搜索
35THEACCOUNTINGREVIEWVol.77Supplement2002pp.35–59TheQualityofAccrualsandEarnings:TheRoleofAccrualEstimationErrorsPatriciaM.DechowUniversityofMichiganIliaD.DichevUniversityofMichiganABSTRACT:Thispapersuggestsanewmeasureofoneaspectofthequalityofworkingcapitalaccrualsandearnings.Oneroleofaccrualsistoshiftoradjusttherecognitionofcashflowsovertimesothattheadjustednumbers(earnings)bettermeasurefirmperformance.However,accrualsrequireas-sumptionsandestimatesoffuturecashflows.Wearguethatthequalityofaccrualsandearningsisdecreasinginthemagnitudeofestimationerrorinaccruals.Wederiveanempiricalmeasureofaccrualqualityastheresidualsfromfirm-specificregressionsofchangesinworkingcapitalonpast,present,andfutureoperatingcashflows.Wedocumentthatobservablefirmcharac-teristicscanbeusedasinstrumentsforaccrualquality(e.g.,volatilityofac-crualsandvolatilityofearnings).Finally,weshowthatourmeasureofaccrualqualityispositivelyrelatedtoearningspersistence.Keywords:qualityofaccruals;earningsquality;estimationerrors;earningspersistence.DataAvailability:Dataareavailablefromsourcesidentifiedinthepaper.I.INTRODUCTIONThispapersuggestsanewmeasureofoneaspectofthequalityofworkingcapitalaccrualsandearnings.Themeasureisbasedontheobservationthataccrualsshiftoradjusttherecognitionofcashflowsovertime,sothattheadjustednumbers(earnings),bettermeasurefirmperformance(e.g.,seeStatementofAccountingConceptsWethankworkshopparticipantsatTheUniversityofArizona,UniversityofChicago,ColumbiaUniversity,IndianaUniversity,UniversityofMichigan,TheAccountingReviewConferenceonQualityofEarningsheldatEmoryUniversity,the2001BigTenResearchConference,andespeciallyDanCollins,BobLipe,MortPincus,KatherineSchipper,PeterWysocki,andtwoanonymousreviewers.ProfessorDechowacknowledgesfinancialsupportprovidedbytheMichaelA.SakkinenResearchScholarFundattheUniversityofMichiganBusinessSchool.ProfessorDichevacknowledgesthefinancialsupportoftheSanfordRobertsonAssistantProfessorshipinBusinessAdministrationandtheNormanAuerbachPricewaterhouseCoopersFacultyFellowship.36TheAccountingReview,2002SupplementNo.1,FASB1978,para.44).Forexample,recordingareceivableacceleratestherecog-nitionofafuturecashflowinearnings,andmatchesthetimingoftheaccountingrecognitionwiththetimingoftheeconomicbenefitsfromthesale.However,accrualsarefrequentlybasedonassumptionsandestimatesthat,ifwrong,mustbecorrectedinfutureaccrualsandearnings.Forexample,ifthenetproceedsfromareceivablearelessthantheoriginalestimate,thenthesubsequententryrecordsboththecashcollectedandthecorrec-tionoftheestimationerror.Wearguethatestimationerrorsandtheirsubsequentcorrectionsarenoisethatreducesthebeneficialroleofaccruals.Therefore,thequalityofaccrualsandearningsisdecreasinginthemagnitudeofaccrualestimationerrors.Ourempiricalmeasureofaccrualqualityistheextenttowhichworkingcapitalaccrualsmapintooperatingcashflowrealizations,whereapoormatchsignifieslowaccrualquality.Theintuitionbehindthismeasureisavailableinboththeoreticalandpracticaltextsthatobservethatthebeneficialroleofaccrualsisreducedbyvariouslimitations,includingestimationerrors.Forexample,Palepuetal.(2000)discussestimationerrorsasafactorthatreducesaccountingquality,andsuggestthatestimationaccuracydependsonfirmchar-acteristicslikecomplexityoftransactionsandpredictabilityofthefirm’senvironment.Wedevelopthisexistingintuitionintoapracticalmeasureofaccrualandearningsquality.Ourapproachtoestimatingthequalityofaccrualsisrelatedtotwostreamsofresearch.First,severalstudiesdocumentthebenefitsoftheaccrualprocess,findingthatearningsisabettermeasureofperformancethantheunderlyingcashflows(e.g.,Dechow1994;Dechowetal.1998;Liuetal.2002).Webuildonthisevidencebyexploringthetrade-offsinherentintheaccrualprocess.Second,anumberofstudiesusemodelsof‘‘discre-tionaryaccruals’’toinvestigatethemanipulationofaccrualstoachieveearningsmanage-mentgoals(seeHealyandWahlen[1999]forarecentreview).Thesestudiesfocusontheopportunisticuseofaccrualstowindow-dressandmisleadusersoffinancialstatements.Thisstreamofresearchsuggeststhatmanagerialintentaffectstheincidenceandmagnitudeofaccrualestimationerrors.Incontrast,wearguethatevenintheabsenceofintentionalearningsmanagement,accrualqualitywillbesystematicallyrelatedtofirmandindustrycharacteristics.Thisdis-tinctionisimportantbecausesuchcharacteristicsarelikelytobebothobservableandrecurring(e.g.,thevolatilityofoperationsissystematicallyrelatedtothepropensitytomakeestimationerrors)ascomparedtothedeterminantsofmanagerialopportunismthatareoftenunobservableand/orsporadic(e.g.,beforestockofferings).Forourpurposes,wedonotattempttodisentangle‘‘intentional’’estimationerrorsfromunintentionalerrorsbe-causebothimplylow-qualityaccrualsandearnings.Wedevelopamodelthatexaminestheoriginationandreversalofworkingcapitalaccrualsinastylizedfirm.Themodelembodiestheintuitionthatthetimingofthefirm’seconomicachievementsandsacrificesoftendiffersfromthetimingoftherelatedcashflows,andthatthebenefitofaccrualsistoadjustforthesecashflowtimingproblems.However,themodelalsorevealsthatthebenefitofusingaccrualscomesatthecostofincludingaccrualcomponentsthati