speeches · February 24, 2004

Speech

William Poole · President

Best Guesses and Surprises CharlotteEconomicsClub Charlotte,NorthCarolina February25,2004 PublishedintheFederalReserveBankofSt.LouisReview,May/June2004,86(3),pp.1-7 Ihave a simple message today—that anyone CONSENSUS OUTLOOK TODAY interestedinmonetarypolicyshouldspend WhatistheconsensusoutlookfortheU.S. less time on economic forecasts and more economytoday?Numerousforecastsareinthe time on implications of forecast surprises. publicdomain,fromgovernmentandprivate If you are in the forecasting business, it makes sources.1Directcomparisonoftheseforecastsis good sense to write at length about the forecast notstraightforward,astherearedifferencesamong and the analysis behind it. For the rest of us, the thevariablesforwhichtheforecastsarepresented, forecast provides the baseline for examining the differencesintheforecastingtimehorizons,and most important policy issues. The true art of differencesinaveraging,withsomeforecasts good monetary policy is in managing forecast presentedonafourth-quartertofourth-quarter surprises and not in doing the obvious things basisandotherspresentedonanannual-average implied by the baseline forecast. overannual-averagebasis. I’llproceedbyoutliningtheconsensusout- Nevertheless,atpresentaremarkablyuniform lookandthenwilldiscusshowIviewthejobof picturefromaperusalofthesevarioussources dealingwithsurprises.I’llemphasizethatthe emergesforthemajoreconomicindicators.Real keyissueisthatmonetarypolicyresponsesto grossdomesticproduct(GDP)isforecasttogrow surprisesoughtnottoberandom,butaspre- inthe4to4½percentrangefromthefourthquar- dictableaspossible.Therearesomeprinciples terof2003tothefourthquarterof2004.Inflation ofgoodresponsesthatmakeiteasierforstudents asmeasuredbytheconsumerpriceindex(CPI) ofmonetarypolicytopredictwhattheFederal isforecastinthe1½to2percentrangeandas Reservewilldo. measuredbytheGDPchainpriceindexisfore- Beforediggingintothesubstanceofmysub- castinthe1to1½percentrangeoverthathorizon. ject,IwanttoemphasizethattheviewsIexpress Theunemploymentrateisforecasttobearound donotnecessarilyreflectofficialpositionsofthe 5½percentbythefourthquarterof2004. FederalReserveSystem.Ithankmycolleagues MycolleaguesaroundtheFederalOpen attheFederalReserveBankofSt.Louisfortheir MarketCommittee(FOMC)tableareonaverage comments—especiallyBobRasche,seniorvice slightlymorebullishthantheabovepicture:The presidentanddirectorofresearch,whoprovided midpointoftherangeofforecastsofrealGDP specialassistance.However,Iretainfullrespon- growthincludedintheMonetaryPolicyReport sibilityforerrors. totheCongresssubmittedtwoweeksagois4½ 1 ApartiallistincludesBlueChipEconomicIndicators(publishedeachmonth);SurveyofProfessionalForecasters(compiledquarterlybythe FederalReserveBankofPhiladelphia);WallStreetJournalForecastingSurvey(publishedearlyJanuaryandJulyofeachyear);Congressional BudgetOffice,BudgetandEconomicOutlook(publishedeachFebruaryandAugust);CouncilofEconomicAdviserseconomicoutlook (publishedeachFebruaryintheEconomicReportofthePresidentandupdatedeachJuly);andFederalReserveSystem,MonetaryPolicy ReporttotheCongress(publishedeachFebruaryandJuly),containingtheeconomicprojectionsoftheFederalReservegovernorsand ReserveBankpresidents. 1 MONETARYPOLICYANDINFLATION percentforthefourthquarterof2004overthe BecauseofthelaginreleasingtheGreenbook, fourthquarterof2003.Themidpointoftherange thisstudyanalyzestheGreenbookforecasting ofinflationforecasts(measuredbythechained recordupthrough1996.Theothercomparisons priceindexforpersonalconsumptionexpendi- includeforecaststhrough2001. tures)inthatreportis1.13percent,andthemid- TheauthorscomparedtheBlueChipfore- pointoftheforecastfortheunemploymentrate casts,theGreenbookforecasts,andtheFOMC inthefourthquarterof2004is5.38percent.I’ll members’forecastsagainstanaive,same-change refertothisforecastasthe“FOMCmembers’fore- forecastbeginningin1980forbothrealoutput cast.”TheforecastreflectsasurveyofFOMC growthandinflation.Threedifferentforecasting members,butisnotanFOMCforecastperse horizonswereexamined:six,twelve,andeight- becausetheCommitteedoesnotdebateandvote eenmonths.3Notsurprisingly,theaccuracyof ontheforecasttomakeitaCommitteeforecast theforecastsdeterioratesastheforecastinghori- assuch.NorisittheBoardofGovernorsstaff zonislengthened.Foraone-year-aheadforecast, forecastpreparedforeachFOMCmeetingand theroot-mean-squaredforecasterror(ameasure reproducedintheGreenbook;theGreenbookis ofthedispersionoftheforecastsaroundthereal- releasedwiththeFOMCmeetingtranscriptonly izedvalue)forrealoutputgrowthisontheorder afterafive-yearlag. of1.4percentagepointsforallthethreesetsof forecastsconsidered.Theroot-mean-squaredfore- Thoseforecasterswhoriskinterestratefore- casterrorforthenaiveforecastisconsiderably casts(theWallStreetJournalForecastingSurvey, larger,ontheorderof2.2percentagepoints. theBlueChipEconomicIndicators,andthe Clearly,theforecastaccuracyoftheforecast- CongressionalBudgetOfficeEconomicOutlook) ersissubstantiallybetterthanthatofthenaive expectTreasurybillratesaround1.2percent, forecast,butstillleavesalotofroomforsurprises. andten-yearTreasurybondratesaround4.6per- Tomakethispointclearintoday’scontext,iffor centeitheronanannualaveragebasisoratthe conveniencewesaythattheGDPgrowthforecast middleof2004.IshouldnotethatFOMCmem- is4percentoverthefourquartersending2004:Q4, bersdonotmakepublicforecastsofinterestrates. thenonestandarderrorleavesuswithaforecast bandof3to6percentgrowthoverthisperiod.If weweretohavea3percentoutcome,everyone HOW RELIABLE IS THE wouldfearthattherecoveryisfaltering;ifwe CONSENSUS OUTLOOK? weretohavea6percentoutcome,themostlikely Thesmalldispersionamongforecaststoday characterizationwouldbethatwehaveaboom isnotunusualandshouldnotbeinterpretedasa onourhands. measureoflikelyforecastaccuracy.Overthe Moreover,keepinmindthatonestandard years,numerousstudieshaveinvestigatedthe deviationoneithersideoftheexpectedvalue forecastaccuracyofprivateforecasters.More doesnotbyanymeansexhausttherangeofpos- recentlyseveralstudieshavecomparedtheaccu- sibleoutcomes.Asaroughapproximation,one racyofboththeFOMCmembers’forecastsand timeoutofthree,theone-year-aheadforecastof theGreenbookforecastswiththoseofprivate realoutputgrowthwillfalloutsidearangeof forecasters.Onerecentanalysiswasproduced plusorminus1.4percentagepointsofthestated byWilliamGavinandRachelMandalinthe forecastnumber.Assumingasymmetricaldistri- researchdepartmentattheFederalReserveBank butionofforecasterrors,whichseemsreasonable, ofSt.Louis.Theirpaperwaspublishedlastyear thereisaprobabilityofabout0.16thatrealout- intheInternationalJournalofForecasting.2 putgrowthoverthenextfourquarterswillexceed 2 WilliamT.GavinandRachelJ.Mandal,“EvaluatingFOMCForecasts,”InternationalJournalofForecasting,2003,19(4),pp.655-67. 3 DetailscanbefoundinTables4and5ofGavinandMandal. 2 BestGuessesandSurprises 6percentandaprobabilityofabout0.16that forecast,inDecember2003,was3.1percent, outputgrowthwillfallshortof3percent. whichisthecurrentlypublishednumberforreal Clearlytherangeoferrorassociatedwiththe growthin2003over2002.Theinitialreleaseof currentstateoftheforecastingartfailstodistin- theconsensusCPIinflationforecastfor2003over guishbetweenareallystrongexpansion—a 2002inJanuary2002was2.4percent.Thisfore- boom—andafalteringrecovery.Andtheaccu- castchangedverylittleoverthefollowing24 racyofinflationforecastsisnotmuchbetter.On months,increasingto2.5percentinmid-2002 aone-year-aheadforecastinghorizon,theroot- andthensettlingdownat2.3percent,theCPI mean-squarederrorofinflationforecastsisin inflationratethatwasrealizedfor2003over2002. therangeof0.75to0.9percentagepoints.This Therelativelylowvariabilityoftheconsensus forecastingrecordisnotmuchbetterthanwould forecastfor2003maskstheheterogeneityamong havebeenachievedwithanaiveforecasting theindividualsurveyrespondentsthatreflects model.AswithrealGDP,thereisasignificant theinherentuncertaintyofeconomicforecasts. probabilitythattheoutcomecouldfalloutside Inearly2002,therangeofforecastsforrealgrowth theone-standard-deviationband.Anditisalso in2003acrosstheBlueChiprespondentswas truethataninflationoutcomeoutsidetheband 2.0to6.0percent.Bythebeginningof2003this wouldcreateconsiderableconcern. rangehadnarrowedtoabout2.5to4.5percent. Publishedforecastsarerepeatedlyupdated Onlyafterthemiddleof2003,whendatafrom onevershorterandshorterforecastinghorizons. sixoftheeightquartersinvolvedinthecompu- TheresultsreportedbyGavinandMandalindi- tationofayear-over-yeargrowthratewereavail- catethatasthehorizonbecomesshortertheuncer- able,didtherangeofindividualforecastsdrop taintysurroundingtheforecastrealizationis below1percentagepoint. reduced—though,perhapssurprisingly,notbya Asimilardispersionisobservedamongthe particularlylargeamount.Theiranalysissuggests individualforecastsofCPIinflationfor2003.At thatforrealGDPgrowththeroot-mean-squared thebeginningof2002therangeofforecastswas forecasterroronaneighteen-monthhorizonis fromlessthan1.0percenttoalmost4.5percent. between1.5and1.9percentagepointswhileat Byearly2003thisrangehadnarrowedtoless asix-monthhorizonitisreducedtoonly1.3per- than1.5toslightlymorethan2.5percent.Itwas centagepoints.Forinflation,theroot-mean- onlyafterSeptember2003thattherangeoffore- squaredforecasterrorattheeighteen-month castsshrunktolessthan1percentagepoint. horizonisbetween1.1and1.3percentagepoints butissubstantiallyreducedtoaround0.5per- centagepointsatasix-monthhorizon. SOME EXAMPLES OF FORECAST Aswegothroughtheyear,theforecastfor SURPRISES 2004willbeupdatedasresultsforeachquarter comein.Anexampleofthisprocessisprovided Forecastsurprises,orforecasterrors,area bythemonthlyBlueChipconsensusforecastfor standardpartofthepolicylandscape.Itisvery 2003.TheinitialreleaseoftheBlueChipforecast easytocriticizeforecastsandextremelydifficult forlastyearwasinJanuary2002;thus,wehave tocomeupwithbetterforecasts.Thefactisthat arecordof24successiveBlueChipforecastsfor goodforecastersproducestate-of-the-artforecasts. 2003.Theinitialforecastwasforayear-over-year Policymakersmustdealwithforecastsurprises. growthrateof3.4percent.Throughthefirsthalf Whatarethesourcesofthosesurprises? of2002theconsensusforecastwasrevisedup Thedifficultyofforecastingturningpoints slightly,reachingapeakof3.6percentinJune. ineconomicactivityismostsignificant.What- Thereaftertheconsensuswasfairlysteadily evercreatesarecessionalsocreatesaforecast reviseddownwardoverthenextyear,reachinga surprise.Forexample,theOctober2000Blue troughof2.3percentinAugust2003.Thefinal Chipconsensusforrealgrowthoverthefive 3 MONETARYPOLICYANDINFLATION quartersending2001:Q4wasforaverysteady thefourquartersendinginthefourthquarterof quarter-to-quarterexpansioninrealGDPinthe eachyearfrom1996through1999.Onaverage rangeof3.3to3.6percentatanannualrate.The theseforecastsunderestimatedrealgrowthby businesscycledatingcommitteeoftheNational 2.1percentagepointsforthesefouryears.The BureauofEconomicResearchlaterdatedacycle rangeofforecasterrorswasfrom2.4to1.9per- peakinMarch2001andatroughinNovember centagepoints.Theerrorswereallinthesame 2001.Actualquarter-to-quarterrealgrowthduring directionandallofsignificantsize.Duringthe thisperiodrangedfrom–1.3to2.1percent.Thus, samefouryearstheCPIinflationforecasterror fivemonthsbeforetheonsetofthe2001recession, averaged0.0percent—rightonthebutton.How- theBlueChipconsensusforecastmissedthe evertheforecasterrorsfortheindividualyears recessioncompletely! rangedfrom–1.2to0.6percentagepoints. MypointhereisnottopickontheBlueChip Thereasonsforforecasterrorsaremany.Some respondents.MycolleaguesontheFOMChad reflectincompleteunderstandingofhowthe nogreaterforesight.IntheminutesoftheFOMC economyworks,suchastheerrorsinprojecting productivitygrowth,orconsumerbehaviorright meetinginOctober2000wecanreadthat, afterthe9/11terroristattacks.Somereflectunpre- “[l]ookingahead,they[FOMCmembers]generally dictableshocks,suchasasharpchangeinenergy anticipatedthatthesofteninginequityprices pricesorthe9/11terroristattacksthemselves. andtheriseininterestratesthathadoccurred Somereflectfinancialdisturbances,suchasthe earlierintheyearwouldcontributetokeeping 1987stockmarketcrash.Whatevermaybethe growthindemandatamoresubduedbutstill reasonsforforecasterrors,theyareafactoflife. relativelyrobustpace.”4 AsecondnoteworthyexampleisOctober 2001.Intheimmediateaftermathofthe9/11 THE POLICY SIGNIFICANCE OF attacks,forecastersturnedextremelybearishon thenear-termprospects.TheBlueChipconsen- FORECAST UNCERTAINTY susforrealgrowthin2001:Q4intheOctober10, Whataretheimplicationsofthedocumented 2001,surveywas–1.3percent,witharangeof uncertaintysurroundingforecastsoffutureecon- forecastsfrom–3.2to0.8percent.TheBlueChip omicactivity?Somedismissforecastsaltogether respondentswereparticularlypessimisticabout andviewthemasirrelevantforpolicybecause prospectsforthemanufacturingsector;thecon- theirerrorsaresolarge.Tome,thatresponseis sensuswasforgrowthof–3.1percent,witha completelywrong.Instead,policyneedstobe rangefrom–7.4to0.6percent.Wenowknow informedbythebestguessesincorporatedin thatin2001:Q4theeconomyreboundedtoa2.1 forecastsandbyknowledgeofforecasterrors. annualrateofgrowthinrealGDP,ledbyanall- Forecasterrorscreaterisk,andthatriskneedsto timerecordrateoflightvehiclesales.Keepin bemanagedasefficientlyaspossible.Andthe mindthatthisGDPgrowthratewasabovethe surprisesthatcreateforecasterrorsalsocreate forecastofeverysingleoneofthe50plusBlue theneedforpolicychangesthatcannotbeantici- Chiprespondentsatthebeginningofthequarter. patedinadvancebecausethesurprisescannot Athirdexampleisthehistoryofrealgrowth beanticipated. andinflationforecastsinthesecondhalfofthe Giventhesizeofforecasterrors,wewillfre- 1990safterthenow-apparentincreaseintrend quentlyobservetheeconomyevolvingalonga productivitygrowth.Considerthemidpointof substantiallydifferentpathfromthatportrayed therangeofforecastsofrealgrowthandinflation byconsensusforecastsonlyashorttimeearlier. byFOMCmemberspreparedeachFebruaryfor Withnewlyavailableinformation,forecasterswill 4 MinutesoftheFOMCMeetingofOctober3,2000,FederalReserveBulletin,January2001,p.23. 4 BestGuessesandSurprises adjusttheirprognostications,andpolicymakers, marketparticipantscouldnothavebeenfore- suchastheFOMC,willadjusttheirviewofthe warnedofsuchactions,theFOMCwouldbe appropriatepolicystance.Iftherevisedviewof derelictinitsresponsibilitiesifitfailedtoact. theappropriatepolicystanceissufficiently Giventheshock,theFOMC’sactionoughtnotto changed,policymakerscanandshouldimple- beasurprise.Therealsurprisewouldariseifthe mentthechangesinpolicysettings,suchasthe FOMCweretodonothinginthefaceofashock intendedfederalfundsrate,thattheybelieveare callingforaction. consistentwiththenewinformation. Acoupleofexamplesareillustrative.Consider Suchpolicyactionsshouldbeimplemented theAsiandebtcrisis,theRussiandefault,and whetherornottheywillcomeassurprisesto thecollapseofLongTermCapitalManagement marketparticipantsandthegeneralpublic.Here (LTCM)inAugust-September1998.Noonefore- itisimportanttobeclearaboutthedistinction sawthiscombinationofevents,norwasthe betweenapolicyandapolicyaction.Apolicy impactoftheseeventsontheliquidityofmajor shouldbeviewedasaruleorresponseregularity financialmarketspredictable.Atthetimeofthe thatlinkspolicyactions,suchasadjustmentsin FOMCmeetingonAugust18,1998,federalfunds theintendedfederalfundsrate,tothestateof futuresforcontractsasfaroutasDecember1998 theworld.Apolicyislikeadecisiontodrive65 weretradingwithinacoupleofbasispointsof milesperhour;giventhatpolicy,apolicyaction theprevailing4.50percentintendedfundsrate. istheadjustmentoftheacceleratortomaintain Nevertheless,bytheconclusionoftheFOMC thetargetspeed.Iftheeffectsofwindandhills meetingonNovember17,1998,theintended onspeedcannotbeanticipated,thenneithercan fundsratehadbeenreducedinthreestepsbya thepolicyactionsofacceleratoradjustments.A totalof75basispoints,includingacutof25 goodpolicyrequiresclarityaboutpolicyobjec- basispointsatanunscheduledconferencecall tivesandasmuchprecisionaspossibleastohow meetingonOctober15. policyactionswillrespondtonewinformation Theseratecutsinthefallof1998wereasur- tobestachievethepolicyobjectives. prisefromthevantagepointofearlyAugust.But Inthemonetarypolicycontext,anticipated therealsurprisewouldhavebeeniftheFOMC policyactionsarenaturallytiedcloselytothe hadtotallyignoredtheRussiandefaultandcol- forecast.Tomaintainthepolicyofachievinglow lapseofLTCM.Holdingtheintendedfundsrate andstableinflation,unanticipatedpolicyactions constantgiventhemarketturmoilwouldnothave mustoftenaccompanyforecastsurprises.Should beenconsistentwiththeFed’sresponsibilityto aninflationaryshockhittheeconomy,forexam- contributetofinancialstability. ple,thatshockandanincreaseintheFOMC’s Icouldwalkthroughnumerousotherexam- targetfederalfundsratewouldbothbesurprises. plestodrivehomethepointthatakeyfeatureof OnnumerousoccasionsIhavestatedmyview monetarypolicyismeasuredandappropriate thattheFOMCshouldcommunicateitsintention responsestotheconstantstreamofsurprises.In aboutmonetarypolicyaspreciselyaspossibleto discussionsofmonetarypolicy,Iwouldliketo getmarketsin“synch”withpolicy.Myview seemuchmoreemphasisonappropriatepolicy shouldnotbeinterpretedtoimplythattheFOMC responsestosurprisesandpotentialsurprises canonlyactafterithas“prepared”marketpartici- andmuchlessemphasisonforecasts.Anover- pantsforachangeintheintendedfederalfunds emphasisonconsensusforecastsmayleadmarket rate.Therewillbetimeswhensignificantunfore- participantstoafalseprecisionintheirviews seeneconomicnewswillberevealedverysud- aboutthefederalfundsrategoingforward.Itis denly—eventsthatcanbeappropriatelydescribed muchmoreproductivetothinkthroughthesorts as“shocks.”If,inthejudgmentoftheFOMC, ofthingsthatmighthappenandtheappropriate suchnewscallsforpolicyactionseventhough responsetosuchevents.Acarefulanalysisof 5 MONETARYPOLICYANDINFLATION riskshelpstopreparethemindfordealingwith event.Whileeconomistscontinuetodebate surpriseswhentheyoccur.Marketparticipants exactlyhow“longandvariable”theresponseof andtheFOMCshouldnotfocusonthepredict- theeconomyistoapolicyaction,thereisacon- abilityofaparticularpathforthefundsrate,but sensusofprofessionalopinionthatittakesat insteadonthepredictabilityoftheresponseof leastseveralmonthsbeforetheeconomyresponds. theFOMCtonewinformationabouttheeconomy. Ofcourse,judgmentisalwaysnecessarytodeter- minewhetheranyparticularshockislikelytobe transitory. PRINCIPLES OF FOMC Sometransitoryshocksoccurbecause“news” RESPONSES TO “SHOCKS” doesnotprovideaccurateinformation.Many datareleasesaresubjecttoseveralrevisions,and Foratleast40years,economistshavebeen oftenthereviseddatarevealadifferentpicture tryingtodevelopaquantitativecharacterization thanthatportrayedbytheinitialrelease.Quarterly ofFOMCpolicyactions—apolicyreactionfunc- GDPdataarerevisedtwiceuntilthe“finalesti- tion.5Areviewpublishedin1990analyzed42 mates,”andthese“finalestimates”aresubjectto publishedexamplesofattemptsatcharacterizing annualbenchmarkrevisionsandcomprehensive FOMCbehavior.6Since1993,theprevalentframe- revisionsatroughlyfive-yearintervals.Fora worktoquantifyFOMCactionisthe“Taylor recentexampleofsignificantdatarevisions,initial rule.”7Noneoftheseeffortshaveachievedtheir measurementofthe2001recessionsuggested objective.8Inmyjudgment,itisnotpossibleat negativerealGDPgrowthinthesecondandthird thecurrentstateofknowledgetodefineaprecise quartersof2001,whereasthecurrentlyavailable reactionfunctionoftheFOMC,andperhapsit datameasurenegativerealgrowthasearlyasthe neverwillbepossible. thirdquarterof2000andforthefirstthreequar- Itispossible,however,todescribesomegen- tersof2001. eralprinciplesthatguideFOMCbehaviorand Themonthlypayrollemploymentdataare thatcanbeappliedbymarketparticipantstoform revisedinthetwomonthsfollowingtheinitial expectationsabouthowtheFOMCwillrespond tonewandunexpectedinformation.Ibelieve releaseandarerevisedagainatthebeginningof thattheseprinciplesarefairlywidelyaccepted, thefollowingcalendaryeartoincorporatebench- butdifferentFOMCmemberswillapplythemin markstotheunemploymentinsurancesystem differentwaysatdifferenttimes.Andtheprinci- records.Withtheinitialreleaseofthepayroll plesalwaysinvolveadegreeofjudgment. employmentnumbersforOctober2003atthe ThefirstprincipleisthattheFOMCwillnot beginningoflastNovember,themeasured respondto“shocks”thatareseenasverytransi- monthlyincreaseof126,000workersgenerated tory.Policyshouldonlyreactto“shocks”that thehopethatthetransitionfromthetwo-year arelongerlasting—highlypersistent.Thereason “joblessrecovery”toaperiodofrapidemploy- forthisprincipleisquitestraightforward—noth- mentgrowthwasathand.TheOctoberdataas ingthattheFOMCcandowilloffsettheimpact currentlyrevisedindicateanincreaseinpayroll ontheeconomyofa“heretoday,gonetomorrow” employmentofonly88,000workers.Wenow 5 AnearlyanalysisisWilliamG.DewaldandHarryG.Johnson,“AnObjectiveAnalysisoftheObjectivesofAmericanMonetaryPolicy,1952-61,” inDeaneCarson(ed.),BankingandMonetaryStudies,Homewood,IL:RichardD.Irwin,1963,pp.171-89. 6 SalwaS.Khoury,“TheFederalReserveReactionFunction:ASpecificationSearch,”inThomasMeyer(ed.),ThePoliticalEconomyof AmericanMonetaryPolicy,NewYork:CambridgeUniversityPress,1990,pp.27-50. 7 JohnB.Taylor,“DiscretionversusPolicyRulesinPractice,”Carnegie-RochesterConferenceSeriesonPublicPolicy,1993,39(0),pp.195-214. 8 AnillustrationofthedeviationsofthepredictedfromactualfundsratefromaTaylorrulewithaconstanttargetinflationrateoverthepast decadecanbefoundinFederalReserveBankofSt.Louis,MonetaryTrends,p.10,availableathttp://research.stlouisfed.org/publications/mt/. 6 BestGuessesandSurprises believethatonlyinJanuary2004didmonth-to- andhouseholdsurveysarequitedistinctstatisti- monthpayrollemploymentgrowthexceed cally,asthesurveys’coverageandmethodsdo 100,000jobs—andonlyananemic112,000at notoverlap.Dataoninitialclaimsforunemploy- that. mentinsurancesupplementthemessagefrom Thepresenceofmeasurementerrorinindi- thetwomainsurveys.Inaddition,theFederal vidualeconomicdataseries,particularlyinthe Reserveaccumulatesawealthofanecdotalinfor- initialreleasesofsuchdata,requiresthatanalysts mationonthelabormarketfromcontactsacross andpolicymakersexaminemultipledatasources thecountry;muchofthisinformationappearsin foraconsistentpictureoftheunderlyingtrends theBeigeBook.Whendatafromdiversesources ineconomicactivity.Thereisnowaytogeneralize pointinthesamedirection,confidenceinthe aboutthisissue.Differentserieshavedifferent directionindicatedisincreased.Conversely,when sourcesoferroranddifferentfrequenciesoflarge thesignalsareconflicting,thereisoftengood revisions.Someseriesaremoresubjecttospecial reasontoreservejudgmentanddelaypolicy disturbances,suchasbadweather,thanothers. action.Analysisofthestrengthofhousehold TheFederalReservehastremendousstaffexpert- demand,businessinvestmentdemand,inflation- iseandaccesstostatisticalagenciesthatpermit arypressures,andallotherkeyelementsofthe ittoformthebestjudgmentspossibleonthese picturecananddoesproceedthesameway. trickyissues. OnceFOMCmembershavereachedacon- Asanaside,letmeofferanotherobservation. clusiononwheretheeconomyisandwhereitis Currently,theFederalReserveenjoysveryhigh heading,therearesituationswherethedecision credibility.Amongotherbenefits,thatcredibility ontheappropriatepolicyresponseisstraightfor- enablestheFedtoreacttoitsbestjudgmentabout wardandothercaseswheretheappropriate whatincomingdatamean.TheFeddoesnothave responseisproblematic.Considersomeexamples toacttomaintainappearances.Forexample,my ofeasycasesfirst.Supposetheeconomyhas impressionisthatthereweretimesinthe1970s shownrobustgrowthwithlowinflationfora whentheFedfailedtoreacttoaccumulatingevi- periodoftime,andinformationaccumulatesthat denceofeconomicweaknessforfearthattoreact leadstoareasonableinterpretationthatbothreal beforeinflationdeclinedcouldbeinterpretedas growthandinflationpressuresareincreasing,or alackofinflation-fightingresolve.Policyactions bothdecreasing.Facedwithsuchinformation, designedprimarilytoattempttoaffectexpecta- centralbankerswithadualmandate(suchasthe tions,eventhoughcontrarytothefundamentals, FOMC)arelikelytorespondbyraisingorlower- ultimatelyincreaseuncertaintyasitbecomes ing,asappropriate,thenominalinterestrate clearthattheactiondidnotfitthefundamentals. target.Whencredibilityishigh,moreover,the Successinbringingdowninflationandkeeping decisionneednotbeaquickone.Butwhenthe itdownmeansthattheFedcanignoreasurgein issueisclear,thecentralbankmustactvigor- priceindicesoranyothertroublinginformation ouslyenoughtoensuremaintenanceofanon- ifitsbestjudgmentisthatthedatareflectasta- inflationaryequilibrium. tisticalaberrationoratransitoryevent. Indeed,suchresponsesare,qualitatively, I’vearguedthattheFederalReservemust exactlythosepredictedbytheTaylorrule.Under analyzethedataforpotentialstatisticalproblems theseconditions,marketparticipantsandprivate andthatitmustdoitsbesttosortouttransitory agentscanlikelyaccuratelyanticipatethedirec- disturbancesfromlongerlastingones.Another tion,ifnotthetimingandmagnitude,ofFOMC dimensionoftheproblemisthatacentralpartof actions.TheFOMCpracticesinceFebruary1994 makingsuchjudgmentsistocollateinformation ofgenerallyrestrictingchangesintheintended fromavarietyofsources.Employmentdatapro- federalfundsratetoregularlyscheduledmeetings videanexcellentexample.Theestablishment andmakingchangesinmultiplesof25basis 7 MONETARYPOLICYANDINFLATION pointshasdemonstrativelyimprovedthepre- mandate.”Inparticular,afundamentalresponsi- dictabilityofthetimingandmagnitudeofchanges bilityenvisionedbythearchitectsoftheFederal intheintendedfundsrateinsuchcases.9 ReserveActwasthatthenewcentralbanking Theappropriatepolicyresponseinother structureavoidrecurrenceofepisodesoffinan- casesislessclear.Supposetheeconomyhas cialinstabilityandbankingpanicssuchasthose shownrobustgrowthwithlowinflationfora thatoccurredregularlyinthelate19thandearly periodoftimeandinformationarrivesthatleads 20thcentury.TheGreatDepressionwasaFederal toareasonableinterpretationthatrealgrowthis Reservepolicyfailureofthefirstorder.Recent decreasingandinflationisincreasing.Ahistori- episodes,suchasthe1987stockmarketcrash, calepisodeofthissortisthe“oilshock”inlate thefinancialmarketupsetinthefallof1998, 1973and1974.Here,onecomponentofa“dual Y2K,and9/11provideevidencethattheFederal mandate”signalsapolicyactioninonedirection Reservehaslearnedlessonsfromthe1930swell andtheothercomponentintheoppositedirection. andcandealeffectivelywithsystemicthreatsto Thisisadilemmacaseinwhichthebehaviorof financialstability. theeconomyispullingthepolicymakerstobe Itisimportanttounderstand,however,that botheasierandtighter.Aweightingofobjectives concernsaboutfinancialstabilityrequirethat andcarefulattentiontolong-runconcernsisnec- theFederalReservesortoutshocksthatraise essary.Evenifacentralbankweretofollowa suchconcernsfromthosethatdonot.Notevery Taylorruleapproachtoimplementingpolicy largeeventcreatesrisksforthefinancialsystem changes,intheabsenceofdisclosureoftheexact asawhole.Thelargestockmarketdeclinethat reactionfunction,outsideobserverswouldbe startedinearly2001didtendtodepresseconomic unabletopredictthepolicyaction.Itisunrealis- activitybut,unlikethecrashof1987,neverraised tictobelievethatacentralbankcanprovidethe issuesofsystemicstability. transparencyrequiredforoutsiderstoaccurately predictpolicyactionsinallsuchcircumstances. Predictabilityinthedilemmacasescanbe CONCLUDING COMMENT improvedandtheappropriatepolicyresponse Forecastuncertaintyisafactoflife.Forecasts facilitatedwhenacentralbankhasacredible arelikenewspapers.Justaslastweek’snewspaper commitmenttomaintaininglowinflationinthe isoflittlevalueinunderstandingtoday’snews, longrun.Inthesecircumstances,thecentralbank lastmonth’sforecastisoflittlevalueindetermin- canlikelypursueshort-runstabilizationobjec- ingtoday’spolicystance.Oldnewspapersand tiveswithoutsignificantinfluenceonexpecta- oldforecastsareprimarilyofhistoricalinterest. tionsoflong-runinflation.Incurrentparlance, Theobviousfactthatweinsistonusingthe inflationaryexpectationsare“wellanchored.”In mostup-to-dateforecastavailableindicatesthat suchenvironments,policyactionsaimedatshort- forecastschange,sometimessubstantially,with runstabilizationarelikelytobemoreeffective. newinformation.Forecastsarevaluableinformu- Underconditionsofhighcredibility,policy latingmonetarypolicy,butitisofcriticalimpor- actionsarelikelymorepredictable. tancethatwenotallowtoday’spolicysettingsto TheFederalReservehaspolicyresponsibili- becomeentrenchedinourminds. tiesbeyondanarrowinterpretationofthe“dual 9 WilliamPoole,RobertH.Rasche,andDanielL.Thornton,“MarketAnticipationsofMonetaryPolicyActions,”FederalReserveBankofSt. LouisReview,July/August2002,84(4),pp.65-93;andWilliamPooleandRobertH.Rasche,“TheImpactofChangesinFOMCDisclosure PracticesontheTransparencyofMonetaryPolicy:AreMarketsandtheFOMCBetter‘Synched’?”FederalReserveBankofSt.LouisReview, January/February2003,85(1),pp.1-9. 8

Cite this document

APA

William Poole (2004, February 24). Speech. Speeches, Federal Reserve. https://whenthefedspeaks.com/doc/speech_20040225_poole

BibTeX

@misc{wtfs_speech_20040225_poole,
  author = {William Poole},
  title = {Speech},
  year = {2004},
  month = {Feb},
  howpublished = {Speeches, Federal Reserve},
  url = {https://whenthefedspeaks.com/doc/speech_20040225_poole},
  note = {Retrieved via When the Fed Speaks corpus}
}