Estimation-and-empirical-properties-of-a-firm-year

Estimationandempiricalpropertiesofafirm-yearmeasureofaccountingconservatismMozaffarKhan,RossL.WattsSloanSchoolofManagement,MassachusettsInstituteofTechnology,50MemorialDr.,E52-325,Cambridge,MA02142,USAarticleinfoArticlehistory:Received17December2007Receivedinrevisedform19August2009Accepted25August2009Availableonline16September2009JELclassification:M41M44G34G38K22D82Keywords:ConservatismAsymmetrictimelinessFirm-yearmeasureInformationasymmetryLitigationabstractWeestimateafirm-yearmeasureofaccountingconservatism,examineitsempiricalpropertiesasametric,andillustrateapplicationsbytestingnewhypothesesthatshedfurtherlightonthenatureandeffectsofconservatism.Theresultsareconsistentwiththemeasure,C_Score,capturingvariationinconservatismandalsopredictingasymmetricearningstimelinessathorizonsofupto3yearsahead.Cross-sectionalhypothesistestssuggestfirmswithlongerinvestmentcycles,higheridiosyncraticuncertaintyandhigherinformationasymmetryhavehigheraccountingconservatism.Eventstudiessuggestincreasedconservatismisaresponsetoincreasesininformationasymmetryandidiosyncraticuncertainty.&2009ElsevierB.V.Allrightsreserved.1.IntroductionAccountingconservatismistheasymmetricverificationthresholdforgainsversuslosses:theverificationthresholdforgainsishigher(Basu,1997).Thisdifferentialverificationrequirementresultsinasymmetricearningstimelinesswithrespecttogainsversuslosses,conservatismflow,thatbuildsupacumulativeunderstatementofnetassets,conservatismstock(Watts,2003a).Conservatismhasbeenanenduringqualitativecharacteristicoffinancialreportingforatleastfivecenturies(Basu,1997),andhasbeenthesubjectofmuchacademicresearch.1Empiricalresearchonconservatismrequiresametricorscalethatcancharacterizebothcross-sectionalandtime-seriesvariationinconservatism.Anumberofmeasuresofconservatismhavebeenproposed(seeWatts,2003b,forasummary),withtheBasu(1997)flowmeasurebeingthemostwidelyused(Ryan,2006).TheBasumeasureisestimatedeitherforanindustry-yearusingacross-sectionoffirmsintheindustryorforafirmusingatime-seriesoffirm-years.Bothestimationmethodshavelimitations.Theindustry-yearmeasureobscurescross-sectionalvariationintheconservatismofindividualfirmfinancialreportsbyassumingallfirmsintheindustryarehomogeneous.TheindividualfirmmeasureobscurestheContentslistsavailableatScienceDirectjournalhomepage::10.1016/j.jacceco.2009.08.002Correspondingauthor.E-mailaddresses:[e-mail](M.Khan),[e-mail](R.L.Watts).1Forexample,Balletal.(2000),GivolyandHayn(2000),andHolthausenandWatts(2001),amongmanyothers.JournalofAccountingandEconomics48(2009)132–150timingofchangesintheconservatismofindividualfirmfinancialreportsbyassumingthefirm’soperatingcharacteristicsarestationary.Manychangesaffectingafirm’sfinancialreportingconservatismarelikelybothtime-andfirm-specific.Anexampleisachangeintheinformationasymmetrybetweeninvestorsandthefirm’smanagerscausedbyafirm-specificreductioningrowthopportunities(LaFondandWatts,2008).Notsurprisingly,researchershaveexpressedademandforafirm-levelmeasureofconservatismthatcanreflectthetimingofconservatismchangesandthevariationofconservatismacrossfirmswithinanindustry.Theseexpressionsarebothexplicit(e.g.,Ryan,2006,p.3),2andimplicit(byresearchersproposingsuchmeasures,e.g.,Callenetal.,2008).Despitethedemand,thereisasyetnowell-articulatedfirm-yearconservatismflowmeasure.Thispaper’sobjectiveistomeetthatdemand.Inthepaperwe:(i)estimateafirm-yearconservatismmeasure;(ii)provideevidenceconsistentwiththemeasurereflectingconservatisminatimelyfashion;and(iii)illustratethemeasure’sapplicationbytestingpreviouslyuntestedconservatismhypotheses.Estimation.Estimationofourfirm-yearmeasure,C_Score,isbasedontheBasu(1997)asymmetricearningstimelinessnotionimpliedbyaccountingconservatism(e.g.,Basu,1997;Watts,2003a;RoychowdhuryandWatts,2007).Thereisalargeliteratureprovidingempiricalevidenceonbothasymmetricearningstimeliness’existence(forasummaryseeWatts,2003b),anditscross-sectionalandinter-temporalvariationaspredictedbythetheoryofconservatisminWatts(2003a).Giventhisevidence,weassumeasymmetricearningstimelinessrepresentsaflowmeasureofaccountingconservatism,andgenerateafirm-yearmeasureofconservatism.Weestimateannualcross-sectionalBasu(1997)regressions,specifyingtheasymmetricearningstimelinesscoefficient(orBasucoefficient)asalinearfunctionoffirm-specificcharacteristics.Thesecharacteristics—size,market-to-bookandleverage—arechosenbecauseconservatismvarieswiththem,boththeoreticallyandempirically(e.g.,LaFondandWatts,2008).Welimittheestimationtothesethreeindependentvariablesforreasonsofparsimony.Usingfewervariablesinestimationmakesmorefirm-yearsandhencelargersamplesavailabletoresearchers.3Thefirm-yearconservatismmeasure,C_Score,iscalculatedbysubstitutingthefirm’ssize,market-to-bookandleverageintotheestimationregressionforthatyear.Section2providesdetails.EmpiricalProperties.ToassessC_Score’seffectivenessinmeasuringconservatism,weexaminewhetheritsempiricalpropertiesareconsistentwithth