsastowhyFOMCvotingbehavior,andbyimplicationmonetarypolicydecisionsattheFed,mightbeaffectedviathechannelsdescribedabovearewellknown.Manyexplanationsappealtotheeco-nomictheoryofbureaucracy,drawingontheFed’ssymbioticrelationshipswithCongressandtheUSPresi-dent.2ThisisespeciallythecasewhenoneconsidershowthenuancesoftheUKpoliticalandparliamentarysystem,andtheparticularinstitutionalframeworkenjoyedbytheBankofEngland,mayshapemonetarypolicybyaffectingvotingdecisionsonthepolicyrate.Forexample,Santoni(1984)drawsonthehistoryoftheBankofEnglandfromitsestablishmentin1694until1932,illustratingthatmonetarypolicyactionsareafunctionofthesetofincentivesconfrontingpolicymakers.Asincentivesdiffernotonlyacrosstime,butacrossinstitutions,theincentivesfacingUKmonetarypolicymakersmaydiffersubstantiallyfromthosefacingtheFed. PublicChoice‘collegial’MPCs,suchastheFOMCandECBGoverningCouncil,theChairmanenjoyslessinfluenceinshapingdecisionoutcomes,andcannotsoeasilyinducemembersto‘fallintoline’.3Accordingly,weproposethatifMPCmemberssharediverseviews,andvoteinaccordancewiththem,thensuchanestimationstrategyiswarranted.4Indeed,thisstrategymaybelooselyviewedasatestoftheextenttowhichtheMPCisindividualistic,asreflectedbythemagnitudeandnumberofstatisticallysignificantfixed-effectsparameters.However,priortotheempirics,wefirstrelateourcontributiontopreviousstudiesofMPCvotingbe-havior,placingparticularemphasisontheinstitutionalandpoliticalfactorsthatcharacterizemonetarypolicymakingintheUK.2Whydissent?Theinternal-externaldistinctionandbeyondThemajorityofcontributionsthatexaminethepolicychoicesofMPCmembersaregearedtowardsexplainingthevotingpatternsofinternalandexternalcommitteemembers(seeforexampleGerlach-Kristen2003,2009;BhattacharjeeandHolly2005;Besleyetal.2008;Brooksetal.2007;HarrisandSpencer2009);and,evenwherefeaturessuchascareerback-groundsareaccountedfor(asinBesleyetal.2008andRiboniandRuge-Murcia2008),theyareintroducedinthecontextofreactionfunctionestimation.Givenitsnotorietyasatopicofscrutinyintheliterature,wefirstoutlinethenatureoftheinternal-externaldistinction,ac-countingforwhythebehaviorofbothofthesegroupsmightbeexpectedtodiffer.Yet,whilethisinstitutionalfeatureisclearlyimportantinexplainingmember’svotes,focusingalmostexclusivelyonitpotentiallyomitsotherfactorswhichmaydrivevotingbehavior.Tothisend,anumberofadditionalconsiderations—mostnotablytheanticipatedeffectsemanatingfromthechannelofappointmentandcareerbackgrounds—areexamined.2.1Theinternal-externaldistinctionAfundamentaldifferencebetweenthefiveinternalandfourexternalmemberscomprisingtheMPCisthat“internalmembershavefull-timeexecutivepositionsinthebank”whereasexternalmembers“workmostlyparttime,andhavenoexecutiveresponsibilities”(Besleyetal.2008:218).5Theinternalmembers,forexample,comprisetheGovernoroftheBank,3Toaccountforhowdifferentconsensualnormsmayinfluencevotingbehavior,thereaderisreferredtoBlinder’s(2007)typologyofmonetarypolicycommittees,whichpartlyexplainsthesignificantlygreaternumberofdissentscastbyBankofEnglandMPCmembersinrelationtoFOMCmembers.4CoincidentwiththeaboveassertionistheobservationthatnotallMPCmembersactinaccordancewithaviewprevalentintheliterature,thatcomparedtoMPCmembersappointedfromwithintheranksofbankstaff,externalMPCmemberschooselowerinterestrates(Gerlach-Kristen2003),andareevencharacterizedbylossfunctionsthataremoresympathetictodeviationsofoutputfrompotential(Gerlach-Kristen2009).Bothofthesefindingssuggestthatexternal(internal)membersmaybemorepredisposedtodissentonthesideofmonetaryease(tightness),somethingwhichinpracticeisclearlynotobservedacrossallmembers.Whilethisargumentprovidessupportforthepresenceofvoterheterogeneity—andthusafixed-effectapproach—itmoregenerallyhighlightsanareainwhichtheMPCliteratureisdeficient,namelythattheinsider-outsiderdistinctionisoverlysimplistic,andasisdiscussedlater,doesnotcaptureotherpotentialinfluencesonvotingbehaviorsuchaschannelofappointmenteffects.5TheBank’spositiononwhetherexternalMPCpositionscanbefullorpart-timeisfoundinarecentdoc-umentbytheTreasuryCommittee(acommitteecomprisedofMPsfromacross-sectionofpoliticalparties,whichhasresponsibilityonbehalfofParliament,forscrutinizingtheconductofmonetarypolicyandMPCappointmentsatregularpublichearings),whichpublishedareportentitledTheMonetaryPolicyCommitteeoftheBankofEngland:tenyearson(18thSeptember2007,HouseofCommonsPaperNo.299).Itstatesthat PublicChoicethetwoDeputyGovernorsforMonetaryPolicyandFinancialStabilityrespectively,andtwoseniormemberswithexecutiveresponsibilitiesfortheBank’soperations,oneofwhomistheBank’sChiefEconomist.Externalmembersareappointedmainlyfromacademiaandthefinancialsector.Explanationsastowhythesegroupsshouldbedistinctintheirvotingbehaviorarenumerous.AsnotedinBuiter(1999),Gerlach-Kristen(2003)andHarrisandSpencer(2009),differencesmayemergeifinternalsshareanorganizationalconsensusor‘Bankview’,thatisdistinctfromtheview(s)ofexternalmembers:inHarrisandSpencer(2009),forexample,suchadeviceispotentiallyresponsibleforthefindingthatinternalmembersaremorelikelytovoteasabloc.Moreover,asinternalsenjoyafive-fourmajorityoverexternals,thisincreasestheirchancesofbeingonthewinningsideofadecision.Relat-edly,externalsmaybeeasilyout-votedateachmeetingifrigorousresearchtobackuptheirargumentsislacking.SuchasituationmayariseifexternalsaredeniedaccesstoadequateresearchresourcesattheBank:forexample,ifexternalmembersaretoprovideanalysisquestioningthefindingsoftheBank’sstandardsuiteofmodels(orindeed,toconductpol-icyexercisesusingBankmodelsbutconditionedondifferentassumptions),resourcesneedtobeprovidedtocatertotheserequirements.AsthereisstrongevidenceofaresourcesdisputedevelopingintheearlyyearsoftheMPC,theresourceissue,anditsproposedim-plicationsforvoting,isnotmoot.Duetospaceconstraints,amoredetailedaccountofthisepisodeisexpoundedinAppendixA.6Thereare,asmightbeanticipated,additionalreasonswhythevotingbehaviorassociatedwithinternalsandexternalsmaybeexpectedtodiffer.InternalswhowishtobuildacareerwithintheBankofEnglandmayfacecareerconcerns,andthusbeincentivizedto‘fallinline’witheachother:here,disagreeingtoomuchwithone’scolleaguesmaybeperceivedbymembersasbeingdetrimentaltocareerprogressionwithintheirorganization.7Incontrast,becausetheprospectofcareerprogressionwithintheBankisnotanissueforexternals,thereexternalmemberswillbegivenpart-timeappointmentsunlessindividualcircumstances,suchasaconflictofinterest,warrantafull-timepositionbeingoffered.Forexample,Dr.SushilWadhwaniwasemployedbyaprivateinvestmentorganization,sobecameafull-timeexternalMPCmemberonthatbasis;ontheotherhand,CharlesGoodhart,aprominentUKmonetaryeconomistworkedfortheBankasapart-timeMPCmemberwhilstmaintaininguniversity-basedresponsibilities.However,eventhen,afulltimepositionshouldbealastresort:thatis,aslongasanexternalmemberisofferedanMPCpositioningoodtime,itshouldbepossibleforadditionalnon-Bankemploymenttobearranged,suchthatamembershouldnotrequireafull-timeposition.6Wearegratefultoananonymousrefereeforhighlightingthepotentialroleofresourcecontrol,giventhatitishereuntounexploredintheBankofEnglandMPCliterature.Nevertheless,thenotionofresourcecontrolasusedherereferstocontrolbyBankexecutivesoverresearchfacilitiesforexternalmembers,whomaywishtoconducttheirownresearchattheBank.ThisisdistinctfromusagereferringtohowtheFedisincentivizedtoconductpolicysoastomaximizethetotalamountresourcesunderitscontrol(ShughartandTollison1983),orhowstaffsizeandmonetarypartisanismareshapedbyaBank’sstatutoryindependence(Vaubel1997).OurexampleperhapscomesclosesttoHavrileskyandGildea(1992),whoemphasizehowFOMCchairmenexertresourcecontroloverfellowFOMCmemberssittingontheBoardofGovernors,andexploitthistoconstraintheir“independent”behavior.7MeadeandStasavage(2008)introducearoleforcareerconcerns,suggestingthatmembersofamonetarypolicycommitteearelesslikelytovoicedissentwhenthetranscriptsofmeetingsareplacedinthepublicdomain.Inrelationtotheabovearguments,theremayalsobearoleforcareerconcerns,particularlyifMPCmembersperceivealinkbetweenvotingbehaviorandreappointment.ThisprospecthasbeensuggestedbymembersoftheUKpoliticalestablishment—Howard(2002)suggeststhatifoneisseekingreappointment,thenvotingforlower,asopposedtohigherinterestratesmaysecureasecondterm.Preliminaryeconometricresults(notpresentedhere),however,donottosupportthisconjecture.Interestingly,itdoesnotseemtohaveaffectedtheappointmentofMervynKingtoGovernoroftheBank,whocastmoredissentingvotesonthesideofmonetarytightnessthananyotherMPCmemberuptothetimeofhis(re)appointment.ThisobservationisseeminglyatoddswithHoward’ssuggestion. PublicChoiceislessincentivetovoteasablocorfallintolinewithothermembers.Indeed,externalsmayevenbeincentivizedtodissenttogainmediaattention,therebyincreasingtheirpublicprofile(Gerlach-Kristen2003).82.2Politicalandnon-politicalappointmentstotheMPCWhilethenatureoftheinternal-externaldistinctionmayincentivizememberstovoteindif-ferentways,itdoesnot,byconstruction,explicitlycapturefundamentaldifferencesinthewaymembersareappointed.ThisisunliketheFOMC,whereBankPresidentsareappointedbytheirReserveBank’sBoardofDirectors,andBoardGovernorsbytheUSPresident.In-ternalmembersareappointedinoneoftwodifferentways.TheGovernorandtwoDeputyGovernorsareallCrownappointments,whichmeansthattheyareappointedbytheQueenontheadviceofthePrimeMinister.Inpracticehowever,theChancelloroftheExchequermayplayasignificantroleinselection.9Withrespecttothetworemaininginternals,al-thoughstatutesprescribethatthesemembersshouldbeappointedbytheGovernoronlyaf-tertheChancelloroftheExchequerhasbeenconsulted,inpractice,theChancellorhaslittlesayinthematter.10AllfourexternalMPCmembersareappointeddirectlybytheChancel-lor.Forthepurposesofsubsequentanalysis,werefertomembersappointedbyeithertheCrownordirectlybytheChancelloraspoliticalappointments,wheretheadjective‘politi-cal’signifiesthattheindividualischosendirectlybyamemberoftheExecutiveBranchofUKgovernment.11Incontrast,thetwointernalmemberschosenbytheGovernorareclassedasnon-politicalappointments.Wenotethatthistypeofdistinction—betweenpoliticalandnon-politicalappointees—isalsodevelopedinthecontributionsofKrause(1994,1996),whoexaminesFOMCvoting.TheobservationthattheoverwhelmingmajorityofMPCmembersarepoliticalappoint-mentsraisesanumberofissues.Ononeargument,thenotionthatgovernmentalinfluenceshapesinterestratedecisionsthroughthepoliticalappointmentschannelishighlyplausible.Cobham(2000)forinstance,arguesthattheMPCappointmentsprocedureisopaqueand8Thecentralbankindependenceliteraturealsosuggeststhatlongertermlengthsincreaseacentralbank’spoliticalindependencefromthegovernment(Grillietal.1991).ThisisofinterestinsofarasboththeGov-ernorandtwoaforementionedDeputyGovernorsareappointedforfiveyearrenewableterms,whereasallotherMPCmembersserveshorterthreeyearrenewableterms.Inenjoyinglongertermlengths,theGovernorandthetwoDeputyGovernorsmaybelesssusceptibletogovernmentalpressuretoreduceinterestrates;thisavenueofresearchisnot,however,pursuedinthisstudy.9InpreviouscorrespondencewiththeSecretaryoftheBankofEngland,itwassuggestedthattherelativeinfluenceofthePrimeMinisterandChancellorinselectingtheGovernorandtwoDeputyGovernorschangesfromadministrationtoadministration,dependingonpersonalitiesandcircumstances.Forinstance,whilenewspaperreportsfromthe1980ssuggestthatGovernorRobertLeigh-PembertonwaschosensolelybyMargaretThatcher,theChancellorissaidtohavemadethedecisioninothercases.WethankPeterRogersforsheddinglightonthisissueforus.Theseinformalaspectsoftheappointmentsprocedurearenotseeminglycapturedbythe1998BankofEnglandAct,anddifferfromthecaseoftheFed,wheretheUSPresidentenjoystotalautonomyinappointingtheChairman.10WethankCharlesGoodhartforclarifyingthismatter.11ThisofcourseassumeseitherthattheChancellorandthePrimeMinistersharecommonpreferencesre-gardingwhomtoappoint.AsthereisnoevidencetosuggestthatPrimeMinisterBlairandChancellorBrownwereatloggerheadsoverBankappointments,theaboveassumptionisnotentirelyunreasonable. PublicChoiceopentopoliticalopportunism.12Similarly,anumberofwitnessesatTreasuryCommittee13hearings,includingex-MPCmembers,haveindicatedthatwhiletheydonotbelievethattheChancellorbehavedinanopportunisticmannerwhenchoosingexternals,theinstitutionalframeworkdoesnotprecludefutureChancellorsfromabusingtheirposition(TreasurySe-lectCommittee2007a:33–35).Relatedly,thefactthatpoliticalappointmentstotheMPCaresubjecttofewerlegislativeconstraintsthanUSpresidentialappointmentstotheFOMC,14maymakeitrelativelyeasierfortheExecutiveBranchofUKgovernmentto‘pack’theMPCwith“eminentlyreliableagentswhoseviewsalignwiththeirown”(HavrileskyandGildea1992:398).Thislineofreasoningsuggeststhatpoliticalappointeesmightexhibitstrongpartisanbehavior,orvoteinawaywhichreflectsanytime-consistentinflationarybiasonthepartofthegovernment.However,asdiscussednext,therearecompellingargumentstosuggestthatchannelofappointmenteffectsmayinfactbesmall—especiallywhenoneconsidershowtheframeworkforUKmonetarypolicymitigatestheso-named“democraticdeficit”associatedwithmonetarypolicymaking.2.2.1ThedemocraticdeficitofmonetarypolicyandtheinflationtargetAsnotedbyTucker(2008),priortooperationaldecisionsonmonetarypolicybeingdele-gatedtotheMPCinMay1997,theUKenjoyednohistoryofnon-electedofficialsassumingresponsibilityformajorpolicydecisions:monetarypolicydidnotsufferfroma‘democraticdeficit’,asitwasundertheaegisofelectedpoliticians,whowereaccountablefortheirac-tionsontheflooroftheHouseofCommons.Goodhart(2006)proposesthatthesettingofaninflationtargetbytheexecutivebranchofUKgovernmenteffectivelymitigatesthedemocraticdeficitthatarosefollowingtheshifttooperationalindependence.Thisisun-liketheUnitedStates,wherethecorrespondingdeficit“ismitigatedbythepoliticisationoftheappointmentsprocess”(GoodhartandMeade2003).Onthisargument,politicalap-pointmentstotheMPCmightnotbeexpectedtoexhibitbehaviorconsistentwithnotionsofpartisanship,astheprocessofselectionitselfisconsiderablylesspoliticized.Accord-ingly,weproposethatpoliticalandconstituencygrouppressureismorelikelytoinfluencemonetarypolicythroughthechoiceoftheinflationtargetitself,asopposedtothroughthechoiceofindividualsselectedtosetpolicy.ThisconjectureisbroadlyconsistentwiththewellknownviewofMcCallum(1995),whoarguesthatbyretaininggoalindependence,thegovernmenteffectivelycontrolsmonetarypolicythroughitspoweroverthegoal:inthecaseofaninflationtargetinginstitution,thisamountstocontrolovertheinflationtargetitself(DuPlessis2005).Indeed,ashasbeenconsistentlyreiteratedbytheChancellor,theUKinflation12InoneTreasurySelectCommitteehearing,RichardLambert(anexternalmemberwhojoinedtheMPCinMay2003)wasaskedwhether‘acoupleofcallstoJapanwasalltherewastobecomingamemberoftheMPC.’(OralevidenceofMr.RichardLambert,takenbeforetheTreasuryCommitteeonMonday16June2003).13TheTreasurySelectCommitteehasnopowertovetoappointmentstotheMPC,evenifitdoesnotapproveofthem.14WhereasintheUS,SenateapprovalisrequiredforappointmentstotheBoardofGovernorsandthepositionofFedChairmanandDeputyChairman,politicalappointmentstotheMPCarenotsubjecttosimilarParlia-mentaryapproval:thehandsoftheChancellorandPrimeMinisterarethussignificantlylessconstrainedwithrespecttothechoiceofappointees.ThisfacetisimportantinsofaraseventhoughtheinstitutionalprovisionrequiringSenateapprovalfornomineesimpedestheabilityoftheUSPresidentfor‘packingtheFed’(Morris2002:82),thereisstrongevidencethatalargenumberpresidentialappointeesbehaveinpartisanfashion.Therefore,onemightinferthatifaUSpresidentboundbyCongressionalconstraintscanstillappointparti-sans,arelativelyunrestrainedUKExecutiveBranchcaneasily‘packtheMPC’. PublicChoicetargetisnotcastinstone,andcanbechangeditifitisdeemedexpedienttodoso:15since1997,thetargethasbeenchangedoncefrom2.5%RetailPricesIndexexcludingMortgageInterestPayments(RPIX)inflationto2%ConsumerPricesIndex(CPI)inflation.16Buildingfurtherontheabovearguments,weproposethatwhileinpracticeitmaybeeasytoappointpartisanstotheMPC,itwouldatthesametimemakelittlesensefortheExecutiveBranchtoappoint(i)memberslackingtherequiredeconomicexpertisetohittheprimaryobjectiveofUKmonetarypolicy—anexplicitlydefinedandtransparentinflationtargetof2%CPIinflation—or(ii)individualswhowouldsystematicallyunderminetheGovernment’sowninflationtargetingframework,throughpursuing,forexample,shorttermoutputgainatthecostofsignificantlyovershootingthetargetitself.Thisisattributabletothefactthattheinflationtarget‘is2percentatalltimes’andrepresents‘therateatwhichtheMPCisrequiredtoachieveandforwhichitisaccountable’.17TheabilitytohittheinflationtargetthusrepresentsthebenchmarkagainstwhichtheperformanceoftheMPCanditsmembers—whoareallindividuallyaccountablefortheirdecisionstoParliament—ismeasured.18Inthissense,thechoiceof,andbehaviorofpoliticalappointeesshouldnotonlybeperceivedasreflectingthe(veryfew)politicalconstraintsonselection,butthenatureoftheinflationtargetingframeworkintowhichMPCmembersareappointed.2.3CareerbackgroundeffectsThenotionthatcareerbackgroundsplayaroleindrivingvotingbehavioriswellexploredintheFOMCliterature.Forinstance,Chappelletal.(1995)findthatpartisanbehaviorispartiallyattributabletothecareerbackgroundsofFOMCmembers,notingthat“experienceingovernment,particularlyattheFederalReserveBoard,isassociatedwithsignificantlystrongerpreferencesformonetaryease”(Chappelletal.1995:130).SimilarfindingsarealsouncoveredinHavrileskyandSchweitzer(1990),Gildea(1990)andAdolph(2005).OfparticularrelevancetothisstudyistheworkofHavrileskyandSchweitzer(1990),who15“TheinflationtargetwillbeconfirmedineachBudget.ThereisavalueincontinuityandIwillhaveproperregardtothat.ButIwillalsoneedtoconsiderthecaseforarevisedtargetatthesetimesonitsmer-its.”RemitfortheMonetaryPolicyCommittee,letterfromtheRt.Hon.GordonBrowntoEddieGeorge,12thJune1997.Availableathttp://www.bankofengland.co.uk/monetarypolicy/pdf/chancellorletter970506.pdf,ac-cessed15June2009.16MovingtoCPIinflationmayprovideamoreobjectiveyardstickagainstwhichpossiblefutureUKentryintotheEurozonemightbejudged—thisisbecausetheCPImeasureissimilartotheHICPmeasureofinflationtargetedbytheEuropeanCentralBank.Thedecisiontoswitchmightthusbeviewedasbeingbotheconomicallyandpoliticallyexpedient.WefurthernotethatwhiletheprimaryobjectiveofUKmonetaryassumestheformofasymmetricinflationtarget,thereisnoreasonwhytheChancellorcannot,shouldhewishtodoso,re-define‘pricestability’asarangeofprices,oreveninqualitative(asopposedtoexplicitlyquantitative)terms.ForapublicchoiceorienteddiscussionofthedemocraticdeficitinthecontextoftheEuropeanCentralBank,theinterestedreaderisreferredtoMueller(2005).17RemitfortheMonetaryPolicyCommittee,letterfromtheRt.Hon.GordonBrowntoMervynKing,21stMarch2007.Availableat:http://www.bankofengland.co.uk/monetarypolicy/pdf/chancellorletter080311.pdf,accessed5thMay2009.18ThisisunliketheFed,whose(dual)targetsareonlyqualitatively,andnotquantitativelydefined(Belongia2007),renderingmonetarypolicyfarlessunderstoodbythepublic.Buildingonthispoint,Buiter(2008)suggeststhatbureaucraticconsiderationsandCongressionalconstraintsmaypreventtheFedfromannounc-inganexplicitinflationtarget,eventhoughitisinprinciple,abletodoto.Thisisbecausetheopaqueandnon-quantifiedobjectivesoftheFedbenefitsCongress,thewhimsofwhomtheFedhastobendtoinordertomaintainitsautonomy. PublicChoiceintroduceageneraltheoreticalframeworkfordissentvoting.19Whilewedonotformallypresenttheirmodelinthissection(seeAppendixBforafullertreatment),wenotethatourempiricaltreatmentofcareerbackgrounds,especiallywithrespecttovariableconstructioninSect.4,drawsheavilyontheirapproach,whichispremisedonthenotionthatthegov-ernmenthasatime-consistentinflationarybias:accordingly,MPCmemberswhosecareerbackgroundslierelativelyclosertocentralgovernment(experienceingovernmentandin-dustry)aredeemedmoreliabletoconformtosuchbias,andthemoreoneconformstoit,thegreaterthepropensitytodissentonthesideofmonetaryease.Conversely,memberswhoseexperiencesarerelativelyfurtherfromcentralgovernment(experienceinacademia,finan-cialorganizationsandatthecentralbank)areconsideredmorepronetodissentonthesideofmonetarytightness.Asisshownlateron,MPCmembers’careerbackgroundsaremarkedbyconsiderablediversity,facilitatingacomprehensiveanalysisofcareerbackgroundeffectsonvotingbehavior.Wedo,however,notethatjustasinHavrileskyandSchweitzer(1990),anumberofjudgementcallsweremadewithrespecttotheimpactofparticularcareerback-groundsonvotingbehavior.However,theseeffectsmightbeexpectedtobesmall,giventhenatureofthedemocraticdeficitandtheinflationtargetingframeworkwithinwhichMPCmemberstakeinterestratedecisions.Inadditiontotheeffectsdiscussedabove,anumberofotherfactorsmayinformthede-cisiontodissent,whichfurthertobeingshapedbycareerexperience,maybedeterminedbymember-specificfactorsthataredifficulttoobserve,andthusdifficulttoproxyfor:theseincludemembers’viewsabouttheunderlyingstructureoftheeconomy,andmembers’infor-mationsets.DifferentMPCmemberswillinvariablyholddifferentviewsabouttheunderly-ingstructureoftheeconomy,andbeexposedandreceptivetodifferentsourcesofeconomicinformation.AsBlinder(2007)notes,evenwhenfacedwiththesameinformation(forex-ample,aspresentedattheBank’smonthly‘pre-MPC’meeting),membersofamonetarypolicycommitteemayhavecontrastingviewsastotheappropriatepolicystance.Asallofthesefactorsrepresentpotentialsourcesofdisagreementregardinginterestratepolicy,thismayleadtodissentvoting.20Anaturalwaytocaptureunobservedmember-specificfactors,19HavrileskyandSchweitzeralsoassumethatthereisautilityanddisutilityassociatedwithdissenting.Utilityarisesfromamemberbelievingitismorallyrighttodissent.Thisexplainswhy,eveninthefaceofpotentialpressuretovotewiththemajorityofMPCmembers,someindividualsmaychoosetomaketheirdifferingopinionknownthroughrecordingadissentingvote.Ontheotherhand,disutilityarisespreciselybecauseoftheneedforcommitteememberstofallintolinewitheachother.Thereissomeevidencetosuggestthismayinpracticehappen:recentworkbyHarrisandSpencer(2009)showsthatcomparedtoexternalmembers,internalsaresignificantlymorepronetovoteasabloc.WhilethistendencymayreflectaninstitutionalconsensusamongstinternalMPCmembers,suchmembersmaydesistfromvotingagainsteachotherbecausetheyworkforthesameorganization:theconjecturehereisthatvotingagainstone’speerstoooftenwillbeviewedinadimlight,andmayaffectcareerprospects,asalreadydiscussedinSect.2.1.Buildingonthisargument,itisnotedthatwhiledissentvotingisnotactivelyencouragedontheMPC,itsindividualisticnature(see,forexample,BlinderandWyplosz2005andBlinder2007)maylessenthestigmaattachedtodissenting:Nakahara(2001),forinstance,citesformerexternalMPCmemberDeAnneJuliusasattributingdissentvotingtoindividuallyaccountability,withoutwhichmemberswould“losetheincentivetomakepublictheirpositionatthevotingstageeveniftheyhadvoicedopposingviewsduringthedebate.”Weproposethatwhileindividualaccountabilitymaymitigatethestigmaattachedtodissenting—asmighttheindividualisticnatureoftheMPC(Blinder2007)—itdoesnoteliminateinaltogether.20InMay2002,SushilWadhwani,anexternalmemberwhoservedontheMPCbetweenJune1999andMay2002claimedthatmonetarypolicywas‘heldtootightbecauseofabiasedforecast’(Wadhwani2002:1),themaininflationprojectionintheMay2002InflationReportassumeda‘higherpass-throughintoprices’thanwasprobable(Wadhwani2002:3),andthelevelofpotentialoutputwas‘toopessimistic’(Wadhwani2002:10).Suchsentimentisperhapsunsurprising:asCharlesGoodhartnotes,theintroductionofexternalMPCmembersintoacommitteeforecastingprocesssuchastheoneunderlyingtheBank’squarterlyInflation PublicChoiceistouseafixed-effectsapproach,aninnovationwhichispursuedaspartoftheestimationstrategyinSect.4.Priortothis,weaddresstheempirics.3DissentingvotesVotingdataareobtainedfromtheMinutesofMPCMeetings,andidentifieswhothedissent-ingvotersareateachmeeting,andwhethertheydissentedonthesideofeaseortightness.Wedefinetwotypesofdissentvoting:dissentfortighterpolicyanddissentforlooserpolicy.Thereareimportantcaveatstothesedefinitions,whichareexpoundedasfollows:(i)Dissentfortighterpolicy:Definedaswhereamembervotesforahighershort-terminterest-ratethantheratechosenbythewinningmajorityofMPCmembers.Amembermayvotefornochangeoradecreaseintheinterest-ratebutstillbeclassedasdissentingfortighterpolicyiftheratechosenbytheMPCislowerthantheirchosenrate.(ii)Dissentforlooserpolicy:Definedaswhereamembervotesforalowershort-terminterest-ratethantheratechosenbythewinningmajorityofMPCmembers.Amembermayvotefornochangeoranincreaseintheinterest-ratebutstillbeclassedasdissentingforlooserpolicyiftheratechosenbytheMPCishigherthantheirchosenrate.WealsofinditusefultodefineasassentingvoteasonethatiscastinagreementwiththewinningmajorityofMPCmembersateachmeeting.Table1documentsthedissentvotingbehaviorcorrespondingtodifferentgroupsofMPCmembers.Inadditiontoiden-tifyingmembersaccordingtotheirinternal-externalstatus(asinGerlach-Kristen2003;HarrisandSpencer2009),wealsotakeintoaccountwhethermembersareappointedbypoliticalornon-politicalmeans.Thisyieldsthreegroups:(i)politicallyappointedinternals;(ii)non-politicallyappointedinternals;and(iii)politicallyappointedexternals.21Thefirstthreecolumnsshowstatisticspertainingtothenumberofassents,anddissentscastontheTable1NumberofdissentingvotescastbytheMPC,August1997–May2007aAssentsEaseTightnessAlldissentsdissentsdissentsAllMembers892(85.4)83(7.9)b70(6.7)153(14.6)InternalMembersP318(92.2)3(0.9)24(7.0)27(7.8)InternalMembersNP214(89.9)7(2.9)17(7.1)24(10.1)ExternalMembersP360(78.3)73(15.5)29(6.2)102(21.7)aResultsbasedondatafrom119meetingsbFiguresinparentheses(·)expressthenumberofassentinganddissentingvotescastasapercentageofvotescastwithineachgroupPDenotespoliticalNPDenotesnon-politicalappointmentReportprojections,is‘inevitablylikelytogeneratesometensionanddisagreements’(Goodhart2001:62).AlsoseeBlanchflower(2006).21Allexternalsarepoliticalappointments. PublicChoicesideofeaseandtightnessrespectively.Thefinalcolumnshowsthetotalnumberofdissent-ingvotescast,irrespectiveofdirection.22Figuresinparentheses(.)expressthenumberofdissentingvotescastasapercentageofallvotescastwithineachgroup.Whenvotesacrossallmembersirrespectiveofsub-groupingsareconsidered,dissent-ingvotescastformonetaryeaseandtightnessaresplitrelativelyevenly(83vs.70votes,AllMembers).Further,irrespectiveofamember’sinternal-externalstatusorthechannelofappointment,allMPCmembersaremostlikelytocastassentingvotes.Thesebroadsimilar-itiesare,however,deceptive.Externalmembersaremorethantwiceaslikelytodissentthanpoliticallyandnon-politicallyappointedinternalmembersrespectively(102vs.27and24votes).23Whereasinternalmembers—irrespectiveofthechannelofappointment—aremorepronetodissentingonthesideoftightness,externalmembersdosoonthesideofmone-taryease.Chi-squaredtestswerealsoconductedtotestfordifferencesbetweenmembersbasedontheirinternal-externalstatusandchannelofappointment.Ourfindings,fulldetailsofwhichareprovidedinAppendixC,suggestthatinternalandexternalmembersdiffersignificantlyintheirvotingbehavior.However,whilepoliticalandnon-politicalappointeeswerefoundtovoteinasignificantlydifferentway,nosignificantdifferencewasfoundinthevotingbehaviorofpoliticallyandnon-politicallyappointedinternalmembers.Thissuggeststhatdifferenceswhichatfirstappearedtobeassociatedwithappointmentchanneleffectsareinfactattributabletointernal-externalfactors.Finally,Table2showsthedissentvotingbehaviorofMPCmembersatanindividuallevel.Outofthefourteenexternalmembersinthesample,sixmembers(Buiter,Goodhart,Budd,Walton,BesleyandSentance)castmoredissentingvotesonthesideoftightnessthanease.Atthislevelofdisaggregation,thereputationforexternalstocastdissentingvotesonthesideofmonetaryeaseseemstobedrivenbyjustfourindividuals:Julius,Wadhwani,NickellandAllsopp.ExcludingBuiter,whosenumeroustightnessdissentsarebalancedbyasignificantnumberofeasedissents,andLambert,whodidnotdissent,theremainingmemberscastonlyamodestamountofeasedissents.Inshort,externalsvaryconsiderablyintheirindividualdissentvotingbehavior.Turningtointernalmembers,twoindividuals—MervynKingandAndrewLarge—mightbeviewedasdrivingtightnessdissents.Moreover,onceKingisomittedfromthesample,thetotalnumberoftightnessdissentscastbyinternalsfalltoafigurebelowthatofexternals(28vs.29votes).Further,whileinternalsarelesspronetocastdissentingvotesperse(putanotherway,theyarefarmorelikelytobeonthewinningsideofadecision),fourmembers(excludingDavieswhocastonlytwovotes)buck22InthefirstdecadeoftheBankofEnglandMPCalmost15%ofvotescastbyitsmembersweredissents.ThisfigureismarkedlyhigherthantheportionofdissentingvotesassociatedwithmonetarypolicycommitteesatcomparableinstitutionssuchastheUSFederalReserve,theBankofJapanandtheSwedishRiksbank.23Onaverage,externalmembersdissentedapproximatelyonceeveryfivevotes.Thisfigureisaroundoneintenforinternalmembers.Interestingly,exactpermutationtestswerealsousedextensivelytotestfordiffer-encesinthedissentvotingbehaviorassociatedwithinternalandexternalmembersateachMPCmeeting.Foreachindividualmeetingninehypotheseswereexamined,andinall121setsofteststatisticsweregenerated:specifically,foreachtypeofdissentingvotedefinedinSect.3(dissentirrespectiveofdirection,dissentfortighterpolicy,dissentforlooserpolicy),wetestedforwhether(1)internalmemberscastsuchvotesmoreoftenthanexternalmembers(one-tailedtest),(2)externalmemberscastsuchvotesmoreoftenthaninternalmembers(one-tailedtest),and(3)whethertherewasanygeneraldifferencebetweengroups(two-tailedtest).Eachhypothesiswastestedagainstanullofnodifferenceinvotingbehavior.Whenjudgedagainstconven-tionallevelsofsignificance,membersdonotdifferintheirdissentvotingbehavior.However,theverysmallsamplesize(MPCmembershiprangedfromfivetonineindividuals)raisedproblemsofstatisticalpower,andspecifically,theprospectoffailingtorejectthenullhypothesiswhenitisfalse(i.e.,atypeIIerror).Moreover,itwasfoundthatforatesttohaveanystatisticalpoweratconventionallevels(power=0.8),allinternalmemberswouldhavetodissentwhileallexternalmembersassented. PublicChoiceTable2NumberofdissentingvotescastbyindividualMPCmembers,August1997–May2007VotesaAssentsEasedissentsTightnessdissentsAlldissentsInternalmembersP,EddieGeorge7272000MervynKing11910601313DavidClementi6157134AndrewLarge4031099RachelLomax4743224JohnGieve1616000InternalmembersNPIanPlenderleith5954235JohnVickers2823055CharlesBean8177404PaulTucker6053167ExternalmembersPWillemBuiter34178917CharlesGoodhart3431033DeAnneJulius453114014SirAlanBudd1814044SushilWadhwani372413013StephenNickell735613417ChristopherAllsopp372611011KateBarker7368415MarianBell3636505RichardLambert3434000DavidWalton129123DavidBlanchflower128404TimBesley96033AndrewSentance85033PPoliticalappointmentNPNon-politicalappointmentaDenotestotalnumberofvotescastbyeachmemberHowardDavies,aDeputyGovernorwhoonlyservedinthefirsttwomeetings,liesoutsidethesampleusedinestimationandisthusomittedfromthetableMervynKingservedasanExecutiveDirectorattheBankbeforebecomingaDeputyGovernorthetrendthatinternalsaremorelikelytodissentonthesideoftightness—namelyBean,Lomax,GieveandGeorge.Notallinternalsarethesame.ThereasonforGeorgecastingnodissentsmaybepreciselyputdowntohisroleasMPCChairmanbetweenJune1997–May2003:hewasneveronthelosingsideofadecision,arguablyduetohispowerastomakethepolicyproposalateachmeeting,coupledwithhismoregeneralinfluenceasGovernoroftheBank.24 PublicChoice4Econometricanalysis4.1ModelspecificationandvariablesWhenvotingonthepolicyproposal,MPCmembersareviewedasbeingfacedwiththreemutuallyexclusivechoices:todissentonthesideofease,toassent,ortodissentonthesideofmonetarytightness.25Anaturalcandidateformodelingsuchbehavioristheorderedprobit(OP)model,∗Zgt=xgtβ+εgt,(1)∗Zgt=−1ifZgt≤γ1,∗Zgt=0ifγ1γ2whereZ∗isastackedm×1vectorof−1s,0sand1scorrespondingtomembers’votestogtdissentforlooserpolicy,assent,ordissentfortighterpolicy.xgtisam×hmatrixcontaininghindependentvariables,andβisa1×hvectorofparameterestimates.Asthexgtmatrixcontainsthosevariableswhichformthebasisofourestimations,wenowturnattentiontotheirconstruction.4.1.1ChannelofappointmentandcareerbackgroundvariablesToaccountforthedifferentmembertypesintroducedinTable1,threeindicatorvariableswerecreated:IntPolequalsoneforapoliticallyappointedinternalmemberandzerooth-erwise;IntNPolequalsoneforanon-politicallyappointedinternalmemberandzerooth-erwise;lastly,ExtPolisassignedaoneforanexternalmemberandazerootherwise.Un-likeinternalmembers,externalsareallpoliticallyappointed.Toavoidthedummyvariabletrap,IntNPolwasomittedduringestimation.Withrespecttoparameterestimates,aneg-ativesignforExtPolwasexpected.Thisreflectsthestylizedfindingsintheliterature—rationalizationsforwhicharediscussedinSects.2and3—thatinternal(external)memberschoosehigher(lower)rates,andaremorelikelytodissentonthesideofmonetarytightness(ease).Political-appointmenteffects,ifpresent,mightalsobeexpectedtoruninthesamedirection:thisisbecauseceterisparibus,weassumethatpartisansappointedbyaleft-of-centerpartyaremorepredisposedtodissentonthesideonmonetaryeasethannon-political24Significantly,itispossiblefortheChairmantoloseavote.Thishappenedforthefirsttimeduringthe100thmeetingoftheMPCinAugust2005,whenGovernorKingwasonthelosingsideofa5–4split.Havingascertainedthepolicystancesofallcommitteemembers,Kingtabledapolicyproposalwhichhethenproceededtovoteagainst.25OurestimationstrategyextendstheoriginalapproachofHavrileskyandSchweitzer(1990)inanumberofimportantrespects.First,inadditiontousingdissentingvotes,weutilizeallvotescastinagreementwiththepolicyproposal(assentingvotes).Assuch,informationcontainedinthevotingrecordisnolongerwasted,makingourapproachconsistentwiththeU.S.FOMCapproachesofChappelletal.(1993,2004,2005).Second,inadditiontocareercharacteristicsdata,ourestimationscontrolforthepresenceofotherfactorswhichdeterminethedecisiontodissent,thusaddressingtheissueofomittedvariablebias.Third,andbuildingonthefirstpoint,throughutilizingassentingvotes,thelargersamplesizepermitsustopursueestimationstrategieswhichconditiononunobservedheterogeneity. PublicChoiceappointees.26Ofcourse,theassertionthatpoliticalappointeesmightbeanticipatedtodis-sentonthesideofmonetaryease,restsontheobservationthatallsuchappointmentsinoursamplehavebeenmadebytheLabourParty.Theelectionofaright-of-centerConservativegovernmentmaychangethisbehavioralassumption.YetgiventherearestrongargumentstosupposethattheselectionprocessforMPCmembersisnotpoliticized(Goodhart2005;GoodhartandMeade2003),anysucheffectisanticipatedtobeatmostsmall.Moreover,ifdifferencesbetweenMPCmembersarenotdeterminedbytheappointmentchannel,itimpliesthattheIntPolparametershouldbestatisticallynodifferentfromzero:liketheomit-tedgroup,itsmembersareallinternals,withitsmembersonlydifferinginhowtheyareappointed.Tocapturecareerbackgroundeffects,aseriesofcovariatesproxyingmembers’careercharacteristicsarealsoconstructed.Careerbackgroundsarecategorizedaccordingtoyearsspentworkinginsixbroadlydefinedcategories:(i)Academia—referstoyearsworkingatauniversityinanacademiccapacity.(ii)Bank—denotesthenumberofyearsemployedattheBankofEngland.(iii)Finance—referstopositionsheldinbankingandfinance.(iv)Government—denotesyearsspentworkinginthecivilserviceorfortheUKGovern-ment.(v)Industry—referstoyearsspentaneconomistinindustry.(vi)NGO—referstonon-governmentalorganizations.Thiscoversbothnationalandin-ternationalindependentresearchorganizationssuchtheOrganizationforEconomicCooperationandDevelopment(OECD),andtransnationalinstitutionssuchastheIn-ternationalMonetaryFund(IMF),WorldTradeOrganization(WTO)andBankforInternationalSettlements(BIS).OurclassificationsystemcoversonlyfulltimepositionsandsecondmentsheldbyMPCmembersuptobutnotincludingtimeworkingontheMPC;excludedfromthecriteriaareallpart-timepositions,specialadvisoryrolesandacademicconsulting.Consequently,alltimeservedontheMPCispurposelyneglected.Whiletheeffectsofcareerbackgroundshaverequiredmakingsomejudgementcalls,vot-ingbehaviorisanticipatedtobeinfluencedinthefollowways.Weassumethatbackgroundsinacademia,finance,attheBankofEnglandandNGOspromotetightnessdissents:inthecaseofacademia,thisreflectsthelargeimpactoftheliteratureontime-consistentmonetarypolicy,andaviewthatexperienceinacademiapromotesindependentthinking,hencelower-ingmembers’susceptibilitytoyieldtoshort-runpoliticalpressures.ExperienceattheBankofEnglandisassumedtoengenderanacuteawarenessoftheinflationaryconsequencesofactivistmonetarypolicy,thuspromotingtightnessdissents.For‘career’centralbankers,26Empiricalsupportfortheconjecturethatleft-wingpartiesaremorelikelytopursueeconomicpolicieswhichboostoutputisexemplifiedintheworkofAlesinaandRoubini(1992).WealsonotethatastheMPCwasnotonlyestablishedbyaLabourgovernment,butonewhichheldpowerforthedurationofthesample,itisnotpossibletotestforvotingdifferencesassociatedwithpoliticalmemberschosenbydifferentUKpo-liticalparties.ThisisunlikestudiesofFOMCvoting,wherethemonetarypolicypreferencesofindividualmembersaremodeledasafunctionoftheparty-politicalaffiliationoftheindividualswhoappointedthem.Suchstudiestypicallycoverperiodsencompassingdifferentpoliticaladministrations,RepublicanandDemo-crat.Forexample,Chappelletal.(1993)findthatthepowerofappointmentprovides‘animportantchannelofsystematicpartisaninfluence’(Chappelletal.1993:209),suchthatDemocratappointeesexhibitsignifi-cantlydifferentvotingbehaviorthanRepublicanappointees.Whileourdataprecludeusfromtestingforsucheffects,itdoesnotpreventtestingfordifferencesbetweenpoliticalandnon-politicalappointments;further,astheLabourPartyispoliticallyleft-of-center,wepositthatsuchpartisaninfluence,ifpresent,willmanifestitselfinMPCmembersdissentingonthesideofmonetaryease. PublicChoiceTable3CareerbackgroundsofindividualMPCmembers,June1997–May2007Careerexperience(years)AcademiaBankFinanceIndustryGovt.NGOInternalmembersEddieGeorgeP0290006MervynKing,NP,P1870000DavidClementiP0220000AndrewLargeP0026850RachelLomaxP0000361JohnGieveP0000280IanPlenderleithNP0320003JohnVickersNP1700000CharlesBeanNP1800070PaulTuckerNP01210000ExternalmembersWillemBuiterP2300000CharlesGoodhartP15170030DeAnneJuliusP400907SirAlanBuddP22040100SushilWadhwaniP8010000StephenNickellP3000000ChristopherAllsoppP3140020KateBarkerP0001600MarianBellP0015030RichardLambertP0035000DavidWaltonP0019020DavidBlanchflowerP2100000TimBesleyP1800000AndrewSentanceP5001500PPoliticalappointmentNPNon-politicalappointmentHowardDavies,aDeputyGovernorwhoonlyservedinthefirsttwomeetingsliesoutsidethesampleusedinestimationandisthusomittedfromthetableMervynKingservedasanExecutiveDirectorattheBank(NP)beforebecomingaDeputyGovernor(P)dissentingonthesideofmonetarytightnessmayalsobeusedtosignaltheircredentialsasbeing‘conservative’or‘inflation-averse’.Finally,theinclusionofpriorexperienceinfi-nanceandNGOsreflectsaviewthatsuchcareersareremovedfromgovernmentalpowerandinfluence.Wealsoproposethattimespentinindustryandgovernmentwillpromoteeasedissents.Inthecaseofindustry,whilerisingpricesmayimplyhigherwageclaimsandthusrisingcostsforthefirm(promptingcallsforthemonetaryauthoritiestobringinflationundercontrolthroughtighteninginterestrates),easedissentsaremorelikelytobepromotedashigherinterestrateshittheabilityoffirmstoinvestandborrow,reduceconsumerexpen-diture,andreducetheinternationalcompetitivenessofproductsforexportthroughexchangerateeffects.Table3documentsthecareerexperiencesofallindividualswhoservedonthe PublicChoiceFig.1MeancareerexperienceofMPCmembers,June1997–May2007MPCbetweenJune1997andMay2007,measuredinthenumberofyearsspentineachcategorypriortojoiningtheMPC.Forestimationpurposes,andinlinewithHavrileskyandSchweitzer(1990),thesevari-ablesaresubsequentlymanipulated:foreachMPCmember,experiencewithineachcareercategoryisexpressedasthedifferencebetweenthenumberofyearsspentworkinginthatcategoryandthecommitteemeanforthatcategory.27Thesenewvariables,whichwelabelAcadD,BankD,FinD,GovtD,IndDandNGOD—wheretheDsubscriptdenotes‘deviationfromthecommitteemean’—varyacrosstimeduetomembers’overlappingterms.Figure1showshowthecommitteemeanforeachcareercategoryhaschangedovertheperiodunderscrutiny:forinstance,meanexperienceattheBankofEnglandhasfallenconsiderablysincetheestablishmentoftheMPC,whileexperienceingovernmentrosemarkedlyduringthefirsteightyears,beforefallingbacktomid-1997levelsatthestartof2006.Bycontrast,theaveragelevelofacademicexperienceremainedrelativelyhighandconstantacrossthewholesampleperiod.TocomplementFig.1,Table4showsthenatureofmembers’overlap-pingterms,andcaptureshowmembershipoftheMPChaschangedovertime.Onaverage,thecompositionoftheMPCchangedapproximatelyeverysixmonths,raisingtheprospectthatoverthesampleperiod,asdifferentpersonalitiesbothenteredandleftthegroup,thedecisionmakingdynamicsofthecommitteeunderwentconsiderablechange.27Careercharacteristicsshouldnotbeinterpretedascareer‘fixed’effectsasamember’scareerexperienceforagivencharacteristicisnotdetrendedbyitsmean.Asthecommitteemeanforagivencharacteristicchanges(i.e.,withtheturnoverofnewmemberswithdifferentcareerbackgrounds)sotoodoesamember’sgivencareercharacteristic.Moreformally,(xgt−xg)isusual,butinourcasewehave(xgt−xt),wherexrepresentsagivencareercharacteristic,xtitsmeanvalueinperiodt,xgisthemeanofthatcharacteristicformemberg,andxgtistheperiodtvalueofthatcharacteristicformemberg. PublicChoiceaExternalb,PlenderleithBuiter,GoodhartdInternalKingPlenderleithPlenderleithPlenderleithPlenderleith,VickersPlenderleith,VickersPlenderleith,VickersBuiter,GoodhartPlenderleith,BeanBuiter,Goodhart,Julius,BuddBuiter,Goodhart,JuliusPlenderleith,BeanBuiter,Goodhart,Julius,WadhwaniBuiter,Goodhart,Julius,BuddBean,TuckerJulius,Wadhwani,Allsopp,NickellBean,TuckerJulius,Wadhwani,Allsopp,NickellBean,TuckerWadhwani,Allsopp,Nickell,BarkerBean,TuckerBean,TuckerBean,TuckerAllsopp,Nickell,BarkerBean,TuckerAllsopp,Nickell,Barker,BellBean,TuckerAllsopp,Nickell,Barker,BellBean,TuckerAllsopp,Nickell,Barker,BellBean,TuckerNickell,Barker,Bell,LambertBean,TuckerNickell,Barker,Bell,LambertBean,TuckerNickell,Barker,Lambert,WaltonBean,TuckerNickell,Barker,Lambert,WaltonNickell,Barker,WaltonBarker,Walton,BlanchflowerBarker,BlanchflowerBarker,Blanchflower,BesleyBarker,Blanchflower,Besley,SentanceDGDGDGDGDGDGDGDGDGDGDGDGDGDGDGDGDGDGDG,Clementi,Clementi,Clementi,Clementi,Clementi,Clementi,Clementi,Clementi,Clementi,Large,Large,Gieve,Gieve,Gieve,Gieve,Gieve,GieveDG,Lomax,LomaxDGDGDGDGDGDGDGDGDGDGDGDGDGDGDGDGDGDGDGDGDG,Davies,King,King,King,King,King,King,King,King,King,King,King,King,KingaGGGGGGGGGGGGGG,Large,Large,Lomax,Lomax,Lomax,Lomax,Lomax,LomaxGGGGGGGGagggg1n2n1n11nGeorgeiGeorgeiGeorgeiGeorge1Georgei1K1K1KGeorge1K++++++++++g12g1g1g1,n2,George1,Georgen1,Georgen2,Georgen11i11,i1,i1,iGeorge1,George1KGeorge1,1K1K1K−+++−−−−−+−+−−−−−−−++Size65membership7890Internal999989George8999998878921364912181722121812121224cThechangingcompositionoftheMPC,June1997–May2007DenotesdeputyGovernorPoliticalappointmentNon-politicalappointmentIncludestheemergencyMPCmeetingof18thSeptember2001MervynKingservedasanExecutiveDirectorattheBankbeforebecomingaDeputyGovernorDenotesGovernor(andhenceMPCChairman)Table4PeriodJun97–Jul97:Aug97:Sep97–Nov97:Dec97–May98:MeetingsJun98–May99:Jun99–May00:Jun00–Sep00:MPCOct00–May01:Jun01–May02:Jun02:ChangeinJul02–Aug02:Sep02:Oct02–May03:MembersJun03:Jul03–Jun05:Jul05–Jan06:Feb06–Mar06:Apr06–May06:Jun06:Jul06–Aug06:Sep06:Oct06–May07:abcdGDG PublicChoice4.1.2AdditionalvariablesTocontrolforeconomicconditions,wealsoconstructTaylor-ruletypecovariatesbasedontheMPC’sin-houseinflationandoutputgrowthforecastsaspublishedintheBank’squar-terlyInflationReport.ForecasthorizonsinlinewithviewsexpressedbytheMPC(BankofEngland1999)thatinterest-ratechangestaketwoyearstomaximallyimpactinflation,andapproximatelyoneyearforoutputarechosen.ThepublishedmodalprojectionsarealsoadjustedfollowingthemethodsuggestedbyGoodhart(2005),28andareexpressedindevia-tionform:specifically,outputgrowthminuspotential(assumedtobe2.4%perannum)andthedeviationofinflationfromtarget.WedenotetheseforecastsπandGDP.29WhileitGGisclearlyimportanttocontrolforeconomicconditions,wealsonotethatitisnotimmedi-atelyobviouswhatthesignsforπandGDPshouldbe.30ToaccountforthepossibilityGGthatdifferenttypesofmembersmightnotrespondequallytochangesinmacroeconomicconditions(basedonthediscussionofmembers’informationsetsandeconomicmodelsinSect.2),IntPol,IntNPolandExtPolwerealsointeractedwiththeTaylor-typevariables.Inall,sixinteractiontermswerecreated:(Intpol×GDPG)and(Intpol×πG);(IntNpol×GDPG)and(IntNpol×πG);and(Extpol×GDPG)and(Extpol×πG).Lastly,tocapturetheimpactunderdifferentMPCchairmen(GovernorsGeorgeandKing),weincludethebinaryvariableChair,whichassumesavalueofone(zero)ifGeorge(King)occupiedthepost.314.1.3Fixed-effectsWenowaddressthepotentialroleforfixed-effects.AstherearerepeatedobservationsforindividualMPCmembers,itispossibletoconditiononunobservedindividualheterogeneity28Goodhartarguesthattheex-postnatureoftheBank’spublishedforecastsdiminishestheirimportanceinexplainingtheMPC’spolicydecisions:thisisbecauseinpracticeMPCmembersreacttoex-anteforecasts(i.e.,conditionedontheinterest-ratesetbytheMPCinthepreviousmonth).Weusetheex-postforecaststoconstructproxyex-anteforecasts.Theseforecastsarepotentiallymuchclosertothoseonwhichindividualvotingdecisionsarebased.29EstimationswerealsoperformedusingTaylor-ruletypevariablesconstructedfromreal-timeconsensusforecastsofGDPgrowthandinflation,obtainedfromHMTreasury’sForecastsfortheUKEconomy.Pub-lishedmonthly,thisisacompendiumofforecastsproducedbycityandindependentforecasters.WenotethatwhilstdecisionsonUKinterestratesareinpartafunctionoftheBank’scentralinflationandGDPpro-jections,theirquarterlynaturemakestheirincorporationintooureconometricframeworkdifficult—thisisbecausetheMPCtakesinterestratedecisionsonamonthlybasis.Forthisreason,andtotakeintoaccountnewforecastinformationavailabletoMPCmembersateachmonthlymeeting,wealsoranregressionsus-ingthesenewvariables.Resultsconsistentlyplacedasmallerparameterontheinflationterm,althoughtheoutputtermwasstillnegativeandofteninsignificant.Allotherparameterswererobusttotheinclusionofthesealternativeforecasts.However,weoptedfortheresultsbasedontheMPC’sBankforecastsduetotheirpurportedimportanceininformingtheMPC’sdecisions.Fullconsensusestimationresultsareavailablefromtheauthorsonrequest.30Wethankananonymousrefereeforraisingthisissue.Inthemonetarypolicyrulesliterature(forexample,Taylor1993),itisusualtoexpectthesignsontheoutputandinflationvariablestobepositive:thiswouldimplythatUKinterestratesshouldbehigher(lower)whenforecastinflationisabove(below)itstargetlevel,andoutputisabove(below)potential.However,inthispaper,wearemodelingdissentingvotes,andnotmembers’interestratechoicesasinBesleyetal.(2008)andHarrisandSpencer(2009).Wearenotestimatingamonetarypolicyrule.Forthisreason,highervaluesofπGandGDPG—whichshouldintheorybeassociatedwithmemberschoosinghigherinterestrates—mightnotnecessarilyimplyasimultaneousdesiretodissentforgreatertightness.31Wenotethattwoadditionalvariables—adummytocapturetheimpactofgender,andareappointmentdummytoproxyfortheroleofcareerconcerns—wereintroduced,butsubsequentlydroppedfromestimationsastheyprovedtobeconsistentlyinsignificant. PublicChoicebyaugmentingequation(1)toincludeanunobservedeffect,αg:∗Zgt=xgtβ+αg+εgt.(3)Thisbegsthequestionofhowtotreattheαg.Whilstnon-linearpaneldataestimationhastraditionallyfocussedontreatingunobservedheterogeneityasrandomduetothe‘incidentalparameters’problem(NeymanandScott1948),recentdevelopmentssuggestthatthenatureofoursamplepermitsafixed-effectsestimationstrategy.Specifically,weareinapositioncontrarytothattypicallyobservedinthepaneldataliterature,withsmallcross-sectionalcomponentrelativetothesample(i.e.,largeTandsmallN).Theoverwhelmingmajor-ityofthe25MPCmembersinthesampleareobservedoverarelativelylongtimeperiod(t=1,...,Ti):otherthanDavies(Ti=2)—whowasremovedfromthesampleduetoanin-sufficientnumberofobservations—thenumberoftimeperiodsrangedfromTi=8(Besley)toTi=121(King).Heckman(1981)suggeststhatatemporalsamplesizeofT=8issuf-ficientforanysignificantfixedTbiastohaveessentiallydisappeared.Greene(2004)pro-videsfurtherevidence,citingasignificantreductioninbiasesfromT=3onwards.Inlightofthesearguments,weincludefixed-effectsdummiesforallMPCmembersbarDavies.ThisamountstorelaxingthecommonlymaintainedassumptionofExgtαg=0,∀g,t.(4)Thebaselineequationonwhichestimationisbasedhencebecomes∗Zgt=xgtβ+αgDg+εgt,(5)whereDgrepresentsadummyvariableformemberg.4.2EstimationresultsEstimatesfortheorderedprobitmodelareshowninTable5.Standarderrorsaregiveninparentheses(.),withcorrespondinglevelsofsignificance,where***,**,*denote1,5and10%levelsrespectively.AICandBICdenotetheAkaikeandBayesianinformationcriteria,whereasmallervaluesuggestsabetterspecification.Model1,whichmodelsdissentvotingusingcareercharacteristicsalone,canbeviewedasa‘baseline’specification,andisanalogoustotheestimationstrategyinHavrileskyandSchweitzer(1990);Models2and3augmentthebaselinespecificationbycontrollingfordifferentmembertypes(IntPol,IntNPolandExtPol),theimpactofeconomicconditions,andChairmaneffects;Model4introducesmember-specificfixed-effects.WhileinModels1,2,and3thejointhypothesisofcareerbackgroundshavingnosignifi-canteffectwasrejectedoutright(p<0.01ineverycase),thedirectionofsomecareerback-groundeffectparametersdidnotconformtopreviouslystatedpriors.Forexample,whereasexperienceattheBankofEnglandwasfoundtopromotetightnessdissents(positive,sig-nificant,andconsistentwithourpredictions),timespentworkinginindustryalsopromotesdissentonthesideofmonetarytightness(significant,positivelysigned,notconsistentwithpredictions).Experienceingovernment,financeandacademiagenerallyhasnosignificanteffectondissentvotingbehavior.However,evenwherecareervariablesarehighlysignifi-cant,theimpactondissentvotingisoftennegligible.32Thesefindingsaresomewhatatodds32Thisassertionisalsobasedoncalculatingmarginaleffects,whicharenotreportedhereduetospaceconstraints. PublicChoiceTable5Determinantsofdissent:orderedprobitestimatesModel1Model2Model3Model4πG–3.167(0.470)∗∗∗–3.285(0.512)∗∗∗GDPG–−0.390(0.101)∗∗∗–−0.276(0.129)∗∗IntPol–0.283(0.144)∗∗0.265(0.150)∗–ExtPol–−0.320(0.154)∗∗−0.424(0.161)∗∗∗–IntPol×πG––1.214(0.833)–IntNPol×πG––1.445(0.972)–ExtPol×πG––5.213(0.661)∗∗∗–IntPol×GDPG––−0.231(0.185)–IntNPol×GDPG––−0.228(0.200)–ExtPol×GDPG––−0.544(0.132)∗∗∗–Chair–−0.113(0.117)−0.109(0.119)0.389(0.165)∗∗AcadD0.001(0.009)0.007(0.009)0.009(0.009)–BankD0.048(0.008)∗∗∗0.043(0.009)∗∗∗0.043(0.009)∗∗∗–FinD0.005(0.009)0.005(0.010)0.007(0.010)–GovD0.019(0.008)∗∗0.010(0.009)0.012(0.009)–IndD0.118(0.018)∗∗∗0.128(0.019)∗∗∗0.132(0.020)∗∗∗–NGOD−0.280(0.036)∗∗∗−0.272(0.041)∗∗∗−0.273(0.042)∗∗∗–George,P–––−0.671(0.261)∗∗Plenderleith,NP–––−0.558(0.264)∗∗∗Clementi,P–––−0.519(0.264)∗∗Vickers,NP–––−0.004(0.319)Bean,NP–––−0.729(0.294)∗∗∗Tucker,NP–––0.045(0.261)Large,P–––0.669(0.274)∗∗∗Lomax,P–––−0.380(0.304)Gieve,P–––−0.431(0.486)Buiter,P–––−0.735(0.287)∗∗∗Goodhart,P–––−0.359(0.324)Julius,P–––−2.145(0.278)∗∗∗Budd,P–––0.237(0.390)Wadhwani,P–––−2.040(0.285)∗∗∗Nickell,P–––−1.032(0.238)∗∗∗Allsopp,P–––−1.763(0.288)∗∗∗Barker,P–––−0.635(0.266)∗∗Bell,P–––−1.433(0.314)∗∗∗Lambert,P–––−0.356(0.355)Walton,P–––0.171(0.467)Blanchflower,P–––−1.897(0.427)∗∗∗Besley,P–––0.862(0.471)∗Sentance,P–––0.996(0.491)∗∗γ1−1.523(0.062)∗∗∗−1.726(0.147)∗∗∗−1.811(0.154)∗∗∗−2.075(0.203)∗∗∗γ2−1.615(0.066)∗∗∗1.660(0.146)∗∗∗1.622(0.149)∗∗∗1.538(0.187)∗∗∗ PublicChoiceTable5(Continued)SUMMARYSTATISTICSAIC1014.159952.434940.889916.531BIC1053.7731016.8071025.0691055.180†Kingistheomittedvariable,Daviesdroppedduetoinsufficientobservations‡Standarderrorsinparentheses(·)DenotesinternalmemberDenotesexternalmember∗∗∗Denotestwo-tailedsignificanceatthe1%level.Noofobs=1045∗∗Denotestwo-tailedsignificanceatthe5%level.Noofobs=1045∗Denotestwo-tailedsignificanceatthe10%level.Noofobs=1045PDenotespoliticallyappointedmemberNPDenotesnon-politicallyappointedmemberwiththeFOMCliterature,wherecareerbackgroundsplayanimportantroleindeterminingvotingbehavior.33Wefurthernotethatthecareerbackgroundparametersareveryrobusttospecificationchange,asevidencedbythemagnitudesandsignsofcorrespondingparameterestimatesacrossmodels1,2and3.ThedecisiontoaugmentthebaselinespecificationinModel1withadditionalcovariatesisalsoclearlydesirablebasedonAICandBICgrounds.Themostnoteworthyinnovationinmodels2and3istocontrolfortheeffectsofdiffer-entmembertypes.Ourfindingssupporttheconjecturethatcomparedtochannelofappoint-menteffects,theinternal-externaldistinctionplaysasignificantlygreaterroleindrivingdis-sentvotingbehavior.Model2suggeststhatwhereaspoliticallyappointedinternals(IntPol)aremorelikelytodissentonthesideoftightnessthannon-politicallyappointedinternals(IntNPol),externalmembers(ExtPol)arerelativelymorepronetodissentonthesideofmonetaryease.However,theparameterestimateforIntPolissmallerinabsolutesizethanforExtPol.Further,thefactthatIntPolispositivelysignedsuggeststhatbeingappointedbytheexecutivebranchofgovernmentpromotestightnessdissents.Thisresultstandsatoddswithourpredictionthatceterisparibus,politicalappointeesshouldeitherdissentonthesideofease,orbehaveinawaythatisnotsignificantlydifferentfromnon-politicallyappointedinternals.Anypositiveappointmentchanneleffects,however,appeartobeoutweighedbyeffectsappertainingtotheinsider-outsiderdistinction,andmorespecifically,thenegativeimpactassociatedwithamember’sexternalstatus:here,theconfluenceofbeingpoliticallyappointedandanexternalMPCmemberresultsanegativelysignedparameterforExtPol.33Asananonymousrefereehassuggested,oneinterpretationofourfindingsisthatourpriorswereatbestspeculative,andmoreover,thatcertainmembercharacteristicsmightnotholdthesamebehavioralpredictionsfortheFedandtheBankofEngland.Ontheonehand,thisobservationisnotentirelyunreasonablegiventheparticularnuancesofeachinstitution.Indeed,ageneralexplanationforsuchdifferencesmaybeattributabletotheroleofinstitutionaldesign,andmorespecifically,howtheassociateddemocraticdeficitismitigated.Sayingthis,whileourchoiceofpriorsinvolvedmakingsomejudgementcalls,itseemedreasonabletosup-posethatpriorssimilartothoseusedinFOMCpapersmightbeapplicabletotheMPC.Independentofthispoint,itwasalsohighlightedthatbecausethetimeframeusedinHavrileskyandSchweitzer(1990)differstoours,itmayhaveaffectedourfindings.Althoughnotexaminedhere,onewaytoaddressthisproblemwouldbetoundertakeacomparativestudyofbothinstitutionsusingFOMCandMPCdataoveracomparableperiod. PublicChoiceTheimpactofeconomicconditionsisalsonotable.WhileπGandGDPGaregenerallyhighlysignificantinmodels2and4,GDPGisnegativelysigned.Thisresultisinlinewithotherfindingsintheliterature,namelyBesleyetal.(2008)andHarrisandSpencer(2009).ThemagnitudeoftheπGparameterissubstantiallygreaterthanthatcorrespondingtoallothermodelparameters,implyingthatdeviationsfromtargetinflationplayanimportantroleindrivingdissent.Model3,whichintroducesinteractionterms,reportssimilarfindings,butthestatisticalsignificanceoftheestimatedparametersisrestrictedtoexternalmembersonly:economicconditionsassumenoroleindrivingdissentforbothpoliticallyandnon-politicallyappointedinternalmembers,forwhomtheassociatedmodelparametersarenotsignificantlydifferenttozero.Thisevidencesuggeststhatthefactorsdeterminingdissentaredifferentforinternalsandexternals.Thefixed-effectsspecificationinModel4excludescareercharacteristicsonthegroundsthatbecausecareerexperienceisbydefinitionspecifictoeachmember,itrepresentsanindividualcharacteristic,whichinadditiontofactorssuchasamember’sinformationsetandmodeloftheeconomy,isimplicitlycapturedbyamember’sfixed-effectsdummy.34Crucially,mostmemberdummiesexhibitedsignificanceatthe5%levelorless,andatestofthehypothesisthatfixed-effectsdummiesarejointlyinsignificantwasrejectedoutright,(χ2=144.46,p=0.000),aswasatestforthejointequalityofcoefficients23(χ2=137.14,p=0.000).Forcompleteness,testsforthejointequalityofcoefficients22withindifferentmembersubgroupswerealsoconducted,andthenullofnodifferenceacrossrejectedineverycase(χ2=20.18,p=0.0005(IntPol);χ2=9.77,p=0.0206(IntNPol);43χ2=102.60,p=0.000(ExtPol)).Mostmemberdummiesarealsonegativelysigned.This13reflectsthefactthatKing(theomitteddummy)castasubstantialnumberoftightnessdis-sentsrelativetoothermembers.Coupledwiththetestsofequalityandjointsignificance,thenumberofstatisticallysignificantdummiesisindicativeofconsiderablevoterhetero-geneity,thusprovidingsupportforBlinder’scontentionthattheBankofEnglandMPCisindividualistic.35Finally,theeffectunderdifferentChairmanisambiguous—parameteres-34Theinclusionofcareercovariatesgeneratedfixed-effectsestimateswhichwereconsiderednonsensical.Apossibleexplanationforthisfindingoutcomeisthatcareercovariatesarehighlycorrelatedwithmem-berdummies,andalthoughtheyvaryacrosstimeforeachmember,maylacksufficientvariationtoavoidcollinearity.Resultsavailablefromtheauthorsonrequest.35Forcompleteness,wealsoperformedestimationstreatingtheαgin(3)asrandom.Wooldridge(2002),forinstance,statesthat‘italmostalwaysmakesensetotreattheunobservedeffectsasrandom’.UsingStata,estimationinthepresenceofAR(1)errorswasalsoimplemented,usingthearoprobitcommandofHeiss(2007).Thislatterstrategyallowsforthefactthat‘justasobservedcovariatescanchangeovertime,sotoocanunobservedinfluencesanddeterminantsoftheoutcomes’(Heiss2007:1).Therandomeffectsmodel(3)thusbecomes∗2Zgt=xgtβ+αgt+εgt,εgt∼iid(0,σε)(3)whereαgt=ραg,t−1+vit,vit∼iid(0,(1−ρ2)σ2).(3)Here,αgtisani.i.d.randomvariablewithzeromeanandvarianceσ2,onlynowtheunobservedheterogeneityobeysastationaryAR(1)processovertimewithcorrelationparameterρ.Itisnoteworthythatifαgt=αg∀t,(3)reducestothestandardrandomeffectsmodel(seeHeiss2007forfulldetails).ResultsareshowninTable6,wherewenotethatforModels5and6(randomeffects)ρisinterpretedastheproportionofvarianceexplainedbythepanel-levelvariance,andforModels7and8,ρrepresentsthecorrelationparameterin(3).Werestrictourattentiontocareerbackgroundcovariates,theestimatedparametersofwhicharehighlysimilartothoseinModels1,2and3:experienceinacademiaandfinanceisnotstatisticallysignificant,whileindustryandNGOexperienceexertthegreatestimpact(whilestillhavingincorrectsigns).Interestingly,AIC PublicChoiceTable6Orderedprobitestimates:randomeffectsandAR(1)errorsRandomeffectsAR(1)errorsModel5Model6Model7Model8πG3.273(0.508)∗∗∗3.085(0.495)∗∗∗4.492(1.085)∗∗∗4.627(1.082)∗∗∗GDPG−0.313(0.121)∗∗∗−0.234(0.123)∗−0.901(0.307)∗∗∗−0.888(0.292)∗∗∗Chair0.326(0.163)∗−0.015(0.135)−0.693(0.400)∗−0.629(0.353)∗AcadD–−0.006(0.010)–0.003(0.028)BankD–0.048(0.010)∗∗∗–0.092(0.031)∗∗∗FinD–−0.016(0.013)–0.007(0.027)GovD–0.029(0.010)∗∗∗–0.035(0.030)IndD–0.147(0.024)∗∗∗–0.217(0.067)∗∗∗NGOD–−0.336(0.050)∗∗∗–−0.535(0.133)∗∗∗γ1−1.699(0.149)∗∗∗−1.745(0.127)∗∗∗−3.210(0.530)∗∗∗−3.297(0.553)∗∗∗γ21.824(0.151)∗∗∗1.881(0.132)∗∗∗2.426(0.390)∗∗∗2.657(0.411)∗∗∗ρ0.352(0.057)∗∗∗0.426(0.074)∗∗∗0.868(0.035)∗∗∗0.807(0.046)∗∗∗SummarystatisticsAIC951.273936.480864.268853.755BIC980.984995.901898.931918.128Standarderrorsinparentheses(·).Noofobs=1045∗∗∗Denotestwo-tailedsignificanceatthe1%level∗∗Denotestwo-tailedsignificanceatthe5%level∗Denotestwo-tailedsignificanceatthe10%leveltimatesrangefromnegativeandinsignificant(Models2and3),topositiveandsignificant(Model4).Intermsofmodelselection,theinformationcriteriadonotselectanymodelunani-mously:accordingtotheBIC,Model2performsbest,whileModel4ispreferredbyAIC.Insomerespects,thisresultisunsurprising:duetoitsasymptoticconsistencyanditsheavypenaltyoncomplexity,BICtypicallyselectsmoreparsimoniousspecifications.Conversely,AICoftenchooseslessparsimoniousspecificationsascomplexityisnotsoheavilypenal-ized,especiallyforsmallormoderatesamplesizes.However,itisnotablethatalternativegoodnessoffitmeasuresnotreportedhere(McFadden’sR2andAdjustedR2,Cox-SnellMaximumLikelihoodR2,Cragg-UhlerR2andMcKelveyandZavoina’sR2)allidentifyModel4ashavingthebestexplanatorypower.5ConclusionAttheoutsetofthispaper,wesetouttoexplainthetypeandfrequencyofdissentingvotescastbyMPCmembers,byappealingtocareerbackgrounds,thechannelofappointment,andunobservedheterogeneity.ThefindingsoftheFOMCliterature,whichestablishesaandBICidentifytheAR(1)model,whichconfirmsthepresenceofhighdegreefirstorderserialcorrelation,assuperiortotheREspecifications. PublicChoiceprominentroleforsuchfactorsindeterminingvotingbehaviorraisedthehereuntoneglectedquestionofwhethervotesonthepolicyrateatmonthlyBankofEnglandMPCmeetingsweresimilarlyinfluenced.Tomotivatetheissue,werationalizedwhyMPCmembersmightbeincentivizedtovoteindifferentways,byproposingthattheinternal-externaldistinctionwastoosimplistic,andpayingparticularattentiontotheinstitutionalfeaturesgoverningUKmonetarypolicy.Whileitwasfoundthatvotingpatternsexhibitconsiderableheterogeneityacrossallmembers,theeconometricevidenceinitiallypresentedsomethingofa‘careerbackgroundpuzzle’—unliketheFOMCliterature,careerexperienceplaysaveryweakroleindeter-miningamember’sdecisiontodissent;moreover,wherecareerbackgroundsaresignificant,theyareoftencounter-intuitivelysigned.AcredibleresolutiontothispuzzleisfoundbyappealingtohowthedemocraticdeficitassociatedwithUKmonetarypolicyismitigated:constituentgrouppressuremaymoreplausiblymanifestitselfthroughthechoiceofthein-flationtargetitself,asopposedtotheindividualschosentosetpolicy.WealsoproposedthatthechannelofappointmentfortheMPCisconsiderablylesspoliticizedthanforitsUScounterpart,conjecturingthatitwouldplayaverylimitedroleindeterminingvotingbehavior.Ourempiricalresultsbroadlysupportthisview,andservetohighlighttherela-tionshipbetweentheUKpoliticalandparliamentarysystem,andtheassociatedmonetarypolicyframework.Again,centraltothisrelationshipisthemitigationoftheso-calledde-mocraticdeficitofmonetarypolicymaking,whichisachievedthroughtheinstitutionofanexplicitinflationtarget.Wedidfind,however,thattheinternal-externaldistinctionplaysanimportantrole,whichmaybeattributabletocareerconcernsandorganizationalfactors.Whiletheroleofdifferentchairmenwasfoundtobeambiguous,thedeviationoftheMPC’sinflationforecastfromtargetwasfoundtoheavilyinfluenceexternaldissents,butnotthoseassociatedwithinternals,whoseparameterestimateswerestatisticallynodifferenttozero.Analternativeapproachwastoassumethatthedeterminantsofdissentarebestcapturedbymember-specificfixed-effects.Econometricestimatesprovidestrongencouragementfortheuseofsuchaninnovation,andsuggestthattheMPCisindeedindividualistic(Blinder2007).IntegraltothisapproachistheviewthatalthoughdemarcatingbetweendifferentMPCmembertypesduringestimationmaybefruitful,suchastrategyfailstocapturethetrueextentofmemberheterogeneity.Finally,futureworkinthisareamightsystematicallycomparethedeterminantsofdissentvotingindifferentmonetaryinstitutions,suchastheUSFederalReserve,SwedishRiksbankandBankofJapan.OurfindingsmayalsohaveimportantimplicationsfortheinstitutionaldesignofpolicyinstitutionssuchastheUSFederalReserve:ifone’saimistoreducetheroleofpoliticsandpartisanshipinshapingvotesontheFOMC—andtherebyUSmonetarypolicydecisions—apotentialmechanismfordoingthismaybetoadoptatransparentandexplicitinflationtargetingregime,asiscurrentpracticeintheUK.Thismay,however,simplyrelocatetheproblemofgovernmentalandconstituentgroupinfluencetothechoiceoftargetitself.Moreover,asmorecentralbanksdelegatemonetarypolicytocommittees,andplacethevotingrecordinthepublicdomain,investigationofthisissuecanonlyheightenourunderstandingofmonetarypolicy,andoftheindividualsandinstitutionsthatshapeit.AcknowledgementsSpenceracknowledgesfinancialsupportfromtheEconomicandSocialResearchCouncil(ESRCPostgraduateStudentshipR42200134224).Thispaperisasubstantiallyre-workedversionof“TheDissentVotingBehaviourofBankofEnglandMPCMembers”,UniversityofSurreyDepartmentofEconomicsWorkingPaperNo.03/06.Authorsarelistedalphabetically.WewouldliketothankAliChoud-hary,AndyDickerson,StephenDrinkwater,HuwEdwards,CarréEmmanuel,AlessandroFlamini,AnthonyGlass,KarligashKenjegalieva,AkiraMurayama,TomWeyman-Jones,twoanonymousreferees,theEditor-in-Chiefofthisjournal,andseminarparticipantsattheDepartmentofEconomics,UniversityofSheffieldforhelpfulcommentsandsuggestions.TheauthorsalsoextendspecialthankstoNicolettaBatiniandWillemBuiterforadditionalclarificationoftheMPCresourcesdisputein1999.Theusualdisclaimerapplies. PublicChoiceAppendixA:ResourcecontrolIn1999adisputeoverMPCresourcesemerged,whichsubsequentlybecamethesubjectofaHouseofCommonsTreasurySelectCommitteereport.Thereport,entitledReportonResearchAssistanceforMonetaryPolicyCommitteeMembers(TreasurySelectCommit-tee1999),centeredontheexperienceofagroupofexternalcommitteemembers(WillemBuiter,CharlesGoodhart,DeAnneJulius,andSushilWadhwani)whowererefusedresearchresourcesthatwererequestedfromtheexecutiveleadershipattheBank:theGovernor(Ed-dieGeorge)andtwoDeputyGovernors(MervynKingandDavidClementi)wereallunitedintheiroppositiontotherequest.36Onbecomingpublic,thedisputereceivedcoverageinthefinancialpressandnewsmedia.Newspapersreportedthatsomeexternalsbelievedhigherproductivitywaseasinglong-terminflationarypressure,butfelttheyhadinsufficientre-sourcestosupporttheirview.37WhilsttheGovernorpubliclydescribedthedisputeasa“storminateacup”,38theTreasurySelectCommitteereportconcludedthatthiswasnotthecase,andwarnedtheBanktobettermanagefuturedisagreements.Althoughthedisputewasultimatelyresolved—suchthatexternalmembersrequiringextraresourcesarenowallocatedonepost-graduateandonegraduateeconomisteach—theprocessbywhichitwassettledisofparticularinterest.Priortobecomingpublic,thematterwastakentotheCourtoftheBankofEnglandforarbitration.Atthetimeofthedispute,theCourtwascomprisedoftheGovernor,thetwoDeputyGovernors,andsixteenindependentnon-executiveDirectors,withthelattersixteenmemberscomprisingasub-committeeinitsownrightcalledNedCo.Althoughneitherbodyplaysaroleintheformulationofmonetarypolicy,boththeCourtandNedcoenjoyimportantroleswithrespecttothegovernanceoftheBank.WhereastheCourthasoverallresponsibilityformanagementoftheBank’saffairs(thisincludesdeterminingBankstrategyandensuringthemostefficientuseoftheBank’sresources),NedCoplaysanimportantmonitoringrole,whichincludesensuringthattheMPC’sprocessesandproceduresworkeffectively.TheroleoftheCourtinattemptingtosettlethedisputehasbeenheavilycriticizedbyWillemBuiter:“DuringmythreeyearsontheMPC,theexternalMPCmembersoncetriedtoinvolvetheCourtinamatterthatoughttohavebeenofinteresttotheCourt,givenitsroleinthegovernanceoftheinstitution.ThiswasduringtheconflictbetweentheexternalmembersoftheMPCandtheBank’sexecutiveleadership,aboutthecreationofadedicatedresearchunittosupporttheexternalMPCmembers,independentoftheBank’sExecutive.TheCourtwasutterlyineffectiveandresolvednothing.”(Buiter2008)TheTreasuryCommitteesinglesoutNedCoforsimilarcriticism,assertingthat:36Ine-mailcorrespondencewiththeauthors,WillemBuiterindicatedthatwhileCharlesGoodhartwantedtopreventaconfrontation,forothermembersthiswasnotthecase.Forinstance,BuiterhimselfclaimstohavehadstrongargumentswiththeGovernorandDavidClementiabouttheissue.37See“BanksaysMPCresearchrowhasbeensettled”,TheGuardian,WednesdayNovember24,1999;alsoseefurtherarticlesinTheIndependentonOctober30th1999,November22th1999andNovember24th1999.Whilewearenotinapositiontosubstantiatealloftheclaimsmadeinthenewspaperreportscitedhere(i.e.,mostoftheeventsdescribedthereintookplacebehindcloseddoors,andmayhavebeenbasedonthetestimonyofanonymoussourcesfromwithintheBank),itisclearfromtheTreasuryCommitteereportandpersonalstatementsmadebyMPCmembers,thatthereportsarenotwithoutsomefoundation.38SelectCommitteeonTreasury,MinutesofEvidence,ExaminationofWitnesses,Question81,Tuesday23rdNovember1999. PublicChoice“Theissuewasonlyfullyresolvedafterithadbeenmadepublicandbecameafocusofoneofourpublichearings.WebelievethattheNon-ExecutiveDirectors[oftheCourt],needtobemuchmorepro-activeinensuringthattheproceduresoftheMPCoperatefairlywithrespecttobothinternalandexternalmembers.”(SelectCommit-teeonTreasury,NinthReport(paragraph40),prepared28thMarch2001.Wordsinsquarebracketsaddedbytheauthors).Althoughthismatterisnotpursuedanyfurtherinthispaper,wesuggestthattheaspectsofresourcecontrolcoveredhere,andtheirimplicationsforvoting,areclearlydeservingoffurtherattentionintheliterature.Forfurtherdetailsrelatingtothisaffair,andtheroleoftheTreasurySelectCommitteeinrelationtotheMPCandUKmonetarypolicyformulation,thereaderisreferredtoTreasurySelectCommittee(2007a)andTreasurySelectCommittee(2007b).AppendixB:TheHavrilesky-Schweitzer(H-S)modelInsettingouttheformalmodel,weenvisageaMPCcomposedofgmembers,eachofwhommayhaveamassedcareerexperiences,fordifferentdurations,injdifferentfields.TheseexperiencesarereferredtobyH-Sasamember’scareercharacteristics.Thepracticalanalogueofthesejcharacteristicswouldbeexperiencesindifferentsectorsorareasoftheeconomy,suchasprivateindustryandfinance.ThesemeasuresaredevelopedinSect.4.1.1.Formemberg,denoteherjthcareercharacteristicasXgj,suchthatXjrepresentstheMPC’smeanforthatcharacteristic.So-called‘careerproximity’tocentralgovernmentisincreasinginXgj−XjsuchthatXgj−Xj>0(<0)promotesdissentsonthesideofmon-etaryease(tightness).AsXgj−Xj>0(<0)becomeslarger(smaller),sotoodoesthepropensitytodissentonthesideofease(tightness).However,giventherearejcharacteris-tics,theextenttowhichagivenMPCmemberdissentsisultimatelyafunctionofhoweachcharacteristicisweighted.Wearenowinapositiontowriteanexpressionformemberg’sutility,namelyUg(Dg)=U(Dg|Xgj−Xj,j=1,2,...,N).(B.1)Itisfurtherassumedthat(B.1)ischaracterizedbyauniqueglobalmaximumthatdefinestheoptimalnumberofdissentingvotes,Dg,suchthat∂U<0,(B.2)∂(Dg−Dg)∂2U<0.(B.3)∂(Dg−Dg)2In(B.1),theutilityachievedbymembergisafunctionofthenumberofdissentingvotescast,thedirectionandnumberofwhichisconditionedbycareerproximityparameters,Xgj−Xj,j=1,2,...,N.Itturnsoutthattheactualnumberofdissentscastbymemberg,Dg,willnotnecessarilyequalthenumberofdissentswhichmaximizeutility.Thisisbecausemembersalsoexperiencedisutility,anexpressionforwhichisgivenbyVg(Dg)=V(Dg|Xgj−Xj,j=1,2,...,N)(B.4) PublicChoicewhichhasauniqueglobalminimumatDg=0.Here,theassumedpropertiesofVg(Dg)∂V>0,(B.5)∂Dg∂2V>0(B.6)∂D2gimplythatasthenumberofdissentsmovesfurtherawayfromzero,thedisutilityfeltbymembergincreasesatanincreasingrate.Itiseasilyshownthatwhenthemarginalutilityofincreasingdissentequalsthemar-ginaldisutilityofincreasingdissent,memberg’snetutilitywillsatisfyanunconstrainedmaximumwhere:∂U∂V=.(B.7)∂Dg∂DgPutanotherway,marginalnetutilitymustbezero.Togleanthenormativeimplicationsofthemodel,H-Sconsidertheconditionsrequiredtoensureamonotonictransformationfromtheweightedcareercharacteristicdifferences,Xgj−Xj,totheactualnumberofdissents,Dg.Duetothenatureofthefirstorderconditionsforutilityanddisutilityin(B.1)and(B.4),theactualnumberofdissentsisnotguaranteedtomapmonotonicallyontocareercharacteristicdifferences.Ensuringsuchatransformationrequirestherestrictionthatthemarginalnetutilityofthejthmemberincreasingdissenttowardsherglobaloptimumisstrictlylessthanthatpertainingtothe(j+kth)member:asHavrileskyandSchweitzerstate,thisholdstheimplicationthat“amemberwithmarginallystrongermoralconvictionsinfavorofdissentingcannotbemarginallymoreeasilycowedbygroup...disapproval”.Thiscanbeformallyprovedasfollows.Inordertopartiallydifferentiate,expressthenetwelfareofmembergasWg=Ug−Vg=Wi(Dg,Xgj−Xj,j=1,2,...,N).Thefirstorderconditionforamaximumisthen∂Wg(Dg,Xgj−Xj,j=1,2,...,N)=0(B.8)∂DgwhichresultsinanoptimalchoiceD=D∗(X−X,j=1,2,...,N).From(B.3)andiigjj(B.6)∂2Wg<0(B.9)∂D2gandthesecond-orderconditionforanoptimumissatisfied.PartiallydifferentiatingagainwithrespecttothedeviationofcharacteristicjdenotedbyXˆgj≡Xgj−Xjyields∂2W∂D∗∂2Wggg+=0.(B.10)∂D2∂Xˆ∂Xˆ∂DggjgjgHencewehave∂2Wg∂D∗g∂Xˆgj∂Dg=−.(B.11)∂2Wg∂Xˆgj∂D2g PublicChoiceTable7Testsofdifferencesbetweengroups,June1997–May2007H0:InternalMembersP=InternalMembersNP=ExternalMembersPχ2(2)p-valueRejectH0?Assentvs.easevs.tighten70.820.000YesEasevs.tighten38.650.000YesAssentvs.dissent37.220.000YesH0:InternalMembersP=InternalMembersNPχ2(1)p-valueRejectH0?Assentvs.easevs.tighten3.610.165NoEasevs.tighten2.630.105NoAssentvs.dissent0.900.343NoH0:InternalMembersP+NP=ExternalMembersPχ2(1)p-valueRejectH0?Assentvs.easevs.tighten69.970.000YesEasevs.tighten36.990.000YesAssentvs.dissent34.650.000YesH0:InternalMembersP=ExternalMembersPχ2(1)p-valueRejectH0?Assentvs.easevs.tighten60.330.000YesEasevs.tighten32.240.000YesAssentvs.dissent29.870.000YesH0:InternalMembersNP=ExternalMembersPχ2(1)p-valueRejectH0?Assentvs.easevs.tighten31.290.000YesEasevs.tighten15.070.000YesAssentvs.dissent33.730.000YesH0:Politicalappointments=Non-politicalappointmentsχ2(1)p-valueRejectH0?Assentvs.easevs.tighten10.550.005YesEasevs.tighten7.220.007YesAssentvs.dissent16.730.000YesPPoliticalappointmentNPNon-politicalappointmentFiguresinparentheses(·)denotedegreesoffreedom PublicChoice∂D∗gFrom(B.11)and(B.9)wethereforehavethat>0andthereisamonotonictransforma-∂Xˆgjtionfromtheeachofthecareercharacteristicdifferences,Xgj−Xj,totheactualnumberof∂2Wg∂∂WgdissentsDg,iff=()>0.Thatis,themarginalnetutilityofthejthmem-∂Xˆgj∂Dg∂Xˆgj∂Dgberincreasingdissenttowardsherglobaloptimumisstrictlylessthanthatpertainingtothe(j+kth)member.Thisprovestheresult.AppendixC:Testingforgroupdifferencesbasedonmembers’internal-externalstatusandthechannelofappointmentBuildingontheanalysisinSect.3,Chi-squaredtestswereextensivelyusedtotestfordifferencesinmembers’votingbehavior,basedonthecriteriausedtoconstructthethreesub-groupsexaminedinTable1.ResultsaredisplayedinTable7.Specifically,differencesassociatedwiththreealternative,buthighlyrelatedcriteriawereexamined:first,thatthepatternofvotesforassent,easeandtightnesswerenodifferentbetweengroups;second,thatthepatternofvotesforeaseandtightnessdissentswerenotsignificantlydifferent;andthird,thatthepatternofvotesforassentinganddissentingvotespersedidnotsignificantlydiffer.Weinitiallyproceededwithajointtestofthenullhypothesisthatthethreesub-groupsintroducedinTable1werenotsignificantlydifferentintheirvotingbehavior(H0:InternalMembersP=InternalMembersNP=ExternalMembersP).Whilethenullofnodifferencewasrejectedatthe1%levelforeachcriterion,subsequenttestsbetweendifferentpairsofsub-groupsrevealedamorecomplexpicture.Mostinterestingly,politicalandnon-politicallyappointedinternalmembers(InternalPvs.InternalNP)didnotsignificantlydifferintheirbehavior(p>0.1ineverycase,indicatingacceptanceofthenullofnodifferencebetweenthesegroups).Therewere,however,statisticallysignificantdifferencesbetweeninternalmembers—whetherpoliticallyappointed,not-politicallyappointed,orcollectively—andexternalmem-bers.Thisfindingprovidessupportforthenotionthatamember’sinternal-externalstatushasasignificantbearingonvotingbehavior.Moreover,theresultsatthefootofTable7—whichindicatehighlysignificantdifferencesbetweenpoliticalandnon-politicalappointments—shouldbeinterpretedwithcaution:suchdifferencesarearguablydrivenbyinternal-externaldistinction.Indeed,whilepoliticallyappointedinternalmembersaregenerallymorelikelytoassentandcastthemajorityofdissentingvotesonthesideoftightness—summingacrossvotesforthesetwogroupsresultedinagroupwhichretainsthedissentvotingcharacteris-Pticsofexternalmembers(notethatPoliticalappointments=InternalMembers+ExternalMembersP,whereasNon-politicalappointments=InternalMembersNPonly).Asanexam-ple,thesheernumberofdissentingvotescastbyexternalmembersledtoanoverwhelmingtendencyforpoliticallyappointedmemberstodissentonthesideofmonetaryease.ReferencesAdolph,C.(2005).Threesimpletestsofcareerinfluencesonmonetarypolicy.Paperpresentedatthean-nualmeetingoftheAmericanPoliticalScienceAssociation,MarriottWardmanPark,OmniShoreham,WashingtonHilton,Washington,DC,September1st,2005.Alesina,A.,&Roubini,N.(1992).PoliticalcyclesinOECDeconomies.ReviewofEconomicStudies,59(4),663–688.BankofEngland.(1999).Thetransmissionmechanismofmonetarypolicy.BankofEngland.Availablefromwww.bankofengland.co.uk. PublicChoiceBelongia,M.T.(2007).Opaqueratherthantransparent:whythepubliccannotmonitormonetarypolicy.PublicChoice,133,259–267.Besley,T.,Meads,N.,&Surico,P.(2008).Insidersversusoutsidersinmonetarypolicymaking.AmericanEconomicReview(PapersandProceedings),98(2),218–223.Bhattacharjee,A.,&Holly,S.(2005).Inflationtargeting,committeedecisionmakinganduncertainty:thecaseoftheBankofEngland’sMPC.CambridgeWorkingPapersinEconomics05/30,UniversityofCambridge.Blanchflower,D.(2006).ReflectionsonmyfirstfourvotesontheMPC.SpeechgivenattheBankofEngland’sAgencyforWales,Wednesday27thSeptember2006.Blinder,A.(2007).Monetarypolicybycommittee:whyandhow?EuropeanJournalofPoliticalEconomy,23(1),106–123.Blinder,A.,&Wyplosz,C.(2005).Centralbanktalk:committeestructureandcommunicationpolicy.Con-ferencepaper,CentralBankCommunication,attheASSAmeetings,Philadelphia.Brooks,R.,Harris,M.,&Spencer,C.(2007).Aninflatedorderedprobitmodelofmonetarypolicy:evidencefromMPCvotingdata.MPRAPaper8509,UniversityLibraryofMunich,Germany.Buiter,W.(1999).Answerstothequestionnaire‘BankofEnglandcommission;issuesforconsideration’.December17th1999.Buiter,W.(2008).AfinancialstabilitycommitteefortheBankofEngland?http://blogs.ft.com/maverecon/2008/06/a-financial-stability-committee-for-the-bank-of-england/,June11,2008.Chappell,H.W.,&McGregor,R.R.(2000).AlonghistoryofFOMCvotingbehavior.SouthernEconomicJournal,66(4),906–922.Chappell,H.W.Jr.,Havrilesky,T.M.,&McGregor,R.R.(1993).Partisanmonetarypolicies:Presidentialinfluencethroughthepowerofappointment.QuarterlyJournalofEconomics,108(1),185–218.Chappell,H.W.Jr.,Havrilesky,T.M.,&McGregor,R.R.(1995).Policymakers,institutions,andcentralbankdecisions.JournalofEconomicsandBusiness,47(2),113–136.Chappell,H.W.Jr.,McGregor,R.R.,&Vermilyea,T.(2004).Majorityrule,consensusbuilding,andthepoweroftheChairman.JournalofMoney,Credit,andBanking,36(3),408–422.Chappell,H.W.Jr.,McGregor,R.R.,&Vermilyea,T.(2005).Committeedecisionsonmonetarypolicy.Cambridge:MITPress.Cobham,D.(2000).RestoringtrustattheTreasury.FinancialTimes.10thAugust2000.DuPlessis,S.A.(2005).Thedemocraticdeficitandinflationtargeting.SouthAfricanJournalofEconomics,73(1),93–104.Gerlach-Kristen,P.(2003).InsidersandoutsidersattheBankofEngland.CentralBanking,14(1),96–102.Gerlach-Kristen,P.(2009).OutsidersattheBankofEngland’sMPC.JournalofMoney,CreditandBanking,41(6),1099–1115.Gildea,J.A.(1990).ExplainingFOMCmembers’votes.InT.Mayer(Ed.),ThepoliticaleconomyofAmeri-canmonetarypolicy(pp.211–228).Cambridge:CambridgeUniversityPress.Goodhart,C.A.E.(2001).Theinflationforecast.NationalInstituteEconomicReview,175,59–66.Goodhart,C.A.E.(2005).TheMonetaryPolicyCommittee’sreactionfunction:anexerciseinestimation.TopicsinMacroeconomics5(1),Article18.Goodhart,C.A.E.(2006).Letter.InThestateofmonetarypolicy.SelectCommitteeonEconomicAffairs,HouseofLords,FirstReport,December18th.Goodhart,C.A.E.,&Meade,E.(2003).Centralbanksandsupremecourts.Mimeo,LondonSchoolofEconomics.Greene,W.(2004).Thebehaviourofthemaximumlikelihoodestimatoroflimiteddependentvariablemodelsinthepresenceoffixedeffects.EconometricsJournal,7(1),98–119.Grilli,V.,Masciandaro,D.,&Tabellini,G.(1991).Politicalandmonetaryinstitutionsandpublicfinancialpoliciesintheindustrialcountries.EconomicPolicy,6,341–392.Harris,M.N.,&Spencer,C.(2009).ThepolicychoicesandreactionfunctionsofBankofEnglandMPCmembers.SouthernEconomicJournal,76(2),482–499.Havrilesky,T.,&Gildea,J.A.(1992).ReliableandunreliablepartisanappointeestotheBoardofGovernors.PublicChoice,73(4),397–417.Havrilesky,T.,&Schweitzer,R.(1990).AtheoryofFOMCdissentvotingwithevidencefromthetimeseries.InT.Mayer(Ed.),ThepoliticaleconomyofAmericanmonetarypolicy(pp.197–210).Cambridge:CambridgeUniversityPress.Heckman,J.(1981).Statisticalmodelsfordiscretepaneldata.InC.Manski,&D.McFadden(Eds.),Struc-turalanalysisofdiscretedatawitheconometricapplications.Cambridge:MITPress.Heiss,F.(2007).LimiteddependentpaneldatamodelswithAR(1)errorcomponentsinStata.Mimeo,Uni-versityofMunich,DepartmentofEconomics,December2007.Howard,M.(2002).SpeechtotheInstituteforPublicPolicyResearch.MondayMarch11,2002. PublicChoiceKeech,W.R.,&Morris,I.L.(1997).Appointments,presidentialpower,andtheFederalReserve.JournalofMacroeconomics,19(2),253–267.Krause,G.A.(1994).FederalReservepolicydecisionmaking:politicalandbureuacraticinfluences.Ameri-canJournalofPoliticalScience,38(1),124–144.Krause,G.A.(1996).AgentheterogeneityandconsensualdecisionmakingontheFederalOpenMarketCommittee.PublicChoice,88(1),83–101.McCallum,B.T.(1995).Twofallaciesconcerningcentralbankindependence.AmericanEconomicReview(PapersandProceedings),85,207–211.McGregor,R.R.(2007).FederalReservetransparency:themorethingschange,themoretheystaythesame?PublicChoice,133,269–273.Meade,E.E.,&Stasavage,D.(2008).Publicityofdebateandtheincentivetodissent:evidencefromtheUSFederalReserve.EconomicJournal,118(525),695–717.Morris,I.L.(2002).Congress,thePresident,andtheFederalReserve:thepoliticsofAmericanmonetarypolicy-making.AnnArbor:UniversityofMichiganPress.Mueller,D.C.(2005).ConstitutionalpoliticaleconomyintheEuropeanUnion.InW.F.ShughartII,&R.D.Tollison(Eds.),Policychangesandpoliticalresponses:publicchoiceperspectivesonthepost9/11world(pp.57–73).Berlin:Springer.Nakahara,N.(2001).SpeechgivenbyNobuyukiNakahara,memberofthePolicyBoardoftheBankofJapan,attheCapitalMarketsResearchInstituteinTokyoonDecember11,2001.Neyman,J.,&Scott,E.(1948).Consistentestimatesbasedonpartiallyconsistentobservations.Economet-rica,39,359–382.Riboni,A.,&Ruge-Murcia,F.J.(2008).Preferenceheterogeneityinmonetarypolicycommittees.Interna-tionalJournalofCentralBanking,4(1),213–233.Santoni,G.J.(1984).Aprivatecentralbank:someoldeEnglishlessons.FederalReserveBankofSt.LouisReview,66,12–22.Shughart,W.F.II,&Tollison,R.D.(1983).PreliminaryevidenceontheuseofinputsbytheFederalReservesystem.AmericanEconomicReview,73(3),291–304.Taylor,J.(1993).Discretionversuspolicyrulesinpractice.Carnegie-RochesterConferenceSeriesonPublicPolicy,38,195–214.TreasurySelectCommittee(1999).ResearchassistanceforMonetaryPolicyCommitteemembers.TreasuryCommitteeReportHC43290,9December1999.TreasurySelectCommittee(2007a).TheMonetaryPolicyCommitteeoftheBankofEngland:tenyearson,Vol.I.TreasuryCommitteeReportHC299-I,publishedon18September2007.TreasurySelectCommittee(2007b).TheMonetaryPolicyCommitteeoftheBankofEngland:tenyearson,Vol.II.TreasuryCommitteeReportHC299-II,publishedon19February2007.Tucker,P.(2008).Remarkson‘Makingmonetarypolicybycommittee’.DiscussantremarksbyMrPaulTucker,ExecutiveDirectorandMemberoftheMonetaryPolicyCommitteeoftheBankofEngland,on“MakingMonetaryPolicybyCommittee”byAlanBlinder,attheBankofCanadaConference“Interna-tionalExperiencewiththeconductofMonetaryPolicyUnderInflationTargeting”,Ottawa,22–23July2008.Vaubel,R.(1997).Thebureaucraticandpartisanbehaviorofindependentcentralbanks:Germanandinter-nationalevidence.EuropeanJournalofPoliticalEconomy,13(2),201–224.Wadhwani,S.B.(2002).TheMPC:somefurtherchallenges.NationalInstituteofEconomicandSocialResearchGovernors’Seminar.NationalInstituteofEconomicandSocialResearch.Wooldridge,J.(2002).Econometricanalysisofcrosssectionandpaneldata.Cambridge:MITPress.