2010The Effect of Private Information and Monitori

TheEffectofPrivateInformationandMonitoringontheRoleofAccountingQualityinInvestmentDecisions*ANNEBEATTY,TheOhioStateUniversityW.SCOTTLIAO,UniversityofTorontoJOSEPHWEBER,MassachusettsInstituteofTechnology1.IntroductionInformationasymmetrybetweenmanagersandoutsidecapitalsupplierscanaffecthowfirmsfinancecapitalinvestments.Agrowingbodyofevidenceindicatesthatbetteraccountingqualitycanreduceinformationasymmetrycostsandreducefinancingconstraints.Consistentwiththispossibility,Bid-dleandHilary(2006)documentthathigheraccountingqualityreducesthesensitivityoffirms’investmenttotheirinternallygeneratedcashflows.Verdi(2006)andBiddle,Hillary,andVerdi(2009)findthataccountingqualityispositivelycorrelatedwithinvestmentforfirmspronetounder-investandisnegativelycorrelatedwithinvestmentforfirmspronetooverinvest.Theimportanceofaccountingqualityoninvestmentinefficiencymaybemitigatedwhenoutsidecapitalsuppliershaveprivateinformationorcandirectlymonitormanagers.Byaccessingprivateinformationandcontrollingmanagerialactions,outsidecapitalsupplierscandirectlyaffectafirm’sinvestments,reducingtheimportanceofaccountingquality.Consistentwiththisidea,BiddleandHilary(2006)comparetheinfluenceofaccountingqualityoninvestmentefficiencyacrosscountries.TheyfindthataccountingqualityinfluencesinvestmentefficiencyintheUnitedStates,butnotinJapan.Theysuggestthatonepotentialexplanationforthiscross-countrydifferenceisthemixofdebtandequityinthecapitalstructuresofU.S.ver-susJapanesefirms.Weextendthisresearchbyexamininghowdifferentsourcesoffinancingaffecttheimportanceofaccountingqualityonfirms’investment–cashflowsensitivity.Directlytestinghowdifferentsourcesoffinancinginfluencethe*AcceptedbyShivaramRajgopal.BeattythanksDeloitte&Toucheforfinancialsupport.WethankJenniferAltamuro,RonDye,BenLansford,TomLys,WaleedMuhana,RickJohnston,ShailPandit,JoePiotroski,ShyamSunder,RickYoung,HelenZhang,andseminarparticipantsatStanfordUniversity,NorthwesternUniversity,andOhioStateUniversity.ContemporaryAccountingResearchVol.27No.1(Spring2010)pp.17–47CAAAdoi:10.1111/j.1911-3846.2010.01000.xeffectofaccountingqualityontheinvestment–cashflowsensitivityischal-lengingbecauseacomparisonoffirmsthatrecentlyobtaineddebtfinancingtothosethatdidnotcouldbeaffectedbytheirabilitytoobtainfinancing.Toalleviatethisproblem,werestrictoursampletofirmsthathaveallrecentlyobtaineddebtfinancingandexploitthedifferencesinaccesstopri-vateinformationandmonitoringthatexistacrossthepublicdebttoprivatelendingcontinuumassuggestedbyDiamond1991.1Weacknowledgethatthesensitivityofinvestmenttointernalcashflowsmaybelowerimmedi-atelyafterobtainingdebtfinancing.However,thispossibilitywouldbiasagainstrejectingourhypotheses.Specifically,weidentifyasampleof1,163firmsontheSecuritiesData-corp(SDC)databasethathaverecentlyraisedcapitalthroughtheissuanceofeitherpublicdebtorsyndicatedbankdebt.Werestrictoursampletofirmsthathaverecentlyobtaineddebtfinancingtoholdconstantthefirmcharacteristicsassociatedwithborrowing.However,withinthesampleoffirmsthathaverecentlyobtaineddebtfinancing,therearelikelytobesig-nificantdifferencesinthecapitalprovider’saccesstoprivateinformationandtheconstraintstheyplaceonmanagerialactions.Diamond’s(1991)theoryimpliesthatpublicdebtholdershavelessaccesstoprivateinformationandarethuslesseffectiveinmonitoringbor-rowersthanbanks.Basedonthesearguments,wepredictthataccountingqualityshouldhavealargerinfluenceonfirms’investment–cashflowsensi-tivityforfirmswithpublicdebtthanforthosewithbankdebt.Second,weexpectthattheeffectofaccountingqualityonfirms’investmentpoli-cieswilldependonwhetherlenderscontractuallyrestrictinvestment.Wepredictthat,whenfirmsfacecontractualrestrictionsoninvestments,over-investmentproblemsarepartiallyaddressed,andaccountingqualitybecomeslessvaluableandlesslikelytoaffecttheinvestment–cashflowsensitivity.Toprovideevidenceonthesehypotheses,weestimateamodelofinvest-mentsoninternalcashflowsseparatelyforthesampleoffirmsthatonlyissuepublicdebtandforthesampleoffirmsthatissuesyndicatedloans.Intheseregressions,weinteractthecashflowvariablewithourmeasureofaccountingqualitytoinvestigatewhetheraccountingqualityinfluencesinvestment–cashflowsensitivities.Forthebanksample,wefurtherinteracttheproductofthecashflowandaccountingqualityvariableswiththeinvestmentrestrictionsvariablestoexaminewhethercapitalexpenditurecovenantsaffecttheinvestment–cashflowsensitivity.Weusebothordinaryleastsquares(OLS)andendogenousswitchingmodelestimationtoaddresspotentialsampleselectionbiasarisingfromthepublicversusbankdebt1.Analternativewouldbetoinvestigatetheroleofrelationshipbankingonthesensitivityofinvestmenttocashflows.TheSDCloandatabasehasmostlysyndicatedloans.Thelackofasignificantnumberof‘‘relationship’’loanslimitsourabilitytoexplorethispossibility.18ContemporaryAccountingResearchCARVol.27No.1(Spring2010)financingchoice.Ourfirst-stagemodelofthedeterminantstoenterthepub-licversusprivatedebtmarketfollowsBharath,Sunder,andSunder2008.Weestimatesecond-stage‘‘regime’’investmentmodelregressionsseparatelyforfirmsthathavepublicversusprivatedebt.Theseregimeregressionsalsocontrolforendogeneityin