2025美聯(lián)儲(chǔ)研究與統(tǒng)計(jì)司的經(jīng)濟(jì)計(jì)量（英）

FinanceandEconomicsDiscussionSeries

FederalReserveBoard,Washington,D.C.ISSN1936-2854(Print)

ISSN2767-3898(Online)

BeyondtheStreetlight:EconomicMeasurementintheDivisionofResearchandStatisticsattheFederalReserve

CarolCorrado,ArthurKennickell,andTomazCajner2025-019

Pleasecitethispaperas:

Corrado,Carol,ArthurKennickell,andTomazCajner(2025).“BeyondtheStreetlight:EconomicMeasurementintheDivisionofResearchandStatisticsattheFederalReserve,”FinanceandEconomicsDiscussionSeries2025-019.Washington:BoardofGovernorsoftheFederalReserveSystem,/10.17016/FEDS.2025.019.

NOTE:StaffworkingpapersintheFinanceandEconomicsDiscussionSeries(FEDS)arepreliminarymaterialscirculatedtostimulatediscussionandcriticalcomment.TheanalysisandconclusionssetfortharethoseoftheauthorsanddonotindicateconcurrencebyothermembersoftheresearchstaffortheBoardofGovernors.ReferencesinpublicationstotheFinanceandEconomicsDiscussionSeries(otherthanacknowledgement)shouldbeclearedwiththeauthor(s)toprotectthetentativecharacterofthesepapers.

BeyondtheStreetlight:EconomicMeasurementintheDivisionofResearchandStatisticsattheFederalReserve*

CarolCorrado,formerlyoftheDivisionofResearchandStatistics**

ArthurKennickell,formerlyoftheDivisionofResearchandStatistics***TomazCajner,DivisionofResearchandStatistics

Thisversion:December27,2024

Abstract

Thispaperwaswrittenfortheacademicconferenceheldincelebrationofthe100thanniversaryoftheDivisionofResearchandStatistics(R&S)oftheFederalReserveBoard.TheworkoftheFederalReserveturnsstronglyonempiricaleffortstounderstandthestructureandstateoftheeconomy,andR&ScanbethoughtofasoperatingalargefactoryfordiscoveringanddevelopingdataandanalyticalmethodstoprovideevidencerelevanttothemissionoftheBoard.Thispaper,assignaledbyitstitle,illustrateshowthemeasurementresearchcomponentoftheR&SfactoryoftenlooksfarbeyondcurrentconventionstomeettheneedsoftheBoard—andhasdonesosinceitsearliestdays.ItwouldtakeafarlongerpapertoprovideacompletehistoryandevolutionofmeasurementactivitiesinR&S;here,weprovideanindicativereviewfocusingonselectedareasfromwhich,webelieve,itiseasytoconcludethatR&Shasbeen—andlikelywillcontinuetobe—animportantinnovatorineconomicmeasurement.

Keywords:Datacollectionmethodsandestimationstrategies;Businesscycles,productivity,andpricemeasurement;Financialaccountsandfinancialdata;theSurveyofConsumerFinances;Blendeddata.

*ThispaperwaspreparedfortheR&SCentennialConference,November8,2023.TheauthorsthankDave

Byrne,whosecontributionsareclosetothatofco-author,andAriGelbard,whoprovidedresearchassistance.TheauthorsaregratefulforhelpfulcommentsfromtheConferencediscussantsforthispaper,ChristopherCarrollandDanielSichel.

**SeniorPolicyScholar,CenterforBusinessandPublicPolicy,McDonoughSchoolofBusiness,Georgetown

University(carol.corrado@)

***AffiliatedScholar,StoneCenter,CityUniversityofNewYorkGraduateCenter(arthur.kennickell@)

PAGE

10

Introduction

Incommemorationofthe100thanniversaryofthemergeroftheDivisionofAnalysisandResearchwiththeOfficeoftheStatisticianin1923tocreatetheDivisionofResearchandStatistics(R&S)oftheFederalReserveBoard,thepaperprovidesanoverviewandsomeinsightsintothesubsequenthistoryandroleofeconomicmeasurementinthedivision.

Inlargepart,R&SoperatesasalargefactoryforproducinginformationrelevanttoitsmissionofsupportingtheBoardthroughanalysisoftheeconomyandtheeffectsofpolicydecisions.Asmall,butimportant,partofthatinformationproductioninvolvesthecreationofprimarydataviasurveydatacollectionasinthecaseoftheSurveyofConsumerFinances.Amuchlargerpartinvolvesusesofawidevarietyofsecondarydatasources,sometimesblendedwithothersources.Somesuchsecondaryusesfeedintothecreationofhigher-levelpublishedstatisticalseries,suchastheFinancialAccountsoftheUnitedStatesortheIndustrialProduction(IP)andCapacityUtilizationsystem,ortothecreationofroutinelyusedindicatorssuchashouseholddebtburdens.Inmanyothercases,theinformationaloutputistheresultofanexerciseinmodelingorestimationforinternaluse,suchaspotentialoutput;othersaresharedwiththepublicviaFEDSNotes,suchasarecentlydevelopedindexofcommoninflationexpectations

.1

Theseestimatesservetoclarifythecurrentstateofsomeaspectsoftheeconomyortotesttheappropriatenessofaparticularviewofeconomicstructure.Inacomplexdataecosystemsuchasthis,itisdifficultatbesttodefineneatlywhatconstitutes“measurement,”thesubjectofthispaper.Itwouldnotbepossibleinthescopeofthispapertoaddresseveryrelevantangle.

Instead,belowwesketchaframeworkofinformationuseanddiscusssomeofthekeyrelatedelementsofmeasurement.

Fromtheperspectiveofthepresent,threemutuallysupportingfactorsaremoststrikingabouttheoverallhistoryofmeasurementinR&S.First,thevarietiesandsourcesofdatahaveexpandedenormouslyovertime.Atonetime,simplyhavingreliablenationalincomeaggregatesandmoneysupplydatarepresentedprogress.Now,thegrowingrangeofrelevantdata—commercialdata,sensordata,administrativedata,anddataofothersortsfrommultiple

1SeeAhnandFulton(2020).

governmentagencies—canseemalmostboundless.Second,computingandrelatedtechnologiestocopewithsuchdatahavegrownroughlyinparallel.WhentwooftheauthorsofthispaperbeganworkingattheBoard,anyoneusingmorethanonemegabyteofcomputermemoryonthemainframe,thentheonlysourceofcomputingotherthanacalculatororpencilandpaper,wasrequiredtopresentabluecardsignedbyadivisionofficertoawindowinthecomputercenter.Evenwithwhatwasthenconsideredanadvancedcomputer,congestion,delays,andsystemcrasheswereoftenanissue.Suchtechnologyseemsalmostmedieval,comparedwithcurrentresources.Third,thesefactorssupportedthegrowingcomplexityofthecontemporaryeconomythatrequiredanexpanded,andmoredetailedandsophisticated,analysisbyBoardstaff.Wewillvariouslyreturntoeachofthesefactorsinthediscussionofspecificanalyticalareas.

AnimportantfacetofmeasurementinR&Shasbeen,andcontinuestobe,theinvestigationoftensionswithinandacrossdatasourcesforinterpretingdevelopmentsintheeconomy,oralternatively,tensionsbetweenatheoreticalpriorandincomingdataorempiricaltestresults.AremarkinapaperbyformerR&SDivisionDirectorDavidWilcoxexpressesthisperspective.

Whenresearcherstestandrejectanimplicationofatheoreticalmodel,theyusuallyassumethatthemodelisinerrorandthatsubsequentinvestigationshouldbedirectedtowardthedevelopmentofalternativemodelsthatmightbetteraccountfortheobservedcharacteristicsofthedata.Theyusuallyspendlittleeffortinvestigatingthecharacteristicsofthedatathemselvesorthesuitabilityofthedataforuseintheapplicationathand.Thispaperreversestheseprioritiesandinvestigatesthesourcedataandestimationmethodsusedtoconstructtheretail-salesandaggregateconsumptiondataintheUnitedStates,searchingespeciallyforimperfectionsthatmighthaveimplicationsfortheoutcomeofempiricalwork.(Wilcox1992,page922)

Thismindset(echoedinourpaper’stitle)playsanimportantroleintheworkthathassupportedanddevelopedthedivision’sstatisticalprogramsreviewedinthepaper.

Someimportantmeasurementsystemspresentalternativeconstructsthatconceptuallydonotdifferbutrathermaketensionsinunderlyingsourcedataandsurveystransparent,e.g.,thediscrepancybetweengrossdomesticproduct(GDP)versusgrossdomesticincome(GDI)is

almosta“feature”ofnationalaccounts.EmploymentfromtheBureauofLaborStatistics(BLS)householdsurveyisadjustedtoitsestablishmentemploymentsurvey’sconcepttoillustratealignmentofitsalternativesourcesforcollectingdataonemployment(BLSdoesnotfeaturethisanalysis,thougharecentwebsitedesignhasmadeiteasierforuserstofind).

Internally,IPislinedupwithgoodsGDP;theIPflow-of-goodssystemwithNIPAinventories;and,atonetimeatleast,savinginthenationalaccountswithitsconceptuallyconsistentcounterpartinthefinancialaccounts.Sortingthroughspecificdiscrepanciesisroutineintheconductofdivisionwork,butmanyofthepuzzlesortensionsthatariseinincomingdataarenothandedtousaspartofastatisticalreleaseorroutineconstruct.Theymustbedetectedandsortedthroughbeforetheycanbeidentifiedasaprobleminmeasurementratherthanaprobleminoureconomicpriors.

Theinfluenceofthis“beyondthestreetlight”mindsetgoesbeyondtheworkenvironmentoftheBoard.DistinguishedalumsofR&Shavetakenlessonsfromstudyingtensionsandpuzzlesindatatoothergovernmentagencies,academia,andthebroadworldofpolicy—exportsthatcouldbesaidtorivaltheinfluenceofthebodyofresearchundertakenbyR&Sperse.

Furthermore,morethanadozenR&SalumsserveorrecentlyhaveservedonadvisorycommitteesforU.S.statisticalagencies,internationalagencies,andcentralbanksintheircapacityasexpertsineconomicmeasurement.ThedivisionnurturesitsinfluenceoneconomicmeasurementresearchandpracticebybeingasponsoroftheNBER’sConferenceonResearchonIncomeandWealth(CRIW).TheCRIWholdsannualresearchconferencesandaworkshopattheNBERSummerInstituteoneconomicmeasurementthatbringtogetherpractitionersandleadingresearchersontopicsofmutualinterest.

ThispaperdoesnotattempttocaptureeveryaspectofeconomicmeasurementinR&Soverthepast100years.Todosowouldnecessitateafarlongerpaper,whichwouldbetediousreadingforallbutthemostdevoteddatageeks.Instead,wereviewfiveareasofworkthatseemtoustobeonesthatarelikelytohavegreatestsalience.ThefirstsectionbrieflyreviewstheIPandcapacitysystem,andthesecondsectioncoversthehistoryofthemeasurementresearchthatworkedtodevelopaclearerpictureofproductivitygrowth.Thethirdreviewsthe

FinancialAccountsoftheUnitedStates(formerlyFlowofFundsAccounts)fromthepioneeringresearchthatledtotheirfounding,theiruses,andthedevelopmentoftheassociateddistributionalaccounts.Thefourthsectionaddressesthedevelopmentofprimarydatasources,largelyfinancialmicrodata,inthedivision,includinganin-depthreviewoftheemergenceofthemodernSurveyofConsumerFinances.AfifthsectiondiscussestherecenteffortsinR&Stocombinenontraditionaldatasourcestogainbetterinsightintonear-termbehavioroftheeconomy.Thatsectionmakesnoteofarecentinitiativetoutilizeabroadspectrumofsometimesnon-traditionaldatasourcestoconstructindicatorsofbusinessconditions,themotivationfortheIPindexmorethan100yearsago,whichiswherewebegin.

TheIPandCapacitySystem

Thecollectionoffinancialmarketandbankingstatisticsbyacentralbankseemsanobviousduty.Lessobviousisthatmanycentralbanksissuenonfinancialeconomicstatisticsgeneratedtosupporttheimplementationofmonetarypolicy,e.g.,theBankofJapanissuesthecountry’spriceindexesforproducers(“corporategoodsprices”),exports,andimportstothisdayinsupportoftheBank’saimofmaintainingpricestability.

TheFederalReserve’searliestnonfinancialstatisticsgrewoutofaneedtomonitorfluctuationsinindustrialactivityandcommodityprices.Inthefirsteditionin1913ofwhatwouldbecomealifetimeofworkonmeasuringandanalyzingbusinesscycles,WesleyClairMitchellpointedouttheneedforanindexofthephysicalvolumeofproductionandtradeasarequisiteforunderstandingthegrowthandfluctuationsineconomicactivity(Mitchell1913).

TheMarch1918FederalReserveBulletinannouncedthatindexesofbusinessconditionswereunderdevelopment(FederalReserveBoard1918).Followingplansoutlinedinsubsequentissues,thestatisticaltablesintheJanuary1919issueoftheBulletinbegantoreportmorethan100differentdataseriesonthephysicalvolumeofproductionandtrade.Theseriescoveredrefinedagriculturalproducts(e.g.,sugarandflour),miningactivity(e.g.,coalandcrudeoil),andproductionbymanufacturesofironandsteel,nonferrousmetals,lumber,textiles,refined

gasoline,andwoodpulpandpaper.In1922theFederalReserveissued“IndexesofTradeandProduction”withdatabeginningin1919(FederalReserveBoard1922).TheindexesaggregatedavailablephysicalproductionvolumeseriesintomajorcomponentsandpresentedatimelysummaryofmonthlyindustrialactivitythatcametobeknownasmonthlyIP.

TheBulletinannouncementsoftheearly1920sgivenohintoftherelationshipbetweentheFederalReserve’seffortstocompileindexesofproductionandrelatedworkonmeasuringbusinesscyclesbyWesleyMitchellandothers.MitchellhadbeenengagedinamultiyearprojectwiththeNBERofthetimetocollectandorganizemonthlydataoneconomicactivity.Inastudyofpricecontrolsandcommodityshortagesduringthefirstworldwarforthe

government’sWarIndustriesBoard,MitchellusedtheNBERdatatoconstructthefirstindexofthephysicalvolumeofproductionandtradefortheUnitedStates(Mitchell1919,pages44-46).

Mitchell’sNBERdatagatheringprojectandworkforthefederalgovernmentundoubtedlyinfluencedtheparallelprojectattheFederalReserve,aswellaseffortselsewhere.OnesuchefforttookplaceatAmherstCollege,conductedbyWalterStewart(Stewart1921).Stewart’s“indexofproduction”appearstohavedirectlyinfluencedtheindextheFedpublishedin1922.WalterStewartleftAmherstthatyearfortheFederalReserveandinNovember1923becamethefirstdirectoroftheFederalReserveBoard’snewlycreatedDivisionofResearchandStatistics.

Bythe1940stheproductionindexincludedallindustriesinmanufacturing,havingintroducedmoredetailedcoverageofnewandgrowingindustrieslikemachinery.Formanyoftheseindustriesnomonthlyphysicalvolumedatawereavailable,andannualcomprehensiveproductionindexesjointlyproducedbystaffoftheFederalReserveandCensusBureauwereusedtobenchmarkmonthlyseriesbasedonproductionworkerhour

s.2

ThebenchmarkindexesbuiltintomonthlyIPwereforthefiveyearchangetomostrecentquinquennialCensusyear,

2Forexample,the1947indexeswereissuedasasupplementtothe1947CensusofManufactures(e.g.,U.S.BureauoftheCensusandBoardofGovernorsoftheFederalReserveSystem,1952)

withdetailedindustriesaggregatedusingvalueaddedinformationfromthepriorCensusyearasweights.

ThejointFederalReserve-CensusBureaueffortconsideredtheimpactofalternativebaseperiodsforthe(implicit)priceweightsusedinaggregation.Whilethefive-yearbenchmarkIndexesusedinIPwerebasedonweightsforthepriorCensusyear,comprehensiveproductionindexesusingcurrentCensusyearweights,and,interestingly,“cross”weightsusinginformationfrombothyearsalsowerereported.Thoughthecross-weightedindexesdidnotexactlyfollowtheFisherformulationnowinwidespreaduseinUSeconomicstatistics,theworkwasanearlyefforttoevaluateandavoidsubstitutionbiasinofficialstatistics.Comparedwithavolumemeasurethatusedpricesinasingleyeartoaggregateitscomponentsinallyears,thefive-yearlinkedbenchmarkindexesusedtoconstructmonthlyIPyieldedamoreaccurateindicatoroftrendsinindustrialactivity.

ThebroadercontextoftheFederalReserve-CensusBureauefforttodevelopcomprehensivemeasuresofthechangeinmanufacturingoutputwas,asinthecaseofMitchell’sinitialphysicalvolumeindexandtheBoard’sintroductionofmonthlyIPin1922,arecognitionoftheneedtomonitorproductionstrippedoftheeffectsofwartimeinflation.

3

Similarly,theBoard’scapacityutilizationmeasuresgrewoutofaneedtomonitorshort-runpressuresonaggregatesupplythatmightaggravateinflationduringthebooming1960s(deLeeuw1962;1966).

deLeeuw’sapproachtothemeasurementofcapacityexploitedsurveydataonutilizationratesinmanufacturingtodeveloptimely,model-basedcapacityindexesconsistentwithmonthlyIP.TheutilizationdatacamefromathenwidelyreportedsurveyconductedbyMcGrawHill.AmajorexpansionandimprovementtothecapacitysystemcamewiththeincorporationoftheCensusBureau’sannualSurveyofPlantCapacityin1989,whichhadcollectedinformationonindustry-levelfactoryutilizationinthefourthquarterofeachyearsincethe1970s.RichardRaddock(1990)providesadescriptionandanalysisofthenewdata;heandCharlieGilbert

3ItshouldbenotedthattheBureauofEconomicAnalysis(BEA)onlybegantopublishannualconstant-dollarGNPinJanuary1951,whichevolvedintoaquarterlyconstant-dollarGNPseriesthatwasinitiallyissuedinDecember1958,about36yearsaftertheBoard’sissuanceoftheinitialmonthlyIPindex.

designedandimplementedtheexpandedsystem.ThemorerigorouslyconstructedmeasureswerefurtherexaminedbyCorradoandMattey(1997)whohighlightedtheireconomicinterpretationandutilityinbothmacroandmicroanalysis.

ManyrefinementstomonthlyIPhavetakenplacesincetheearlydays,mostofwhicharedescribedinresourcesavailableontheBoard’swebsite.Theintroductionofannual(vsquinquennial)chainingintheproductionbenchmarkswasanespeciallynotablestep,describedinCorrado,Gilbert,andRaddock(1997).Thearticlealsosetsouthowratios(e.g.,capacityutilization)maybepreservedinvolumeindexes.Inanotherkeyimprovement,theestimationofmonthlyIPnowmakesuseofcomprehensive,quarterlysurveydataonplantutilizationcollectedbytheCensusBureau(QPC)andfundedbytheBoard

.4

JohnStevensandNormanMorinshepherdedtheCensusBureau’sintroductionofthenewsurveyintothecapacitymeasures,andCharlieGilbertimplementeditsuseintheestimationofmonthlyIP.Theideahereisthatquarterlychangesinutilizationmainlyreflectchangesinoutput,andincludingthisinformationintheestimationofmonthlyIPdiminishesthesystem’srelianceoninputdataasindicatorsofproductionatahighfrequency.

Alltold,aquantumimprovementinthemonthlyandannualprocessingoftheBoard’sIPindextookplacesincethelate1980s.CharlieGilbertwasresponsibleformostoftheseachievements,asspelledouttoattendeesathisrecentretirementpartyafter40yearsattheBoard.(Toobadwedonothavearecordingofthosetributes!)Ofcourse,manyR&SstaffmembersalsocontributedtotheremarkableautomationofthepresentIPandcapacityutilizationsystem,includingoneofthedivision’sstatisticians,BillCleveland,andeconomistsEricBartelsman,NormanMorin,andKimBayard,tonamebutafew.

AreviewoftheemergenceofthemodernIPandcapacityutilizationsystemwouldnotbecompletewithoutconsideringthedivision’sworktokeeppacewiththerapidevolutionofinformationtechnology(IT),thesubjecttowhichwenowturn.

4Thenewsurvey,theQuarterlySurveyofPlantCapacityUtilization(QPC)replacedtheannualsurveyasapilotsurveyinthefourthquarterof2007andhascontinuedasaquarterlysurveysincethen.Thisnewquarterlysurvey,likeitspredecessorannualsurvey,alsocollectsinformationonplantoperatinghours.

ITOutput,InvestmentandProductivity

ResearchcontributionsbyR&SstafftothemeasurementofITindustriesandinvestmentandtheirimpactsonproductivityarereviewedinthissection.Webeginthisreviewinthe1990s,wheninterestintheeconomicimplicationsofthe“tech”sectorsoared.BEAhadintroducedquality-adjustedpricesforcomputersinthenationalaccountsin1986(Coleetal.1986),afterwhichITequipment—commonlyunderstoodtoincludecommunicationequipmentaswellascomputers—emergedasoneofthefastestgrowingsegmentsofaggregatedemand.

Interestinunderstandingtheroleofquality-changeininflationmeasuresinthedivisionalso

intensifiedinthe1990s.ThesubjectofqualitychangehadbeeninthespotlightinthedivisionwhenCharlesParteeasR&SDivisionDirectorestablisheda“CommitteeonPricesandPrice

Measurement”composedofacademicconsultantsin1965.Anowclassiccollectionofresearch

paperseditedbyZviGrilicheswasaproductofthiscommittee(Griliches1971).Weknowof

littleelseinthedivisiononpriceindexesandqualitychangepriortothiseffort.

5

Webeginwiththewell-knownpolicydilemmafacedbyChairmanGreenspaninthelate1990s

anddiscusshowITandmeasurementofpricessurfacedintheordinarybusinessnewsofthe

time.

Productivityinthe1990sand“TheManWhoKnew”

The1990sproductivityrevivalintheUnitedStatesbeganinthefirsthalfofthedecadebutdiscernmentofthechangetookholdslowlyamongprofessionaleconomists.Productivityanalystsdidnotreachbroadconsensusthataggregateproductivitygrowthpickedupinthemid-1990suntiltheturnofthedecade,whenanaccelerationininformationtechnology(IT)

capitaldeepeningandfastertrendsinnewlyavailableproductivitydatabecameapparent(e.g.,(OlinerandSichel2000).Bycontrast,theFed’scallthatlaborproductivitygrowthwas

5Saveforthisgoodie:AsnotedbyGrilichesinhisclassicpaperonqualitychangeinautomobiles,whichwasreprintedintheFedvolumecitedabove,theearliestconstructionofahedonicpriceindexwasbyAndrewCourtwhileemployedasaneconomistbytheAutomobileManufacturersAssociationinDetroitinthe1930s(Court1939).FilesdiscoveredwhilecleaningoutanofficeintheIndustrialOutputsectioncirca1990revealedthatCourtcorrespondedwithstaffintheoldBusinessConditionsectionforcommentsonhishedonicapproachtomeasuringandanalyzingpricechangesforautomobiles.

accelerating—madebyChairmanGreenspaninargumentstotheFederalOpenMarketCommittee(FOMC,orcommittee)atitsSeptember1996meeting—wasremarkablyearly.

Thiseventinmonetarypolicyhistoryanditsbeneficialaftermathfortheeconomyhasbeenchronicledinthepopularpressandinbooks,includingBobWoodward’sTheMaestro(Woodward2000)andSebastianMallaby’sbiographyofAlanGreenspan,TheManWhoKnew(Mallaby2016).Alesser-knownaftermathisthatinsidetheBoard,theeventpromptedR&StoexpendresourcesonbettermeasuringandanalyzingtheITsectoroftheU.S.economy.

ConsiderfirstthepuzzlethatsurfacedinU.S.macrodatainthesummerof1996:Wagerateswererisingbuttheincreaseinlaborcostswasnotboostinginflation.Thisconundrumcouldbeexplainedbyanaccelerationinlaborproductivity(e.g.,duetoefficiencygainsfromIT-inducedinnovationsthatweremuchdiscussedatthetime)—butaboostinproductivitywasnotapparentintheavailableofficialdata.Thissuggestedtomonetarypolicycommentators(andsomecommitteemembers)thatinterestratesneededtomovehighertostaunchalooming

pickupininflation,aviewthatwasalsobuiltintothestaffforecast.

ToChairmanGreenspanthislineofanalysisdidnotaccordwiththefactthatcorporateprofitsremainedstrong.Inanenvironmentofrisingwagesandstableprices,profitsusuallyweaken,sohecommissionedR&Sstafftoexplorethisobservation.Theobjectivewastotakea“deepdive”anddevelop,totheextentpossible,disaggregateestimatesoflaborproductivityandunitcostsformanufacturingandnon-manufacturingindustriesandforcorporationsversusotherlegalformsofbusiness.Tosortthroughthisrequest,itwasnecessarytoapproximateathree-way

(industry,legalformoforganization,time)decompositionoftheavailabledataonproductivity

andunitcostsintheeconomy—atoughchallengerequiringdetailnotthenfeaturedinthe

industrydatapublishedbytheBureauofEconomicAnalysis(BEA).

Theensuingwork,portionsofwhichwereissuedasastaffstudyandlaterpublished(CorradoandSlifman1999),generatedtwoprimaryconclusions.OneconclusionwasthatofficialGDPgrowthwaslikelyunderstated—thestatisticaldiscrepancyhadswungsharplystartingin1993—andproductivitycalculatedusingincome-sidedata(GDI)wasincreasingfasterthantheheadline

product-side(GDP)figuresuggested;the1997EconomicReportofthePresidentlaterestimatedthatthisgapimpliedthatlaborproductivityfrom1994to1996grew1.3percentagepointsperyearfasterthanofficiallyreported.Theotherconclusionwasthat,usingincome-sidedata,the

inferredprofitsofbusinessesinservicesindustries(corporateandnoncorporate)werewellmaintainedwhiletheirlaborproductivitywasestimatedtobefalling.Howcouldthese

businessesremainprofitablewhiletheirefficiencymeasuredasrealoutputperhourwas

outrightdropping?Oneanswertothisquestionwasthatthepricemeasuresusedtoobtainrealoutputforthesebusinesseswereseverelybiased.

Usingtheseobservations,ChairmanGreenspanconvincedthecommitteethatproductivitywasgrowingfasterthanreportedinofficialdataandthathikingthefederalfundstargetratewasnotwarranted.Thepossibilitythatproductivitymaybemis-measuredduetounderstated

servicespricesorstructuralchangenotcapturedinnationalaccountswasfamiliargroundforcommitteemembersinviewofthepublicdiscourseonmonetarypolicyandtechnologyatthetime.Thisdiscoursecoveredissuesrangingfromtechnicalissuesofpricemeasurementand

discrepanciesinnationalaccountstothe“irrationalexuberance”ofthestockmarketandthepossibilityemergingfromresearchusingcompany-leveldatathatan“ITrevolution”wasunderway.

MeasurementintheNews

ChairmanGreenspanfamouslyincludedacritiqueoftheCPIasabiasedmeasureofpricechangeincongressionaltestimonyinJanuary1995—acritiquedrawnfromaninfluentialstaffanalysis(Lebow,Roberts,andStockton1994).

6

OnDecember4,1996,agroupofeconomistsappointedbytheSenateFinanceCommitteereportedontheirstudyoftheCPI,which

estimatedthattheindexoverstatedinflation