Speech

The Role of Anecdotal Information in Monetary Policy TheBankofKoreaInternationalConferenceonMonetaryPolicyinanEnvironmentofLowInflation Seoul,Korea ConferenceSession“GeneralDiscussion” June16,2006 I’llusemydiscussiontimetomakeasimple Todaycentralbanksareconcernedwithtwo- argument—that in a low inflation environ- sidedriskwithrespecttotheirinflationobjective. ment, further improvements in reducing Ontheonehand,noonewantstoseeanupward the variances of inflation and employment driftofinflation,eithersuddenorgradual,and will require increased attention to informal, or ontheotherhand,nocentralbankwantsasys- anecdotal, information. Before developing that temicdeflationaryenvironment.IntheUnited argument, I offer the usual Fed disclaimer that Statesitislikelythattheinflationratebottomed the views I express here are mine and do not outin2003,aroundthetimethattheFOMC necessarilyreflectofficialpositionsoftheFederal expressedconcernthat“...theprobabilityofan Reserve System. I thank my colleagues at the unwelcomesubstantialfallininflation,though Federal Reserve Bank of St. Louis for their com- minor,exceedsthatofapickupininflationfrom ments, especially Robert Rasche, but I retain full itsalreadylowlevel.”1 responsibility for errors. Now,approximatelythreeyearslater,aftera Centralbankersaroundtheworldarecur- periodofstrongeconomicgrowthinmanyparts rentlyfacedwithachallengenotexperiencedin oftheworldanddemanddrivenincreasesin over40years:maintaininglowandstableinflation energyprices,theperceivedriskhasshiftedin nowthatithasbeenachievedformorethana theotherdirection.Thecurrentassessmentof passingquarterortwo.Intheaftermathofthe GreatInflation,theprimaryobjectiveofmonetary theFOMCissummarizedinitsstatementfollow- policywasclearanddistinctlyone-sided:Bring ingitsmeetinginMay2006: inflationdowntoachievealowandstablelong- Asyet,therun-upinthepricesofenergyand termrateofinflation.Intheearly1990s,inflation othercommoditiesappearstohavehadonly subsidedtolevelsthatarewidelyregardedas amodesteffectoncoreinflation,ongoingpro- roughlyconsistentwithpricestability,andthe ductivitygainshavehelpedtoholdthegrowth short-runvolatilityofinflationbecamesubstan- ofunitlaborcostsincheck,andinflation tiallysmaller.Thisobservationappliestocoun- expectationsremaincontained.Still,possible triesthatadoptedformalinflationtargetsandto increasesinresourceutilization,incombina- countriessuchastheUnitedStateswheremone- tionwiththeelevatedpricesofenergyand tarypolicyinflationobjectivesareassertedonly othercommodities,havethepotentialtoadd qualitatively. toinflationpressures.2 1 FOMCPressRelease,May6,2003. 2 FOMCPressRelease,May10,2006. 1 MONETARYPOLICYANDINFLATION Meetingthechallengeofsustainingalow- betweenthecontrolobjective—here,theinflation inflationenvironmentmayrequirenew rate—andtheinstrumentsthatapolicymaker approachestopolicyanalysis.Policymakers hasavailabletoaffecttheeconomy.Yearsago understandthattheiractionsdonotgenerate SolowandKarekenprovidedastraightforward immediateresponsesintheeconomy,whichis illustrationofthisproposition: whytheymustlookoutintothefuture.Perfect Supposethatbyheroic(andperhapseven foresightmaybeausefulconstructfortheoretical cyclical)variationinthemoneysupplyand analyses,butpolicymakersinpracticemustmake itsrateofchange,theFederalReservemanages theirdecisionswithfullknowledgeofthenormal deftlytocounteralldisturbingimpulsesand rangeofforecasterrors.Economicforecastsare tostabilizethelevelofeconomicactivity rootedinstatisticalmodelsestimatedfromhis- absolutely.Thenanobserverfollowingthe Friedmanmethodwouldseepeaksandtroughs toricaldata,andthemodelforecastsarethen ofmonetarychangeaccompaniedbyasteady modified,oftensubstantially,throughtheappli- levelofaggregateactivity.Hewouldpresum- cationofexpertjudgment.Experienceindicates ablyconcludethatmonetarypolicyhasno thatforecastinginflationishardlypreciseand effectsatall,whichwouldbepreciselythe thatforecastingmodelsarenotrobust. oppositeofthetruth.5 StockandWatsonsummarizetheirexperi- Theobservationthatcorrelationsarechang- mentswithnumerousforecastingmodelsfor ingordisappearingdoesnotmeanthattheecon- inflationacrosssevenindustrialeconomies: omyhasfundamentallychanged.Inparticular,it Theseforecastingsuccesses,however,areiso- islikelythatthecorrelationbetweenthegrowth latedandsporadic...Forexample,theseasonal ofmonetaryaggregatesandtheinflationrate(or no-changeforecastworkswellintheUnited evennominalincomegrowth)willbesmallin Statesinthesecondperiodbutpoorlyinthe lowinflationenvironments.Yetcentralbankers first,asimilarpatternasinCanada(butthe whofailtomonitorthegrowthratesofmonetary oppositepatternasintheUnitedKingdom).3 aggregatesdosoattheirownperil.Historyillus- Theonlysetofpredictorsthatusually tratesthatrapidandacceleratingmonetarygrowth, improvesupontheAR[autoregressive]fore- positiveornegative,isarecipeforthedemiseof castsisthemeasuresofeconomicactivity. thelowinflationregimeintoinflationordeflation. Forexample,theIP[IndustrialProduction] Justbecausealowinflationenvironmenthasbeen andunemploymentgapsbothimproveonthe established,centralbankerscannotprintmoney AR(orarelittleworsethantheAR)forboth withoutrestraint.Largecorrelations,then,pro- periodsforCanada,GermanyandtheUnited videevidencethatthecentralbankhasfailedto States.Evenforthesepredictors,however,the exploitrelevantinformation;aspolicybecomes improvementisneitheruniversalnoralways moreeffective,correlationstendtowardzero. stable.4 Inalowinflationenvironment,thestability Economistshavelongunderstoodthatcontrol ofexpectationsoflong-runinflationiscertainly theoryimpliesthatthemoresuccessfulacentral one,andperhapsthesinglemostimportant,ele- bankisinachievingandmaintainingalowinfla- mentinthecontinuedsuccessofthelowinflation tionenvironment,thesmallerarethecorrelations policy.Offirstimportanceisthatsuchexpecta- 3 J.H.StockandM.W.Watson,“ForecastingOutputandInflation:TheRoleofAssetPrices,”JournalofEconomicLiterature,September,2003, 41,p.805. 4 J.H.StockandM.W.Watson,“ForecastingOutputandInflation:TheRoleofAssetPrices,”JournalofEconomicLiterature,September,2003, 41,p.808. 5 R.M.SolowandJ.Kareken,“LagsinMonetaryPolicy”,inA.Ando,E.C.Brown,R.M.SolowandJ.Kareken,eds.,“LagsinFiscaland MonetaryPolicy,”inCommissiononMoneyandCredit,StabilizationPolicies,EnglewoodCliffsNJ:Prentice-Hall,1963,p.16. 2 TheRoleofAnecdotalInformationinMonetaryPolicy tionsremain“wellcontained”or“wellanchored.” respecttotheannouncedinflationtarget,then Inrationalexpectationsmodelsofthemacro- long-terminflationexpectationsofprivateagents economy,monetarypolicyrulesspecifiedwith willbecomedisconnectedfromthetarget,and anominalinterestrateinstrumentrequireacon- theeconomywillbegintodragitsnominalanchor. stantexpectedlong-term(orequilibrium)expected Centralbankersmaybereluctanttotakedeci- rateofinflationtoassuretheexistenceofan sivepolicyactionsasactualinflationapproaches equilibriumrateofinflation.Thefamiliar“Taylor aperceivedboundaryofpricestabilitybecause rule”containsa≠*term,thedesiredrateofinfla- oftheirimpreciseforecastsofinflationandeco- tionofthemonetaryauthorities.Notethatinthe nomicactivity.Facedwithanuncertainviewof Taylorrulethetermis≠*not≠*.6Insuchmodels, thefuture,thenaturaltendencyofpolicymakers t ≠*isthe“anchor”onwhichthelong-term istowaitforfurtherinformationonthestateof expectedinflationrateofprivateagentsisbased. theeconomy.Intheabsenceofdecisivepolicy Iftheinflationobjectiveistime-varying,oris actions,centralbankersmaybeabletostabilize perceivedbyprivateagentsastime-varying,then long-terminflationexpectationsbyclarifying long-terminflationexpectationsbecomeunglued theirvisionofpricestability.Thetextinthe andtheinflationratewillnotremainlowand minutesoftheMay2003FOMCmeetingisan stable;indeeddependingonhowthedesired exampleofsuchaclarification: inflationrateofthepolicymakersisperceived, Giventhepressureofaconsiderableamount inflationcanbecomeaself-fulfillingexplosive ofunusedresources,anyadversedevelopments process. thathelddowneconomicexpansionwould Suchmodelsprovideinsightintohowto increasetheprobabilityoffurtherdisinflation. conductmonetarypolicythatwillsuccessfully Memberscommentedthatsubstantialaddi- sustainalowandstableinflationenvironment: tionaldisinflationwouldbeunwelcome Themonetaryauthoritiesmustclearlycommu- becauseofthelikelynegativeeffectsoneco- nicatetheirinflationpolicyobjectives.Thecom- nomicactivityandthefunctioningoffinancial municationmustbesymmetric:Privateagents institutionsandmarkets,andtheincreased difficultyofconductinganeffectivemonetary mustunderstandwhatratesofinflationare policy,atleastpotentiallyintheeventthe unacceptablyhighandwhatareunacceptably economywassubjectedtoadverseshocks.7 lowtothecentralbank.Centralbanksthat announceexplicitnumericinflationobjectives Notethattheseminutesprovidesymmetryto goalongwaytowardssatisfyingthiscommuni- theconcernsofFOMCparticipantsaboutinflation: cationobjective.However,announcementsmust theyrevealthattheimplicitinflationobjective beconfirmedbydeeds.Centralbanksmust oftheCommitteeevaluateddownsideaswellas demonstratethattheyarepreparedtoactdeci- upsiderisk. sivelyagainstsustaineddeviationsfromtheir Adifficultquestionfacingcentralbankersis announcedobjective.Further,topreservethe howtojudgewheninflationexpectationsremain credibilityoftheannouncedinflationobjective, anchored.Onewaytoapproachthisquestionis changesintheannouncedobjectivemustbe toaskwhatcriticaldatamightbeexpectedto undertakensparinglyandinfrequently,andcer- showdifferentcharacteristicswheninflation tainlynotindirectionsawayfromlowratesof expectationsare“wellanchored”comparedwith inflation.Ifthecentralbanklosescredibilitywith whentheyarecomingunglued.Mostsuchdata 6 Thereisanongoingdebateinthemonetarypolicyliteratureover“instrumentrules”versus“targetingrules.”Seeforexample,B.T.McCallum andE.Nelson,“TargetingversusInstrumentRulesforMonetaryPolicy,”FederalReserveBankofSt.LouisReview,September/October2005, 87(5),pp.597-612andL.E.O.Svensson,“TargetingversusInstrumentRulesforMonetaryPolicy:WhatIsWrongwithMcCallumandNelson?” FederalReserveBankofSt.LouisReview,September/October2005,87(5),pp.613-26.Inbothapproachesitisimportantthatthepolicy rulesspecify≠*,not≠* t . 7 Minutes,FOMCMeeting,May6,2003. 3 MONETARYPOLICYANDINFLATION willrequiredirectorindirectmeasuresofinflation drawbackofanecdotalinformationisthatthere expectations.Surveydataoninflationexpecta- isnoscientificbasisforthesample.Yettheaccu- tionsprovidesomeevidence,buttypicallyhave mulationofforward-lookinganecdotalinforma- limitationssuchassmallsamplesize,nonscien- tionatcriticaltimescanbeinformative.An tificsampledesign,andarestrictedpopulation examplecanbedrawnfromtherecentlyreleased fromwhichsurveyrespondentsaredrawn.The transcriptsoftheFOMCmeetingsofOctober, introduction,overthepastdecadeinanumberof November,andDecember,2000.Atthattime, countries,ofinflation-indexedlong-termgovern- thebestinferencefromstatisticalforecasting mentdebtinstrumentsandtheemergenceofliq- modelswasthateconomicgrowthintheU.S. uidsecondarymarketsinsuchsecuritiesoffers wouldgraduallyslowfromtheveryhighrateof additionalinformation. thefirsthalfoftheyeartoratesthatwereregarded Inaneconomywithwell-anchoredinflation asmoresustainable.Yet,alsoatthattime,more expectations,itcanbeexpectedthatthoseexpec- andmoreFOMCparticipantswerereportingsto- tationswouldnotbehighlyresponsivetochanges riesindicatingsharplyslowingconditionsfrom intheobservedrateofinflation.Alimitationof aneverincreasingnumberofrespondents.We surveydatainthisrespectisthathistoricallythey nowmeasurerealgrowthinthesecondhalfof havebeencharacterizedbystrongpersistence. 2000aslessthan1percent(annualrate),with Forexample,inthe34quarterlySurveysof negativegrowthinthethirdquarter.Inthis ProfessionalForecastersconductedbytheFederal instance,theanecdotesgaveabetterearlywarn- ReserveBankofPhiladelphiasincethesecond ingsignaloftheturninactivitythandidthe quarterof1998,theaverage10-year-aheadfore- forecastingmodels. castofCPIinflationhasdifferedfrom2.5percent Asimilarsituationmayprevailtoday.Statis- ononlyfiveoccasions.Onthosefiveoccasions, ticalstudiestodetectpass-throughfromrecent therangeoftheaverageforecastwasfrom2.3 energypriceincreaseshavefailedtoshowsignifi- percentto2.6percent. canteffectsinU.S.pricedata,butstoriesabout Incontrast,withhighlyliquidmarketsfor widespreadpass-througharebecomingincreas- bothnominalandindexedgovernmentdebtof inglycommon.Wemay—andIemphasize“may” comparablematurity,itispossibletoobserveboth becausemypurposeistomakeageneralpoint realinterestratesandinflationcompensation.If andnottoconductafullanalysisofthecurrent inflationexpectationsarewell-anchored,the situation—facemoreinflationpressurethancur- correlationbetweenchangesinyieldsonnominal rentlyshowsupinformaldata.Thegeneralpoint securitiesandindexedsecuritiesshouldbequite isthatthemoresuccessfularepolicymakersin high.Attheoppositeendofthespectrum,if exploitingregularitiesinformaldata,themore inflationexpectationsbecomeunanchored,then theywillhavetorelyonanecdotalinformation thecorrelationbetweenchangesintheseyields tomakefurtherprogressinstabilizinginflation. shouldbemuchlower.Intheextremeofhighly Refiningcollectionandanalysisofanecdotal unstableinflationexpectations,changesinnomi- informationpromisesearlierpolicyresponsesto nalyieldswillbedominatedbychangesininfla- changingeconomicconditions.TheFederal tionexpectationsandthecorrelationofchanges Reservealreadydevotessubstantialresourcesto innominalandrealyieldswillapproachzero. makingjudgmentaladjustmentstoitsformal Intheabsenceofprecisestatisticalforecast- forecastingmodel,butmyhunchisthatthere ingmodels,anotherpotentiallyusefulsourceof willbesignificantimprovementsinpolicyanaly- informationtoassessthestabilityofinflation sisfromstrengtheningoursourcesofanecdotal expectationsandthelikelycourseofthereal information. economyisreal-timeanecdotalinformation.The 4

William Poole (2006, June 15). Speech. Speeches, Federal Reserve. https://whenthefedspeaks.com/doc/speech_20060616_poole

@misc{wtfs_speech_20060616_poole,
  author = {William Poole},
  title = {Speech},
  year = {2006},
  month = {Jun},
  howpublished = {Speeches, Federal Reserve},
  url = {https://whenthefedspeaks.com/doc/speech_20060616_poole},
  note = {Retrieved via When the Fed Speaks corpus}
}