speeches · February 12, 2002

Speech

William Poole · President

The Role of Anecdotal Information in Fed Policymaking CenturyClubBreakfast OlinSchoolofBusiness WashingtonUniversityinSt.Louis St.Louis,Missouri February13,2002 There was a time when the Federal DespitethejokesaboutFedpolicy,Ithink Reserveencouragedapublicpercep- fewpeopletodayhaveanydoubtabouttheFed’s tion of the mystique of policymaking. objectives.Policymakersemphasizeandreempha- That isnottheFed’sviewtoday,butmy sizetheimportanceofachievinglowandstable reading of press commentary suggests that the inflationandthattheFedwillacttothefullextent old perception has not disappeared. There is, ofitspowers,consistentwiththatobjective,to however, a difference. Today, rather than refer- cushionfluctuationsinincomeandemployment. ring to the mystique of policy, people are more FinancialmarketsunderstandthattheFedispre- likely to refer to policy as “obtuse,” “incompre- paredtoactdecisivelyintimesofnationalemer- hensible,” or“deliberatelyconfusing.”Chairman gencyandfinancialmarketdistress,asevidenced Greenspan sometimes takes delight in saying byFedactionsinresponsetotheterroristattacks such things as “Senator, if you understand what lastSeptember11,totheRussiandefaultandnear I just said I must have made a mistake.” Althoughmystiquehasturnedintofodder collapseofLongTermCapitalManagementinthe forcartoonists,anditishealthythatwesmileat latesummerandfall1998,andthestockmarket ourselvesfromtimetotime,thereisaserious panicin1987. issueinvolved.Beginningabout35yearsago, Myaimthismorningistoilluminateapart developmentsinmacroeconomictheorybeganto ofthepolicyprocessthatis,Ibelieve,notvery makeclearthattheperformanceoftheeconomy wellunderstood—theuseofinformal,oranec- dependscriticallyonmarketexpectationsabout dotal,informationinthepolicyprocess.I’lldis- howeconomicpolicymakerswillactinthefuture. cussthenatureofthisinformationandhowwe Asourunderstandingoftheseissueshasdeep- useit.I’llstart,though,byoutliningtheroleof ened,ithasbecomeclearthatoneofthekey formaldata. dimensionsofasuccessfulmonetarypolicyis Beforeproceeding,Iwanttoemphasizethat thatthepolicymakersneedtohavewell-defined theviewsIexpressherearemineanddonot goalsandaclearplanastohowtheywillgoabout necessarilyreflectofficialpositionsoftheFederal achievingthegoals.Bothoftheseneedtobe ReserveSystem.Ithankmycolleaguesatthe understoodinthemarkets;otherwise,theecon- omyfacestheequivalentofabrokenplayinfoot- FederalReserveBankofSt.Louisfortheirassis- ball,wheresomemembersofateamthinkone tanceandcomments,especiallyHowardWall, playisbeingrunandothersthinkanotherplay Researchofficer,butIretainfullresponsibility isbeingrun. forerrors. 1 MONETARYPOLICYANDINFLATION THE ROLE OF FORMAL DATA Togetanideaofthescaleofthedifficultyof usingformaldata,let’stakealookatthetimeliness AND ITS PITFALLS anduncertaintyofestimatesofGDP,themost Toavoidmisunderstanding,Imustbeginby comprehensivemeasureofeconomicconditions. emphasizingthatthebasicpictureoftheeconomy Justovertwoweeksago,theadvanceGDPestimate comesfromtheformaldatapublishedbystatisti- wasreleased,indicatingthatrealGDPhadrisen calagencies.Amongotherthings,thedatainclude atanannualrateof0.2percentinthefourth thenationalincomeaccounts;labormarketsta- quarterof2001.Abouttwoweeksfromnowwe tisticsonemployment,unemploymentandrelated willseepreliminaryGDPestimatesforthefourth measures;priceandwagestatistics;industrial quarterandamonthafterthatwewillseefinal productionandcapacityutilization;financial estimates. marketstatisticsonmonetaryvariables,interest Since1978,two-thirdsoftherevisions rates,securitypricesandbankingmarkets;and betweentheadvanceestimateandthefinalesti- internationaltradeandcapitalflows.Wehave mateofrealquarterlygrowthwerebetween–0.6 availablesimilardataoneconomiesaroundthe and0.9percentagepoints.Thismeansthatthe world. likelyrangeofthefinalestimateoffourth-quarter Ourbasicknowledgeoftheeconomydepends realGDPgrowth—whichwewon’tseeuntilthe ontheformaldata.Weusethesedatainourecono- metricmodels.Formaldatahavemanyadvan- endofthisquarter—isbetweenminus0.4and tages.Weknowthestatisticalproceduresbywhich plus1.1percent.Tocompoundtheproblem,the thedataarecreatedandhavehistoricalrecords, so-calledfinalestimateisn’tthelastestimate. wellbackintothe19thcenturyinmanycases, Everysummer,inJulyorAugust,thefinalesti- fromwhichwecanstudyregularitiesofeconomic matesarerevisedandeverythreeyearsweget behavior. majorrevisions.Since1978,latestestimateshave Iliketothinkofinformaldataasproviding differedfromfinalestimatesbyanaverageof1.2 insightintotheformaldata.Theinsightruns percentagepointsineitherdirection.Thus,the acrossseveraldimensions,includingtimeliness latestestimatefor1980,say,changesovertime. andpotentialmeasurementandevenconceptual Asaconsequence,economistsliketosaythat errorsintheformaldata. historyisneverwhatitusedtobe!Inprinciple, Considertimeliness.Whenmakingmonetary theestimateskeepgettingbetterasthestatisti- policydecisions,membersoftheFederalOpen ciansfindimprovedsourcedata,refineestima- MarketCommittee(FOMC)needtoknowasmuch tionmethods,andimproveunderlyingconcepts. aspossibleaboutcurrentandfutureeconomic Whenoneconsiderstheenormoustaskof conditions.Unfortunately,theformaldataon estimatingthesizeoftheU.S.economy,these whichwerelylagcurrenteconomicconditions. problemsmightseemsmall.But,formakingmone- Forexample,theFedhasanincompletepicture tarypolicydecisions,theycanmakeacritical oftheeconomy’stwomostimportantindicators, difference.Infact,therangeofuncertaintyover growthandinflation.TheBureauofEconomic growthratescanimplyoppositeshort-runmone- Analysisreleasesformalestimatesofgrossdomes- tarypolicyresponses.Giventheuncertainty,itis ticproduct(GDP)withlagsofamonthormore. oftenbestforpolicymakerstosittight,waiting Moreover,thedataaresubjecttofrequentand fortheuncertaintytoberesolvedbynewinfor- majorrevisions.Priceindexesalsoareproduced withalagandaresensitivetofactorsthatmay mationandreviseddata.Whatthismeans,obvi- betemporary,suchasfluctuatingenergyprices, ously,isthatsometimesitisclearinhindsight andmeasurementerror.Giventheseproblems,it thatpolicyactionshouldhavecomesooner,or iseasytoseehowmonetarypolicymakinghas eveninadifferentdirection. oftenbeenlikenedtodrivingacarwithablacked- Ofcourse,officialgrowthandinflationdata outwindshieldandfogged-upsidewindows. arenotallthatwehavetogoon.Mostfinancial 2 TheRoleofAnecdotalInformationinFedPolicymaking dataareveryup-to-date,andfuturesmarkets HOW WE OBTAIN ANECDOTAL allowustopeerintothefuture—oratleastinto INFORMATION markets’expectationsofthefuture.Inaddition, somerealeconomicdata,suchasinitialunem- TheFedgathersitsanecdotalinformation ploymentclaims,autoandsteelproduction,and fromawiderangeofsources.Directorsofthe electricityconsumption,areavailableeveryweek. FederalReserveBanksandtheirBranchesprovide Becausethesedataareusedtoconstructtheoffi- writteneconomicreportsofconditionsintheir cialGDPestimates,theycanprovidepartialpic- regions.ReserveBankpresidentsandeconomists turesofcurrent-quarterGDP. travelaroundtheirDistrictsmeetingwithbusiness Otherdataareusedasleadingorcoincident peopleandbankersdiscussingconditionsintheir indicators.Economistsandanalystsrelyonpast industries.ReserveBanksmaintainanetworkof patternsoftheseindicatorstoprovideinsightinto industrycontactswhoarecontactedonaregular thecurrentbusinesscyclephase.Averageweekly basisinadvanceofFOMCmeetings. hoursinmanufacturingarealeadingindicator Wemakeadditionalefforttomaintaincontacts becausefirmstendtoadjustworkhoursbefore inbellwetherindustries,suchasfreightandtrans- increasingordecreasingtheirworkforce.Non- port,whoseactivityiscloselyrelatedtototal agriculturalemploymentisacoincidentindica- economicactivity.Wealsopaycloseattentionto torbecauseittendstoriseandfallwithGDP.In therealestateindustry,wherethelevelofactivity addition,somedataareusedtoidentifyturning mightbeagoodindicatoroftheconfidencethat pointsintheeconomy.Forexample,manyanalysts peoplehaveinthefuture.Afterall,formostpeo- followtheratioofinventoriestosalesbecauseit plethepurchaseofahomeisthelargestfinancial hastendedtopeakatthesametimethattheecon- commitmentthattheywillevermake.Iftheyare omyisinatrough. willingtocontinuebuyinghomeswhenthe WhiletheFedreliesheavilyonformaldata economyisslowing,ashasbeentruerecently, andsophisticatedstatisticalmethodsforanalyz- theymustbereasonablyconfidentabouttheir ingthedata,staffandpolicymakersalikespend personaleconomicoutlook. alotoftimecollectingandusinganecdotalinfor- Inaddition,oureyesandearsarealways mationthatwegatherfromanextensivenetwork open,lookingforemergingeconomictrends.A ofcontacts.Thisanecdotalinformationhelpsus well-knownexampleofthishands-onapproach toseewhatisgoingonintheeconomyalmostas isthatthepresidentoftheMinneapolisFedhas itishappening.Also,becauseitiscollectedfrom beenknowntomakeregularvisitstolocalshop- thepeoplewhoareactuallymakingday-to-day pingmallstocountthecarsintheparkinglots.I businessdecisions,ithelpsustounderstandwhy routinelymakeanumberofphonecallstobusi- trendsinthedataareoccurring. nesscontactsbeforeFOMCmeetings.Iseekspe- FortherestofthistalkIamgoingtodiscuss cializedinformationhighlydependentoncurrent therolethatanecdotalinformationplaysinFed policymaking.Iwilloutlinethewaysinwhich circumstances.Forexample,duringoneofmy wegatherthisinformationandthendescribethe tripsafterSeptember11,Istruckupaconversa- variouswaysthatweuseit.Iwillalsodiscuss tionwithaSouthwestflightattendantandlearned brieflysomerecentevidencethatanecdotalinfor- thattheairlinewascontinuingtohireandtrain mationaddsvaluebeyondwhatwegetfromother newflightattendants.Thatinformationreinforced sources.Andfinally,Iwilltouchonsomeofthe whatIknewfrompressreports,thatSouthwest dangersandpitfallsofrelyingtooheavilyon wasnotcuttingflightsandhadanoptimistic anecdotalinformation. viewofthefuture. 3 MONETARYPOLICYANDINFLATION HOW WE USE ANECDOTAL sivelyforlabor.Myinterpretationofthisinforma- tionwasthatfirmswereconvincedthatinflation INFORMATION wouldremainlowandthattheydarednotlet Thevastamountofanecdotalinformation wagecostsincrease,becausetheywereunlikely collectedthroughouttheFederalReserveSystem tobeabletorecoverthosecostsinhigherprices. isusedforavarietyofpurposes.Mostsystemati- Foraspecificexample,theownerofamanu- cally,itisusedtoproducethe“Summaryof facturingfirminLouisvilletoldushowhewas CommentaryonCurrentEconomicConditions”— abletoexpandemploymentandproductioneven commonlyknownastheBeigeBook—whichis thoughmostofhistraditionalworkforce—prime- publishedtwoweeksbeforeeveryFOMCmeeting. agedmen—werealreadyemployed.Thischallenge ToproducetheBeigeBook,eachFederalReserve ledhimtorethinkhisproductionprocesstomake BankgathersinformationaboutitsDistrictthrough itabettermatchfortheworkersthatwereavail- anetworkofcontacts.The12Districtreportsare abletohim.Theresultwasthat,wheneverpos- collectedtogetherbyanassemblingBank,whose sible,hisproductionmethodswerechangedto staffpreparesthenationalsummaryofeconomic reducetherequirementforphysicalstrength. conditions.TheBeigeBook,bytheway,isavail- Weheardmanysimilarstoriesabouthow ableonthewebsitemaintainedbytheBoardof firmswereprovidingbasicskills,makingtheir Governors. workschedulesmoreflexible,providingtrans- Theanecdotalinformationcollectedmakes portationfortheirworkers,andsoforth,tocope itswayintoFOMCmeetings,whereFedgover- withtherapidlychangingnatureoftheirwork- norsandReserveBankpresidentspresenttheir force.Withoutthisfirst-handknowledgeoffirms’ viewsontheeconomicoutlook.Inadditionto abilitytorespondtocompetitivechallengesand theiruseinassessingthestateoftheeconomy, newenvironments,theFedmightnothaveknown theanecdotesmightbeusedtoillustrateapoint, thatunemploymentcouldkeepfalling,atleast thusaddingimpacttothecomments.Forexample, foratime,withoutinflationbeingignited. aBankpresidentcouldsaythat“themarketfor Ournetworkofcontactsisalsousefulto constructionmaterialinmyDistrictcontinuesto identifyemergingtrends.Forexample,wellin betightandpricesarerising.”Or,hecouldsay advanceofitactuallyoccurring,wehadagood thesamethingandadd,“Thesituationissotight ideathatfirms’healthinsurancepremiumswould thatwehavehadreportsoftruckloadsofdrywall increaseatdouble-digitratesin2001.Weknew beinghijacked.”Theadditionoftheanecdote thisincreasewascomingbeforeitappearedin (whichhappenstobeanactualonefromanother officialdatabecauseourcontactstoldusinmid- District)addsmoretothereportthancouldsev- 2000aboutthehealthinsurancecontractsthey eralchartsortables. weresigningfor2001. Anecdotalinformationalsocanbeusedto Anothergoodexampleofanecdotalinforma- confirmortohelpunderstandongoingtrends tioncamefromoneofourBranchdirectorswho thatarisefromtheformaldata.Forexample,dur- notedinthesummerof2000thatloandemand ingthelate1990s,theunemploymentratefell athisbankwasfallingandthatotherfirmsinhis wellbelowwhatmanypeoplethoughtwasthe levelwhereinflationwouldstarttotakeoff.If areawerebeginningtoexperienceproblems.This wehadreliedonlyontheformaldata,theFed informationwasimportantbecause,atthetime, mighthaveoverlookedwhatfirmsandworkers theeconomywasgrowingrapidlyandnearlyall weredoingtodrivedowntheunemployment forecastsindicatedthatrapidgrowthwouldcon- rateandhowtheywererespondingtotightlabor tinue.Nevertheless,reportsofthissortcontinued markets.Forexample,welearnedfromourcon- tosurfacethroughouttherestof2000andinto tactsinbusinessesthatcompanieswerewilling 2001,helpingtheFedtogetaheadoftherecession toleavepositionsunfilledratherthanbidaggres- bystartingtoloweritsfederalfundsratetarget 4 TheRoleofAnecdotalInformationinFedPolicymaking earlylastyear,eventhoughofficialGDPdata VALUE ADDED availableatthetimesuggestedthattheeconomy Ihavealreadymentionedsomeoftheways wasstillgoingstrong. thatanecdotalinformationaddsvaluetoFed Whilemostoftheanecdotalinformationcol- policymaking,butmycommentsthemselvesare lectedbytheFedisusedtosupplementother onlyanecdotal.Recently,though,economists informationatourdisposal,forsomeone-time withintheFederalReserveSystemhavetriedto events,theanecdotalreportsbecometheprimary usemoretechnicalmethodstoevaluatetheBeige sourceofinformation.Intheseinstances,standard Bookasanindicatorofpresentandfutureeco- dataarenotreliableguidesbecausehistoryhas nomicactivity.Thefirstsuchstudywasdoneat notrecordedapatternforhowtheeconomyis theMinneapolisFedandfoundthattheBeige likelytorespond.Anexampleofsuchanevent Bookhasbeenanaccuratepredictorofrealgrowth wastheseriesofterroristattacksonSeptember11, inthecurrentquarter.Theyalsofound,however, whichhadimmediateanddramaticeconomic thattheBeigeBookdidnotimproveuponprivate consequences,althoughwehadnohistorytouse sectorforecastsofrealgrowth.Thestudycon- topredictwhattheseconsequencesmightbe. cludedthattheBeigeBook’svalueisnotinfore- castingeconomicactivity,butinreflectingthe Nonetheless,wewereabletouseournetworkof economy.Inotherwords,theBeigeBookwas contactstogetagoodideaofthesectorsthatwere foundtoaddvaluebyprovidinginsightand affectedthemost,weeksbeforeanyformaldata contextnotfoundinformalforecastingmodels, wereavailable. whilenotimprovingontheperformanceofthese WefoundoutveryquicklythattheFed’s models. injectionofliquidityintothebankingsystemhad Morerecently,researchfromtheDallasFed beensuccessful,inthatfewbanksreportedhav- hasfoundstrongerempiricalsupportfortheBeige ingliquidityproblemsdespitethenear-complete Bookasapredictivetool.Thisstudyfoundthat shutdownoffinancialmarkets.Wealsofound thenationalsummaryoftheBeigeBookhassig- thatretailsalescametoahaltinthetwotothree nificantpredictivecontentforcurrentandfuture daysaftertheattacksbutsurgedbacktonear- quarterlygrowth.Further,itfoundthattheBeige normallevelsbytheweekend,thatmanufacturers Bookhaspredictivecontentbeyondwhatispro- intheDistrictwereanticipatingthattheywould videdbyprivateforecasts.Ofparticularinterest bereducingtheiroutputbyanaverageof10per- isthat,accordingtothisstudy,theBeigeBook cent,andthatautosalesfortheperiodmightbe appearstohavebeenbetterthanalternative downbyasmuchas50percent.Withinafew methodsatidentifyingturningpointsinthe economy. weeks,ourcontactstoldusthatautosalesin Anotherpotentialsourceofaddedvalueis Octoberwereinfactstrong,inresponsetothe fromtheBeigeBook’s12regionalsummaries. zero-interestfinancingincentiveofferedbyauto ThedecentralizednatureoftheBeigeBookmeans manufacturers.Allofthisinformationwasvital thattheFedhasaninstrumentfordetecting intheweeksimmediatelyfollowingtheattacks, regionaldifferencesinthebusinesscycle.Business whentheFedhadtoreactveryquicklywhile cyclefluctuationsarenowthoughttobemore navigatingtheunchartedwatersofSeptember heterogeneousacrossregionsandsectorsthan andOctober.Indeed,basedonanecdotalreports theyusedtobe.Hence,onehearsreferencestoa andexperience,butwithoutanysubstantial “rollingrecession”thataffectsdifferentregions amountofformaldataapplyingtotheperiod withgreatestseverityatdifferenttimes.Stateand afterSeptember11,theFOMCcuttheintended regionaldata,however,aremuchlesscomplete federalfundsrateonSeptember17andagainon thannationaldata.Forexample,grossstateprod- October2. uctdataareproducedwithatwo-yearlag.Inthis 5 MONETARYPOLICYANDINFLATION void,theBeigeBookcanhelppinpointfocal bestofshape.Ithenaskedthemiftheiranswers pointsofsucharollingdownturnorarolling werereallywhattheythoughtwouldhappenor recovery.Infact,theDallasFedstudysuggests iftheyinsteadreflectedwhattheyhopedwould that,takenasawhole,theregionalsectionsof happen.Severalofthemthenadmittedthattheir theBeigeBookaddpredictivepowerbeyond outlookwasprobablymorehopethanexpecta- whatthenationalsummaryprovides.Further, tions,andadjustedtheiranswersaccordingly. someoftheregionalsections—includingthat Thebiasesoftheeconomistcollectingand producedbytheSt.LouisFed—havebeenuse- analyzingtheanecdotalinformationmayalso fulontheirowninpredictingGDPgrowthone meanthatitisnotrepresentativeofgeneraleco- quarterahead. nomicconditions.Forexample,theeconomist mighttendtopaymoreattentiontoanecdotes thatfithisorherpreviouslyheldbeliefs.Asa DANGERS AND PITFALLS consequence,theoverallimpressionthatiscon- veyedfromtheanecdotesin,forexample,the Thereareanumberofdangersandpitfalls BeigeBook,mighttendtoreflecttheeconomist’s inherentwithanecdotalinformation,soagreat personalviews.Itmightalsobethattheoddor dealofcareshouldbetakeninusingit.Forone quirkyanecdotesaretheonesthathavethemost thing,despitetheeffortthattheFedputsintoit, influencebecausetheyarethemostinteresting, thenumberofcontactsissmallfromastatistical eventhoughtheymightnotberepresentativeof standpoint,andtheyarenotselectedrandomly. generaltrends. Theytendtobeinbusinessesthatarefamiliarto adirectorofaFedBank,whohavevoluntarily agreedtoserveasacontact,orwhosemanageror SUMMARY AND CONCLUSIONS ownerhasbeenaskedtoserveonaFedBank’s advisoryboard.Becauseofthisselectionprocess, Becauseanecdotalinformationisinherently numerousbiasescanarise.Forexample,perhaps unscientific,theFedwillcontinuetorelymost thetypeofpersonwhowouldserveasacontact heavilyonformalmethodswhenmakingmone- orbeamemberofanadvisoryboardwouldalso tarypolicydecisions.Useofformalstatisticsis tendtobemoresuccessfulthantheaveragebusi- animportantdiscipline.Nonetheless,because nessperson.Ifso,thentheinformationthatthe thesemethodsprovideafarfromperfectpicture Fedreceiveswouldtendtounderrepresentfirms oftheeconomy,theFedshouldcontinuetouse thataremorelikelytobeexperiencingdifficulties. anecdotalinformationtohelpfillthegaps.Anec- Also,theresponsesmightreflectthebiases dotalinformationimprovesuponourunderstand- ofthecontactsratherthanbeaccuraterepresen- ingofwheretheeconomyisandwhereitmight tationsofconditions.Thisbiaswouldnotarise begoing,mostnotablybyprovidinginformation throughanyconsciousmisrepresentation,but aheadofformaldata.Theprocessofgathering perhapsthroughthetendencyforsuccessful theinformationputsusindirectcontactwith businesspeopletobemoreoptimisticthanthe peopleactuallymakingday-to-dayeconomic averageperson.Anexampleofthisoccurreddur- decisions.Theinformationforcesustoquestion ingarecentlunchatourBankwhenwehosteda theformaldataandprovidesaviewoftheecon- groupofresidentialhousingdevelopers.Thefirst omythatformalmethodssimplymiss. timeIwentaroundthetableaskingthemfortheir Thisconstantprocessoftestingtheformal outlookonfutureconditionsintheirindustry, dataagainsttheanecdotalreports,andviceversa, nearlyeveryoneofthemwasquiteupbeat.This strengthensourunderstandingofbothtypesof nearunanimitywassurprisingbecausethiswas information.IknowthatIdidnotunderstandthe notlongafterSeptember11andmostofthem scaleandimportanceoftheeffortwhenIcameto alsofeltthattheoveralleconomywasnotinthe theFedaboutfouryearsago,andIsuspectthat 6 TheRoleofAnecdotalInformationinFedPolicymaking fewobserversoutsidetheFedappreciatetherole ofanecdotalinformationinthemonetarypolicy process.ThatiswhyIthoughtthistopicdeserves someattention,andIhopeI’vebeensuccessful inexplainingtoyouhowtheprocessworks. 7

Cite this document

APA

William Poole (2002, February 12). Speech. Speeches, Federal Reserve. https://whenthefedspeaks.com/doc/speech_20020213_poole

BibTeX

@misc{wtfs_speech_20020213_poole,
  author = {William Poole},
  title = {Speech},
  year = {2002},
  month = {Feb},
  howpublished = {Speeches, Federal Reserve},
  url = {https://whenthefedspeaks.com/doc/speech_20020213_poole},
  note = {Retrieved via When the Fed Speaks corpus}
}