Speech

Expectations TheTwenty-SecondHenryThorntonLecture DepartmentofBankingandFinance CityUniversityBusinessSchool London,England, November28,2000 PublishedintheFederalReserveBankofSt.LouisReview,March/April2001,83(2),pp.1-10 It is a great honor for me to be here tonight plete,thanthoseonwhichhehasfoundit to present the Twenty-Second Henry necessarytoremark.Futureinquiriesmay Thorntonlecture.Inpreparingthislecture, possiblypursue,withadvantage,somepar- it has been fascinating to read parts of ticulartopicsonwhichhehasfeltacertain degreeofdistrust. Thornton’s great book, An Enquiry into the Itmaynotbeirrelevantorimproperto Nature and Effects of the Paper Credit of Great observe,thatthepresentworkhasbeenwritten Britain,publishedin1802,andF.A.Hayek’sintro- byapersonwhosesituationinlifehassupplied duction to the 1962 reprint of Paper Credit. I informationonseveralofthetopicsunder recall reading Thornton years ago, but remember discussion…1 littleofit.Rereadinghimtoday,Icertainlyappre- Asonenowpursuingthe“practicalbusiness” ciate Thornton’s insights to a far greater extent ofcentralbanking,IcanrelateeasilytoThornton’s than I did when I first read his book. I have also disclaimer.IwouldjustaddthatIvaluethecon- found it instructive to read several previous versationsonthesesubjectswithmycolleagues Thornton lectures. I’ll refer to these lectures and attheSt.LouisFed,especiallyRobertH.Rasche, to Thornton himself on several occasions this butthatIamresponsiblefortheviewsexpressed. evening. Theseviewsdonotnecessarilyreflectofficial ItisstandardpracticeforFederalReserve positionsoftheFederalReserveSystem. officials,withtheexceptionoftheChairman,to Almosteveryaspectofhumanbehavioris begineverypublicpresentationwithadisclaimer. conditionedbyexpectations.Indeed,adistin- Thorntonhimselfwroteadisclaimerintheintro- guishingfeatureofhumansamongalllivingthings ductionofhisbook,andIwilladopthisdisclaimer isthathumans,toanunmatcheddegree,calculate asmyownforthislecture.Thorntonwrote: behaviorinlightofpossiblefutureoutcomes.I cannotdiscussthewholeofhumanbehaviorin That[thiswork’s]leadingdoctrinesarejust, onelecture,orinonelifetime.Eventhetopicof thewriterfeelsaconfidentpersuasion.That expectationsinamacroeconomicscontextis itmayhaveimperfections,andsome,perhaps, overlybroad;Iwillconcentrateratherunsystem- whichgreatercareonhispartmighthavecor- aticallyonaspectsofthistopicthatareofspecial rected,hecannotdoubt.Buthetrusts,thata manwhoismuchoccupiedonthepractical interesttomebecauseofmycurrentresponsibil- businessoflife,willbeexcusedbythepublic, ities.Iwilldiscussissuesfromtheperspective ifheshouldpresenttothematreatiseless ofcentralbankingproblems,butmuchofwhatI elaborate,and,inmanyrespects,moreincom- sayappliestootherareasofgovernmentpolicy. 1 F.A.vonHayek(1962,p.69):firstpublishedin1939,thisbookcontainsHayek’sintroduction,Thornton’sAnEnquiryintotheNatureand EffectsofthePaperCreditofGreatBritain,twoofThornton’sspeechesintheHouseofCommonsin1811,andothermaterials. 1 MISCELLANEOUS By“rationalexpectations”Imeanthatmarket bankcredibility.Thethirdisinflationaryexpec- outcomeshavecharacteristicsasifeconomic tations.Thefourthistheextenttowhichthe agentsareactingonthebasisofthecorrectmodel marketcanpredictcentralbankactions.Iwill ofhowtheworldworksandthattheyuseall connectthesetopicstoproducewhatIhopewill availableinformationindecidingontheiractions. beacoherentaccountofcertainexpectations Thatinformationincludesprobablefuturemone- issuesfromtheperspectiveofapracticingcentral tarypolicyactionsand,moregenerally,how banker. monetarypolicyactionsarelikelytodependon Idonotdoubtthatexpectationsaresometimes variouspossiblestatesoftheeconomy.Expecta- nonrational.Mymaintheme,however,isthat tionsmaybenonrationalinaninfinityofways.2 wecentralbankersshouldnotbesmuginassess- Almosteveryeconomistisfamiliarwiththecolor- ingourpresumablysuperiorunderstandingof fullanguageKeynesusedtodescribehisview whatexpectationsoughttoprevail.Weneedto onhowsecurityvaluesweredetermined.Inone reflectonourpossibleroleincreatingandsustain- ofhismoresuccinctstatements,Keynes(1936, ingexpectationsthatweregardasnonrational, p.154)saidthat,“Aconventionalvaluationwhich andonthepossibilitiesforpursuingpoliciesthat isestablishedastheoutcomeofthemasspsychol- yieldmarketoutcomesclosertothosereflecting ogyofalargenumberofignorantindividualsis rationalexpectations. liabletochangeviolentlyastheresultofasudden fluctuationofopinionduetofactorswhichdo notreallymakemuchdifferencetotheprospec- WHAT DO PANICS TELL US tiveyield.”Keynesandmanyothershaveviewed ABOUT EXPECTATIONS? expectationsasbeingdrivenbyemotionand effortstoridemarkettrendswithoutregardto Suddenandunpredictablechangesinmarket underlyingvalues.Popularcommentaryonbond, sentimentcreateproblemsforallsortsofbusi- stock,commodity,andforeignexchangemarkets nesses.Hayek,inhisintroductiontoThornton’s oftenfocusesonpresumedpatternsinthedata, PaperCredit,quotesfromacontemporaryaccount suchasresistanceandsupportlevels,thatmake ofanincidentThorntonhadtofacein1810. notheoreticalsenseandarecompletelyunsup- [Thornton]wasonhisroadwithhisfamilyto portedbycarefulempiricalinvestigation. Scotland.Itwasatimeofseverepressureupon Ineconometricmodels,economistshaveoften banksandtradinginterests…Thebankinwhich usedadaptiveexpectations,whicharesimple, Mr.Thorntonwasapartnerfeltthepressure, andsimple-minded,extrapolationsofthepast. andfeltitseverely,justaftertheirmostable Adaptiveexpectationsaretheantithesisofthe partnerhadleftLondonfortheNorth.Had emotionalprocessKeynesemphasized.Adaptive Mr.Thorntonknownwhatwasimpending,he expectations,asaveragesofrecentobservations, wouldnothaveabsentedhimself.Thenews reachedhimonhisroutetoScotland,and changerelativelysmoothlyandcontinuously. causedhimsomeembarrassment.Toreturn Theyareunaffectedbynewsitemsperse;ifnews fromajourneyundertakenandgenerally movesthemarket,theadaptiveexpectationincor- known,wouldhavespreadrumorswhich poratesonlyafractionoftheunexpectedprice mighthavebroughtontheverycrisisthatwas adjustmentintoexpectedfutureprices. tobefeared.Thiscourse,therefore,couldnot Iwilltakeupfourtopics.Thefirstiswhat bethoughtof.Hedecidedtocontinuehisjour- wecanlearnaboutexpectationsfrombanking ney,butheopenedhimselfinconfidenceto panicsand,moregenerally,fromsharpdistur- onevaluedfriend,andstatedhiswishthat bancesinfinancialmarkets.Thesecondiscentral somethousandsofpoundsmightbeplacedat 2 Iusetheword“nonrational”ratherthan“irrational”becausethelattersometimescarriesconnotationsthatIdonotintend.Expectations maydepartfromfullrationalitywithoutbeing“crazy,”“silly,”“emotional,”or“stupid.” 2 Expectations thedisposalofhispartnersinthebank.No Whatisthenatureofexpectationsinapanic? soonerwasthehintgiventhanitwasmetby Isthedistinctionbetweenrationalandnonrational amplesupport.Fundspouredinfromall expectationshelpfulhere?Ithinkwemustlook quarters—Wilberforce,withgenerousardour, attwoissues:Oneisthatsolvencymaynotbe hasteningtoleadtheway;andthemoney cleareventothebestinformed,mostrationally cameinsuchaflood,thathisbanksawitself calculatingobserver;theotheristhattheproblem liftedabovethesandsonwhichitwassettling, issometimesjustinformedversusincompletely andfloatedintodeepwaterswithabundant informedexpectations. resources.(p.27) Academicbattlesoverrationalexpectations Thisincidentisinterestingbecauseitfocuses haveoftenfocusedonrationalexpectationsversus ontheproblemofmanagingmarketexpectations. expectationsdrivenbyemotionorafailureto Fromacentral-bankingperspective,theissues calculatesensibly.However,Ithinkthatpanics havebeenquitewellunderstoodsinceWalter largeandsmallaresometimesdrivenbythelack BagehotpublishedLombardStreetin1873.A ofcompleteinformation,andinthosecasesthe centralbankcanresolveabankingpanicbypro- policyissueisrelativelysimple. vidingliquiditytosolventbanks. Consideranincidentduringthebankingcrisis Let’slookatthenatureoftheexpectations inthestateofRhodeIslandin1990-91.Iwason issuewhenfinancialpanicstrikes.Theplaceto thefacultyofBrownUniversityandlivedin startiswiththisquestion:Aretherumorssparking Providence,RhodeIsland,atthattime.Anumber thecrisistrue?Intheincidentrecountedabove, ofstate-charteredcreditunionsandsavingsbanks wereinsuredbyaprivatedepositinsurancecom- therumorwasuntrue.Thebankwassolventand pany.Oneofthesecreditunions,bytheway,was hadaccesstoamplesourcesofliquidfunds;once theBrownUniversityEmployees’CreditUnion. itmarshaledthefunds,theproblemwassolved. AsIrecallthechronologyofevents,inNovember Inothercases,ofcourse,rumorsaretrue.Inthe 1990oneofthesavingsbanksdiscoveredalarge fallof1998,forexample,LongTermCapital embezzlement,whichledtoitsfailure.Thatfail- Management(LTCM)wasseverelyoverextended. urenearlywipedouttheassetsofthedeposit Thefirmwasindeedindangerofbeingunable insurancecompany,whichinturnledtowide- tomeetitsobligations,andmarketparticipants spreadconcernaboutthesafetyofdepositsin wererighttoquestionitssolvency.Moreover, otherinsuredinstitutions.Thecrisiswasreported theobligationsoutstandingweresolargethat intheProvidencenewspaperdayafterday.CNN significantmarketdisruptionmighthave sentareportertocoverthestory,andthereporter occurredhadthefirmdefaulted. wentoncamerastandinginfrontofalocalbank— Acentralbankfacesseveralissuesincases theOldStoneBank.Thenextday,followingthe likeLTCM.Withoutaction,marketpricesmay CNNreport,therewasarunonOldStoneBank. declinesomuchthatathinlycapitalizedfirm OldStonewasfederallyinsuredandhadnothing goesunder.Butinterventionmayhavetheunde- whatsoevertodowiththecrisisofthelocally sirableeffectofproppingupaninstitutionthat insuredcreditunionsandsavingsbanks. failedtomeetthemarkettest.Thisistheproblem WasitrationalforOldStone’sdepositorsto ofmoralhazard;otherfirmsmaybetoncentral pulltheirfundsoutofthebank?Thoseofus bankinterventioninsimilarcasesinthefuture involvedinbankingandfinancemighteasilysay andtherebymanagetheiraffairsinawaythat thatsuchbehaviorwasirrationalbecausethatbank increasestheprobabilityofacrisis.Bagehot’s wasfederallyinsured.ButasIreflectonmyown solutiontothemoralhazardproblemwasforthe behaviorinareaswhereIamlesswellinformed, centralbanktolendatapenaltyrateofinterest. Iamnotsosurethatjudgmentissound.For Marshalingprivatelenderswhoaccepttherisk example,whentherecentpublicityconcerning worksthesameway. Firestonetireshitthenewspapers,Iwentoutto 3 MISCELLANEOUS mygaragetolookatthetirestofigureouthow thanalleviatingit.Sometimespanicsaredriven mycarswereequipped. byrumorsthatturnouttobesubstantiallyaccu- Informationiscostly,andourbrainshave rate.Insuchcircumstances,thoseinauthority onlyafinitenumberofcellstoholdinformation. mayattempttoalleviateoravoidpanicbygloss- Whenaneventorrumorbringsanissuetopublic ingovertheseverityoftheproblem.Doingso attention,manypeoplewillinevitablyandappro- mayhelptomanageaparticularincident,butat priatelyreactonthebasisofhighlyimperfect thecostofdamagingthelong-runcredibilityof information.Thereactionsmaybeperfectlysen- theauthorities. sible—rational,ifyouwill—giventhelimited Aparticularlyclear,andexpensive,example informationathand.Givenincompleteinforma- ofthisprocesswastheU.S.savingsandloan(S&L) tion,Ithinkitiscompletelyrationalfordepositors industry.Fromthemid-1960stothelate1980s, topullfundsoutofasuspectbank.Indeed,in theU.S.governmentandregulatorybodiestook theEighthHenryThorntonLecture,KarlBrunner numerousstepstodealwiththeinstitutionaland arguedthatmoneyitselfexistsbecauseithelps financialweaknessesofnumerousS&Ls.The toalleviateinformationproblems.Iagreewith processculminatedina$150billiongovernment Brunnerthatthefull-informationversionofthe bailoutoftheFederalSavings&LoanInsurance rationalexpectationshypothesisprovidesvalu- Corporation(FSLIC).Congresscloseddown ableinsightsforcertainproblemsbutisincapable FSLICandtheregulatoryagency,theFederal ofexplainingsomeimportantphenomena. HomeLoanBankBoard.Thepoliticalcareersof ReturningtothecaseinRhodeIsland,therun severalmembersofCongressweredamagedor onOldStoneBankwasquicklyhaltedthrough endedbythevoters.Iamconvincedthatthegov- thespreadofaccurateinformation.Thebank ernmentcouldhaveavoidedthisentiremessifit itselfandbankingauthoritiesemphasizedtothe hadrequiredmarketvalueaccountingforS&Ls publicthatOldStonewasfederallyinsuredand fromthebeginning. hadnoconnectiontothestatewidebankingcrisis. Providinginformationprospectively,aswith Manypaniccasesinpracticereflecthighly marketvalueaccounting,isperfectlyfeasiblein incompleteinformation.Giventhecostsofobtain- manycases.IntheRhodeIslandbankingcrisis, inginformation,Ithinksituationsofthiskind, andothers,partoftheproblemhasbeenthat whicharenotuncommon,providecompelling depositorsgenuinelybelievedthattheirdeposits evidenceagainstapure,full-informationversion wereperfectlysafe—assafeasthecurrencyin oftherationalexpectationshypothesis.Notonly theirwallets.TheRhodeIslandincidentwasnot aresomemarketparticipantspoorlyinformed, unique;theUnitedStateshasalonghistoryof whichisobvious,butmarketoutcomescanreflect failureofprivateandstatedepositinsurance poorlyinformedviews.However,itisessential funds.3Ifagovernmentcanstandardizethedefi- thatwenotequateexpectationsbasedonincom- nitionofScotchwhiskey,whycan’titstandardize pleteinformationwithexpectationsthatarehope- thedefinitionof“deposit”?Giventhatdepositors lesslyemotionalandirrational;provisionof havesooftenbeenconfusedinthepast,whynot informationdoeshaveobservableeffectson reservetheword“deposit”intheUnitedStates marketoutcomes.Fromapolicyperspective,that foraliabilityinsuredbytheU.S.government? meansthatprovisionofaccurateinformationis Alongthesamelines,intheUnitedStates thefirstlineofdefenseincasesoffinancialpanics. weneedtoclarifytheextentofthefederalguar- Ifthisargumentseemsalmostself-evident, anteefortheliabilitiesofgovernmentallyspon- weneedtorememberthatfromtimetotimecen- soredenterprises(GSEs).Althoughthelegal tralbanks(andgovernmentauthoritiesmore situationdiffersfromoneenterprisetoanother, generally)havecontributedtotheproblemrather theliabilitiesofGSEsoftencarrynoexplicit 3 SeeEnglish(1993). 4 Expectations guarantee,yetthemarketpricestheseobligations towardfederalguaranteesiscurrentlyilldefined, asiftherewereafederalguarantee.Basedonpast andnowIwanttoturnmoreexplicitlytomone- practiceandcontinuingdebate,marketpartici- tarypolicy. pantshaveeveryreasontoassignarelativelyhigh Imustsaythatthereisamazinglylittleaca- probabilitytoafederalbailoutshouldaGSEcome demicresearchprovidingsolidguidanceasto closetodefaultingonitsobligations.Similar whatIoughttodotohelpdefineamorecoherent issuessurroundthe“toobigtofail”doctrine monetarypolicy.Iamnotimplying,ofcourse, appliedtolargeprivatefinancialinstitutions. thatIbelievethatFedpolicyisincoherenttoday. Ifamarketcrisisemergedonedaybecause WhatIamsayingisthatresearchshowinghow investorscametobelievethatthefederalgovern- wecandobetter,orevenjustcharacterizingmore mentwaspreparedtoletoneormoreofthese accuratelythepolicyfollowedinrecentyears,is firmsfail,wouldthecrisisbethefaultofnonra- surprisinglythin.Researchonmonetarypolicy tionalexpectationsorofgovernmentpolicythat reactionfunctionsseemsquiteunfruitfultodate. failedtoclarifytheissue? Amongthosewhohaveworkedonthisissue,I Theappropriategovernmentroleinguaran- thinktheviewisnearlyunanimousthatinrecent teeingfinancialobligationsisacomplexissue, yearsFederalReservepolicyhasbeenbetterthan andIdon’tintendtoexplorethemeritsofvarious anyproposedexplicitpolicyrule.Thatmeans positionshere.ButIdofeelstronglythatthe thatnoonehasbeenabletowritedownapolicy governmentitself,notthemarket,isresponsible rulethataccuratelycharacterizesFedpolicy. ifmarketexpectationsoverapotentialdefault Thisobservationhasadirectimplicationfor seememotionallydrivenandvolatile.IhopeI’m researchintotherationalityofexpectations.The wrong,butI’mwillingtospeculatethattheissue keyideaoftherationalexpectationshypothesis willremainunresolvedintheUnitedStatesuntil isthatthemarketformsexpectationsbasedon athreatenedoractualdefaultforcestheissue. estimatesofmodelparametersthatmatchthetrue TheUnitedStatesdidnotaddresstheS&Lissue modelparameters.Nooneshouldbesurprisedif untilitbecametoolargetoignore.Thepolitical economistshavedifficultyconfirmingtheration- responseislikelytodependheavilyonthefacts, alityofmarketexpectationsaboutinflation,for orperceivedfacts,atthetime,especiallyclaims example,ifeconomistscannotevencharacterize aboutwhowillbehurtbywhateverdecisionis Fedpolicywithmuchaccuracy.Whyshould madeandwhois“atfault”andthereforedeserves economistsjudgethemarketbystandardsthey tobepunished.NeitherInormarketexpertswho themselves,withalltheirknowledgeoftheory knowmoreaboutthesematterscanformconfi- andeconometrics,cannotmeet?Indeed,thisline dentexpectationsaboutoutcomesinsuchcases. ofargumentopensupthepossibilities(i)that ButIwanttoreiteratethattheissuessurround- themarketmaybehaveasifitwereabletochar- inggovernmentguaranteescanandshouldbe acterizepolicycorrectlyand(ii)thateconomists’ addressedbeforeacrisisstrikes. testsofrationalexpectationsfailbecauseecono- Therationalexpectationsrevolutioninmacro- mistsfailratherthanbecausemarketsfail.Inthe economicsmadeclearthatthedistinctionbetween lastsectionofthislecture,I’lldescribesomerecent policyandpolicyactionsiscritical.Policyreflects researchattheSt.LouisFedsuggestingthatmar- thegeneralregularityofbehaviorofpolicymakers ketsinfactunderstandrecentFedpolicyfarbetter overtime;policyactionsaretheindividual thaneconomistsdo. responsescasebycase.Wheneverpolicymakers I’vearguedthatmarketpanics,andinexpli- believethatmarketexpectationsareirrational, cablechangesinassetpricesmoregenerally,may policymakersoughtfirsttolookintothemirror notreflecttheirrationalitythatmanyeconomists andaskwhetherpolicyiscoherent.Marketexpec- seemtoassume.Panicsmayarise,atleastinpart, tationsaboutpolicycannotbecoherentifpolicy fromthefailureofpolicymakerstofollowclear isnotcoherent.I’vesuggestedthatU.S.policy andcoherentpolicies.Everyoneagreesthat,in 5 MISCELLANEOUS general,assetpricesoughttochangewhenpolicy Centralbankershavenotalwaysappreciated changes.Ifpolicyisilldefined,thennooneshould theimportanceofcredibility.Torelateapersonal besurprisedwhenassetpriceschangeasmarket examplefromthe1970s,whileonthefacultyof perceptionsaboutprospectivepolicychange. BrownUniversityIhadmanycontactswith Theseperceptionswillbeweaklyheldandare FederalReserveofficials.Asinflationcontinued thereforesubjecttochange,perhapsevenabrupt overthecourseofthedecade,Ibecameincreas- change,becauseitisnotrationaltohavefirm inglyskepticaloftheFederalReserve’sprofession viewsaboutpolicywhenpolicyisilldefined. ofallegiancetothegoaloflowinflation.Isaid,in Anobjectiontothisviewmightbethatit effect,tosomeofmyFedfriends,“Idon’tbelieve providesnoexplanationofthetimingofpanics you.”IthinktheywereinsultedbywhatIsaid, andsharpchangesinassetprices.Butthisobjec- butthemarketsincreasinglydidnotbelievethe tionisunconvincing.Ifanaccurateempirical Fedeither.Althoughcriticismfrommanydiffer- model—whetheraneconomicorapsychological entdirectionsisafactoflifeforcentralbankers, model—oftimingexisted,thenthemarketwould theyshouldtakesuchcriticismseriously.Atthe usethatinformationtoseektheprofitimplied. sametime,theyshouldbecarefulnottoassume Anuncontestableimplicationofrationalexpec- thatcommentsreflectinggeneralesteemforthose tationstheoryandevidenceisthatthereareno inofficenecessarilyareavoteofconfidencein easyprofitstobehadinassetmarkets.Panicsand thepoliciesbeingpursued. marketcrisesmustbeunpredictable.Tome,asa Thereisnowanextensiveliteratureoncentral policymaker,theimplicationofinexplicableand bankcredibility;Icanhardlyclaimtobefamiliar unpredictablepanicsandassetpricechangesis withallofit.Butwhatdoesstrikemeaboutthis notthatweneedanew,nonrationalexpectations literature,asusefulasitis,isthatitdoesnotgo approachtounderstandingexpectations.Instead, veryfarinprovidingspecificadvicetocentral weneedtoexaminehowpolicybodiescanmore banksaboutbuildingcredibility.Thepractical effectivelytransmitaccurateinformationtothe problemIfaceisintryingtodecidehow,ifatall, marketandhowpolicycanbemademorecoherent toreacttothelatestreleaseofemploymentdata, andreliable.Inshort,policymakersneedtoreal- inflationdata,andthesteadyflowofotherinfor- locatetheirthinkingtimemoretolookinginward mationofallkindsdaybyday.Theproblemisto atwhattheydoandlesstolookingoutwardat makeindividualpolicyactionsadduptoacoher- whatmarketsdo. entpolicy.Tobecredible,thecentralbankmust besuccessfulinachievingitsstatedgoals.To deliveronthesegoals,thecentralbankmust knowhowtorespondtothesteadyflowofinfor- WHERE DOES THIS CREDIBILITY mation,anditsresponsestothisinformationmust COME FROM? makesenseaspolicy.Thatis,everycentralbank Thisdiscussionleadsnaturallytothebroader needsamonetarypolicystrategyinwhichthe subjectofcredibility.Marketsshouldvieweco- goalsareclearandthepolicyactionstoachieve nomicpolicyintermsofaruleorregularityof thegoalsarewelldefined. behavior.Marketsinterpretindividualpolicy Manymarketparticipantshavegreatexpertise actionsinthecontextoftheirconsistencywith inmonetarymatters,andtheyformreasoned thepolicy,giventhefactsofthecurrentsituation. judgmentsabouttheperformanceofcentralbanks. Ifauthoritiesmisleadthepublicinaparticular Wemaycalltheviewthatemerges“reasoned situation,thenpublicconfusionordistrustwill credibility.”Butthereisanotheraspectofcredi- makeitmoredifficultforpolicymakerstodeal bilitythatarisesfromthefactthatmostofany withthenextcrisis.Itisimportanttoemphasize individual’sviewsandexpectationscomenot theenormousbenefitofcentralbankcredibility frompersonalstudyandinvestigationbutfrom inallareasinwhichitexercisesitspowers. acceptanceofviewsoftrustedauthorities,or 6 Expectations experts.Noonehasthetimetobeexpertabout Whataretheseobservableconsequences?I everything.Relianceonexpertsisaconsequence willspeakonlytothesituationintheUnited ofthecostlinessofinformation.Ifacentralbanker States,whereIknowthehistoryanddataindetail. isatrustedauthority,hisorherviewonawide Ithinkthattherearemanysuchobservablecon- rangeofeconomicissues,includingmanyfar sequencesandthatoneofthemisthesustained removedfrommonetarypolicy,willcarrygreat favorablesurpriseintheunemploymentrate. weight.Becausetrustedexpertsdiffer,andwe UnemploymentaslowastheratetheUnited allfacetheproblemofpickingwhichexpertsto Stateshasenjoyedinrecentyearscouldnothave believe,overtimeacentralbankcandevelop occurredwithoutentrenchedexpectationsof specialcredibilityamongcompetingauthorities. continuinglowinflation.Intheconventional Wemaycallthisgeneraltrustofacentralbank Phillipscurve,therateofinflationdependson “institutionalcredibility.” expectedinflationandthegapbetweentheactual Credibilityinbothitsdimensionsisearned, andnaturalratesofunemployment.Anecdotal orlost,dayinanddayout,overbigissuesand reportsfromemployersandsystematicinforma- small,andisnotcompartmentalized.Inother tionsuggestthattheU.S.labormarkethasbeen words,acentralbankcannotbedistrustedinone stretchedabnormallytightforseveralyearsnow. areaofitsoperationsandretainhighcredibility Ithinkthebestexplanationofhowthesetight inotherareas. labormarketconditionscancontinueisthat Thevalueofcredibilityisparticularlyclear expectationstrumpthegap.Firmsarejustnot inacrisis.Wheninformationishighlyincomplete willingtobidaggressivelyforlabortofillempty andthetruestateofaffairsmurky,itisextremely positionsbecauseseniormanagementdoesnot valuableforsocietyifthemarketscanlooktothe believethathigherwagescanbepassedonin centralbankasatrustedauthorityandacceptits higherprices.Expectationsofcontinuinglow judgmentsandactions.Ifthecentralbankis inflationdominatetheoutcome. indeedwellinformedandcompetent,itscredi- Thatismytentativehypothesisanyway,but bilityinthemarketswillobviouslymakeitstask becauseIdonothaveresearchresultstosupport fareasier. itatthistimeI’llnotpursuethematterfurther In1985,MichaelParkinpresentedtheSeventh excepttoofferonemoreobservation.Mostecon- HenryThorntonLecture.Histitlewas,“Inflation omistsbelieve,Ithink,thattherationalexpec- Expectations:FromAdaptivetoRationalto…?” tationshypothesisisextremelyvaluablein Asapartofhisinsightfulreviewofexpectations understandingoutcomesinauctionmarkets— issues,Parkindiscussesthefailureofinflation likethoseforequities,bonds,foreignexchange, expectationstofallpromptlywiththechangein andcommodities—butisoflimitedapplication U.K.monetarypolicyintheearly1980s.Hecon- inthelabormarket.Thelabormarket,sotheargu- cludedthat,giventhehistoryandtheincentives mentgoes,isdominatedbyinstitutionalbehavior, toinflate,“itisnotrationaltoexpect,andact attitudesconcerningequity,andslowadjustment uponthebasisof,alowrateofinflation”(p.13). tochangingconditions.Ineconometricmodels BoththeUnitedStatesandtheUnitedKingdom ofthelabormarket,adaptiveexpectationsseem boreheavycoststoreestablishexpectationsof toworkwellenough.WhatI’msuggestingisthat lowinflationandcentralbankcredibility. theU.S.unemploymentratehasdepartedfrom Centralbanksaroundtheworldtodayenjoy theconventionalestimateofthePhillipscurve highcredibilitycomparedwiththesituation becausethatestimatefailedtoaccountadequately only20yearsago.Justastherewereobservable fortheroleofrationalinflationaryexpectations marketconsequences—deeprecession—of inthelabormarket.Thetheoryofrationalexpec- impairedcredibilityintheUnitedStatesand tationsprovidesguidanceinunderstandingeco- UnitedKingdomintheearly1980s,thereare nomicbehaviorinallpartsoftheeconomy,not observablemarketconsequencestoday. justinauctionmarkets. 7 MISCELLANEOUS WHAT DO WE MAKE OF willnotnecessarilybidupinflationexpectations whenandifpolicygoesastray.Highcredibility INFLATION EXPECTATIONS? meansthatthemarkettruststheFederalReserve’s Thorntonhadaclearunderstandingofthe policyjudgments.Thatbeingthecase,theFederal distinctionbetweenthenominalandrealrateof Reservecannotreliablyextractinformationfrom interest.InaspeechbeforetheHouseofCommons dataonexpectationsabouttheappropriateness in1811,henotedthefollowing:“If,forexample, ofcurrentpolicyactions. amanborrowedofthebankathousandpounds Itislogicallypossiblethatpolicyactionsare in1800,andpaiditbackin1810,havingdetained inconsistentwithsustainedlowinflationatthe itbymeansofsuccessiveloansthroughthat sametimethatthemarketsimplytruststheFed period,hepaidbackthatwhichhadbecome anddoesnotperformaseparateanalysisofpolicy worthlessby20or30percentthanitwasworth actions.Whyshouldanyofus,onanymatter, whenhefirstreceivedit.Hewouldhavepaidan engageinacostlyinvestigationwhenwecan interestof50poundsperannumfortheuseof insteadsimplyacceptthejudgmentofatrusted thismoney;butiffromthisinterestwerededucted authority?Theanswerisobvious:Iftheauthor- the20poundsor30poundsperannum,which ityiscompletelytrusted,andifseparateconfir- hehadgainedbythefallinthevalueofthemoney, mationoftheinformationiscostly,thenthe hewouldfindthathehadborrowedat2or3 cost-efficientthingtodoissimplytoacceptthe percent,andnotat5percentasheappearedto authority’sjudgment. do”(Hayek,pp.335-36). Letmesummarizethisanalysis.Theexpected Athoroughunderstandingofthedistinction rateofinflationoverafive-year,orlonger,horizon betweenrealandnominalinterestratesisagreat isadirectmeasureofcentralbankcredibility advanceincentralbankingpracticeoverthelast regardinginflation.Atanygiventime,monetary 35years.We’vefinallycaughtupwithThornton. policy—policy,notpolicyactions—mayormay IntheUnitedStates,atleast,inthemid-to-late notbeconsistentwithlong-terminflationexpec- 1960s,thepracticalimportanceofthedistinction tations.Eventually,ofcourse,expectationsand betweenrealandnominalinterestrateswasnot policymustbeconsistentbecauseoneorthe appreciated.Risinginterestratesinthelate1960s otherwilladjust. weremisinterpretedasevidenceofamorerestric- Failuretounderstandthispointcouldfoster tivemonetarypolicy,when,infact,nominalrates policymistakes.Whencredibilityishigh,asitis werenotevenkeepingupwiththeincreasein intheUnitedStatestoday,inflationexpectations inflationexpectations. willbeslowtoadjust.Actualinflation,influenced Comparedwith35yearsago,theFederal byexpectedinflation,mayalsobeslowtoadjust. Reservetodayhasaccesstofarmoredataon Therefore,expectedinflation,certainly,andactual expectations.Withinflation-indexedbondsout- inflation,probably,arepoorguidesastothe standing,wehaveday-by-dayevidenceonthe appropriatenessofmonetarypolicyintheshort behaviorofthespreadbetweenconventionaland run.Similarly,wheninflationexpectationsare indexedbonds.Surveyinformationiswidely highandcredibilitylow,thecentralbankhasthe available.Iwatchthesedatacloselybecausethey twinproblemsofgettingpolicyturnedaroundto provideclearevidenceofthecentralbank’ssuc- beconsistentwithlowerlong-runinflationand cessinmaintainingcredibilityinachievingsus- ofadjustingpolicyascredibilitybuildsovertime. tainedlowinflation. IftheFedcannotrelyonactualandexpected Thelogicofthecredibilityargument,however, inflationtojudgetheappropriatenessofcurrent suggeststhatinflationexpectationsdatadonot policy,becausethesemeasuresaredominatedby providedefinitiveevidenceaboutwhethermon- themarket’sassessmentofFedcredibility,what etarypolicyitselfisontrack.GiventhattheFed canitrelyon?Weneedtoconcentrateonthe enjoysveryhighcredibilitytoday,themarkets underlyingdeterminantsofinflationandearly 8 Expectations warningsigns.Therateofmoneygrowth,spreads targetrateisthefederalfundsrate.So,I’llassume infinancialmarkets,thesupply-demandbalance thatourrealcentralbankimplementspolicy acrossindustriesingeneral,andthebehaviorof actionsthatway. specificpriceslikelytoleadoverallinflationare I’vegivenyouaverysimpledescriptionof relevant.Theaimofpolicyshouldbetoactbefore whatthecentralbankwantstodoanditsproce- changesininflationappear;clearly,oncethese dureforpursuingitsobjective.Giventheassumed changesdoappear,thetaskofrestoringcredibility nirvanastateofmonetarypolicy,thecentralbank andreversingalltheadjustmentsthatfirmsand doesitsjobefficiently.BythatImeanthatit householdshavestartedtomakebecomesmore respondssensiblytoalltheambiguitiesandprob- difficult. lemsrealcentralbanksface.Asnewinformation arrives,thecentralbankefficientlyprocessesits significanceandadjustsitstargetfortheovernight WHAT IS THE SIGNIFICANCE OF rateasrequiredtoachieveitspolicygoals.Given theinherentgapsinknowledgeanddata,some- MARKET PREDICTIONS OF timesthecentralbankwillacttooquicklyortoo CENTRAL BANK POLICY? slowly,bytoomuchortoolittle.Butmypresump- I’mnowgoingtobringthevariousstrandsof tionisthatthecentralbankcanavoidcumulative mydiscussiontogether.MycolleagueRobert errorsandrecoverfrompolicymisstepswithout RascheandIhavebeenpursuingalineofresearch missingitsobjectives. onthepredictabilityofmonetarypolicyactions. Participantsinfinancialmarketswillunder- Thepaper(PooleandRasche,2000)isavailable standwhatthecentralbankisdoing.Tounder- intheworkingpaperssectionoftheSt.LouisFed standmarketoutcomesinthissetting,oneother Website;itwillbepublishedintheJournalof observationisneeded.IntheUnitedStates—I’m FinancialServicesResearch.I’lloutlinethebasic notsureaboutthesituationelsewhere—thecentral ideainthatpaperandthenconnectittothe bankhasnosignificantinformationaladvantage argumentofthislecture. overthemarket.TheFedandthemarketsreceive Considerastateofmonetarypolicynirvana governmentstatisticaldataatessentiallythesame intheworldweactuallylivein.Thatis,ifthe time.TheFeddoeshaveanadvantageoverthe centralbankdidasgoodajobasyoucanimagine marketinthatithasaverylargestaffanddoes intoday’sworld—aworldwithmanygapsin obtainanecdotalinformationnotgenerallyavail- knowledge,datainaccuracies,andallthereal able.However,individualfirmshavemuchmore problemsrealcentralbanksface—whatwould extensiveinformationabouttheirownmarkets weobserve? thantheFeddoes.Ithinkitisapproximately Let’ssupposethatyouthinkameasuredCPI correct,andcertainlyappropriateatthelevelof inflationrateof1percentperyearisoptimaland theoreticalmodeling,toassumethatthemarkets thatyoubelievethecentralbankcanoffsetsome andtheFedreceivethesameinformationatthe financialandrealdisturbancestocushionfluctu- sametime. ationsinoutputandemploymentwithoutcom- Marketparticipantshaveampleincentiveto promisingtheinflationobjective.Thisisashort formaccurateexpectationsaboutcentralbank descriptionofwhatIbelieve,butyoucansubsti- policyactions.Howaccuratearethoseexpecta- tuteyourownspecificationformine. tionslikelytobe?Givenmyassumptions,the Themarketwill,induetime,learnofthe marketoughttobeveryaccurateinpredicting policyobjectiveandthepolicyactionsdesigned policyactions.Themarketandthecentralbank toachievethatobjective.Realcentralbanksalmost getthesamedataatthesametime;themarket withoutexceptionimplementpolicybysettinga understandsthepolicyobjectivesandthepolicy targetforashort-terminterestrate,usuallyan actionsappropriatetoachievetheobjectives.As overnightbankrate.IntheUnitedStates,that newdataarrive,themarketshouldinterpretthe 9 MISCELLANEOUS datathesamewaythecentralbankdoes,atleast anythingclosetotheaccuracyofthepredictions mostofthetime,andreachthesameconclusion inthefederalfundsfuturesmarket.Thereisan aboutthesignificanceofthedata. importantresearchagendaimpliedbythisobser- RascheandIhaveexploredthishypothesis vation.WeneedadeeperunderstandingofU.S. fortheUnitedStates.Ourresearchisongoing, monetarypolicytoincreasetheprobabilityof butatthistimewecanreportthatasofthelast extendingrecentpolicysuccessesintotheindef- fewyearsthemarkethasbeenquiteaccuratein initefuture. forecastingFedpolicyactions.Since1988,when Thisexperiencealsoshowsthatthecentral tradingopenedinthefederalfundsfuturesmar- bankcanchangewhatitdoestopromotemore ket,wehavehadaverydirectreadingonmarket accuratemarketexpectations.Bydisclosingpolicy expectationsaboutFedpolicy.Since1994,that decisionsquicklyandbyconfiningpolicyactions markethaspredictedpolicyactionsquiteaccu- toregularlyscheduledmeetings,theFOMChas ratelyonthewhole. madepossibleimprovedmarketforecastsof Itisinstructivetonotethat1994wasawater- monetarypolicyactions.Thechangeinpractice shedyear.InFebruary1994,theFederalOpen inFebruary1994illustratesthepointIemphasized MarketCommittee(FOMC)—theFed’smain earlier—thatthecentralbankcanimprovethe monetarypolicybody—firstbegantoreleasea accuracyofinformationavailabletothemarket. policydecisionaboutitsfederalfundsratetarget Ibelievethatthesimplestepofpromptdis- immediatelyfollowingtheFOMCmeeting.Before closureinFebruary1994alsoimposedavaluable thattime,themarketlearnedofpolicyactions disciplineontheFOMCitself.Byconfiningmost byobservingopenmarketoperationsconducted policyactionstodaysofFOMCmeetings,the bytheOpenMarketDeskattheNewYorkFed. Committeemadeitsownbehaviormorepredict- Moreover,before1994,policyactionsoccurred able.Now,everyoneknowsthatapolicyaction moreoftenbetweenregularlyscheduledFOMC atanothertimeisspecial.TheFOMCmustthink meetingsthanatthemeetings.SinceFebruary carefullyaboutwhetheritwantstosendaspecial 1994,almostallpolicyactionshavebeentaken messagebychangingpolicybetweenmeetings atregularFOMCmeetings. and,ifitdoes,whatthemessageis.Whatthe AlthoughtheFOMCadjuststhetargetfederal centralbankdoeswillshapeexpectations;for fundsratemostoftenbyonly25basispoints,it thecentralbanktobeabletopredictitseffects sometimeshasmadelargeradjustments.Butthese onexpectations,itsownbehaviormustbeas adjustmentshavebeenwellpredictedbythe regularaspossible. market. Thisevidenceshowsconclusivelythatitis possibleforacentralbanktopursueahighlypre- WHAT SHOULD THE AGENDA dictablepolicy,inthesensethat,giventheavail- FOR CENTRAL BANKS BE? ableinformationatthetimeofapolicymeeting, themarketcanpredictthepolicyaction.Policy Therationalexpectationsrevolutionin actionscannotbepredictedfarinadvancebecause macroeconomicschangedforeverhowwethink theinformationdrivingpolicydecisionscannot abouteconomicpolicy.Weknowthatunder- bepredictedfarinadvance.But,asinformation standingmarketsrequiresthatweunderstand accumulatesbeforeapolicymeeting,themarket marketexpectationsaboutmonetarypolicy.We andthecentralbankcanconvergeonacommon knowthatthedistinctionbetweenpolicyactions interpretationoftheinformation. andpolicyitselfisofcentralimportance.We Moreover,themarketiswellaheadofecono- knowthatexpectationsarenotalwaysfully mistsinunderstandingthisprocess.Iknowofno rational,butIhavebeenatpainstoarguethat econometricmodelsthatpredictboththetiming someoftheproblemscausedbynonrational andthemagnitudeofFedpolicymoveswith expectationsarecorrectable. 10 Expectations Iknowofnopolicymodelsindicatingthat policyinstrumentsbeadjusted?Thatis,what theeconomyworksbetterwhenmarketsarekept shouldcentralbanksdoandwhenshouldthey guessingaboutmonetarypolicy.Thepresumption doit?Ofmycentralbankcolleagues,Iaskthat mustbethatmarketparticipantsmakemoreeffi- wespendmoretimefocusedondefininggeneral cientdecisions—decisionsthatmaximizeeco- policyrules,orregularities,withinwhichwewill nomicgrowthbyminimizingthewastageof fitindividualpolicyactions.Bothenterprises resourcesfromexpectationalerrors—whenmar- promisesignificantimprovementsintheaccu- ketscancorrectlypredictcentralbankactions. racyofmarketexpectationsandthestabilityof Thatdoesnotrequirethatcentralbankersand markets. marketparticipantsbeabletoforecasttheunfore- castable,butthattheyhaveacommonunderstand- ingofthestrategygoverningpolicyactions. REFERENCES I’vesuggestedalargeagenda—onethatis Bagehot,Walter.LombardStreet:ADescriptionof indefinitelylarge—forcentralbanksandgovern- theMoneyMarket.London:1873. ments.Weneedtofocusonareaswheremarket expectationsarehazybecausegovernmentpolicy English,William.“TheDeclineofPrivateDeposit itselfisormaybeilldefined.Theseincludethe InsuranceintheUnitedStates.”Carnegie-Rochester natureofgovernmentguarantees,monetarypolicy ConferenceSeriesonPublicPolicy,June1993,pp. objectives,andthestrategytoreachthoseobjec- 57-128. tives.Someofthethingsweneedtoexaminemay appearterriblymundane.Forexample,Ithink Hayek,F.A.von,ed.AnEnquiryintotheNatureand thattheFOMCprobablymeetsmoreoftenthan EffectsofthePaperCreditofGreatBritain(1802) necessary.Marketinterestrateshaveampleroom [reprint].NewYork:AugustusM.Kelley,[1939] tofluctuateforanygivenfederalfundsrate,and 1962. itisrarethatanythinghappenswithintheusual sixweeksbetweenFOMCmeetingstorequirea Keynes,JohnMaynard.TheGeneralTheoryof reassessmentofpolicy.Ifthemarketsandthe Employment,InterestandMoney.NewYork: centralbankreallydohaveacommonunder- HarcourtBrace,1936. standingofmonetarypolicy,itishardformeto Parkin,Michael.“InflationExpectations:From believethatoutcomesforthe10-yearbondrate, AdaptivetoRationalto…?”SeventhHenryThornton say,willdependonwhetherthepolicymeetings Lecture,CityUniversityBusinessSchool,London, occuronceamonthoronceaquarter.However, 14November1985. eachmeetingisanobjectofspeculation;the marketwouldbebetterservediftraderswould Poole,WilliamandRasche,RobertH.“Perfectingthe concentrateonthefundamentalsbehindpolicy Market’sKnowledgeofMonetaryPolicy.”Working decisionsthanonthemeetingitself.Mypointis Paper2000-010A,FederalReserveBankofSt.Louis, notactuallytotakeafirmpositionontheminor April2000;JournalofFinancialServicesResearch issueofthemeetingschedulebutinsteadtopoint (forthcoming). outthatallsortsofthingsshouldbediscussedas possiblewaystoimprovethemarket’sunderstand- ingofmonetarypolicy. Ifinishwithapleatobothacademicsand centralbankers.Ofacademics,Iaskthatresearch addressthisquestion:How,veryexplicitly,should 11

William Poole (2000, November 27). Speech. Speeches, Federal Reserve. https://whenthefedspeaks.com/doc/speech_20001128_poole

@misc{wtfs_speech_20001128_poole,
  author = {William Poole},
  title = {Speech},
  year = {2000},
  month = {Nov},
  howpublished = {Speeches, Federal Reserve},
  url = {https://whenthefedspeaks.com/doc/speech_20001128_poole},
  note = {Retrieved via When the Fed Speaks corpus}
}