01.The-Cash-Flow-Sensitivity-of-Cash

THEJOURNALOFFINANCE•VOL.LIX,NO.4•AUGUST2004TheCashFlowSensitivityofCashHEITORALMEIDA,MURILLOCAMPELLO,andMICHAELS.WEISBACH∗ABSTRACTWemodelafirm’sdemandforliquiditytodevelopanewtestoftheeffectoffinancialconstraintsoncorporatepolicies.Theeffectoffinancialconstraintsiscapturedbythefirm’spropensitytosavecashoutofcashflows(thecashflowsensitivityofcash).Wehypothesizethatconstrainedfirmsshouldhaveapositivecashflowsensitivityofcash,whileunconstrainedfirms’cashsavingsshouldnotbesystematicallyrelatedtocashflows.Weempiricallyestimatethecashflowsensitivityofcashusingalargesampleofmanufacturingfirmsoverthe1971to2000periodandfindrobustsupportforourtheory.TWOIMPORTANTAREASOFRESEARCHincorporatefinancearetheeffectsoffinancialconstraintsonfirmbehaviorandthemannerinwhichfirmsperformfinancialmanagement.Thesetwoissues,althoughoftenstudiedseparately,arefunda-mentallylinked.AsoriginallyproposedbyKeynes(1936),amajoradvantageofaliquidbalancesheetisthatitallowsfirmstoundertakevaluableprojectswhentheyarise.However,Keynesalsoarguedthattheimportanceofbalancesheetliquidityisinfluencedbytheextenttowhichfirmshaveaccesstoexternalcapitalmarkets(p.196).Ifafirmhasunrestrictedaccesstoexternalcapital—thatis,ifafirmisfinanciallyunconstrained—thereisnoneedtosafeguardagainstfutureinvestmentneedsandcorporateliquiditybecomesirrelevant.Incontrast,whenthefirmfacesfinancingfrictions,liquiditymanagementmaybecomeakeyissueforcorporatepolicy.Despitethelinkbetweenfinancialconstraintsandcorporateliquidityde-mand,theliteraturethatexaminestheeffectsoffinancialconstraintsonfirmbehaviorhastraditionallyfocusedoncorporateinvestmentdemand.1Inanin-fluentialpaper,Fazzari,Hubbard,andPetersen(1988)proposethatwhenfirmsfacefinancingconstraints,investmentspendingwillvarywiththeavailabilityofinternalfunds,ratherthanjustwiththeavailabilityofpositivenetpresent∗AlmeidaisatNewYorkUniversity.CampelloandWeisbachareattheUniversityofIllinois.Wethankananonymousreferee,RickGreen(theeditor),ViralAcharya,DanBernhardt,MattBillett,LongChen,TedFee,JonGarfinkel,JohnGraham,CharlieHadlock,JarradHarford,HarrisonHong,GeorgePennacchi,EricRasmussen,Ren´eStulz,ToniWhited,andJeffreyWurglerfortheirveryhelpfulsuggestions.CommentsfromseminarparticipantsattheNovember2002NBERCor-porateFinanceMeeting,DukeUniversity,GeorgiaStateUniversity,IndianaUniversity,LouisianaStateUniversity,NewYorkUniversity,NewYorkFederalReserveBank,UniversityofIllinois,andUniversityofIowaarealsoappreciated.1SeeHubbard(1998)foracomprehensivesurvey.SomerepresentativereferencesareHoshi,Kashyap,andSchartstein(1991),Whited(1992),CalomirisandHubbard(1994),GilchristandHimmelberg(1995),andLamont(1997).17771778TheJournalofFinancevalue(NPV)projects.Accordingly,oneshouldbeabletoexaminetheinflu-enceoffinancingfrictionsoncorporateinvestmentbycomparingtheempiricalsensitivityofinvestmenttocashflowacrossgroupsoffirmssortedaccordingtoaproxyforfinancialconstraints.Recentresearch,however,hasidentifiedseveralproblemswiththatstrategy.Therobustnessoftheimplicationspro-posedbyFazzari,Hubbard,andPetersenhasbeenchallengedontheoreticalgroundsbyKaplanandZingales(1997),PovelandRaith(2001),andAlmeidaandCampello(2002),whiletherobustnessofcross-sectionalpatternspresentedintheirempiricalwork(andinthesubsequentliterature)hasbeenquestionedbyKaplanandZingales,Cleary(1999),andEricksonandWhited(2000).Alti(2003)furtherdemonstratesthatbecausecashflowscontainvaluableinfor-mationaboutafirm’sinvestmentopportunities,thecross-sectionalpatternsreportedbyFazzari,Hubbard,andPetersencanbeconsistentwithamodelwithnofinancingfrictions(seealsoGomes(2001)).Thisargumentcastsdoubtontheverymeaningoftheempiricalcashflowsensitivitiesofinvestmentre-portedintheliterature.Inthispaper,wearguethatthelinkbetweenfinancialconstraintsandafirm’sdemandforliquiditycanhelpusidentifywhetherfinancialconstraintsareanimportantdeterminantoffirmbehavior.Wefirstpresentamodelofafirm’sliquiditydemandthatformalizesKeynes’intuition.Init,firmsanticipat-ingfinancingconstraintsinthefuturerespondtothosepotentialconstraintsbyhoardingcashtoday.Holdingcashiscostly,nonetheless,sincehighercashsavingsrequirereductionsincurrent,valuableinvestments.Constrainedfirmsthuschoosetheiroptimalcashpolicytobalancetheprofitabilityofcurrentandfutureinvestments.ThispolicyisincontrasttothatoffirmsthatareabletofundalloftheirpositiveNPVinvestments:Financiallyunconstrainedfirmshavenouseforcash,butalsofacenocostofholdingcash(i.e.,theircashpoliciesareindeterminate).Thestarkdifferenceintheimpliedcashpoliciesofconstrainedanduncon-strainedfirmsallowsustoformulateanempiricalpredictionabouttheeffectoffinancialconstraintsonfirms’financialpolicies.Ourmodelsuggeststhatfinancialconstraintsshouldberelatedtoafirm’spropensitytosavecashoutofcashinflows,whichwerefertoasthecashflowsensitivityofcash.Inparticular,financiallyunconstrainedfirmsshouldnotdisplayasystematicpropensitytosavecash,whilefirmsthatareconstrainedshouldhaveapositivecashflowsensitivityofcash.Assuch,thecashflowsensitivityofcashprovidesathe-oreticallyjustified,empiricallyimplementablemeasur