riskless5555ss

DFYIF~MAY/JUNE1Q95StephenC.
CetchethisprofessorofeconomicsatOhioStateUniversity.
TheauthorthanksMargaretMoryMcConnellforableresearchassistance,ondAllenBerger,BenBernanke,AnilKoshyap,NelsonMark,AlanViardandtheparticipantsattheconferenceforcommentsandsuggestions.
TheauthoralsoexpressesgratitudetotheNationalStienceFoundationandtheFederalReserveBankofClevelandforfinancialandresearchsupport.
DistinguishingTheoriesoftheMonetaryTransmissionMechanismStephenG.
Cecchettiyairaditionalstudiesofmonetarypolicy'simpactontherealeconomyhaveIfocusedonitsaggregateeffects.
BeginningwithFriedmanandSchwartz(1963),modernempiricalresearchinmone-taryeconomicsemphasizestheabilityofpol-icytostabilizethemacroeconomy.
Butcasualobservationsuggeststhatbusinesscycleshavedistributionalimplicationsaswell,Onewayofcastingthedebateovertherelativeimportanceofdifferentchannelsofmonetarypolicytransmissionistoaskifthesedistrib-utionaleffectsaresufficientlyimportanttowarrantclosescrutinyThepointcanbeunderstoodclearlybyanalogywithbusinesscycleresearchmoregenerallyIfrecessionswerecharacterizedbyaproportionatereductionofincomeacrosstheentireemployedpopulation—forexample,everyoneworked39ratherthan40hoursperweekforafewquarters—theneconomistswouldpaysubstantiallylessattentiontocycles.
Itistheallocationoftheburdenorbenefitoffluctuations,withsomeindividualsfacingmuchlargercoststhanothers,thatisofconcern,Therearetwowaysforanecon-omisttoaddressthisproblem.
Thefirstistoattempttostabilizetheaggregateeconomy,thetraditionalfocusofpolicy-orientedmacroeconomics,Thesecondistoaskwhythemarketdoesnotprovidesomeformofinsurance.
Therecentdebateoverthenatureofthemonetarytransmissionmechanismcanbethoughtofinsimilarterms.
AccordingtotheoriginaltextbookIS-LMviewofmoneychangesinpolicyareimportantonlyinsofarastheyaffectaggregateoutcomes.
Onlythefluctuationintotalinvestmentisimportantsincepoliciesonlyaffecttherequiredrateofreturnonnewinvestmentprojects,andsoitisonlytheleastprofitableprojects(economy-wide)thatarenolongerfunded.
Butsincethemostprofitableprojectscontinuetobeundertaken,therearenodirectefficiencylossesassociatedwiththedistributionalaspectsofthepolicy-inducedinterestrateincrease,Incontrast,the"lending"viewfocusesonthedistributionalconsequencesofmone-tarypohcyactions.
Byemphasizingacombi-nationofcapitalmarketimperfectionsandportfoliobalanceeffectsbasedonimperfectassetsubstitutability,thisalternativetheorysuggeststhepossibilitythatthepolicy'sinci-dencemaydiffersubstantiallyacrossagentsintheeconomyFurthermore,thepolicy'simpacthastodowithcharacteristicsoftheindividualsthatareunrelatedtotheinherentcreditworthinessoftheinvestmentprojects.
Anentrepreneurmaybedeemedunworthyofcreditsimplybecauseofacurrentlylownetworth,regardlessofthesocialreturntotheprojectbeingproposed.
Itisimportanttounderstandwhethertheinvestmentdeclinescreatedbymonetarypolicyshiftshavetheserepercussions,nInthisessay,Iexaminehowonemightdeterminewhetherthecross-sectionaleffectsofmonetarypolicyarequantitativelyimpor-tant.
Mygoalistoprovideacriticalevaluationofthemajorcontributionstotheliteraturethusfar.
Thediscussionproceedsinthreesteps.
Istartinthefirstsectionwithadescriptionofageneralframeworkthatencompassesallviewsofthetransmissionmechanismasspecialcases,therebyhigh-hghtingthedistinctions.
Inthesecondsection,Ibeginareviewoftheempiricalevidencewithanassessmentofhowresearcherstypi-callymeasuremonetarypolicyshifts,ThefollowingtwosectionsexaminethemethodsThefrancialaccelerator,irwhichtheimpacteaianestmeetofsmallinterestchangesismognifiedbybalancesheeteffects,is(Isoanimportantpartofmanydiscussionsofthelendingniew.
FEDERALRESERVESANKOFSt.
LOUIS83llFYIE~MAY/JUNEI9952Seelngersall(1987)foreconpletedescfptianofthisproblem.
[allowingthetreditnnalfrancialeconomicsapproach,Ibaneavoirhddiscussingdemandandsupplyexplicitly.
Irstead,theonsetdemandsorederinedIromtheideaofarbitragerelaflnnshipsamongniloftIneassets.
Theseincludethelimitedparficipa-tanmodelsbasederLucas(1990).
SeethesurveybyFexrst(1993),aswellasthesummaryinChdstiannandlichenhoaml1992Lusedfordifferentiatingbetweenthetheories.
Studiesfallintotwobroadcategoriesdependingonwhethertheyuseaggregateordisaggregatedata.
Thethirdsectiondiscussestheaggregatedata,whilethefourthsectiondescribestheuseofdisaggregatedata,Aconclusionfollows.
MONfl4Ufl(POUcYT.
TWBIRY/T;tj~yflT;T;rcVt:rcfl51(4~(OnewayofposingthefundamentalquestionassociatedwithunderstandingthemonetarytransmissionmechanismistoaskhowseeminglytrivialchangesinthesupplyofanoutsideassetcancreatelargeshiftsinthegrossquantityofassetsthatareinzeronetsupplyI-lowisitthatsmallmovementsinthemonetarybase(ornonborrowedreserves)translateintolargechangesindemanddeposits,loans,bondsandothersecurities,therebyaffectingaggregateinvest-mentandoutputThevariousanswersnothispuzzlecanbeunderstoodwithintheframeworkorigi-nallyproposedbyBrainardandTobin(1963).
Theirparadigmemphasizestheeffectsofmonetarypolicyoninvestorportfolios,andiseasytopresentusingtheinsightsfromFama's(1980)seminalpaperontherelation-shipbetweenfinancialintermediationandcentralbanks.
Fama'sviewoffinancialintermediariesisthelimitofthecurrenttypeoffinancialinnovation,becauseitinvolvesthevirtualeliminationofbanksasdepositoryinstitu-tions.
Thesetupfocusesonaninvestor'sportfolioprobleminwhichanindividualmustchoosewhichassetstoholdgiventhelevelofrealwealth.
Labelingtheportfolioweightonassetiasw,,andtotalwealthasWthentheholdingofasset,i—theassetdemand—isjustX,=w1WIngeneral,theinvestorisdividingwealthamongrealassets—realestate,equityandbonds—andoutsidemoneyEachassethasstochasticreturn,~,,withexpectation~andthevectorofassetreturns,~,hasacovariancestructureF.
Givenautilityfunc-tion,aswellasaprocessforconsumption,itispossibletocomputetheutilitymaximizingportfolioweights.
Thesewilldependonthemeanandvarianceofthereturns.
~andF,themomentsoftheconsumptionprocess,callthesep.
r,andavectoroftasteparametersthatIwilllabel4,andassumetobeconstants.
Theutilitymaximizingassetdemandscanbeexpressedas=w*,('~,fl~Thisrepresentationmakesclearthanassetdemandscanchangefortworeasons.
Changesineitherthereturnsprocess(~,Formacroeconomicquantities(~,W)willaffecttheXIs,3Atthemostabstractlevel,financialintennediariesexisttocarryouttwofunctions.
First,theyexecuteinstructionstochangeportfolioweights.
Thatis,followingachangeinoneorallofthestochasticprocessesdrivingconsumption,wealthorreturns,theinterme-diarywilladjustinvestors'portfoliossothattheycontinuetomaximizeutilityInaddition,ifoneinvestorwishestotransfersomewealthtoanotherforsomereason,theintermediarywilleffectthetransaction.
WhatismonetarypolicyinthisstylizedsetupForpolicytoevenexist,somegov-ernmentauthority,suchasacentralbank,mustbethemonopolysupplierofanominallydenominatedassetthatisimperfectlysubsti-tutablewithallotherassets.
Iwillcallthisasset"outsidemoney"Inthecurrentenvi-ronment,itisthemonetarybase.
ThereisasubstantialliteratureonhowthedemandforoutsidemoneyarisesendogenouslyinthecontextofthetypeofenvironmentIhavejustdescribed.
4Butinaddition,asFamaemphasizes,theremaybelegalrequirementsthatforceagentstousethisparticularassetforcertaintransactions.
Reserverequirementsandtheuseofreservesforcertaintypesofhankclearingsareexamples.
Withinthisstylizedsetup,apolicyactionisachangeinthenominalsupplyofoutsidemoneyForsuchachangetohaveanyeffectsatall,(1)thecentralhankcontrolsthesupplyofanassetthatisbothindemandandforwhichthereisnoperfectsubstitute,and(2)pricesmustfailtoadjustfullyandinstantaneouslyOtherwise,achangeinthenominalquantityofoutsidemoneycannothaveanyimpactontherealinterestrate,andwillhavenorealeffects.
But,assumingthatFEDERALREEERVIRANKOFST.
LOUIS84DF~IF~MAY/JUNE'995thepolicymakercanchangetherealreturnontheassetthatismonopolisticallysupplied,investors'portfolioweightsmustadjustinresponsetoapolicychange.
Theviewoffinancialintermediariesthatisimplicitinthisdescriptionservestohigh-lighttheBrainardandTobin(1963)insightthatmonetarypohcycanbeunderstoodbyfocusingsolelyontheendogenousresponseofinvestorportfolios.
Understandingthetransmissionmechanismrequiresacharac-terizationofhowassetholdingschangeinresponsetopolicyactions.
Second,eventhoughthereneedbenobanksasweknowthem,therewillsurelybeintermediariesthatperformtheserviceofmakingsmallbusinessloans.
Theagencycostsandmonitoringproblemsassociatedwiththistypeofdebtwillstillexist,andspe-cialistsinevaluationwillemerge.
Whiletheywillhavesuchloansasassets,theymostlikelywillnothavebankdepositsasliabihties.
Suchentitieswillbebrokers,andtheloanswillhebundledandsecuritized,Withthisasbackground,itisnowpossibletosketchthetwomajorviewsofthemonetarytransmissionmechanism.
Thereareanumberofexcellentsurveysofthesetheories,includingBernanke(1993a),GertlerandGilchrist(1993),KashyapandStein(l994a)andHubbard(1995).
Asaresult,Iwillberelativelybriefinmydescriptions.
flflEMON.
EYVIEVIThefirsttheorycommonlylabeledthemoneyview,isbasedonthenotionthatreductionsinthequantityofoutsidemoneyraiserealratesofreturn,5This,inturn,reducesinvestmentbecausefewerprofitableprojectsareavailableathigherrequiredratesofreturn—thisisamovenuentalongafixedmarginalefficiencyofinvestmentschedule.
Thelesssubstitutableoutsidemoneyisforotherassets,thelargertheinterestratechanges.
Thereisnorealneedtodiscussbanksinthiscontext,Infact,thereisnoreasontodistinguishanyofthe"other"assetsininvestors'portfolios.
Intermsofthesimpleportfoliomodel,themoneyviewimpliesthattheshiftinthew~sforalloftheassetsexcludingoutsidemoneyareequal.
Animportantimphcationofthistradi-tionalmodelofthetransmissionmechanisminvolvestheincidenceoftheinvestmentdecline.
Sincetherearenoexternalitiesormarketimperfections,itisonlytheleastsociallyproductiveprojectsthatgounfunded.
Thecapitalstockismarginallylower.
But,giventhatadeclineisgoingtooccur,theallocationofthedeclineacrosssectorsissociallyefficient.
Thistheoryactuallypointstoameasureofmoneythatisrarelystudied,Mostempiri-calinvestigationsofmonetarypolicytrans-missionfocusonM2,butthelogicoftheportfolioviewsuggeststhatthemonetarybaseismoreappropriate.
Itisalsoworthpointingoutthatinvestigatorshavefounditextremelydifficulttomeasureeconomicallysignificantresponsesofeitherfixedorinven-toryinvestmenttochangesininterestratesthatareplausiblytheresultofpolicyshifts,Infact,mostoftheevidencethatisinterpretedassupportingthemoneyviewisactuallyevi-dencethatfailstosupportthelendingview.
THELIEHDffNOViEW:BALANCE~.
w~rr:.
_;EPECTSThesecondtheoryofmonetarytrans-missionisthelendingview,6Ithastwoparts,onethatdoesnotrequireintroductionofassetssuchashankloans,andonethatdoes.
Thefirstissometimesreferredtoasthebroadlendingchannel,orfinancialaccelerator,andemphasizestheinnpactofpolicychangesonthebalancesheetsofborrowers.
Ithearssubstantialsimilaritytothemechanismoper-atinginthemoneyview,becauseitinvolvestheimpactofchangesinth~realinterestrateoninvestment.
Accordingtothisview,therearecreditmarketimperfectionsthatmakethecalcula-tionofthemarginalefficiencyofinvestmentschedulemorecomplex.
Duetoinformationasymmetriesandmoralhazardproblems,aswellasbankruptcylaws,thestateofafirm'sbalancesheethasimplicationsforitsabilitytoobtainexternalfinance.
Policy-inducedincreasesininterestrates(whicharebothrealandnominal)cancauseadeteriorationinlerminalogyhasthepotenflnltocreateconfusionhere.
Ihavecho-senthetroditianaltermforthistextbookIS-tMar'narrow'moneynew.
Idanotmeantoimplythatthisisthe'monetaist'viewafthetransmissionmechanism.
follawKashyapandStein's11994alterminologyratherthanthemorecommoncreditxiewtoemphasizetheimportanceofloansintheerarsmissianmechanism.
lerrankeandGerfier(1989,19901pravidetheoriginalthearet'calunderpinningsforthisview.
FEDERALRESERVESANKOFST.
LOUIS85DFYIIE~MAY/JUNE1905lernunke,GertlerandGilchriso(1994)refertothisnsafinancioloccebruro-sinceitcoasessmallchangesininterestratesnohavepotentiallylargeeffectsoninvest-mentandoutput.
°ltmaybeparticularlydifficultoxdis-tagaishtheseeffectsfromthosethatarisefromoaryiagcyclicnlityofdifferentfirms'solesandprofitability.
oSeeJames(1987)foradiscussionoftheoniquenessofbarkloans.
mxWithnominalrigidity,adecreaseinoutsidemoneyreducesthepricelevelslowly,andsotherealreturntoholdingmoneyincreases.
Thischanneloftransmissionrequiresthatinoestorsshiftawayfromloansinresponse.
KoshyapandStein(1994b)pointoutthatlargebankscarissue(Osinawaythatinsulatestheirhalencesheetsfromcontractorindeposits,hatsmallbarkscannot.
Solongassmallbanksareanimportantsourcexlfundsfarsamehnnk'dependertfirms,therewillsoIlbeahonklendingchannel.
Inotherwards,forhonklendingtobeanimportantpartofthetraesmis-sianmechanism,creditmarketimpemfectonsmestbeimpartaetfarhonks.
thefirm'snetworth,bybothreducingexpectedfuturesalesandincreasingtherealvalueofnominallydenominateddebt,Withlowernetworth,thefirmislesscreditworthybecauseithasanincreasedincentivetomisrepresenttheriskinessofpotentialprojects.
Asaresult,potentiallenderswillincreasetheriskpre-miumtheyrequirewhenmakingaloan.
Theasymmetryofinformationmakesinternalfinanceofnewinvestmentprojectscheaperthanexternalfinance.
Thebalancesheeteffectsimplythattheshapeofthemarginalefficiencyofinvestmentcurveisitselfafunctionofthedebt-equityratiointheeconomyandcanbeaffectedbymonetarypolicy7Intermsofasimpletext-bookanalysis,policymovesboththeISandtheLMcurves.
Foragivenchangeintherateofreturnonoutsidemoney(whichmaybetherisklessrate),alenderislesswillingtofinanceagiveninvestmentthemoredebtapotentialborrowerhas.
Thispointstotwocleardistinctionsbetweenthemoneyandthelendingviews—thelatterstressesboththedistributionalimpactofmonetarypolicyandexplainshowseeminglysmallchangesininterestratescanhavealargeimpactoninvestment(thefinancialaccelerator).
Returningtotheportfoliochoicemodel,thepresenceofcreditmarketimperfectionsmeansthatpolicyaffectsthecovariancestructureofassetreturns,Asaresult,thew7swillshiftdifferentiallyinresponsetomonetarytighteningastheperceivedriski-nessofdebtissuedbyfirmswithcurrentlyhighdebt-equityratioswillincreaserelativetothatofothers.
8Thesecondmechanismarticulatedbyproponentsofthelendingchannelcanbedescribedbydividingthe"other"assetsininvestors'portfoliosintoatleastthreecate-gories:outsidemoney,"loans"andalltheothers.
Next,assumethattherearefirmsforwhichloansaretheonlysourceofexternalfunds—somefirmscannotissuesecurities.
0Dependingonthesolutiontotheportfolioallocationproblem,apolicyactionmaydirectlychangeboththeinterestrateandthequantityofloans,Itisnotnecessarytohaveaspecificinstitutionalframeworkinmindtounderstandthis,Instead,itoccurswheneverloansandoutsidemoneyarecomplementsininvestorportfolios;thatis,whenevertheportfolioweightonloansisanegativefunctionofthereturnonoutsidemoneyforgivenmeansandcovariancesofotherassetreturns.
noTheargumenthastwoclearparts.
First,thereareborrowerswhocannotfinancenewprojectsexceptthroughloans,andsecond,policychangeshaveadirecteffectonloansupplyConsequentlythemostimportantimpactofapolicyinnovationiscross-sectional,asitaffectsthequantityofloanstoloan-dependentborrowers.
Mostoftheliteratureonthelendingviewfocusesontheimplicationsofthismechanisminaworldinwhichbanksaretheonlysourceofloansandwhosehabilitiesarelargelyreservabledeposits.
Inthiscase,areductioninthequantityofreservesforcesareductioninthelevelofdeposits,whichmustbematchedbyafallinloans.
Theresultingchangeintheinterestrateonoutsidemoneywilldependonaccesstoclosebankdepositsubstitutes.
Butthecontractioninbankbal-ancesheetsreducesthelevelofloans.
Lowerlevelsofbankloanswillonlyhaveanimpactontherealeconomyinsofarastherearefirmswithoutanalternativesourceofinvestmentfunds.
Asatheoreticalmatter,itisnotnecessarytofocusnarrowlyoncontemporarybanksintryingtounderstandthedifferentpossiblewaysinwhichpolicyactionshaverealeffects.
AsIhaveemphasized,bankresponsestochangesinthequantityofreservesarejustonemechanismthatcanleadtoacomple-mentaritybetweenoutsidemoneyandloans.
AspointedoutbyRomerandRomer(1990),totheextentthatthereexistreadysubstitutesinbankportfohosforreservabledepositssuchasCDs,thisspecificchannelcouldheweaktononexistent,mnButitremainsarealpossi-bilitythattheoptimalresponseofinvestorstoapolicycontractionwouldbetoreducethequantityofloansintheirportfolios.
TheportfoliochoicemodelalsohelpstomakeclearthatthemannerinwhichpolicyactionstranslateintoloanchangesneednotFEDERALRESERVERANKOFSY.
LOUIS86IIE~IF~MAY/JUNE1995bearesultofloanrationing,althoughitmaynzAsStiglitzandWeiss(1981)originallypointedout,aformofrationingmayariseinequilibriumasaconsequenceofadverseselection,Butthepresenceofalendingchanneldoesnotrequirethattherebeborrowerswillingtotakeondebtatthecur-rentpricewhoarenotgivenloans.
Itariseswhentherearefirmswhichdonothaveequivalentalternativesourcesofinvestmentfundsandloansareimperfectsubstitutesininvestors'portfolios.
Obviously,thecentralbankcantakeexplicitactionsdirectedatcontrollingthequantityofloans.
Again,loweringthelevelofloanswillhaveadifferentialimpactthatdependsonaccesstofinancingsubstitutes.
Butthemechanismbywhichexplicitcreditcontrolsinfluencetherealeconomyisadifferentquestion.
tm°"54cr645SSxc~xx55455456.
;woVV~TsOTVET~.
CONBiiXlOTIONSDistinguishingbetweenthesetwoviewsisdifficultbecausecontractionarymonetarypolicyactionshavetwoconsequences,regardlessoftherelativeimportanceofthemoneyandlendingmechanisms.
Itbothlowerscurrentrealwealthandchangestheportfolioweights.
nxAssumingthattherearerealeffects,contractionaryactionswillreducefutureoutputandlowercurrentrealwealth,reducingthedemandforallassets.
Inthecontextofstandarddiscussionsofthetransmissionmechanism,thisisthereductionininvestmentdemandthatarisesfromacyclicaldownturn.
8Thesecondeffectofpolicyistochangethemeanandcovarianceofexpectedassetreturns,Thischangesthew55's.
Inthesimplestcaseinwhichtherearetwoassets,outsidemoneyandeverythingelse,theincreaseinthereturnonoutsidemoneywillreducethedemandforeverythingelse.
Thisisareduc-tioninrealinvestment.
ThelendingviewimpliesthatthechangeinportfolioweightsismorecomplexandinanimportantwayTheremaybesomecombinationofbalancesheetandloansupplyeffects.
Thisimmediatelysuggeststhatlookingataggregatesforevidenceoftherightdegreeofimperfectsubstitutabilityortimingofchangesmaybeverydifficult,Whatseemspromisingistofocusontheotherdistinctionbetweenthetwoviews—thelendingview'sassumptionthatsomefirmsaredependentonloansforfinancing.
Inadditiontodifferencesstemmingfromtherelativeimportanceofshiftsinloandemandandloansupplythelendingviewalsopredictscross-sectionaldifferencesarisingfrombalancesheetconsiderations,Thesearealsolikelytobetestable.
Inparticular,itmaybepossibletoobservewhether,giventhequalityofpotentialinvestmentprojects,firmswithhighernetwortharemorelikelytoobtainexternalfunding.
Again,themajorimplicationsarecross-sectional.
EMPIRICALtBeforediscussinganyempiricalexami-nationofthemonetarytransmissionmecha-nism,twoquestionsmustbeaddressed.
First,donominalshocksinfacthaverealeffectsUnlessmonetarypolicyinfluencestherealeconomyitseemspointlesstostudythewayinwhichpolicychangeswork.
Second,howcanwemeasuremonetarypolicyInordertocalculatetheimpactofmonetarypolicyweneedaquantitativemeasurethatcanreliablybeassociatedwithpolicychanges.
HereItakeupeachoftheseissues.
Inthefollowingsection,1willweightheevi-denceontherealeffectsofmoneyThisisfollowedbyadiscussionofwaysinwhichrecentstudieshaveattemptedtoidentifymonetaryshocks.
%COTtLc-ntcCLilc~54t45444554~5rcr54544944—55'5/~~<'<455~564—ModerninvestigationoftheimpactofmoneyonrealeconomicactivitybeganwithFriedmanandSchwartz(1963).
Inmanyways,thisisstillthemostpowerfulevidenceinsupportoftheclaimthatmonetarypolicyplaysanimportantroleinaggregatefluctua-tions.
Throughanexaminationthatspanned52Sincetheremustbefirmsthanoreloan-dependent,thereissf11someformofrationinginthesecuritymarket.
xxSeetamerandRamer(1993)foraconcisediscussionofrecentepisodesinwhichtheFederalReservehasattemptednachungethecompositonofhonkholoncesheetsthroughmeansotherthansnundardpolicyactions.
HThechangeinportfolioweightscanriseeitherfromanycombinationofachargeinthereturnontheoetsideasset,achangeinthecovariancestructrrexfreturns,orshiftintheconsumptionprocess.
~Ingeneralequilibrium,thereisanoffsettingeffectthatarisesfromtheincreaseintheieteresorate.
Allotherthingsequal,thiswouldincreasesavingandthereforeinvestment.
lxwecanbefaidyconfidentthatsolongasmaneoarypolicyighteniagcancaeseareces-sian,theimpactoftheincomeandwealthdeclineswillhelargeenoughthatinoestmentwillfoil.
FEDERALRESERVESANKOFSY.
LOUIS87HtYIF~MAY/JUNE199556Theequinalentapeneconomyobservationisthatinsmallopeneconomies,enchorgeratesmoveinresponsetochangesinpolicy.
xxBoschenandMills(19921describearelatedtechnique.
~°SeeHamilton11994)foracompletedescriptorafthemethodalogy.
~Theenactmeasrresandsamplefol-lowthoseofKrshyapandStein(1994rl,whokindlysuppliedthedata.
numerousmonetaryregimes,theyarguethatapparentlyexogenousmonetarypolicyactionsprecededoutputmovements.
Recentresearchersusemoresophisticatedstatisticaltoolstostudythecorrelationsbetweenmoneyandincome.
This"money-incomecausality"literatureislargelyinconclu-sive,becauseitfailstoestablishconvincinglyeitherthatmoney"caused"outputorthereverse.
Intheend,thetestssimplyestablishwhethermeasuresofmoneyforecastoutput,notwhetherthereiscausation.
Giventhatoutsidemoney—themonetarybase—islessthan10percdntofthesizeofM2,itisnotsurprisingthateconomistsfindthesimul-taneityproblemsinherentinthequestiontoodauntingandgiveup.
Twopiecesofevidenceseemreasonablypersuasiveinmakingthecasethatmoneymatters.
First,theFederalReserveseemstoheabletochangethefederalfundsratevir-tuallywithoutwarning.
(Iamnotarguingthatthisisnecessarilyagoodidea,justthatitispossible.
)Intheveryshortrun,thesenominalinterestratechangescannotbeassociatedwithchangesininflationaryexpectations,andsotheymustrepresentrealinterestratemovements,Suchrealinterestratechangesalmostsurelyhaveanimpactonrealresourceallocations,raThesecondpieceofevidencecomesfromtheexaminationoftheneutrahtyofmoneyinCecchetti(1986,1987).
Inthosepapers,Iestablishthatoutputgrowthissignificantlycorrelatedwithmoneygrowthatlagsofupto10years!
Thereareseveralpossibleinterpretationsofthesefindings,buttheystronglysuggestthatmonetaryshockshavesomethingtodowithaggregaterealfluctuations.
<515551515<5555~6<5554<~5<Anumberofauthorshavearguedconvincinglythatpolicydisturbancescannotbegaugedbyexaminingmovementsinthemonetaryaggregates.
Thereasonisthatthevarianceintheinnovationstobroadmeasuresofmoneyareacombinationofendogenousresponsestorealshocks(KingandPlosser,1984)andshiftsinmoneydemand(BernankeandBlinder,1992).
Therehavebeentworeactionstothefactthatmonetaryaggregatesprovidelittleinsightintopolicyactions.
BothbeginbylookingatthefunctioningoftheFederalReserveandexamininghowpolicyisactuallyformulated.
Thefirst,duetoBernankeandBlinder(1992),notethatthefederalfundsrateistheactualpolicyinstrumentthatisusedonaday-to-daybasis,Thissuggeststhatinnovationstothefederalfundsratearelikelytoreflect,atleastinpart,policydisturbances.
Themainjustificationfortheirconclusioncomesfromexaminingtheinstittitionsofhowmonetarypolicyiscarriedout.
RomerandRomer(1989)suggestasecondmethod.
ByreadingtheminutesoftheFederalOpenMarketCommittee(FOMC)meetings,theyhaveconstructedaseriesofdatesonwhichtheybelievepolicybecamecontractionary.
nT:55Tounderstandtheshortcomingsofthesetwoapproaches,Iwilldescribehoweachisused.
Inthefirst,researchersbeginbyspecify-ingavectorautoregression.
Forthepurposesoftheexample,IwillusetheformulationinBernankeandBlinder's(1992)SectionIVTheyemployasix-variablespecificationwiththetotalcivilianunemploymentrate,thelogoftheCPI,thefederalfundsrate,andthelogofthreebankbalancesheetmeasures,allinrealterms:deposits,securitiesandloans.
Theassumptionisthatthefederalfundsrateisa'policy"variable,andsoitisunaffectedbyallothercontemporaneousinnovations,mnFollowingBernankeandBlinder,Iesti-matetheVARwithsixlagsusingseasonallyadjustedmonthlydatumOFigures1and2plotsomeinterestingresultsfromthisVAR.
Thefirstfigureshowstheestimatedresidualsfromthefederalfundsrateequation.
ThesolidverticallinesareNationalBureauofEconomicResearch(NBER)referencecyclepeaksandtroughs,whilethedashedverticallinesaretheRomerandRomerdates,intend-edtoindicatetheonsetofcontractionaryFEDEEALREliEVEBANKOFST.
LOUiS88IIE~I[t~MAY/JUNE199$monetarypolicyepisodes.
Thisserieslooksextremelynoisyanditishardtoseehowitcouldrepresentpolicychanges.
The1979-82periodistheonlyonewithlargepositiveornegativevalues.
AlthoughitissurelythecasethatthereareunanticipatedpolicychangesbothwhentheFederalReserveactsandwhenitdoesnot,onewouldexpectsmallnormalshockswithoccasionalspikes.
Ifdecisionsarereallythisrandom,thereissomethingfundamentallywrongwiththepolicymakingapparatus.
Furthermore,sincethefederalfundsrateitselfistheequilibriumpriceinthereservesmarket,giventechnicalitiesofthewaythatmonetarypolicyisactuallycarriedout,themarket-determinedlevelofthefundsrateisnotapolicyinstrument.
20ThesecondfigureshowstheresponseofthelogoftheCPItoapositiveonepercentagepointinnovationinthefederalfundsrate.
Tounderstandhowthisiscomputed,beginbywritingthevectorautoregressionasAa)y,=~x'whereAU.
)isamatrixofpolynomialsinthelagoperatorL(L'Yx=yr,),y1isthevectorofvariablesusedintheestimation,andC15meanzeroindependent(butpotentiallyhet-eroskedastic)error.
Thefirststepistoesti-matethereducedformversionofequation1byassumingthatnocontemporaneousvari-ablesappearontheright-handsideofanyequations(A(0)=I),Thisresultsinanesti-mateA(L)alongwithanestimatedcovariancematrixforthecoefficientestimates—callthis(2.
TheimpulseresponsefunctionsareobtainedbyinvertingtheestimatedlagpolynomialB(L)=Butthepointestimateoftheimpulseresponsefunctionisnotreallyenoughtoallowustoreachsolidconclusions,Itisalsoimportanttoconstructconfidenceintervalsfortheestimates.
Therearetwowaystodothis.
ThefirstinvolvesthetechniquethathasbeencalledMonteCarloIntegration.
ThisisaBayesianprocedurethatinvolvespresumingthatthedistributionofthevectoroferrorsinequation1—thec's—isi.
i.
d.
normal.
22Toavoidmakingsuchstringentassumptions,IchoosetoestimateconfidencebandsusinganalternativetechniquegroundedEsfimatedInnovationstotheFederalFundsRatesPercentagepoints34<4'21hi-:44:0-4654320—l—2—3656880inclassicalstatistics.
Thedeltamethodisthesimpleproce-durethatcomesfromnotingthatiftheesti-matesofthecoefficientsinlagpolynomialsareasymptoticallynonnallydistributed,thenanywell-behavedfunctionoftheseparameterswillalsobeasymptoticallynormallydistrib-uted.
StackingalloftheparametersinA(L)andcallingtheresult0,thenIT(o—o)~N(0,O).
Itfollowsthatanyfunctionoftheseparameters[(0)—forexample,theimpulseresponsefunction—willbeasymptoticallynormallydistributed,83198920(hisentirediscnssiorignoresthepossibilitythatont)cipaSmona-terypolicymatters—somethingthanresearchersshouldconsiderbringingintothediscussion.
Thisisalsodis-cussedinHamilton(1994).
xxSeeloon(1990).
FEDEEAI.
RESEEVEBANKOFST.
LOUIS—S19596271747786ResponseoftheLogCPItoaChangeintheFederalFundsRate02468101214Horizoninmonths161820222489DF~IF~MAY/JUNE199513Atestforwhetheroilofthereaetonsorejointlyzerorejectsotthepercentlevelforthehrstsevenmonths,andthenfailstorejectThetestforwhetheralltheeffectsorezerosimnitrreouslyforoIl24monthsfoilsturejectwithop-valueof0.
/f.
onThereisofurtherreasontoniewthismeasureofpolicywithsameskepticism.
Becauseofthelorgenumberofparametersestimated,theregressionsareusuallyorerfit-ted.
Asaresnlt,theynormallyhoveverypooroat-of-samplefortcastiegproperties.
25SeeFriedman(1990)forodiscos'sionofthestrategyofusingpolicy-makers'statementstogangetheiractions.
o=~'S~'do'dOwhichcanbeestimatednumericallyTheresultplottedinFigure2wasfirstpointedoutbySimsandisknownasthe"pricepuzzle.
"ParadoxicallytheVAResti-matesimplythatmonetarypolicycontrac-tionsleadtopriceincreases!
Asisclearfromtheestimatedstandard-errorbands,thispriceriseissignificantlypositiveforapproximatelythefirstyearAftertwoyears,however',itisnotpossibletorejectthehypothesisthatafundsrateincreasehasnoeffectonthepricelevel.
2'ThestandardconclusionisthattheVARismisspecifiedinsomewayOnestrongpos-sibilityisthatthefundsrateisnotexogenousinthewaythatisrequiredforthisidentifica-tiontobevalid,andsotheseinnovationsdonotaccuratelyreflectpolicymovements.
24Myconclusionsmaybetooharshforthefollowingreason.
AsBenBernankepointedoutintheconference,theestimatedinnova-tionsarethesumoftruepolicyinnovation,policyresponsestoomittedvariables,andmoregeneralspecificationerrorsintheVAR.
Asaresult,onewouldexpectthemtobenoisyFurthermore,aspointedoutbyAdrianPagan,sinceoneisprimarilyinterestedintheimpulseresponsefunctions—theimpactofunanticipatedpolicyonoutput,pricesandthelike—thenitmaybeimmaterialthattheestimatedpolicyinnovationsarenoisyevenifthetrueinnovationsarenot.
fflaKim-eras~sd.
R-orrserDatesTheRomerandRomerdateshavebeenbothwidelyusedan4extensivelycriticized.
2'Theysufferfrombothtechnicalandsubstan-tiveproblems.
First,theyarediscrete.
Presumablypolicychangeshavebothanintensityandatiming.
Ignoringthesizeofpolicychangesmusthaveanimpactonresults.
Second,RomerandRomerchoosetofocustheirinquiryonlyonpolicycontrac-tions,becausetheyfeelthatexpansionsweremoreambiguous.
Sincemostmodelspredictsymmetricresponsestopositiveandnegativemonetaryinnovations,thisstrategythrowsoutinformation.
Butthemainissueistheexogeneityofthepolicyshifts.
ItisdifficulttobelievethattheactionsoftheFOMC,asreportedintheminutesofthemeetings,aretrulyexoge-nousevents,Therehavebeentworesponsestothis.
First,HooverandPerez(1991)pro-videalengthydiscussionofwhyRomerandRomer'smethodsarenotcompellinginiden-tifyingoutputfluctuationsinducedbyexoge-nousmonetaryshocks.
Takingaslightlydifferentapproach,Shapiro(1994)examineswhethertheFOMCisrespondingtochangesineconomiccondi-tions,andsothereissomereactionfunctionimplicitinpolicyHeestimatesaprobitmodelfortheRomerandRomerdatesusingmeasuresofinflationandunemployment,bothasdeviationsfromacarefullyconstructedtargetlevel,asdeterminants.
Figure3repro-duceshisestimatesoftheprobabilityofadate,withtheverticallinesrepresentingthedatesthemselves.
Theunanticipatedpolicyactionis1minustheestimatedprobabilityAsisclearfromthefigure,severalofthedateswerelargelyanticipated,andthereweresomeperiodswhenpolicyshiftswerethoughttobelikelyandthendidnotoccur.
Overall,Shapiro'sresultssuggestthatthestandardinterpretationofthedummyvari-ablesasexogenousisincorrecttovaryingdegreesovertime.
Thereseemstobenowaytomeasuremonetarypolicyactionsthatdoesnotraiseseriousobjections.
Giventhis,itmightseemdifficulttoseehowtoproceedwiththestudyofdifferenttheoriesofthetransmissionmechanismn.
Buttheliteratureproceedsintwodirections.
ThefirstusesthesemeasuresdirectlyinanattempttogaugetheinfluenceofpolicychangesdirectlyTheconclusionsofthesestudiesmustbeviewedwithsomedegreeofskepticism.
Thealternativeapproachistonotethatinvestmentdeclinesaccountforthemajorshareofoutputreduc-tionsduringrecessions.
IfoneisabletoshowthatthedistributionofthecontractionininvestmentiscorrelatedwithvariablesFEOEEALRESEEVEBANKOFST.
LOUIS'h(j{o]-i[o])~N(O,O~),where90IIi~IF~MAY/JUNE1995relatedtoafirm'sbalancesheetanditsaccesstobankloans,thenthisstronglysuggeststheexistenceofalendingchannel.
r"9~4<5'445144554.
44M~w4#i4455ttt*5E44.
10*Numerousstudieshaveusedaggregatedatainanattempttodistinguishthechannelsofmonetarytransmission.
Thisliteraturecanbedividedintothreecategories:Thefirstlooksattherelativeforecastingabilityofdifferentquantityaggregates;thesecondstudiesdifferencesinthetimingoftheresponseofaggregatequantitiestopresumedpolicyshocks;andthethirdexaminesthebehaviorofinterestrates.
Beforeexaminingtheworkonquantities,Iwilldiscusstheuseofinterestratedata.
2'Asisclearfromthediscussioninthefirstsection,thelendingviewdoesallowformovementsinmarketinterestrates.
Furthermore,thesemovementsareinthesamedirectionasthosepredictedbythemoneyview,andtheirmagnitudedependssolelyonthedegreeofsubstitutabilitybetweenoutsidemoneyandvariousotherassets.
Wherethetwoviewsdifferisintheirpredictionsformovementsintheinterestrateonloans.
Butsincethereiscurrentlynosecondarymarketforthesesecurities,itisimpossibletodeterminetheinterestrateontheseloans.
2'Thisimpliesthatmarketinterestratesareofvirtuallynouseinthisexercise.
Thereisnosenseinwhichthebehaviorofinterestratescouldservetodistinguishbetweenthemoneyorlendingviews.
Inowturntotheworkonquantities.
Inthefollowingsection,1examinetestsinvolvingtherelativeforecastingabilityoftneasuresofmoneyandcredit.
Thisisfollowedbyadis-cussionofpapersthatemphasizeaggregatetianingrelationships.
55—r—-so4',Ketaamerarecr.
sstma.
i~a.
mtyAnumberofpapershaveexaminedtheabilityofdifferentfinancialaggregatestoforecastoutput(orunemployment)fluctua-tions.
Ramey(1993)isarecentexample.
Themainmethodologyhereistoaskwhethermeasuresofcreditareinformativeaboutfutureoutputmovements,oncemoneyhas~wssrassassaEstimatedProbabilityofaRomerandRomerDate(ShapiroprahitmodelUSinghi!
lotionandvnemploymnt)0.
6i'm55'sr5''<'5H'5'5''5'S'5,5('55i'ii'iAI~rr'19535659626568717477808386891992beentakenintoaccount.
TheproblemwiththisisthatcreditisusuallyjustabroadermeasureofmoneyToputitslightlydifferentlythebalancesheetidentityofthebankingsystemimpliesthatbankassetsequalbankliabilities.
AsBernanke(1993b)pointsout,monetaryaggregatesareameasureofbankhabihties,whilecreditaggregatesaremeasuresofhankassets.
Sincethesearecalculatedslightlydifferentlythe)'willnotbeidentical.
Butitisthesetechnicalmeasurementdiffer-encesthatarelikelytoaccountforthediffer-encesinforecastingabilitynotanythingaboutthetransmissionmechanism.
Moregenerallythemainfindingisthatcreditlagsoutput.
Unfortunately,thistellsusnothingaboutthetransmissionmechanism.
Theaggregatedatadoshowthataggregatecreditiscotantercyciicco!
,hutitiseasytofindexplanationsforthisthatareconsistentwiththelendingview.
Forexample,KiyotakiandMoore(1993)presenaamodelinwhichindi-vidualsmustcontinuetoservicecreditevenafterincomefalls,andsocreditfallsafterincomeeventhoughitisthefundamentalsourceoffluctuations.
Intheend,itisdiffi-cultnoseehowaggregatetimingrelationshipscantellusanythingatallaboutthewayinwhichmonetarypolicyaffectsrealactivity2'Aggregate;Da.
siggReIatia~rsshipsTheseconduseofaggregatedatahasbeentoexaminetheresponseofvariousrEDERALRESERVEBANKOFST.
LOUIS0.
50.
40.
30.
20,10Mirxn,RowerandWeil11994)stxdypre-WoridWarIIinterestratesinonottempttoaddresstheseqnestmrs.
inthepresencemlrotioning,thereistieaddedcowplicatinrthatone'aoaldneedobservationsantheshodonvpriceforaloomtoabar'roworwhoisdeemednottohecreditworthygivertheoconomicennironwent.
Obviously,thereisnoeasywaytairfersuchaprice.
oxThispointisalsomodebyBemnrnke,GerfierandGilchnist(1994).
91H[Yl[~MAY/JUHE1995financialquantitiestopolicyinnovations.
ReturningtoBernankeandBlinder(1992),theystudywhetherbankloansandsecuritiesresponddifferentlytofederalfundsrateinnovations.
29Thestandardmethodologyistocalculatetheimpulseresponsesforthetwovariablesandnotethattheylookdifferent.
Figure4reportsthecommonfinding,calcu-latedusingthesix-variableBernankeandBlinderVARestimatedoverthe1959-90sample.
Inresponsetoapositive1percentagepointinnovationinthefederalfundsrate,theunemploymentraterisesbynearly0.
1percentagepointafterone-and-a-halfyears,whilebanksecuritiesfall0.
07percentandloansdecline0.
02percent.
Securitiesfallbothbyalargeramountandmorequicklythanloans.
ButpointestimatesoftheseimpulseresponsesdonottelltheentirestoryInFigure5,1plotthepointestimateandtwostandarderrorbandsforthedifferencebetweentheimpulseresponseforloansandsecurities.
Thisallowsanexplicittestofwhetherthesetwoassetsareimperfectlysub-stitutableinresponsetotheshock.
Thedif-ferencesareindividuallygreaterthanzeroinonlyafewmonths,andajointtestofthefirst24monthsoftheimpulseresponse,whichisasymptoticallydistributedasaChi-squared,hasap-valueof0.
70.
Myconclusionisthatreduced-formvec-torautoregressionsarenearlyincapableofprovidingconvincingevidenceofadifferentialimpactoffederalfundsrateinnovationsonvariouspartsofbankbalancesheets.
Theseresultsarebasedontheestimationofalargenumberofparameterswitharelativelysmallamountofdata—thisVARhas237parametersand354datapoints—andsotheestimatesarefairlyimprecise.
'0Butevenifoneweretofindthattheimpulseresponsesdifferedsignificantlythiswouldonlybearonthesubstitutabilityoftheassets,andnotdirectlyonthevalidityofthelendingview,Boththepricesandquantitiesofperfectsubstitutesmusthavethesamestochasticprocess,andsofindingthatthisparticularpartialcorrelationisdifferentwouldbeevidenceofimperfectsubstitutabil-ityAsBernankeandBlinder(1992)makeclearindiscussingtheirfindings,thisisanecessarybutnotasufficientconditionforthelendingviewtohold.
Itisnotpossible,usingreduced-formestimatesbasedonaggregatedataalone,toidentifywhetherbankbalancesheetcontractionsarecausedbyshiftsinloansupplyorloandemand.
Whatisneededisavariablethatisknowntoshiftonecurvebutnottheother.
Kashyap,SteinandWilcox(1993)alsoprovideevidencebasedonaggregatetiming.
Theycomparetheresponseofbankloanstothatofcommercialpaperissuancefollowingpolicyinnovations.
Theyfindthatmonetarypolicycontractionsseemtodecreasethemixofloansrelativetocommercialpaper.
Borrowersthatcanmoveawayfromdirectbankfinancefollowingatighteningappeartodoso.
BothFriedmanandKuttner(1993),andOlinerandRudebusch(1993)takeissuewiththesefindingsandshowthatchangesinthemixareduetoincreasesintheamountofcommercialpaperissuanceduringarecession,butthatthequantityofbankloansdoesnotchange.
Inaddition,OlinerandRudebuschshowthatoncefia'msizeistakenintoaccount,andtradecreditisincludedinthedebtofthesmallfirms,themixoffinancingisleftunaf-fectedbypolicychanges.
Itisworthmakinganadditionalpointaboutthecommercialpapermarket.
First,Post(1992)documentsthatallcommercialpaperratedbyaratingagencymusthaveabackupsourceofliquiditywhichisgenerallyabanklineofcreditorastandbyletterofFEDERALRESERVEBANKOFST.
LOUIS.
2~At4~~44<,<<~s'5D5Asvs~xx1havetriedtwovatantsoftheBemankeandBlinderVARmethodthatmightseempromisingwaysofaddressingtheproblemofneasor-ingmonetarypolicychanges.
Inthefirst,IsabsttemtedthefendsrateforgetusreportedinSellan11994)fartheactualfundsrate.
Thishasverylittleimpactontheresults,asthefedcomesextremelyclosetohittingthetargets.
Second,Imodethealternativeextremeassomptionthattheliedsrateitselfisexoge-eGos.
Thishasverydramaticeffectsantheresnlts,asBernarkemadBlinder'sconclusionsorecorny-pletelyonsvpported.
Ifallmove'mentsinthefundsroteareassnmedtorepresentevogennaspolicyactions,itwoeldbeeotreme-lydifficulttoclaimthatloansandsecuritiesrespondeddifferentlytopolicyshifts.
92HF~II[~MAY/JUNE'995credit.
Thismeansthatcommercialpaperisanindirectliabilityofbanks,albeitonethatisnotontheirbalancesheet.
Furthermore,Calomiris,HimmelbergandWachtel(1994)suggestthatincreasesincommercialpaperissuanceareaccompaniedbyanincreaseintradecredit.
Thismeansthatapohcycon-tractionmaysimplycauseare-shufflingofcreditbyforcingbankstomoveliabilitiesoffoftheirbalancesheetsuchthatlargefirmsissuecommercialpaperinordertoprovidetradecredittosmallfirmsthatwouldhaveotherwisecomefrombanks.
~4'5y45iS4.
555°0005s4°Os5°<'~4555555005505%000t%05v50554fiajir<Thefirstsetofpaperstriestogaugetheimportanceofcapitalmarketimperfectionsoninvestment,andsoisrelatedtothebalancesheeteffectsdescribedinthefirstsection.
Thesecondset,whichisfairlysmall,examinestime-seriesvariationincross-sectionaldatainanattempttocharacterizethedistributionaleffectsofmonetarypolicydirectlyIwillbrieflydescribeeachofthesestrategies.
5<5<55<—',o—50-n5ie-aeuritoqcftyThegeneralfindinginthisliteratureisthatinternalfinanceislesscostlythanexternalfinanceforfirmsthathavepooraccesstoprimarycapitalmarkets.
Theempiricalstudiesfallintotwocate-gories.
Thefirstexaminesreduced-formcorrelations,whilethesecondlooksdirectlyattherelationshipbetweenthecostandexpectedreturntoamarginalinvestmentproject—theyestimatestructuralFulerequations.
Redu~cesd-FtrrroCarreIatiónsFazzari,HubbardandPetersen(1988)—10DifferenceBetweenLoanandSecurityResponsetoFederalFundsRateShockfpersentagechangeatannualratewithtwostandard.
doviadenbends)2520IS10S0pioneeredthetechniqueofdividingfirm-leveldataintogroupsusingmeasuresthoughttocorrespondtotheprojectmonitoringcostscreatedbyinformationasymmetries,andthenseeingifthecorrelationbetweeninvest-mentandcash-flowmeasuresvariesacrossthegroups.
Thefindinginawiderangeofstudiesisthatinvestmentismoresensitivetocash-flowvariablesforfirmswhohavereadyaccesstooutsidesourcesoffunds.
3tThemainissueininterpretingtheseresultsiswhetherthecharacteristicsofthefirmusedtosplitthesampleareexogenoustofinancingdecisions.
Measuresoffirmsize,dividendpolicies,bondratingsandthehkemayberelatedtothequahtyofinvestmentprojectsafirmhasavailable,andsolenderdiscriminationmaynotbeaconsequenceofasymmetricinformation.
Thereareseveralexamplesinwhichresearchersidentifypotentiaflyconstrainedfinnsbasedoninstitutionalcharacteristics,andsotheendogeneityproblemsaremitigated.
Iwillmentiontwo,Hoshi,KashyapandScharfstein(1991)findthatinvestmentbyJapanesefirmsthatweremembersofakeiretsu,orindustrialgroup,wasnotinflu-encedbyliquidityeffects.
Usingdataonindividualhospitals,CalemandRizzo(1994)findthatinvestmentdependsmoreheavilyoncash-flowvariablesforsmall,single-unithos-pitalsthanforlarge,network-affiliatedones.
Inthemostconvincingstudyofthistwe,CalomirisandHubbard(1993)study25lernanke,GertlenandGilchrist(1994)surveythelargenumberofstudiesthatusethisappmach.
FEDERALRESERVEBARKOFST.
LOUIS024681012141618202224Horizoninmonths93HF~IF~MAY/JUNE1Q95'°Oliaerandtadebesch(1994)andBernanke,GertlerandGilchnist11994)obtainsimilarresultsesingtheOftdataaswell.
theundistributedcorporateprofitstaxin1936and1937toestimatethedifferencesinfinancingcostsdirectlyfromfirms'responsestotheinstitutionofagraduatedsurtaxintendedtoforceanincreaseinthedividendpayoutrate.
Theirresults,holdinginvestmentopportunitiesfixed,arethatinvestmentspendingisaffectedbythelevelofinternalfundsonlyforthosefirmswithlowlevelsofdividendpaymentsandhighmarginaltaxrates.
Furthermore,thesetendedtobesmallerandfastergrowingfirms.
brructuratnWiththeappro-priatedata,itisthenpossibletoseewhethertheseEulerequationshold.
HubbardandKashyap(1992)isaninterestinguseofthistechnique.
FollowingtheworkofZeldes(1989)onconsumption,theyexaminewhethertheabilityofagriculturalfirmstomeetthisfirst-orderconditiondependsontheextentoftheircollaterizablenetworth.
Theyfindthatduringperiodswhenfarmershavehighnetworth,andsohavebetteraccesstoexternalfinancing,theirinvestmentbehaviorismorelikelytolookasifitisunconstrained.
Theseinvestmentstudieshavebeenverysuccessfulinestablishingtheexistenceofcapitalmarketimperfectionsaswellastheirlikelysourceininformationasymmetriesarisingfrommonitoringproblems.
Whiletheworkhaslittletosayaboutmonetarypolicydirectlyitdoesprovideanexcellentcharacterizationofthedistributionaleffectsofchangesinthehealthoffirms'balancesheetsregardlessofthesource.
thne-ziseryesLCo.
taeroc.
teThestrategyinthesecondsetofstudiesistousethecross-sectionaldimensiontoidentifythetransmissionmechanism,Thegoalistodeterminewhetherthereductioninloansduringmonetarycontractionsisacon-sequenceofshiftsinloandemandorloansupplyMyconclusionisthatthesestudiesfailtoestablishthedesiredresultinacon-vincingwayInstead,theyprovidefurtherevidenceofcapitalmarketimperfections.
Threemajorstudiesusedataonmanu-facturingfirms.
Inthefirst,GertlerandHubbard(1988)findthattheimpactofcashflowoninvestmentishigherduringrecessionsforfirmsthatretainahighpercentageoftheirearnings.
Thesecond,byKashyap,LamontandStein(1992),showsthatduringthe1981-82recession,theinventoriesoffirmswithoutreadyaccesstoexternalfinancefellbymorewhentheirinitiallevelofinternalcashwaslower.
Ontheotherhand,theinventoryinvestmentbehavioroffirmswithreadyaccesstoprimarycapitalmarketsshowednoevidenceofliquidityconstraints.
Inthethird,GertlerandGilchrist(1994)usetheQuarterlyFinancialReportforMantefacturingCorporations(QFR)todividefirmsintoasset-sizecategoriesandfindthatsmallfirmsaccountforadispropor-tionateshareofthedeclineinmanufacturingfollowingamonetaryshock,'2BothKashyapandStein(1994b)andPeekandRosengren(forthcoming)focusonthebehavioroflendersratherthanborrow-ers.
Byexaminingthecyclicalbehaviorofbanks,KashyapandSteinhopetofindevi-dencefortheimportanceofloansupplyshifts,Thestrongestresultintheirpaperisthat,followingamonetarycontraction,thetotalquantityofloansheldbysmallbanksfallswhilethatoflargebanksdoesnot.
Bycontrast,PeekandRosengrenstudyNewEnglandbanksduringthe1990-91recessionandfindthatpoorlycapitalizedbanksshrinkbymorethanequivalentinstitutionswithhighernetworth.
Myinterpretationisthatbothoftheseshowthatthecapitalmarketimperfectionscommonlyfoundtoapplytomanufacturingfirmsapplytobanksaswell.
Therearetwodifficultiesinherentinanyattempttoestablishthattheimportanttrans-missionrhechanismformonetarypolicyshocksisthroughbankloansupplyshifts.
First,asdescribedatlengthinthesecondsection,thereistheproblemofempiricallyidentifyingmonetarypolicyBeyondthis,thereisthesubtletyofdistinguishingloanFEDERALRESERVEBANKOFST.
LOUIS94ll1~I[~MAY/JUNE1995supplyshiftsfromthebalancesheeteffects.
Istheobservedreductioninloansaconse-quenceoftheircomplementaritywithoutsidemoneycausedbythestructureofthebankingsystem,orisittheresultofchangesintheshapeofthemarginalefficiencyofinvestmentschedulebroughtonbythebalancesheeteffectsKashyap,LamontandStein(1992)suggestonepossiblewayofdistinguishingthesepossibilities.
Ifonecanfindareces-sionaryperiodthatwasnotprecededbyamonetarycontraction,andshowthatinterestratesrosebutthatbankdependencewasirrelevanttoindividualfirms'experiences,thiswouldmeanthatbanksareresponsibleforthedistributionaleffectsinducedbymonetaryshocks.
Unfortunatelysuchevidenceisnotreadilyavailable.
sSSs5555SS<55<5<'05<5<55555<45#Thismeansthatmonetarypolicyshiftshaveanimportantdistributionalaspectthatcannotheaddressedwithinthetraditionalmoneyview.
Itisthesmallerandfastergrowingfirmsthatbearadisproportionateshareoftheburdenimposedbyarecession.
Sincethesearelikelytohefirmswithhighlyprofitableinvestmentopportunities,thishasimportantimplicationsforsocialwelfare.
Notonlyarerecessionsassociatedwithaggregateoutputandinvestmentdeclines,butthedeclinesareinefficient.
Beyondthis,thereistheissueofdistin-guishingthetwopartsofthelendingview.
DowecareifwecandistinguishchangesininvestmentopportunitiesresultingfromfinancialacceleratoreffectsfrombankloansupplyshiftsperseDoestheconclusionhaveimplicationsfortheactualconductofmonetarypolicyIbelievetheanswerisyes.
Ifthecomplementarityofbankloansandoutsidemoneyariseslargelyasaresultofthefinancialregulatoryenvironment,then,withfinancialinnovationandliberalization,theseeffectsarelikelytobecomelessimportantovertime.
Withtheintroductionofinterstatebankingandthedevelopmentofmoresophisticatedinstrumentsaimedattradingpoolsofloans,itisonlythebalancesheeteffectsthatwillremain,Asaresult,itisimportanttoknowwhichisthemoreimpor-tantchannelofmonetarypolicytransenission.
Bernanke,BenS.
"HowImportantistheCreditChannelintheTransmissionofMonetaryPolicyAComment,"Carnegie-RochesterConferenceSeriesonPublicPolicy39(December1993)pp.
47-52.
CreditintheMacroeconomy,"FederolReserveBookofNewYorkQuarterlyReview(spring1993)pp.
50-70.
ondMarkGertler.
"FinancialFragilityandEconomicPerformance,"QuarterlyJournalofEconomics(February1990),pp.
87-114.
andMarkGerfier"AgencyCost,NetWorthondBusinessFluctuatians,"TheAmericanEconomicReview(March1989),pp.
14-31.
-andAlanS.
Blinder.
"TheFederalFundsRateandtheChannelsofMonetaryTransmission,"TheAmericanEconomicReview(September1992),pp.
901-21.
Markt3erflerandSimonGilchrist.
"TheFinancialAcceleratorandtheFlighttoQuality,"NationalBureauafEconomicResearchWorkingPaperNo.
4789(July1994).
Brainard,WilliamC.
,andJamesTobin.
"FinancialIntermediariesandtheEffectivenessofMonetaryCantrals,"TheAmericanEconomicReview(May1963),pp.
383-400.
Boschen,JohnF.
,andleonardO.
Mills.
"TheEffectsofCountercyclicalMonetaryPolicyanManeyandInterestRates:AnEvaluationofEvidencefromFOMCDocuments,"workingpaper(1992),CollegeofWilliamandMary.
Calem,Pool,andJohnA.
Rizza.
"financingConstraintsandInvestrenent:NewEvidencefromHospitalIndustryData,"workingpaper(1994),YaleUniversity,SchoolofMedicine,DepartmentofEpidemiologyandPublicHealth.
Calamiris,CharlesW.
,CharlesP.
HimmelbergandPaulWachtel.
"CommercialPaperandCorporateFinance:AMicroecanomicPerspective,"workingpaper(April1994),SternSchoolofBusiness,NewYorkUniversity.
andR.
GlennHublmrd.
"laxPolicy,InternalFinance,andInvestment,"InternalFinanceandInvestmentEvidencefromtheUndistributedCorporatePrafitsTaxof1936-37,NotionalBureauofEconomicResenrthWorkingPaperNo.
4288(March1993).
Cecchetti,StephenC.
"TestingShortRunNeutrality:InternationalEvidence,"ReviewofEconomicsandStotisllcs(february1987)pp.
135-40.
FEDERALRESERVEBANKOFST.
LOUIS95IIE~I[~MAY/Sum1995TestingShortRunNeutrality,"JournalofMonetarylames,Christopher'SomeEvidenceontheUniquenessofBunkEconomics(May1986)pp.
409-23.
loans,"JournalofFinancialEconomics(1987),pp.
217-36.
Christiano,Lawrence,andMartinS.
Eirhenbaum.
"LiquidityEffectsandKoshyop,krilK.
,andJeremyC.
Stein.
"MonetaryPolicyandBanktheMonetaryTransmissionMechanism,"TheAmericanEconomiclending,"inN.
GregoryMankiw,ad.
,MonetaryPolicyUniversityofReview(May1992)pp.
346-53.
ChicagoPressfortheNotionalBureauofEconomicResearch,1994,pp.
221-62.
Doon,ThomasA.
User'sMonual,RATSVersion3.
10.
VAREconometrics,andTheImpactofMonetoryPolicyonBank1990.
BalanceSheets,"workingpaper(March1994),UniversityofFama,EugeneF.
"BankingintheTheoryofFinnnce,'JournalofChicogo,GraduateSchoolofBusiness.
MonetaryEconomics(January1980)pp39~57JeremyC.
SteinondDavidW.
Wilcox.
"MonetaryPolicyondCreditConditions:EvidencefromtheCompositionofExternalFazzari,StevenM.
,R.
GlennHubhordandBruceC.
Petersen.
Finance,"TheAmericanEconomicReview(Morch1993),pp.
78-98.
"FinancingConstrointsandCorporateInvestment,"BrookingsPopersonEconomicActivity(1988:1)pp.
141-95,,OwenA.
LomontandJeremyC.
Stein.
"CreditConditionsandtheCyclicalBehaviorofInventories:ACaseStudyofthe1981-Feurst,Timothy.
"LiquidityEffectModelsandTheirImplicationsfor1982Recession,"NotionalBureauofEconomicResearchWorkingMonetaryPolicy,"workingpaper(April1993),BowlingGreenStatePaperNo.
42tI(November1992).
University,DepartmentofEconomics.
King,RobertG.
,andCharlesI.
Plosser.
"Money,CreditandPricesinAFriedman,BenlaminM.
"Discussionof'NewEvidenceontheMonetaryRealBusinessCycle,"TheAmericanEconomicReview(June1984),TransmissionMechanism,"BrookingsPapersonEconomicActivitypp.
363-80.
(1990:1),pp.
204-9.
King,StephenR.
"MonetaryTransmission:ThroughBankLoansorBankLiabilities"JournalofMoney,CreditandBonking(AugustandKennethN.
Kuttnec"EconomicActivityandtheShort-1986),29g,3O3.
TermCreditMarkets:AnAnalysisofPricesandQuantities,"BrookingsKiyotoki,Nobuhiro,ondJohnMoore.
"CreditCycles,"workingpaperPapersonEconomicActivity(1993:2),pp.
193-266.
(March1993),LondonSchoolofEconomics.
Friedmon,Milton,andAnnaSchwartz.
AMonetaryHistoryoftheUnitedLucas,RobertE.
,Jr.
"LiquidityandInterestRates,"JournalofEconomicStates,1867-1960.
PrincetonUniversityPress,1963.
Theory(1990),pp.
237-64.
Gertler,Mark,andSimonGilchrist.
"MonetaryPolicy,BusinessCycles,Miran,Jeffrey,ChristinaD.
RomerandDavidWeil.
"HistoricalandtheBehavioralSmallManufacturingFirms,"QuarterlyJaurnolofPerspectivesontheMonetaryTransmissionMechanism,"inN.
GregoryEconomics(May1994)pp.
309-40.
Monkiw,ed.
,MonetaryPolicy.
UniversityofChicagoPressfortheNotionalBureauofEconomicResearch,1994,pp.
263-306.
andTheRoleofCreditMarketImperfectionsMorgan,DonaldP.
"TheLendingViewofMonetaryPolicyandBankintheMonetarytransmissionMechanism:kgumentsandEvithnce,"LoanCammitanents,"workingpaper(December1992),FederalScandinavianJournalofEconomics(1993)pp.
34-64.
ResemeBankofKansasCity.
andR.
GlennHubbard.
"FinancialFactorsinBusinessOliner,Stephen,andGlenn0.
Rudebusch.
'IsthereaBankCreditFluctuations,"NationalBureauofEconomicResearchWorkingPaperChannelforMonetaryPolicy"BoardofGovernorsoftheFederalNo.
2758(November1988).
ReserveSystem,FinanceandEconamicsDivisionDiscussionPaperNo.
93-8(february1993).
Hamilton,James0.
TimeSeriesAnalysis.
PrincetonUniversityPress,1994.
Peek,Joe,andEricRosengren.
"TheCapitalCrunch:NeitheroBorrowerHoover,Kevin0.
,andStephenJ.
Perez.
"PostHacErgoPrupterHocNoraLenderBe,"JournalofMoney,CreditandBankingOnceMore:AnEvaluationof'DoesMonetaryPolicyMailer'inthe(forthcoming).
SpiritofJamesTobin,"JoumalofMonetaryEcanomics(februaryPost,MitchellA.
"TheEvolutionoftheU.
S.
CommercialPaperMarket1992),pp.
3-24.
Since1980,"FedemlReserveBulletin(December1992),Hoshi,Takeo,AnilK.
KashyopandDavidScharfttein.
"Corporatepp.
879-91.
Structure,LiquidityandInvestment:EvidencefromJapanesePanelRamey,Valerie.
"HowImportantistheCreditChannelintheData,"QuarterlyJournalofEconomics(February1991),pp.
33-60,TransmissionofMonetaryPolicy"Camegie'RochesterConferenceSeriesonPublicPolicy(December1993),pp.
1-45.
Hubbord,R.
Glenn.
"IsTherea'CreditChannel'forMonetaryPolicy"thisReview(May/June),pp.
63-82.
Romer,Christina0.
,andDavidH.
Romer,"CreditChannelsarCreditActionsAnInterpretationofthePostwarTransmissionMechanism,"andAnilK.
Kashyop.
"InternalNetWorthandtheinChongingCapitalMarkets:ImplicationsforMonetaryPolicy.
InvestmentProcess:AnApplicationtoU.
S.
Agriculture,"JaurnalofFederalReserveBankofKansasCity,1993.
PoliticotEconomy(June1992)pp.
506-34.
andNewEvidenceontheMonetaryIngersoll,JonathanF.
TheoryofFinancialDecisionMaking.
RowmanTransmissionMechanism,"BrookingsPapersonEconomicActivityandtittlefield,1987.
(1990:1),pp.
149-213,PEPE&ALRESflVISANKOFST.
LOUIS96t1F~IIj~MAY/JU*U'995andDaesMonetaryPolicyMailerANewSfiglitz,JasephE.
,andAndrewWeiss.
'CreditRationinginMarketswithTestintheSpititofFriedmanandSchwartz,"NationalBureauatImperfectlnfarmntian,"TheAmericanEconomicReview(June1981),EconomicResearchMocroecanomicsAnnual.
MITPress,1989,pp.
393-410.
121-10.
Strongin,Steven.
"TheIdentificationofMonetaryPalicyDisturbances:Sargent,Tharnns1.
MacroeconomicTheory,SecondEdition.
AcademicExplainingtheLiquidityPuzzle,"FederalReserveBankofChicagaPress,1987.
WorkingPaperNa.
92-27(November1992).
Sellan,GordonH.
,Jr.
"MeasuringManetoryPolicy,"workingpaperZeldes,StephenP'ConsumptionandliquidityConstraints:An(August1994),FederalReserveBankofKansasCity.
EmpiricalInvestigation,"JournalofPoliticalEconomy(Apdl1989),pp.
305-46.
Shapiro,Matthew0.
FederalReservePolicy:CausenndEffect,inN.
GregoryMankiw,ed.
,MonetaryPolicy.
UniversityofChicagoPressfartheNotionalBureauofEconomicResearch,1994,pp.
30T34.
Sims,Christopher"InterpretingtheMncraeconomicTimeSeriesfacts:TheEffectsofManetoryPolicy,"EuropeanEconomicReview(May1992),pp.
975-1000,FEDflALRESflVESANKOfST.
LOUIS97