Financial Reporting Quality, Structural Problems and the Informativeness of Mandated

JournalofBusinessFinance&Accounting,40(3)&(4),318–349,April/May2013,0306-686Xdoi:10.1111/jbfa.12015

FinancialReportingQuality,StructuralProblemsandtheInformativenessofMandatedDisclosuresonInternal

Controls

ALOKE(AL)GHOSHANDYONGGYULEE*

Abstract:EvenbeforefirmsreportinternalcontrolweaknessesundertheSarbanes–OxleyAct(SOX),theyarecharacterizedbystructuralproblems,arepronetointernalcontrolweaknesses,andhavelowfinancialreportingquality.Ifthestockmarketincorporatesmuchofthisinformationduringthepre-disclosureyears,investorsarelesssurprisedwhenfirmssubsequentlyreportinternalcontrolweaknessesunderSOX.Wefindthatforthepre-disclosureperiod,firmsreportinginternalcontrolweaknessesunderSOX,(1)hadstructuralproblems,(2)werepronetointernalcontrolproblems,and(3)hadlowfinancialreportingquality.Further,weprovidedirectevidencethatstockpricesduringpre-disclosureyearsincorporatemuchoftheinformationaboutstructuralproblems,thelikelihoodofinternalcontrolweaknesses,andlowreportingquality.Finally,wefindthatmanyofthesevalue-relevantfactorsarenotrelatedtoannouncementperiodreturnswhenfirmseventuallydisclosesuchproblemsunderSOXandthatlimitednewinformationaboutstructuralproblemsisgeneratedaroundthisdate.Ourresultsprovideacompellingexplanationforthemutedstockpricereactionaroundthemandatorydisclosuredate.

Keywords:Sarbanes–OxleyActof2002,internalcontrolweakness,marketreactions,financialreportingquality

1.INTRODUCTION

Understandingthedeterminantsoffinancialreportingqualityisoneofthefundamen-talissuesinaccountingresearch(CostelloandWittenberg-Moerman,2011).Factors

∗The

authorsarefromtheStanRossDepartmentofAccountancy,ZicklinSchoolofBusiness,BaruchCollege,CityUniversityofNewYork.Anearlierversionofthispaperwascirculatedunderthetitle“TimelinessandmandateddisclosuresoninternalcontrolsunderSection404.”WeacknowledgehelpfulcommentsfromSaraBerman,PeterJoos,RyanLafond,MartienLubberink,ShaminMashruwala,HugoNurnberg,JohnO’Hanlon,KenPeasnell,StephenPenman,GilSadka,JanSweeney,BobTucker,YuanZhang,JerryZimmerman,participantsattheKatholiekeUniversiteitLeuvenaccountingresearchworkshopandparticipantsattheAmericanAccountingAssociationMeetings.(PaperreceivedAugust,2011,revisedversionacceptedDecember,2012).

Addressforcorrespondence:Aloke(Al)Ghosh,StanRossDepartmentofAccountancy,ZicklinSchoolofBusiness,BaruchCollege,CityUniversityofNewYork,55LexingtonAvenue,NewYork,NY10010,USA.e-mail:Aloke.Ghosh@baruch.cuny.edu

󰀃C2013BlackwellPublishingLtd,9600GarsingtonRoad,OxfordOX42DQ,UKand350MainStreet,Malden,MA02148,USA.

318

INFORMATIVENESSOFINTERNALCONTROLDISCLOSURES

319

thatgovernfinancialreportingqualityincludeinternalandexternalmonitoringmechanisms.Whilethereisextensiveresearchonthelinkagesbetweendisparatemonitoringmechanismsandreportingquality,theliteratureoninternalcontroloverfinancialreportingandinternalmonitoringmechanisms(GohandLi,2011;andJohnstoneetal.,2011)hasgrownconsiderablyoverthelastdecadelargelyasaconsequenceofregulatoryinterventionintheUS(seeSchneideretal.,2009forareview).ThedisclosuresundertheSarbanes–OxleyActof2002(SOX)providevaluabledataforresearcherstouseexploringareasthatwerepreviouslylimitedbecauseinternalcontroldatawerenoteasilyavailable.1

Theacademicliteratureoninternalcontroloverfinancialreportingcanbebroadlyclassifiedintotwocategories.2Onebranchfocusesonthecausesofinternalcontrolproblemsandthekeyattributesassociatedwithweakinternal-controlfirms(Ashbaugh-Skaifeetal.,2007;Doyleetal.,2007a;Ognevaetal.,2007;NaikerandSharma,2009;andHoitashetal.,2009).Anotherbranchexaminestheconsequencesofinternalcontroldisclosuresincludingstockmarketreactions(Beneishetal.,2008;andHammersleyetal.,2008),earningsquality(Doyleetal.,2007a;Ognevaetal.,2007;Ghoshetal.,2010;andGohandLi,2011),andthecostofcapital(Ashbaugh-Skaifeetal.,2007;Ognevaetal.,2007;andBeneishetal.,2008).Inparticular,priormarket-basedstudiesfindamutedstockmarketreactiontotheannouncementofweaknessesininternalcontrols(e.g.,Beneishetal.,2008;andHammersleyetal.,2008).Inthisstudy,weinvestigateexplanationsofwhythestockmarketreactiontointernalcontroldisclosuresismutedbydrawingontheliteratureonthecausesofinternalcontrols.Becausestructuralproblemsleadtointernalcontrolproblems,andmanyofthestructuralproblemsarepricedbyinvestorspriortofirmsreportingsuchproblems,internalcontroldisclosuresprovidelimitednewinformation.Thus,wecontributetotheinternalcontrolliteraturebyprovidingadirectlinkagebetweenthestudiesthatexaminethecausesofinternalcontrolsandthosethatfocusontheconsequencesofinternalcontrols.

Thedisclosuresontheeffectivenessofinternalcontrolsoverfinancialreporting,requiredunderSections302and404(§302and§404,respectively,hereafter)ofSOXareexpectedtoprovidenewinformationaboutthereliabilityoffinancialstatements(SEC,2003).3Contrarytotheseclaims,stockmarketreactionaroundthetimeofthemandateddisclosuresoninternalcontrolsisgenerallyweak.Forinstance,Hammersleyetal.(2008)andBeneishetal.(2008)findthatthemeanabnormalstockreturnsforfirmsreportingmaterialweaknessesininternalcontrolsunder§302

1TheUSCongresspassedtheSarbanes–OxleyActwhichmandatedtheevaluationanddisclosuresoftheeffectivenessofinternalcontrolsbypublicfirmsandanexternalauditofthemanagement’srepresentationoftheeffectivenessofinternalcontrol(KinneyandShepardson,2011).

2Internalcontroloverfinancialreportingisaprocesswhichprovidesreasonableassuranceregardingthereliabilityoffinancialreportingandthepreparationoffinancialstatements.Becausefirmswithinternalcontrolproblemshavemorethanaremotelikelihoodthatamaterialmisstatementoftheannualorinterimfinancialstatementswillnotbepreventedordetected,financialstatementsarelessreliablewhenfirmshaveweakinternalcontrols(PCAOB,2004).

3Becauseweakinternalcontrolsleadtolowfinancialreportingquality,disclosuresofinternalcontrolweaknessesunderSOXmightprovidenewinformationaboutthefirms’financialreportingquality.Similarly,sinceweakinternalcontrolsareassociatedwithpersistentstructuralproblems,internalcontrolweaknessesunderSOXmightindicatethatthesefirmshavefundamentalproblemswhichwerepreviouslyunknowntothemarket.

󰀃C

2013BlackwellPublishingLtd

320

GHOSHANDLEE

areeconomicallysmall(between0.5and2%),butstatisticallysignificant.4Beneishetal.(2008)findnoevidencetosuggestthatthestockmarketreactsnegativelytomaterialweaknessesininternalcontrolsunder§404.Incontrast,Ashbaugh-Skaifeetal.(2009)findastatisticallysignificantstockmarketreaction,butthemagnitudeislessthan1%.Themediannumbersaregenerallyinsignificantornotreportedinthesestudies.Finally,whileafewstudiesfindarelationshipbetweenstockmarketreactionandtheseverityoftheinternalcontrolproblems(Hammersleyetal.,2008),othersfindnosuchevidence(Beneishetal.,2008).

Weprovideseveralinter-relatedexplanationsforthemutedstockmarketreac-tiontotheannouncementofinternalcontrolproblemsunderSOX.First,firmswithweakinternalcontrolsareassociatedwithstructuralproblemsrangingfromchangesintheorganizationalstructure,complexity/scopeofoperations,intricatebusinesstransactions,andoperatingindiversemarkets(Ashbaugh-Skaifeetal.,2007).Becausefirmstendtodevelopstructuralproblemsovertime,andremediationoftheseproblemscanbeprolonged,someofthestructuralproblemsassociatedwithfirmshavingweakinternalcontrolsmighthavepre-datedthemandateddisclosureperiod.Therefore,investorsmightbeabletoassesswhetherfirmshavestructuralproblemsfromobservableattributesincludingchangesinorganizationalstructure,firmperformance,financialhealth,andcomplexityofbusinesstransactions.

Second,itispossibleforinvestorstodiscriminate,atleastpartially,betweenfirmswithweakandstronginternalcontrolsevenintheabsenceofmandateddisclosures.Priorstudiesfindthatfirmswithmaterialweaknessesininternalcontrolsaresmaller,lessprofitable,growingrapidly,morecomplex,takelargerestructuringcharges,andhavefrequentauditorchanges(Doyleetal.,2007a;andKrishnan,2005).Therefore,investorscanrationallyassesswhichfirmsaremorelikelytoencounterinternalcontrolproblemsbasedonobservablefirmattributesoverthepre-disclosureperiod.

Third,firmswithweakinternalcontrolshavelowfinancialreportingqualitybecauseoferrorsintheestimationofaccrualsandmanagementinjectedbiases(Ashbaugh-Skaifeetal.,2007;andDoyleetal.,2007b).Thelowfinancialreportingqualityisalsolikelytohavepre-datedthemandateddisclosureperiod.IffirmsreportinginternalcontrolweaknessesunderSOXwerealsopronetosuchproblemspriortothedisclosureperiod,earningsqualityisalsoexpectedtobelowforthisperiod.Further,someofthestructuralproblemsthatarekeyreasonsforfirmsdevelopinginternalcontrolproblemsarealsofactorsthaterodeearningsquality.

Ourfundamentalhypothesisisthatmostoftheinformationaboutstructuralproblems,latentinternalcontrolproblems,andlowfinancialreportingqualityisexpectedtobeincorporatedinthestockpriceduringthepre-mandateddisclosureyears.Therefore,marketparticipantsarelesssurprisedwhenfirmssubsequentlyreportinternalcontrolweaknessesunderSOX.Figure1providesachartpresentingthelinkagesamongfirm-specificattributes,thelikelihoodofhavinginternalcontrolproblems,andfinancialreportingquality,andtheireffectsonthestockpriceforthepre-andpost-disclosureperiods.

4Usingthemedian(mean)marketvalueofequityreportedforthesampleoffirmswithmaterialweaknessesininternalcontrolsinBeneishetal.(2008)asabenchmark,meanabnormalreturnsofaround2%overathree-daywindowtranslateintoalossofUS$55,522(US$5,428)forshareholders,whichappearstobeeconomicallysmall.

󰀃C

2013BlackwellPublishingLtd

INFORMATIVENESSOFINTERNALCONTROLDISCLOSURES

321

Figure1

FirmswithInternalControlWeaknesses:LinkagesbetweenDisclosureand

Pre-DisclosurePeriods

Firms reporting internal control weaknesses under SOX

Pre-disclosure period (2002)

Disclosure period (2003–2007)

Explanations for weak stock market reaction to disclosures on internal controlsInternal Control Problems-High probabilityStock Market Pricing-Some market reactionto structural problems, high probability of internal control problems, and low financial reporting qualityFinancial ReportingQuality-LowStructural Problems-FirmattributesDisclosures on Internal Controls-Weak stock market reactionWeprovideinsightsintotherelatedhypothesesusingthefollowingmethodology.First,weexaminewhethertheattributesthatareassociatedwithinternalcontrolweaknessespre-datethedisclosuresunderSOX(“firmattributestest”).Second,wetestwhetherfirmsreportinginternalcontrolproblemsunderSOXarepronetointernalcontrolproblemsforthepre-disclosureperiodcomparedtootherfirmswithoutsuchproblems(“controlproblemstest”).5Third,weinvestigatewhetherfirmsreportinginternalcontrolweaknessesunderSOXhadlowfinancialreportingqualityforthepre-disclosureyears(“reportingqualitytest”).

Finally,andmoreimportantly,weexaminewhetherinvestorsincorporatedfirmattributes,highlikelihoodofinternalcontrolproblems,andlowfinancialreportingqualityintosecuritypricesforthepre-disclosureyears(“marketpricingtest”).WedecomposethetimeperiodleadinguptotheannouncementofinternalcontrolproblemsunderSOXintothreesub-periods:(1)pre-disclosureyear,i.e.,2002;(2)theinterimperiodbetween2002andtheyearbeforetheannouncementdate;and(3)theannouncementdate.Ifthestockmarketincorporatedmuchoftheinformationbeforetheannouncementdate,thestockpriceoverthepre-disclosureandinterimyearsisexpectedtobeassociatedwithpre-existingproblemsincludingstructuralproblems,latentinternalcontrolproblems,andlowfinancialreportingquality.Further,thesefactorsareunlikelytobeassociatedwiththestockmarketreactiontotheannouncementofinternalcontrolweaknessesifsuchinformationhasalreadybeenincorporatedinthestockprice.

Ourresultsarebasedonasampleof672distinctfirmsthatreportedmaterialweaknessesunder§302or§404forfiscalyears2003–2007.PriorstudiesdocumentthatfirmsreportinginternalcontrolweaknessesunderSOXwereassociatedwithcomplexoperations,ageofthefirm,poorperformance,poorfinancialhealth,

5Wemodeltheprobabilityofinternalcontrolproblemsbasedonfirmattributes,suchascomplexity,performance,financialhealth,size,growth,auditorcharacteristics,restatement,andlitigiousindustries,usingdatafromthemandateddisclosureperiod(2003–2007)andthenestimatean‘implied’probabilityofinternalcontrolproblemsforthepre-disclosureyear(2002)usingtheestimatedcoefficientsandthefirm-specificvaluesforthatyear.SeeSection3fordetails.

󰀃C

2013BlackwellPublishingLtd

322

GHOSHANDLEE

firmsize,rapidgrowth,frequentauditorchanges,highauditfees,Big6auditors,morerestatements,industrieswithfrequentlawsuits,externalfinancingneeds,andinstitutionalownership.Consistentwiththeattributestest,wefindthatthefactorsthatdistinguishbetweenfirmsreportinginternalcontrolweaknessesunderSOXandthosewithoutsuchproblemsgenerallydiscriminatethetwosetsoffirmsalsoforthepre-disclosureyear.Alsoconsistentwiththecontrolproblemstest,wefindthatfirmsreportingweaknessesininternalcontrolunderSOXhadsignificantlyhigherimpliedinternalcontrolprobabilitiesforthepre-disclosureyearcomparedtofirmswithoutsuchweaknesses.Further,weshowthatthepre-disclosurefinancialreportingqualityislowerforfirmsreportinginternalcontrolweaknessesunderSOX,comparedtothosewithoutsuchproblems.

Moreimportantly,ourmarketpricingtestindicatesthatinformationregardingfirmattributes,internalcontrolissues,andfinancialreportingqualityhavebeenincorporatedinstockpricespriortothemandateddisclosureperiod.Wefindthatthelikelihoodofinternalcontrolproblemsandstructuralproblemsisnegativelyassociatedwithabnormalreturnsforthepre-disclosureyear(2002).Wealsofindevidenceindicatingthatthelikelihoodofinternalcontrolproblemsisassociatedwithabnormalreturnsduringtheinterimperiod.Incontrast,thesevalue-relevantfactorsarenotrelatedtothree-dayabnormalreturnsaroundthedisclosuresofmaterialweaknesses.Therefore,ourresultssuggestthatmostofthevalue-relevantinformationassociatedwithinternalcontrolproblemsisobserved,oranticipated,andincorporatedinthestockpriceoverthepre-disclosureperiod.Controllingforthepriorperiodinformation,limitednewinformationisrevealedaroundtheannouncementofinternalcontrolproblemswhichexplainstherelativelyweakmarketreaction.

Finally,wealsoexamineasub-sampleofforeignfirmscrosslistedintheUS.Becauseforeigncross-listedfirms(e.g.,ADRfirms)arealsosubjecttotheinternalcontroldisclosuresunderSOX,eventhoughsuchdisclosuresmaynotberequiredintheirhomecountries,ADRfirmsprovideapowerfulsettingtoexaminetheinformationcontentofthemandateddisclosuresoninternalcontrols.Similartothefullsampleresults,wefindthatstockpricesincorporateinformationaboutstructuralproblemsevenbeforeADRfirmsreportinternalcontrolproblemsunderSOX.

Ourstudyaugmentstheliteratureinseveralways.First,althoughinternalcontroliswidelyacceptedasakeydeterminantoffinancialreportingquality(e.g.,Peasnelletal.,2005),muchoftheresearchislimitedtotheUSbecauseofdatalimitations.However,insightsfromUSstudies,includingours,canpotentiallyinformothercountriesontherelativebenefitsofregulatinginternalcontrols.Second,researchstudiesoninternalcontrolstendtoprovideuniqueinsightsanalyzingdatafromthepost-SOXperiod(i.e.,themandatedperiod).Incontrast,weemphasizethepre-SOXperiod(i.e.,thepre-mandatedperiod)tobetterunderstandthedisclosureconsequencesinthepost-SOXperiod.Third,theliteratureontheinternalcontrolsfocusesoneitherthecausesorconsequencesofweakinternalcontrolswithoutacknowledgingacausallinkbetweenthetwostreamsofliterature.Webridgethisgap.Finally,akeyimplicationofourstudyisthatearningsquality/structuralproblemsandinternalcontrolweaknessesareendogenous.However,priorstudiesfailtoaccountforthisendogeneity,whichlimitsourabilitytodrawunambiguousconclusionsfromsuchstudies.

Therestofthepaperisorganizedasfollows:Section2providesthebackgroundontheissueofinternalcontroloverfinancialreportingandpresentsourhypotheses.

󰀃C

2013BlackwellPublishingLtd

INFORMATIVENESSOFINTERNALCONTROLDISCLOSURES

323

Section3describesourresearchdesign.Section4explainsourdataselectionproce-dure.Section5reportstheempiricalresults.Section6summarizesandconcludes.

2.INTERNALCONTROLOVERFINANCIALREPORTING

(i)Background

Internalcontroloverfinancialreportingisbroadlydefinedasaprocesscarriedoutbyacompany’sboardofdirectors,management,andotherpersonnelthatisdesignedtoprovidereasonableassuranceregarding(1)theeffectivenessandefficiencyofafirm’soperations,(2)thereliabilityoffinancialreporting,and(3)compliancewithapplicablelawsandregulations.Internalcontroldeficienciesarisefromdesignflawsoroperatingproblems.A“significantdeficiency”isaninternalcontroldeficiencythatcouldadverselyaffectafirm’sabilitytoinitiate,record,process,andreportfinancialdata.A“materialweakness”existswhenasignificantdeficiencyinoneormoreoftheinternalcontrolcomponentsprecludesthefirmfromdetectingandpreventingamaterialmisstatementinitsfinancialstatementsonatimelybasis.

Thedebatesurroundinginternalcontroloverfinancialreportingisnotnew(Kinneyetal.,1990).In1978,theCohenCommissionrecommendedthatmanage-mentprovideareportoftheirassessmentofthefirm’sinternalcontrols(TheCohenCommission,1978).TheTreadwayCommissionof1987(TheTreadwayCommission,1987)andtheCommitteeofSponsoringOrganizationsoftheTreadwayCommission(COSO,1992)alsomadesimilarrecommendations.Allthreereports(Cohen,Tread-wayandCOSO)recognizedthevalueofmanagementreportsoninternalcontrols(Hermanson,2000).

In1991,theUSHouseofRepresentativespassedabillrequiringmanagementtoprovideareportoninternalcontrols,accompaniedbytheauditor’sassessmentofthemanagementreport,butthebilldidnotpasstheSenate.PriortotheSarbanes–OxleyActof2002,theFDICImprovementActof1991requiredthatmanagementoflargebanksreportoninternalcontrolandthatauditorsattesttomanagement’sassessment.Basedonasurveyofninedifferentfinancialusergroups(bankers,brokers,directors,executives,analysts,institutionalinvestors,individualinvestors,CPAsandinternalauditors),Hermanson(2000)concludesthatrespondentsagreeaboutthevalueofvoluntarymanagementreportsoninternalcontrols,butareneutralabouttheroleofmandatorymanagementreportsoninternalcontrolsinenhancingdecision-making.

Morerecently,theSECapprovedrulesimplementing§404requiringmanagementandindependentauditorsofpubliccompaniestoevaluateandreportontheeffective-nessofafirm’sinternalcontroloverfinancialreporting.6TheSecuritiesandExchangeCommission(SEC)claimsthat“aneffectiveinternalcontroloverfinancialreportingisnecessarytoproducereliablefinancialstatementsusedbyinvestors”(SEC,2003).In2004,thePublicCompanyAccountingOversightBoard(PCAOB,2004)issuedAuditingStandardNo.2thatdescribedtherelatedauditorattestationrequirements.Inaddition,§302requiresthatmanagementevaluatetheeffectivenessofdisclosure

6Intheoriginalproposal,firmswererequiredtocomplywith§404forfiscalyearsendingonorafterSeptember15,2003(SEC,2003).However,thecompliancedatewasextendedforaccelerated(non-accelerated)filerstothefiscalyearendingonorafterJune15,2004(July15,2007).

󰀃C

2013BlackwellPublishingLtd

324

GHOSHANDLEE

controlandprocedures,andindicatesignificantchangesininternalcontrolsincethelastForm10-KorForm10-Q(Ashbaugh-Skaifeetal.,2007).Thus,firmsreportinginternalcontrolproblemsunder§404in2004andonwardsmightalsohavereportedsimilarproblemsin2003under§302.

(ii)InformationContentofInternalControlDisclosuresunderSOX

Theprimepurposeofinternalcontroloverfinancialreportingistofostertheprepa-rationofreliablefinancialstatements.Effectiveinternalcontrolinvolvespeopleatalllevelsoftheorganization,includingthosewhomaintainaccountingrecords,prepareanddisseminatepolicies,monitorsystems,andfunctioninavarietyofoperatingroles(Ernst&Young,2005).Effectiveinternalcontrolsimprovethereliabilityoffinancialstatements,andreducetheriskofmaterialmisstatementsandfraud(KinneyandMcDaniel,1989).7

Oneoftheprincipalmotivationsforthe§404disclosuresabouttheeffectivenessofinternalcontroloverfinancialreportingisthatthenewdisclosurerequirementswillenablemoretimelyidentificationofinternalcontrolweaknessesandhowsuchdisclosuresmightinfluencethereliabilityofearnings(Deloitte&Toucheetal.,2004;andPCAOB,2004).Theseclaimssuggestthatmandateddisclosuresoninternalcontrolsystemsareexpectedtogeneratesignificantnewinformationaroundthetimeoftheannouncementofinternalcontrolweaknesses.

Severalrecentstudiesexaminetheinformationcontentofthedisclosuresoninternalcontrolsystemsunder§302and§404byconcentratingonthestockpricereactionstotheannouncementofthesedisclosures.Theresultsgenerallyprovideweakeconomicandstatisticalevidencesupportingthecontentionthatthedisclosuresoninternalcontrolsareinformative.Hammersleyetal.(2008)findthatthemeansize-adjustedreturnsforfirmsdisclosinginternalcontrolweaknessesunder§302ofSOXarenegativeandlessthan1%onthedayofthedisclosures.Eventhoughtheresultsarestatisticallysignificant,theeconomicmagnitudeofthestockmarketreactiononthedayofthedisclosuresissmall.Further,themagnitudeofthemedianresultsisevensmallerandtheresultsarenotstatisticallysignificant.

Otherstudiesalsoexaminingthesamequestionarriveatsimilarconclusions.Beneishetal.(2008)findthatcumulativeabnormalreturnsoverathree-daywindowaroundthedisclosuresofmaterialweaknessesunder§302arearound2%,whichisstatisticallysignificant.Similarly,DeFrancoetal.(2005)findthatthemeancumulativeabnormalreturnsaroundthedisclosuresofinternalcontrolweaknessesunder§404arebetween0and2%withvaryingdegreesofstatisticalsignificancedependingontheeventwindowandthedefinitionofabnormalreturns.Neitherstudyreportsthemedianresults.

Thus,incontrasttotheclaimsthatthemandateddisclosuresoninternalcontroloverfinancialreportingunderSOXprovidevaluablenewinformation,priorstudiesprovideonlymodestevidenceofnewinformationbeingrevealedtoinvestorsaroundthetimeofthemandateddisclosures.

7Acentralpurposeoftheassessmentofinternalcontroloverfinancialreportingistoidentifymaterialweaknessesthathavemorethanaremotelikelihoodofcausingamaterialmisstatementinfinancialstatements(SEC,2005a).

󰀃C

2013BlackwellPublishingLtd

INFORMATIVENESSOFINTERNALCONTROLDISCLOSURES

325

(iii)ExplanationsforWeakMarketReactionsaroundInternalControlDisclosures

Ourstudyinvestigatesseveralinter-connectedexplanationsforthelackofstrongstockmarketreactiontodisclosuresoninternalcontrolsystemsunderSOX.First,firmsreportinginternalcontrolweaknessesunderSOXaretypicallycharacterizedashavingstructuralproblems.Thus,investorscangetnewinformationaboutthenatureandextentofstructuralproblemsfromdisclosuresontheeffectivenessofinternalcontrols.However,analternativeexplanationisthatfirmswithweakinternalcontrolshadsimilarproblemsforthepre-disclosureyears.Becauseinformationaboutstructuralproblemsisaninnatevalue-relevantfactor,investorsarelikelytohaveincorporatedsomeofthisinformationinthestockpriceduringthepre-disclosureyearswhentheproblemsfirstarose.

FirmsreportinginternalcontrolproblemsunderSOXaretypicallyidentifiedasthosehavingstructuralproblemsbecauseofchangesintheorganizationalstructure(i.e.,mergersandacquisitionsorrestructuringactivities),complexityandscopeofthefirm’soperations,intricatebusinesstransactions,andoperatingindiverseindustriesorinternationalmarkets(Ashbaugh-Skaifeetal.,2007;andDoyleetal.,2007a).Becausestructuralproblemstendtodevelopoveraperiodoftime,someofthecharacteristicstypicaloffirmswithweaknessesininternalcontrolsmightbedescriptiveforthepre-disclosureyearsaswell.8Thus,investorscangaugetheextentofstructuralproblemsfromobservablefirmattributes.

Second,firmsreportingweaknessesininternalcontrolsunderSOXmaynotcomeasasurprisetothemarketifinvestorshadanticipatedsomeoftheinternalcontrolproblems.Priorresearchfindsthatfirmswithinternalcontrolweaknessesareassociatedwithsmallsize,rapidgrowth,poorperformance,complexoperations,weakfinancialhealthandhighrisk(Ashbaugh-Skaifeetal.,2007;andDoyleetal.,2007a).Becausetheseobservablefirmattributestendtobestationaryovertime,investorscangaugetheextentofinternalcontrolproblemsduringthepre-disclosureperiod.

Third,firmsreportinginternalcontrolweaknessesunderSOXareexpectedtohavelowfinancialreportingquality.Internalcontrolweaknessescanaffectthereliabilityofearningsintwoprincipalways(Ashbaugh-Skaifeetal.,2007;andDoyleetal.,2007b).Thefirstisthatrandomorunintentionalerrorsinaccountingaccrualsestimationsareexpectedtobelargerinfirmswithinternalcontrolweaknessesthanthosewithoutsuchweaknessesbecauseoflackofadequatepolicies,trainingordiligencebycompanyemployees.Anotherreasonisthataccountingaccrualsmightbemisstatedintentionallybymanagementtomanageearningswheninternalcontrolsarenoteffective.Whileannouncementofinternalcontrolweaknessesmightconveyinformationaboutreportingquality,muchoftheinformationmightbeanticipatedifinvestorsrecognizedthatthesefirmswerepronetointernalcontrolproblemsduringthepre-disclosureyears.Becausestructuralproblemsdeterminefinancialreportingquality,reportingqualitymightbelowifthesefirmsalsohadstructuralproblemsduringthepre-disclosureyears.

Finally,ourfundamentalhypothesisisthatmuchoftheinformationcontainedinthemandateddisclosuresoninternalcontrolsoverfinancialreporting(structuralproblems,pronenesstocontrolproblems,lowreportingquality)isincorporatedinthestockpricesduringthepre-disclosureyears.Hence,whenfirmswiththesetypes

8Priorstudiesalsofindthatthesefirmattributesnegativelyaffectstockpricesandthevaluationofearningsbecauseofconcernsaboutthequalityofthefirmanditsfinancialreports.

󰀃C

2013BlackwellPublishingLtd

326

GHOSHANDLEE

ofproblemsannounceweaknessesininternalcontrolsunderSOX,marketsarelesssurprisedbythedisclosureofsuchnews.9Theserelatedpredictionsandthelinkagesamongfirm-specificattributes,internalcontrolweaknesses,financialreportingquality,andstockmarketbehavioraresummarizedinFigure1.

3.RESEARCHDESIGN

(i)FirmAttributesandtheProbabilityofInternalControlWeaknesses

SimilartoAshbaugh-Skaifeetal.(2007)andDoyleetal.(2007a),wemodelthelikelihoodoffirmsreportinginternalcontrolweaknessesunderSOXusingthefollowinglogisticregression:

Controlweakness=β0+β1Segments+β2Charges+β3Foreign+β4Age+β5ROA

+β6Loss+β7Z-Score+β8Size+β9ExtremeGrowth

+β10AuditChange+β11AuditFee+β12Big6+β13Restatement+β14Litigation+β15EquFin+β16DebFin+β17InstOwn

+β18CEOChair+β19GIndex+β20CFOChange+Industry+v

(1)

whereControlweaknessisanindicatorvariablethatequalsonewhenfirmsreportinternalcontrolproblemsandzerootherwise.

Equation(1)includesthefollowingtwentydeterminantsofinternalcontrolproblemsthatalsoindicatefirms’structuralproblems:Segmentsisthenumberofbusinessorgeographicsegmentsreportedbyafirm;Chargesiscodedoneifafirmhasrestructuringchargesinthecurrentyearorintheprevioustwoyears,andzerootherwise;Foreignisanindicatorvariableequaltooneifafirmhasnon-zeroforeigncurrencytranslationandzerootherwise;AgeisthedecilerankofthenumberofyearsthatafirmhasCRSPdata;ROAisincomebeforeextraordinaryitemsdividedbytotalassetsatthebeginningoftheyear;Lossisanindicatorvariablethatequalsonewhenincomebeforeextraordinaryitemsisnegativeforthecurrentyear;Z-ScoreisthedecilerankofAltman’s(1968)Z-scorewithhighscoresindicatinglessdistressrisk;Sizeistheaverageofmarketvalueofequity(inUS$billions)atthebeginningandendofthecurrentyear;ExtremeGrowthisanindicatorvariablethatequalsoneifsalesgrowthfallsinthehighesttercile,andzerootherwise;AuditChangeisanindicatorvariablesettooneifafirmchangesitsauditorforthecurrentfiscalyear,andzerootherwise;AuditFeeisexpressedperUS$1,000oftotalfirmvalue(marketvalueofequityandbookvalueofdebt);Big6isanindicatorvariablethatequalsoneiftheauditorisoneoftheBig6auditfirms,andzerootherwise;Restatementiscodedoneifafirmhasarestatement,andzerootherwise;LitigationiscodedoneifafirmisinalitigiousindustryhavingSICcodes2833–2836,3570–3577,3600–3674and7370,andzerootherwise;EquFiniscodedoneifafirm’sequityissuanceisgreaterthan10%ofitstotalassetsinthenextyear,andzerootherwise;DebFiniscodedoneifafirm’sdebtissuanceisgreaterthan10%ofitstotalassetsinthesubsequentyear,andzero

9Consistentwithourclaims,SECCommissionerPaulAtkinsasks,“Shouldthese(internalcontrol)disclosurestriggerasignificantmarketimpact?Aretheseproblemsalreadypricedintothestock?”(SEC,2005b).DeFrancoetal.(2005)alsoacknowledgethatitisplausiblethatinvestorsanticipatesomeoftheinternalcontrolproblemsdisclosedunderSOX.

󰀃C

2013BlackwellPublishingLtd

INFORMATIVENESSOFINTERNALCONTROLDISCLOSURES

327

otherwise;InstOwnisthepercentageofsharesheldbyinstitutionalinvestorsdividedbythenumberofinstitutionsthatownthestockasofthefiscal-year-end;CEOChairiscodedoneiftheCEOalsoservesaschairoftheboard,andzerootherwise;GIndexisacompositeofthetwenty-fourvariables,addingonepointifanyoftheprovisionsispresent,whereahigherscoreindicatesmorerestrictionsonshareholderrightsoragreaternumberofanti-takeovermeasures(seeGompersetal.,2003);andCFOChangeiscodedoneifthereisachangeinthechieffinancialofficer’s(CFO)positionandzerootherwise.

Drawingontherecentstudies,wecategorizeourdeterminantsofinternalcontrolweaknessesintotengroups:(1)complexity(Segments,Charges,ForeignandAge),(2)firmperformance(ROAandLoss),(3)financialhealth(Z-Score),(4)size(Size),(5)growth(ExtremeGrowth),(6)auditorcharacteristics(AuditChange,AuditFee,andBig6),(7)restatements(Restatement),(8)industrieswithhighlitigationrisk(Litigation),(9)relianceonexternalfinancing(EquFinandDebFin),and(10)managementandgovernanceattributes(InstOwn,CEOChair,GIndexandCFOChange).

Firmswithcomplexoperationsaremorelikelytohaveinternalcontrolweaknessesbecausecomplexbusinesstransactionsimposegreateraccountingapplicationrisks.Thelikelihoodoffirmsreportinginternalcontrolproblemsishigherforfirmswithpoorperformancebecausesuchfirmsmaynothaveadequateresourcesforpropercontrols.Goodinternalcontrolsrequirebothtimeandresourceswhichmaynotbethepriorityforfirmstryingtoimproveprofitability.Firmswithpoorfinancialhealthmaynothaveenoughresourcestodevotetowardseffectiveinternalcontrolsystems.Firmsizeisinverselyrelatedtotheinternalcontrolproblemsbecauselargerfirmsaremorelikelytohaveresourcestodevotetowardsinternalcontrols.Firmswithmoregrowthopportunitieshaveahigherlikelihoodofreportingmaterialweaknessesbecauseofpossibleproblemswithinternalcontrolsprocesses.Higherauditriskmightbeassociatedwithahigherlikelihoodofreportingmaterialweaknesses,leadingtoagreaterfrequencyofauditorchangesandhigherauditfees.

WhiletheBig6auditfirmsaremorelikelytouncoverinternalcontrolweaknesses,firmswithinternalcontrolproblemsmightavoidusingbigauditfirms.Therefore,anexanterelationshipbetweenBig6andControlweaknessisambiguous.Managers’incentivestodiscoverandreportinternalcontrolweaknessesarehigherforfirmswithrestatementsbecausefirmsaremorelikelytobeforthcomingaboutcontrolproblemswhenthequalityoftheirfinancialstatementshasbeenquestionedbymarketregulatorsorauditors(Ashbaugh-Skaifeetal.,2007).Firmswithhighlitigationriskhavegreaterincentivestodiscloseinternalcontrolproblemsinatimelymannertominimizepotentialsharepricedeclineswhichoftenleadstoshareholderlawsuits(Ashbaugh-Skaifeetal.,2007).Firmsissuingequityordebtarelesslikelytodiscloseinternalcontrolweaknesses(HermansonandYe,2009).Higherinstitutionalownershipprovidesmanagerswithgreaterincentivestodiscoveranddiscloseinternalcontrolproblemsbecauseofincreasedmonitoring.WhentheCEOservesaschairoftheboard,managementislesslikelytodiscloseinternalcontrolweaknessesbecausemonitoringisnotasstrong.AhigherG-indexindicatesmorerestrictionsonshareholderrightsormoreentrenchedmanagement,inwhichcasemanagementislesslikelytodisclosematerialweaknesses.Finally,anewCFOhasstrongerincentivestodiscloseinternalcontrolproblemsandthenblametheprioradministrationforsuchproblems.

󰀃C

2013BlackwellPublishingLtd

328

GHOSHANDLEE

Weestimateequation(1)usinginternalcontroleffectivenessdataasreportedunder§302or§404requirements.Usingthecoefficientestimatesfromthelogisticregression,wecomputetheprobabilityoffirmshavinginternalproblemsusing:

Pr(ICW)i,t=eXβ/(1+eXβ)

(2)

whereβisthevectorofcoefficientsestimatedusingdataforfirmsreportingmaterialweaknessesand2004controlsample.Xisthematrixofindependentvariablesidentifiedinequation(1).

Priorstudiesfindthatobservablefirmcharacteristicsareassociatedwiththelikelihoodoffirmshavinginternalcontrolweaknesses.Totheextentthattheestimatedcoefficientsfromequation(1)andfirmattributesarestationaryovertime,itisreason-abletoexpectthatcapitalmarketscangaugesomeoftheinternalcontrolproblemsevenintheabsenceofmandateddisclosuresbasedonfirm-specificcharacteristics.Toaddressthisissue,wefirstexaminewhetherthefirmattributesthatareassociatedwithinternalcontrolweaknessespre-datethedisclosuresunderSOX(“firmattributestest”).WethentestwhetherfirmsreportinginternalcontrolproblemsunderSOXarealsopronetointernalcontrolproblemsforthepre-disclosureperiodcomparedtootherfirmswithoutsuchproblems(“controlproblemstest”).Ourprobabilitymeasure,Pr(ICW),forthepre-disclosureperiodisexpectedtobehigherforfirmsreportinginternalcontrolproblemsundermandateddisclosuresthanthosewithoutsuchproblems.

(ii)ConstructsforFinancialReportingQuality

Weusetwomeasuresoffinancialreportingquality.ThefirstmeasureisAccrualsQualityasdefinedinDechowandDichev(2002).Thismeasureistheabsolutevalueoftheresidualsfromthefollowingequation:

󰀅WCt=β0+β1CFOt−1+β2CFOt+β3CFOt+1+β4󰀅Salest+β5PPEt+μt

(3)

Theresidualsfromtheregressionmeasuretheextenttowhichcurrentworkingcapitalaccruals(󰀅WC)donoteffectivelymapintopast,presentorfuturecashflows(CFO).FollowingFrancisetal.(2005),wealsoincludethecurrentyearchangeinsales(󰀅Sales)andthecurrentyearlevelofproperty,plantandequipment(PPE).Weestimatetheaboveregressionforeachyear-industrycombination.

ThesecondmeasureisPerformance-MatchedAbnormalAccruals,whichisAbnormalAc-crualsmodifiedbasedonportfolio-matchingapproach,asinFrancisetal.(2005).AbnormalAccrualsistheabsolutevalueofabnormalaccrualsestimatedfromthemodifiedJones(1991)modelasdescribedinDechowetal.(1995).Specifically,weestimatethefollowingregression:

Accruals=β0+β1󰀅Sales+β2PPE+η

(4)

whereAccrualsareincomebeforeextraordinaryitemslessoperatingcashflowfromcontinuingoperations,󰀅Salesischangesinsales,andPPEisthegrossvalueofplant,propertyandequipment.Tocontrolforcross-sectionaldifferencesinfirmsize,we

󰀃C

2013BlackwellPublishingLtd

INFORMATIVENESSOFINTERNALCONTROLDISCLOSURES

329

scaleallthevariablesinequation(4),includingtheintercept,bytotalassetsatthebeginningoftheyear.Weestimatetheequationforeachyear-industrygroupingandtheresultingcoefficientsareusedtoestimatethefirm-specificnormalaccrualsasfollows:

β1(󰀅Sales−󰀅AR)+󰀁β2PPEPredictedaccruals=󰀁β0+󰀁

(5)

where󰀅ARisthechangeinaccountsreceivablebetweenthecurrentyearandthe

precedingyear,scaledbybeginningassets.ConsistentwithDechowetal.(1995),weadjustthereportedrevenuesofthesamplefirmsforchangesinaccountsreceivabletocaptureanypotentialaccountingdiscretionarisingfromcreditsales.WemeasureAbnormalAccrualsastheabsolutevalueofthedifferencebetweenAccrualsandPredictedaccruals.

Toimplementtheportfolio-matchingapproach,wesortcompaniesineachtwo-digitSICusingROAforeachindustryandthenidentifythemedianabnormalaccrualineachquintile.Toobtaintheperformance-adjustedabnormalaccrual,wesubtractthemedianabnormalaccrualoftherelevantquintilefromeachfirm’sabnormalaccrual.Forbothmeasuresoffinancialreportingquality,ahighervalueindicatesalowerqualityoffinancialreporting.

TotheextentthatfirmsdisclosinginternalcontrolweaknessesunderSOXwerepronetocontrolissuesforthepre-disclosureperiod,financialreportingqualityisexpectedtobelowforthisperiod.Financialreportingqualitymightalsobelowasadirectoutcomeofthestructuralproblemsduringthepre-disclosureperiod.WeinvestigatethesepossibilitiesbytestingwhetherfirmsreportinginternalcontrolweaknessesunderSOXhadlowfinancialreportingqualityforthepre-disclosureyears(“reportingqualitytest”).

(iii)InvestorPerceptionsofStructuralProblems,ControlProblemsandReporting

Quality

Themainobjectiveofthisstudyistoexplorewhetherstockpricesincorporatestructuralproblems,potentialinternalcontrolweaknesses,andfinancialreportingqualityconcernsevenbeforefirmsdisclosesuchproblemsunderSOX.Accordingly,wedecomposethetimeperiodleadinguptotheannouncementofinternalcontrolproblemsunderSOXintothreesub-periods:(1)pre-disclosureyear,i.e.,2002;(2)theinterimperiodbetween2002andtheyearbeforetheannouncementdate;and(3)theannouncementdate.Ifthestockmarketincorporatedmuchoftheinformationbeforetheannouncementdate,thestockpriceoverthepre-disclosureandinterimyearsisexpectedtobeassociatedwithstructuralproblems,latentinternalcontrolproblems,andlowfinancialreportingqualityoverthisperiod.Further,thesethreefactorsareunlikelytobeassociatedwiththestockmarketreactiontotheannouncementofinternalcontrolweaknesses.(a)Pre-disclosureYear(2002)

Totestwhetherstockpricesinthepre-disclosureperiod(year2002)incorporatestructuralproblems,potentialcontrolweaknesses,andreportingqualityconcerns,we

󰀃C

2013BlackwellPublishingLtd

330

GHOSHANDLEE

estimatethefollowingregression:10

CARPre=β0+β1Pr(ICW)Pre+β2SP-ScorePre+β3AccrualsQualityPre

+β4󰀅EarningsPre+v

(6)

whereCARPreiscompoundedbuy-and-holdreturnlessthereturnonthevalue-weightedmarketportfolioovera12-monthperiodendingthreemonthsafterthe2002fiscalyear-end,11SP-ScorePreisthesumof14binarysignalsconstructedforourfirsteightgroupsoffirmattributevariablesasofthe2002fiscalyear-end,AccrualsQualityPreisameasureofaccrualsqualityfor2002,and󰀅EarningsPreisthechangeinannualearningsbeforeextraordinaryitemsduring2002,scaledbymarketvalueofbeginningequity.TocomputeSP-Score,wefirstconvertSegmentsandAuditFeeintobinaryvariablesandsetthemequaltooneifthevaluefallsinthehighesttercileofthedistributionforthatyearandzerootherwise.Similarly,weconvertAge,ROA,Z-Score,andSizeintobinaryvariablesandsetthemequaltooneifthevaluefallsinthelowesttercileofthedistributionforthatyearandzerootherwise.Theremainingvariables,i.e.,Charges,Foreign,Loss,ExtremeGrowth,AuditChange,Big6,RestatementandLitigation,arealldefinedasbinaryvariables.Thus,SP-ScorecantakeavaluebetweenzeroandfourteenwithahigherSP-Scoreindicatingmoreseverestructuralproblems.Ifpotentialinternalcontrolweaknesses,structuralproblems,andfinancialreportingqualityconcernswerepricedbythemarketinthepre-disclosureperiod,β1,β2andβ3inequation(6)areexpectedtobenegative.12(b)InterimPeriod

Toexaminewhetherstockpricesintheinterimperiod(i.e.,theperiodbetween2002andoneyearbeforetheannouncementofinternalcontrolweaknessesunderSOX)reflecttheinformationaboutstructuralproblems,potentialcontrolweaknessesandreportingqualityconcerns,weestimatethefollowingregression:

CARInterim=β0+β1Pr(ICW)Interim+β2SP-ScoreInterim+β3AccrualsQualityInterim

+β4󰀅EarningsInterim+v

(7)

wherethesubscriptInterimindicatesobservationsforeachfirm-yearduringtheinterimperiod.Iftheinformationaboutpotentialinternalcontrolweaknesses,structuralproblemsandfinancialreportingqualityconcernswasincorporatedbyinvestorsintheinterimperiod,β1,β2andβ3inequation(7)areexpectedtobenegative.

10Weusereturnsasthedependentvariablebecausesomeofourexplanatoryvariablesareflow(change)variables.Totheextentthatourexplanatoryvariablescapturethelevelofthefirms’structuralproblems,stockpricecouldalsobeusedasthedependentvariable.Wegetconsistentresultsusingstockpriceregressions,buttheconcernisthatstockpriceregressionsarepronetoomittedvariablesproblems.

11WhencomputingCAR,wedonotusearisk-adjustedmeasurebecausedoingsocouldmechanicallyresultinanegativeCARforfirmswithinternalcontrolproblems.Thisisbecausefirmswithinternalcontrolproblemstendtohavehighersystematicrisk(Ashbaugh-Skaifeetal.,2009).

12WedonotincludebothPr(ICW)andSP-Scoreinthesameregressionbecausethesetwovariablesareconstructedbasedonthesameinformationwhichresultsinahighcorrelation.

󰀃C

2013BlackwellPublishingLtd

INFORMATIVENESSOFINTERNALCONTROLDISCLOSURES

331

(c)AnnouncementDate

Weinvestigatethemarketreactionstotheannouncementofinternalcontrolweak-nessesusingthefollowingregression:13

CAR=β0+β1SP-ScorePre+β2󰀅SP-ScoreInterim+β3SP-Score+β4AccrualsQualityPre(8)

whereCARisthree-dayabnormalreturnscumulatedaroundtheannouncementdateofinternalcontrolweaknesses(days–1to+1),󰀅SP-ScoreInterimisthedifferenceinSP-Scorebetweentheyearbeforetheannouncementandthepre-disclosureyear(2002),SP-ScoreisSP-Scoreforthedisclosureyear,󰀅AccrualsQualityInterimisthedifferenceinAccrualsQualitybetweentheyearbeforetheannouncementand2002,Accrual-sQualityisAccrualsQualityforthedisclosureyear,andEarningsSurpriserepresentsthedifferencebetweenincomebeforeextraordinaryitemsforthemostrecentquarterbeforethedisclosuredateandthecorrespondingnumberfourquarterspriorscaledbytotalmarketvalueofequityfourquartersprior.

Iftheinformationaboutstructuralproblemsandfinancialreportingqualityfromthepre-disclosureperiodswereincorporatedinthestockpricebeforetheannounce-mentdate,thepre-disclosureperiodfactorsareunlikelytobeassociatedwiththestockmarketreactionaroundtheannouncementofinternalcontrolproblemsunderSOX.Therefore,β1,β2,β4andβ5areexpectedtobeinsignificant.However,iftheannouncementsofinternalcontrolproblemscontainnewinformation,β3andβ6areexpectedtobesignificantbecausethestockmarketanticipatesyear-endinformationaboutstructuralproblemsandfinancialreportingquality.

4.DATAANDSAMPLESELECTION

+β5󰀅AccrualsQualityInterim+β6AccrualsQuality+β7EarningsSurprise+v

WeobtainoursampleoffirmswithnewdisclosuresoninternalcontroloverfinancialreportingfromAuditAnalytics(AA),asubsidiaryoftheIVESIncgroup.Inadditiontothe§302and§404data,AAprovidesdetailedauditfirmrelateddata.AA’sDisclosureControls(SOX302)andInternalControls(SOX404)datafilescontaindetailedinformationontheeffectivenessofafirm’sinternalcontrols.Ifinternalcontrolisfoundtobeineffective,theunderlyingcausesoftheineffectivenessarealsoreportedinAA.AArecordsseveralcategoriesofcausesofinternalcontrolineffectiveness.WefindthatinternalcontrolweaknessesandGAAP/FASBapplicationfailuresdominatethesampleoffirmswithineffectiveinternalcontrols(100%and95%,respectively).WeobtainaccountingdatafromCompustatandstockreturninformationfromCRSP.DataforinstitutionalinvestorsareretrievedfromtheThomsonReutersdatabase.InformationabouttheboardchairandgovernanceindexisobtainedfromtheRiskMetricsdatabase.WeobtainCFOrelatedinformationfromtheExecuComp.14

13Intheanalysisofmarketreactionstotheannouncementofinternalcontrolweaknesses,weexcludethethreevariablesrelatedtothelikelihoodofreportinginternalcontrolweaknessesforthepre-disclosureyear,interimperiodandthedisclosureyear.Theprobabilityofreportinginternalcontrolproblemsinthedisclosureyearisonebyconstructionforfirmsthatdisclosedinternalcontrolweaknesses.

14WeuseanapproachsimilartoChavaandPurnanandam(2010)toensurethatwedonotmissCFOswithvarioustitlessuchasVicePresident-finance.Werunatextsearchontheannualtitlestringandcaptureallexecutiveswith“finance”,“treasurer”or“controller”inthetitlestring.Inafewcases,ifwefindmorethan

󰀃C

2013BlackwellPublishingLtd

332

GHOSHANDLEE

Table1

SampleSelectionandDistribution

PanelA:SampleSelectionProcedure–FullSample

Firmsannouncingmaterialinternalcontrolweaknessesunder§302or§404,as

identifiedbyAuditAnalyticsforfiscalyearsendinginJan2003throughDecember2007Less:

FirmsnotcoveredbyCompustat

Firmswithday0returnsnotavailablefromCRSPFirmsinfinancialindustries(SIC=6000s)

Firmswithoutatleast15observationsforeachindustry-yearcombination,whereindustryisidentifiedbythe2-digitSICcodeFirmswithinsufficientCompustatdataFirmswithout2002data

Finalsampleofdistinctfirmsthatdisclosedamaterialweakness

Firmsin2004withsufficientCompustatandCRSPdatameetingtheabove

industry-relatedrequirementsLess:

FirmswithidentifiedmaterialweaknessdisclosuresFirmswithoutaninternalcontroldisclosureFinalsampleofcontrolfirms

PanelB:SampleDistributionbyYearandIndustry–FullSampleIndustry(SICCodes)20032004200520062007MiningandConstruction(1000–1999)Food(2000–2111)

Textilesandprinting/publishing(2200–2799)Chemicals(2800–2824,2840–2899)Pharmaceuticals(2830–2836)

Extractive(1300–1399,2900–2999)Durablemanufactures

(3000–3999exc.3570–3579&3670–3679)Transportation(4000–4899)Utilities(4900–4999)Retail(5000–5999)

Services(7000–8999excluding7370–7379)

Computers(3570–3579,3670–3679,7370–7379)OthersTotal

112231753491966

4311712127121153631691293

2764936563231640184

2553713023121023103

91626143520

3,669

(1,716)(346)(252)(35)(310)(338)6724,172(672)(1,444)2,056Total

1.3%2.4%3.9%2.1%5.2%3.0%

25161217126

18928.1%355.2%263.9%7511.2%6710.0%15823.5%20.3%672100.0%

Table1,PanelAoutlinesoursampleselectionprocess.Webeginwithasampleof3,669distinctfirmsthatdisclosedmaterialweaknessesunder§302or§404forfiscalyearsendinginJan2003throughDecember2007.Weeliminatethefollowingobservations:1,716firmsnotcoveredbyCompustat,346firmsforwhichnostockpricedataareavailableonCRSPonthedaytheinternalcontrolweaknessisdisclosed,252firmsinfinancialindustries,35firmswithfewerthan15observationsforeachindustry-yearcombinationrequiredforthecomputationofabnormalaccruals,and310firmsnothavingsufficientCompustatvariablestoconductouranalyses.Wefurtherremove

onepersonwiththefinancetitleinthetopfivemanagers,wegivepreferenceto“chieffinancialofficer”,“CFO”or“VP-finance”over“controller”or“treasurer”.Ifthisstilldoesnoteliminateduplicates,wetaketheexecutivewiththehighestcompensationastheCFOofthefirm.

󰀃C

2013BlackwellPublishingLtd

INFORMATIVENESSOFINTERNALCONTROLDISCLOSURES

333

allfirmswithmissingobservationsin2002.Thissampleselectionprocedureresultsinafinalsampleof672distinctfirmsthatdisclosedamaterialweaknessduringoursampleperiod.ThecontrolsampleconsistsofallfirmslistedonCompustatthatdisclosedin2004ashavingeffectiveinternalcontroloverfinancialreporting.Atotalof2,056firmsmeetthisrequirement.

PanelBofTable1detailstheindustrycompositionofoursampleof672firmswithmaterialweaknesses.Themostcommonindustryisdurablemanufactures(28.1%),whichisfollowedbycomputers(23.5%)andretail(11.2%).PanelBalsorevealsthatoursamplefirmsareconcentratedbetween2005through2007becausemostofthesefirmsmadethefirstdisclosureofinternalcontrolproblemsunder§404.15

5.EMPIRICALRESULTS

(i)FirmAttributesTest

PanelAofTable2reportsthedifferencesinfirmattributesbetweenfirmsreportinginternalcontrolproblems(ICW),underthenewdisclosures,andthosewithoutsuchproblems(Control).Italsopresentstestsofdifferencesbetweenthetwogroupsbasedont-testsandWilcoxonrank-sumtests.

ThedescriptivestatisticsinPanelAaregenerallyconsistentwithpriorstudies(e.g.,Ashbaugh-Skaifeetal.,2007;Doyleetal.,2007a;andHermansonandYe,2009).Wefindthat,comparedtocontrolfirms,ICWfirmshavemorerestructuringactivities,havemoreforeignoperations,andareyounger.ICWfirmsalsohavesignificantlylowerROAandreportlossesmorefrequently.WhileSizeissignificantlysmallerforICWfirms,thedifferenceinExtremeGrowthisnotconsistentwithpriorstudies.Further,ICWfirmschangeauditorsmorefrequently,paymoreauditfees,andarelesslikelytohaveBig6auditfirmsrelativetocontrolfirms.RestatementandLitigationaresignificantlyhigherforICWfirms.Additionally,ICWfirmsaremorelikelytoissueequity.Finally,ICWfirmshavehigherinstitutionalownership,arelesslikelytohavetheCEOalsoservingaschairoftheboard,havegreatershareholderrights(lowerG-index),andaremorelikelytohaveCFOchanges.

InPanelB,wecomparethedifferencesinstructuralproblemsbetweenthetwogroupsoffirmsforthepre-disclosureperiod(2002).Wefindthatahalfofthefirmattributesaresignificantlydifferentbetweenthetwogroupsoffirmsthisperiod.Overall,theresultsfromTable2suggestthat,becausestructuralproblemsassociatedwithinternalcontrolweaknessespersistovertime,capitalmarketscandrawinferencesaboutinternalcontrolproblemsfromfirmcharacteristicsevenbeforefirmspubliclydisclosuresuchproblems.

Forcapitalmarketstodrawinferencesaboutinternalcontrolproblems,oneunderlyingassumptionisthatdifferencesinstructuralproblemsforfirmswithinternalcontrolproblemspersistovertime,i.e.,firmattributesare“sticky”.WedirectlytestthisassumptioninTable3bycomparingfirm-specificattributesbetweenthepre-andpost-disclosureperiodsforICWfirms.Moreover,significantshiftsinsomeoffirmattributesmightalsocausefirmstodevelopinternalcontrolweaknesses.Wefind

15Eventhoughallfirmshadtocomplywith§302fromAugust29,2002,wefindnofirmsinoursampledisclosingweaknessesininternalcontroloverfinancialreportingforthisyear.Therefore,weviewfiscalyear2002asthepre-disclosureperiod.

󰀃C

2013BlackwellPublishingLtd

334

GHOSHANDLEE

Table2

ComparisonofFirmAttributesbetweenMaterialWeaknessFirmsandControl

Firms

PanelA:DisclosurePeriodVariableSegmentsChargesForeignAgeROALossZ-ScoreSize

ExtremeGrowthAuditChangeAuditFeeBig6

RestatementLitigationEquFinDebFinInstOwnCEOChairGIndexCFOChangeSegmentsChargesForeignAgeROALossZ-ScoreSize

ExtremeGrowthAuditChangeAuditFeeBig6

RestatementLitigationEquFinDebFinInstOwnCEOChairGIndexCFOChange

PredictedDifference

+++−−+−−+++±+++++−−++++−−+−−+++±+++++−−+

ICWFirmsMeanMedian

n

ControlFirmsMeanMedian

n

DifferenceMean

Median

5.3745.0006725.2745.0002,0560.1000.000

***

0.000***0.1250.0006720.0820.0002,0560.043

0.3200.0006720.2790.0002,0560.041**0.000**16.43212.00067217.87912.0002,056−1.447**0.000

***

−0.037***−0.0430.0106720.0010.0482,056−0.044

0.4540.0006720.2530.0002,0560.200***0.000***1.1460.9936721.1220.9672,0560.0240.026

***

−0.256***1.2960.2646723.5870.5202,056−2.291

0.2800.0006720.3240.0002,056−0.044**0.000**0.1980.0006720.0770.0002,0560.121***0.000***4.3632.8456721.8780.9892,0562.485***1.8560.8511.0006720.9011.0002,056−0.050***0.000***0.4640.0006720.0480.0002,0560.417***0.000***0.3170.0006720.2790.0002,0560.038*0.000*0.2100.0006720.1500.0002,0560.060***0.000***0.0950.0006720.1040.0002,056−0.0080.000

***

0.002***0.0090.0085070.0070.0061,5230.002

0.6201.0001500.7771.000728−0.157***0.000***8.6589.0002379.1959.000951−0.536***0.000***0.2540.0002400.0820.0007220.172***0.000***5.3360.0910.22513.027−0.0600.4661.0771.0550.3480.1992.8090.9210.0480.3170.2050.1000.0100.7698.6600.117

5.0000.0000.0008.0000.0070.0000.8950.1950.0000.0001.3541.0000.0000.0000.0000.0000.0081.0009.0000.000

6725.1335.0002,0560.2036720.0960.0002,056−0.0066720.2110.0002,0560.01467215.87910.0002,056−2.852***672−0.0280.0272,056−0.032***6720.3550.0002,0560.111***6721.0950.9512,056−0.0186722.7990.3572,056−1.744***6720.3370.0002,0560.0126720.1900.0002,0560.0106722.2390.7732,0560.570***6720.9061.0002,0560.0156720.0280.0002,0560.020**6720.2790.0002,0560.038*6720.1520.0002,0560.053***6720.1020.0002,056−0.0024520.0080.0061,4810.002***1730.8191.000684−0.0502829.1699.000951−0.510***1880.0850.0006470.032

0.0000.0000.000−2.000***−0.020***0.000***−0.056−0.162***0.0000.0000.581***0.0000.000**0.000*0.000***0.0000.002***0.0000.000**0.000

PanelB:Pre-disclosurePeriod(2002)

󰀃C

2013BlackwellPublishingLtd

INFORMATIVENESSOFINTERNALCONTROLDISCLOSURES

335

Table2(Continued)

Notes:

PanelApresentsthedifferencesinfirmcharacteristicsbetweenfirmswithinternalcontrolweaknesses(ICW)andfirmswithoutinternalcontrolweaknesses(Control)inthedisclosureperiod.PanelBpresentsthedifferencesinfirmattributesbetweenthesetwogroupsoffirmsforthepre-disclosureperiod(year2002).Eachcontinuousvariableiswinsorizedatthetopandbottom1%.Variablesaremeasuredasfollows:Segments=Thenumberofreportedbusinessorgeographicsegments.

Charges=Anindicatorvariableequalto1ifafirmhasrestructuringchargesinthecurrentyearorintheprevioustwoyears,and0otherwise.

Foreign=Anindicatorvariableequalto1ifafirmhasanon-zeroforeigncurrencytranslation,and0otherwise.

Age=DecilerankofthenumberofyearsthatafirmhasCRSPdata.

ROA=Incomebeforeextraordinaryitemsdividedbytotalassetsatthebeginningoftheyear.

Loss=Anindicatorvariableequalto1ifafirm’sincomebeforeextraordinaryitemsisnegative,and0otherwise.

Z-score=DecilerankofAltman’s(1968)Z-scorewithhighscoresindicatinglessdistressrisk.

Size=Averageofmarketvalueofequity(in$billions)atthebeginningandendofthecurrentyear.

ExtremeGrowth=Anindicatorvariableequalto1ifyear-over-yearindustry-adjustedsalesgrowthfallsintothetopquintile,and0otherwise.

AuditChange=Anindicatorvariableequalto1ifafirm’sauditorchangedduringthefiscalyear,and0otherwise.

AuditFee=Auditfeesexpressedper$1,000oftotalfirmvalue(marketvalueofequityandbookvalueofdebt).

Big6=Anindicatorvariableequalto1iftheauditorisoneoftheBig6auditfirms,and0otherwise.Restatement=Anindicatorvariableequalto1ifafirmhasarestatement,and0otherwise.

Litigation=AnindicatorvariablecodedoneifafirmisinalitigiousindustryhavingSICcodes2833–2836,3570–3577,3600–3674,and7370,and0otherwise.

EquFin=Anindicatorvariablecodedoneifafirm’sequityissuanceisgreaterthan10%ofitstotalassetsinthenextyear,and0otherwise.

DebFin=Anindicatorvariablecodedoneifafirm’sdebtissuanceisgreaterthan10%ofitstotalassetsinthenextyear,and0otherwise.

InstOwn=Thepercentageofsharesheldbyinstitutionalinvestorsdividedbythenumberofinstitutionsthatownthestockasofthefiscal-year-end.

CEOChair=AnindicatorvariablecodedoneiftheCEOalsoservesaschairoftheboard,and0otherwise.GIndex=Acompositeofthetwenty-fourvariables,addingonepointifanyoftheprovisionsispresent,whereahigherscoreindicatesmorerestrictionsonshareholderrightsoragreaternumberofanti-takeovermeasures(Gompersetal.2003).

CFOChange=Anindicatorvariablecodedoneifafirmexperiencesachangeinitschieffinancialofficer,and0otherwise.

***,**,*denotessignificanceat<0.01,<0.05and<0.10levels,respectively,fortwo-tailedtests.

thatSegments,ROA,Loss,AuditChange,Litigation,EquFin,DebFinandGIndexarenotsignificantlydifferentwhencomparingfirmswithinternalcontrolproblemsbetweenthepre-andpost-disclosureperiods,suggestingthatthesevariablesarestickyandthusmightpredictweaknessesininternalcontrols.

However,wealsofindthatsomeattributesarenotsticky,i.e.,therearesignificantchangesinattributesbetweenpre-andpost-disclosureperiods.Foreignissignificantlyhigherinthepost-disclosureperiodrelativetothepre-disclosureperiodwhichsuggeststhatICWfirmsexpandforeignoperations.16CEOChairislowerinthepost-disclosureperiodwhichcouldcapturestructuralchangesincorporateroomsaroundthistime.CFOChangeissignificantlyhigherinthepost-disclosureperiodwhichsuggeststhatCFOsaremostlylikelytobereplacedsubsequenttoreportingofinternalcontrol

16Manyfirms,includingICWfirms,expandedforeignoperationsoverthepost-disclosureperiodbecauseofglobalopportunities.Therefore,theincreaseinforeignoperationsforICWfirmscanalsobeexplainedbythechangingopportunitiesintheglobaleconomy.

󰀃C

2013BlackwellPublishingLtd

336

GHOSHANDLEE

Table3

StickinessofFirmAttributeswithinMaterialWeaknessFirms

PRE(2002)

VariableSegmentsChargesForeignAgeROALossZ-ScoreSize

ExtremeGrowthAuditChangeAuditFeeBig6

RestatementLitigationEquFinDebFinInstOwnCEOChairGIndexCFOChange

Mean5.3360.0910.22513.027–0.0630.4661.0750.9380.3480.1992.8420.9210.0480.3170.2050.1000.0100.7698.6600.117

Median5.0000.0000.0008.0000.0070.0000.8950.1950.0000.0001.3541.0000.0000.0000.0000.0000.0081.0009.0000.000

n672672672672672672672672672672672672672672672672452173282188

POST(DisclosureYear)Mean5.3740.1250.32016.432–0.0430.4541.1431.1920.2800.1984.5050.8510.4640.3170.2100.0950.0090.6208.6580.254

Median5.0000.0000.00012.0000.0100.0000.9930.2640.0000.0002.8451.0000.0000.0000.0000.0000.0081.0009.0000.000

n672672672672672672672672672672672672672672672672507150237240

Difference(=POST–PRE)Mean0.0370.034**0.095***3.405***0.019–0.0120.068*0.255*–0.068***–0.0011.662***–0.070***0.417***0.0000.004–0.004–0.001*–0.149***–0.0010.137***

Median0.0000.000**0.000***4.000***0.0040.0000.098**0.069***0.000***0.0001.491***0.000***0.000***0.0000.0000.0000.0000.000***0.0000.000***

Notes:

VariablesdefinitionscanbefoundinthenotetoTable2.

problemsinthepost-disclosureperiod.Finally,wealsofindsignificantincreasesinCharges,AuditFeeandRestatementinthepost-disclosureperiod.WhilewedocumentchangesincertainfirmattributesinTable3,weareunabletocommentwhetherthesechangesaretheoutcomeorcauseofinternalcontrolproblems.

(ii)ControlProblemsTest

Inadditiontotheunivariateanalysis,wealsomodelthelikelihoodoffirmsreportinginternalcontrolproblemsbasedonfirmsattributesidentifiedinTable2.Weestimatealogisticregressionwherethedependentvariableisanindicatorvariablewhichequalstooneifthefirmdisclosedamaterialweaknessininternalcontrol,andzerootherwise.TheresultsarereportedinPanelAofTable4.Wepresentfourmodelsbecausedataforsomevariablesarelimited.Thefirstmodelincludes14firmattributesbutwithoutincludingindustryindicators,whileinthesecondmodelweadditionallyincludeindustryindicators.Inthethirdmodelweincludethreeadditionalattributes(EquFin,DebFin,andInstOwn)andinthefourthmodelweincludethreeadditionalgovernanceattributes(CEOChair,GIndexandCFOChange).17

17Tominimizetheeffectofoutliersinthelogisticregression,wewinsorizeeachcontinuousvariableatthetopandbottom5%.Ourconclusionsremainunchangedwhenwedonotwinsorizethatdata.

󰀃C

2013BlackwellPublishingLtd

󰀃C

Table4LikelihoodofInternalControlWeaknessesp-value0.000***0.056*0.036**0.5390.037**0.001***0.004***0.091*0.1400.3100.000***0.000***0.6820.000***0.023**−4.6060.0340.359−0.0130.2751.7660.499−0.041−0.0300.1440.7350.299−0.0772.8730.384−4.8020.0320.272−0.0060.2912.6600.755−0.033−0.0260.0970.5710.307−0.3182.9090.4160.378−0.0239.5420.000***0.062*0.050**0.5750.034**0.003***0.006***0.1120.1430.2530.000***0.000***0.6580.000***0.013**0.000***0.1280.2110.8390.053*0.000***0.000***0.2720.2670.5070.004***0.000***0.1130.000***0.022**0.039**0.9120.466Estimatep-valueEstimatep-valueEstimatep-value0.1650.1130.3110.7470.6440.2050.017**0.5300.4590.2590.2760.003***0.6540.000***0.4080.5720.3290.072*0.010**0.9790.140Yes0.294670.03184.7122,0302013BlackwellPublishingLtd

PanelA:LogisticRegressionfortheDisclosurePeriodVariablePredictionEstimate−2.7930.0340.377−0.0140.2641.9720.520−0.037−0.0290.1240.7600.298−0.0712.8320.276InterceptSegmentsChargesForeignAgeROALossZ-ScoreSizeExtremeGrowthAuditChangeAuditFeeBig6RestatementLitigationEquFinDebFinInstOwnCEOChairGIndexCFOChangeIndustryindicatorsPseudoR2LikelihoodRatioPercentConcordantnNo0.288877.62684.6812,728Yes0.293893.85684.8822,728INFORMATIVENESSOFINTERNALCONTROLDISCLOSURES

+++−−+−−+++±+++++−−+−3.7920.0780.4740.0250.1743.4971.4650.0570.043−0.450−0.9350.4690.8544.006−0.424−0.433−0.711193.014−0.1920.0100.687Yes0.469248.07392.071596PanelB:PredictedProbabilityforthePre-disclosurePeriod(2002)Controln672Mean0.169Median0.106n2,056Mean0.041***DifferenceMedian0.028***ICWVariableMeanMedianPr(ICW)0.2090.135337

Notes:VariablesdefinitionscanbefoundinthenotetoTable2.338

GHOSHANDLEE

Inthefirsttwomodels,thepseudoR2isaround29%forbothmodels,whichissubstantiallyhigherthantheR2reportedbypriorstudies.Wealsofindthatpercentconcordantexceeds84%inbothspecifications.Therefore,ourmodelseemstoperformreasonablywellinpredictinginternalcontrolproblemscorrectly.18

Individualcoefficientsaremostlyconsistentwiththeresultsfrompriorstudies.ThecoefficientsonSegmentsandChargesarepositive,suggestingthatthelikelihoodoffirmsreportinginternalcontrolweaknessesincreaseswithbusinesscomplexity.ThecoefficientonROAissignificantlypositive,whichissurprising,butthecoefficientonLossissignificantlypositiveasexpected.Becausetheproportionoffirmswithlossesisrelativelyhighinoursample,theassociationbetweenLossandmaterialweaknessesdominatesthatbetweenROAandmaterialweaknesses.Also,thecoefficientonAgeisnotconsistentwithpriorstudies.ThesignificantlynegativecoefficientonZ-scoreconfirmsthatmaterialweaknessfirmsarefinanciallydistressed.WealsofindthatICWfirmschangeauditorsmorefrequentlyandpaymoreauditfees.ThecoefficientsonRestatementandLitigationarealsosignificant,consistentwithpriorresearch.Ontheotherhand,thecoefficientsonForeign,Size,ExtremeGrowthandBig6areinsignificant.Ourfindingssuggestthattheexplanatorypowerofthesevariablesmighthavedeclinedovertime.

Thethirdmodelincludestheinstitutionalownershipvariable,InstOwn,andtheexternalfinancingvariables,EquFinandDebFin.IncludingInstOwnreducesthesamplesizebyabout26%.Resultsaresomewhatweakerforthisspecification;onlyoneofthethreeadditionalvariablesissignificant.ThefourthmodelincludesCEOChair,GIndexandCFOChange.Thenumberofobservationsdeclinesbyabout71%relativetothethirdspecification.Mostoffirmattributesarenolongersignificantinthisspecificationpossiblybecauseofthesizebiasfromrequiringgovernancedata.

Overall,theresultsfromPanelAofTable4demonstratethatoursecondmodelperformsreasonablywellinpredictinginternalcontrolproblemscorrectlyaswellascapturingthedeterminantsofsuchproblems.Fortheremainderofthisstudy,werelyonthesecondmodelwhencomputingPr(ICW)andSP-Score.

InPanelBofTable4,wecomputean“impliedprobability”forICWfirmsforthepre-disclosureperiodbasedontheestimatedcoefficientsfromthesecondspecificationofthelogisticregressioninPanelAandfirm-specificattributesfor2002.Wefindthat,forfirmsreportinginternalcontrolproblemsunderSOX,themean(median)impliedprobabilityis20.9%(13.5%)forthepre-disclosureperiod,whereasthemean(median)impliedprobabilityforthecontrolsampleforthecorrespondingperiodis16.9%(10.6%).Thedifferencesinimpliedprobabilitiesbetweenthetwosetsoffirmsaresignificantatthe1%level.Thus,firmsreportinginternalcontrolproblemsunderSOXalsohadahigherprobabilityofreportingsuchproblemsin2002.OurresultsfromTable4areconsistentwiththeassertionthatmarketparticipantscandiscriminatebetweenfirmswithhighandlowlikelihoodofhavinginternalcontrolproblemsevenintheabsenceofmandateddisclosures.

18Wefurthercheckthepredictiveaccuracyofourmodelusinga“cross-validation”approach.Specifically,werandomlydividethesampleof2,728firms(672ICWfirmsand2,056controlfirms)intoananalysis-sampleandholdout-sampleofequal-size.Wethencomparetheclassificationaccuracybetweenthetwosubsamples.Wefindthattheclassificationaccuracyfortheanalysissampleis79.0%andthatfortheholdoutsampleis82.6%.Thisenhancesourconfidenceintheaccuracyofourpredictionmodel.

󰀃C

2013BlackwellPublishingLtd

INFORMATIVENESSOFINTERNALCONTROLDISCLOSURES

339

Table5

QualityofFinancialReporting

PanelA:DisclosurePeriodVariable

AccrualsQuality

Performance-MatchedAbnormalAccruals

Mean0.1190.097

ICWFirmsMedian0.0760.051

n672672

Mean0.1180.073

ControlFirmsMedian0.0710.042

n2,0562,056

DifferenceMean0.0020.023***

Median0.0040.009***

PanelB:Pre-disclosurePeriod(2002)AccrualsQuality

Performance-MatchedAbnormalAccruals

0.1460.115

0.0730.052

672672

0.1270.104

0.0660.046

2,0562,056

0.018*0.011

0.008*0.006*

Notes:

PanelApresentsthedifferencesinfinancialreportingqualitybetweenfirmswithinternalcontrolweaknesses(ICW)andfirmswithoutinternalcontrolweaknesses(Control)inthedisclosureperiod.PanelBpresentsthedifferencesinfinancialreportingqualitybetweenthesetwogroupsoffirmsforthepre-disclosureperiod(year2002).Thevariablesarewinsorizedatthetopandbottom1%.Ourmeasuresoffinancialreportingareasfollows:

AccrualsQuality=TheabsolutevalueoftheresidualsfromtheDechowandDichev(2002)accrualsqualitymeasure,asadjustedbyFrancisetal.(2005).SeeSection3formoredetails.

Performance-MatchedAbnormalAccruals=AbnormalAccrualsmodifiedbasedonFrancisetal.’s(2005)portfolio-matchingapproach,whereAbnormalAccrualsistheabsolutevalueofabnormalaccrualsestimatedfromthemodifiedJones(1991)modelasdescribedinDechowetal.(1995).SeeSection3formoredetails.***,**,*denotessignificanceat<0.01,<0.05and<0.10levels,respectively,fortwo-tailedtests.

(iii)ReportingQualityTest

Table5,PanelAreportsthedifferenceinfinancialreportingqualitybetweenICWfirmsandcontrolfirmsforthedisclosureperiod.Wefindthatthemean(median)Performance-MatchedAbnormalAccrualsis0.097(0.051)forICWfirmswhilethecorre-spondingnumberforthecontrolfirmsis0.073(0.042).Themean/mediandifferencebetweenthetwogroupsoffirmsissignificantatthe1%level.However,AccrualsQualitydoesnotsignificantlydifferbetweenthetwogroupsoffirms.

InPanelBofTable5,weexaminewhetherICWfirmshavelowearningsqualityevenbeforefirmsreportsuchproblemsunderSOX(pre-disclosureperiod).Ourresultsprovidesomeevidenceconsistentwiththeconjecturethatthepre-disclosurefinancialreportingqualityisalsolowforICWfirmscomparedtocontrolfirms.Forthepre-disclosureperiod,themean(median)AccrualsQualityis0.146(0.073)forICWfirmswhilethecorrespondingnumberforcontrolfirmsis0.127(0.066).Themean/mediandifferencebetweenthetwogroupsoffirmsissignificantatthe10%level.WealsofindthatthemedianPerformance-MatchedAbnormalAccrualsissignificantlylargerforICWfirmsthancontrolfirms.ThisprovidessomesupportforthepredictionthatICWfirmsarelikelytohavelowfinancialreportingqualityinthepre-disclosureperiod.

(iv)MarketPricingTest

Inthissub-section,weexplorewhetherstockpricesreflectinformationaboutin-ternalcontrolproblems,firmattributesandreportingqualityineachofthethree

󰀃C

2013BlackwellPublishingLtd

340

GHOSHANDLEE

sub-periods,i.e.,(1)pre-disclosureyear(2002),(2)theinterimperiodbetween2002andtheyearbeforetheannouncementdate,and(3)theannouncementdate.(a)Pre-disclosureYear(2002)

Table6providestheresultsformarketpricingofpotentialcontrolweaknesses,firmattributes,andreportingqualityforthepre-disclosureperiod.Wepresentfiveestimationsoftheregressionmodeldescribedinequation6.ThecoefficientonPr(ICW)Preissignificantlynegative(β=–0.495,p-value<0.001)inthefirstmodel.Thesamecoefficientisalsosignificantlynegative(β=–0.522,p-value<0.001)inModel4whenweadditionallyincludetheproxyforreportingquality,AccrualsQuality.OurresultssuggestthatstockpricesincorporateinformationaboutpotentialcontrolweaknessesevenbeforefirmsreportinternalcontrolproblemsunderSOX.

InModel2,wefindthatthecoefficientonSP-ScorePreisnegativeandsignificant(β=–0.034,p-value=0.009).Thesamecoefficientisalsosignificantlynegative(β=–0.037,p-value=0.005)inModel5whenweadditionallyincludeAccrualsQuality.Thus,stockpricesappeartoincorporateinformationaboutstructuralproblemsevenbeforefirmsreportinternalcontrolproblemsunderSOX.However,AccrualsQualityPreisnotsignificantlynegativeinanyregression.(b)InterimPeriod

Wealsotestwhetherstockpricesincorporateinformationaboutpotentialcontrolweaknesses,structuralproblems,andreportingqualityintheinterimperiod,whichistheperiodbetween2002andoneyearbeforefirmsfirstdisclosecontrolproblemsunderSOX.IncontrasttoTable6,Table7incorporatesannualobservationsformultipleyearsforeachfirmwhichincreasesthenumberofobservations.

InModel1,wefindthatthecoefficientonPr(ICW)Interimissignificantlynegative(β=–0.227,p-value=0.001).Thesamecoefficientisalsosignificantlynegative(β=–0.224,p-value=0.001)inModel4whenweadditionallyincludeAccrualsQuality.Thus,theresultsfromTables6and7suggestthatthenegativeinformationaboutafirm’slatentinternalcontrolproblemsisincorporatedinstockpricesduringtheinterimperiodandthepre-disclosureyear(2002).Inotherspecifications,however,wefindthatthecoefficientonSP-ScoreInterimisinsignificantandAccrualsQualityInterimismarginallysignificantinModel3.Theseresultsimplythattheinformationcontainedinthesevariablesisgenerallynotnegativelyreflectedinstockpricesduringtheinterimperiod.Overall,theresultsreportedinTable7,aswellasthosereportedinTable6,provideevidencethatstockpricesreflectinformationaboutthelikelihoodofinternalcontrolweaknessesandstructuralproblemsduringthepre-disclosureyear.Theresultsalsosuggestthatsomenewinformationaboutthelikelihoodofinternalcontrolweaknessesisgeneratedduringtheinterimperiod.(c)AnnouncementDate

Wenextexaminewhetherstockreturnsaroundthedisclosuredateofinternalcontrolproblemareassociatedwithneworpriorperiodinformationaboutcontrolweaknesses,i.e.,structuralproblemsandreportingqualityconcerns.

󰀃C

2013BlackwellPublishingLtd

󰀃C

Table6MarketPricingoverthePre-disclosureYear(2002)Model2Estimate−0.082−0.0340.651Yes0.1156540.000***0.2040.597Yes0.1086540.1800.000***0.2680.651Yes0.1276540.009***0.833−0.2480.524−0.228−0.5220.077*0.000***0.5530.000***p-valueEstimatep-valueEstimatep-valueModel3Model4Model5Estimate−0.110p-value0.7782013BlackwellPublishingLtd

Model1VariableEstimatep-value−0.188−0.4950.6250.000***INFORMATIVENESSOFINTERNALCONTROLDISCLOSURES

InterceptPr(ICW)PreSP-ScorePreAccrualsQualityPre󰀅EarningsPreIndustryindicatorsAdjustedR2N0.657Yes0.1246540.000***−0.0370.2680.647Yes0.1186540.005***0.080*0.000***Notes:ThistablereportsparameterestimatesfromthefollowingOLSregressionforthepre-disclosureyear(2002):CARPre=β0+β1Pr(ICW)Pre+β2SP-ScorePre+β3AccrualsQualityPre+β4󰀅EarningsPre+ν.Eachcontinuousvariableiswinsorizedatthetopandbottom5%.Thevariablesaremeasuredasfollows:CARPre=Compoundedbuy-and-holdreturnlessthereturnonthevalue-weightedmarketportfolioovera12-monthperiodendingthreemonthsafterthe2002fiscalyear-end.Pr(ICW)Pre=Thepredictedprobabilityofreportinginternalcontrolweaknessesbasedontheestimatedcoefficientsfromthelogisticregression(Model2inTable4)andthevaluesoftheindependentvariablesfrom2002.SP-ScorePre=Thesumofbinarysignalsconstructedforeachofthefirst14structuralproblemvariablesinTable2forthepre-disclosureyear.SeeSection3formoredetails.AccrualsQualityPre=TheabsolutevalueoftheresidualsfromtheDechowandDichev(2002)accrualsqualitymeasureasadjustedbyFrancisetal.(2005)for2002.SeeSection3formoredetails.󰀅EarningsPre=Thechangeinannualearningsbeforeextraordinaryitems,scaledbymarketvalueofbeginningequity,for2002.***,**,*denotessignificanceat<0.01,<0.05and<0.10levels,respectively,fortwo-tailedtests.341

342

Table7MarketPricingOvertheInterimPeriodModel2Estimate−0.160−0.004*Model1p-value0.6590.6020.000YesYes0.2281,566YesYes0.2291,566***Model3Estimate−0.1340.711−0.093−0.224−0.1471.3670.7960.001***0.1130.000***YesYes0.2351,566p-valueEstimatep-valueModel4Model5Estimate−0.125p-value0.731VariableEstimatep-value−0.132−0.227***0.7150.001***1.356−0.1561.3550.0930.000***InterceptPr(ICW)InterimSP-ScoreInterimAccrualsQualityInterim󰀅EarningsInterimIndustryindicatorsYearindicatorsAdjustedR2N1.3640.000YesYes0.2341,566−0.003−0.1531.358GHOSHANDLEE

YesYes0.2291,5660.7360.1020.000***󰀃C

Notes:ThistablereportsparameterestimatesfromthefollowingOLSregressionfortheinterimperiod(i.e.,theperiodbetween2002andoneyearbeforetheannouncementofinternalcontrolweaknessesunderSOX):CARInterim=β0+β1Pr(ICW)Interim+β2SP-ScoreInterim+β3AccrualsQualityInterim+β4󰀅EarningsInterim+ν.Eachcontinuousvariableiswinsorizedatthetopandbottom5%.Thevariablesaremeasuredasfollows:CARInterim=Compoundedbuy-and-holdreturnlessthereturnonthevalue-weightedmarketportfolioovera12-monthperiodendingthreemonthsaftereachfiscalyear-endduringtheinterimperiod.Pr(ICW)Interim=Thepredictedprobabilityofreportinginternalcontrolweaknessesbasedontheestimatedcoefficientsfromthelogisticregression(Model2inTable4)andthevaluesoftheindependentvariablesfromeachyearduringtheinterimperiod.SP-ScoreInterim=Thesumofbinarysignalsconstructedforeachofthefirst14structuralproblemvariablesinTable2foreachyearduringtheinterimperiod.SeeSection3formoredetails.AccrualsQualityInterim=TheabsolutevalueoftheresidualsfromtheDechowandDichev(2002)accrualsqualitymeasureasadjustedbyFrancisetal.(2005)foreachyearduringtheinterimperiod.SeeSection3formoredetails.󰀅EarningsInterim=Thechangeinannualearningsbeforeextraordinaryitems,scaledbymarketvalueofbeginningequity,foreachyearduringtheinterimperiod.***,**,*denotessignificanceat<0.01,<0.05and<0.10levels,respectively,fortwo-tailedtests.2013BlackwellPublishingLtd

INFORMATIVENESSOFINTERNALCONTROLDISCLOSURES

343

Table8

MarketPricingaroundDisclosuresofInternalControlProblems

Model1

Variable

InterceptSP-ScorePre

󰀅SP-ScoreInterimSP-Score

AccrualsQualityPre

󰀅AccrualsQualityInterimAccrualsQualityEarningsSurpriseIndustryindicatorsAdjustedR2N

Estimate0.0140.002−0.001−0.002

p-value0.7050.2610.7090.291

Model2Estimate−0.001

p-value0.969

Model3Estimate0.0050.0020.000−0.0020.0040.0330.0280.051

p-value0.8850.2510.8030.1740.8340.1100.1810.254

0.059

Yes−0.006654

0.194

0.0070.0320.0270.059

Yes−0.003654

0.7290.1110.1880.184

Yes0.000654

Notes:

ThistablereportsparameterestimatesfromthefollowingOLSregressionforthefullsamplearoundthedateofinternalcontroldisclosure:CAR=β0+β1SP-ScorePre+β2󰀅SP-ScoreInterim+β3SP-Score+β4AccrualsQualityPre+β5󰀅AccrualsQualityInterim+β6AccrualsQuality+β7EarningsSurprise+ν.Eachcontin-uousvariableiswinsorizedatthetopandbottom5%.Thevariablesaremeasuredasfollows:

CAR=Cumulativeabnormalreturnsovertheeventwindow[days–1,1],whereday0isthedateofinternalcontroldisclosureandabnormalreturnsaredefinedasthedifferencebetweenrawreturnsandvalue-weightedmarketreturns.

SP-Score=Thesumofbinarysignalsconstructedforeachofthefirst14structuralproblemvariablesinTable2.SeeSection3formoredetails.SP-ScorePre=SP-Scorefor2002.

󰀅SP-ScoreInterim=DifferenceinSP-Scorebetweentheyearbeforetheinternalcontroldisclosureand2002.AccrualsQuality=TheabsolutevalueoftheresidualsfromtheDechowandDichev(2002)accrualsqualitymeasure,asadjustedbyFrancisetal.(2005).SeeSection3formoredetails.AccrualsQualityPre=AccrualsQualityfor2002.

󰀅AccrualsQualityInterim=DifferenceinAccrualsQualitybetweentheyearbeforetheinternalcontroldisclosureand2002.

EarningsSurprise=Forthemostrecentquarterbeforethefirstdateofinternalcontroldisclosure,thedifferencebetweenincomebeforeextraordinaryitemsforthecurrentquarterandincomebeforeextraordinaryitemsannouncedfourquartersprior,scaledbytotalmarketvalueofequityfourquartersprior.

***,**,*denotessignificanceat<0.01,<0.05and<0.10levels,respectively,fortwo-tailedtests.

InTable8,weconsiderthreevariablesforeachofourtwoconstructs(i.e.,structuralproblemsandfinancialreportingquality)relatedtothepre-disclosureyear,theinterimperiod,andthedisclosureyear.WefindthatthecoefficientsonSP-ScorePreand󰀅SP-ScoreInterimarebothinsignificantinModels1and3.Thisfinding,combinedwiththeresultsfromTables6and7,suggeststhatmuchoftheinformationaboutstructuralproblemsisincorporatedinthestockpricepriortotheannouncementofinternalcontrolproblemsand,therefore,stockreturnsaroundthedisclosuredatearenotassociatedwiththepriorperiodinformation.Inaddition,SP-Scoreisnotsignificantinanyregressionspecification,implyingthatthedisclosureofinternalcontrolproblemsprovideslittlenewinformationabouttheextentofafirm’sstructuralproblems.

InModels2and3,weuseourproxiesforfinancialreportingqualityfromthethreeperiodsasthemainexplanatoryvariables.Wefindthatthecoefficientsonallthreevariablesareinsignificant.Theseresultssuggestthatreportingqualityconcerns

󰀃C

2013BlackwellPublishingLtd

344

GHOSHANDLEE

inthedisclosureyearaswellasthoseintheprioryearsconveylittleinformationtothemarketwhenafirmannouncesinternalcontrolweaknessesunderSOX.TheinsignificantcoefficientonAccrualsQualityimpliesthatnonewadverseinformationaboutfinancialreportingqualityisconveyedtothemarketaroundtheannouncementdate.19

Overall,theresultsfromTables6,7and8revealthatmostofthevalue-relevantinformationaboutlatentinternalcontrolproblemsisincorporatedinthestockpriceoverthepre-disclosureyearandtheinterimperiod.Thispartiallyexplainstheweakstockmarketreactionstothemandateddisclosuresoninternalcontrol.

(v)Out-of-SampleTest

Wecomplementourmarketpricingtestbycomparingthree-daycumulativeabnormalreturnsaroundtheannouncementdatesofinternalcontrolweaknessesbetweenfirmswithhighPr(ICW)andthosewithlowPr(ICW),wherePr(ICW)iscomputedusingobservationsin2002asinPanelBofTable4.Iftheinformationaboutthelikelihoodofinternalcontrolweaknesseswaspricedduringthepre-disclosureperiod,firmswithlowPr(ICW)arelikelytobeassociatedwithmorenegativereactionaroundtheannouncementofinternalcontrolweaknessdisclosuresunderSOXrelativetofirmswithhighPr(ICW).

Consistentwiththisprediction,inuntabulatedresults,wefindthatthemeancumulativeabnormalreturnsfor168firmsinthelowestquartileofPr(ICW)is–0.73%,whichisnegativeandsignificantatthe10%level.Incontrast,themeancumulativeabnormalreturnsfor168firmsinthehighestquartileofPr(ICW)is–0.15%,whichisnotsignificantlydifferentfromzero.ThisresultimpliesthatthefirmswithlowPr(ICW)experiencenegativemarketreactionswhentheinternalcontrolproblemisannounced,whereasthefirmswithhighPr(ICW)donot.Theout-of-sampletestsstrengthenourargumentthatmuchoftheinformationassociatedwithinternalcontrolproblemsispricedduringthepre-disclosureperiodandlimitednewinformationisrevealedwhenfirmseventuallydisclosetheirproblems.

(vi)SensitivityAnalysis

Mostfirmsdonotannounceinternalcontrolproblemsviaaseparatepressreleaseoraseparate8-Kfilings.Instead,theyincludesuchinformationontheirfinancialstatements.WeusethefinancialstatementfilingdatewiththeSEC(10-Kor10-Q)tomeasurestockmarketreactionforourmarketpricingtestsaroundthedisclosuredate.However,usingthefilingdatetomeasurestockmarketreactiontointernalcontrolproblemsisnoisyasfirmsdiscloseothervaluerelevantinformationintheirfinancialstatements.Therefore,asasensitiveanalysis,werepeatouranalysesforasub-sampleoffirmswithuniquelyidentifiedinternalcontrolproblemannouncementdates.

WesearchFactivaPressReleaseServiceforpressreleases(articles)containingthewords“internalcontrol”or“materialweakness”within12monthsbeforethedisclosuredateidentifiedbyAuditAnalyticswhichisinvariablythefinancialstatementfilingdate.Wefind85firmsthatseparatelyannounceinternalcontrolproblemsviaapress

19WeuseAccrualsQualityastheproxyforfinancialreportingqualityinthemarketpricingtests.WealsousePerformance-MatchedAbnormalAccrualsandobtainqualitativelysimilarresults.

󰀃C

2013BlackwellPublishingLtd

INFORMATIVENESSOFINTERNALCONTROLDISCLOSURES

345

release.Becausesmallerfirmsmightreportinternalcontrolproblemsviaseparate8-Kfilingswithouthavingapressrelease,wealsosearchtheLexisNexisAcademicdatabasefor8-Kfilingsusing“internalcontrol”or“materialweakness”asthesearchcriteria.Thissearchresultsin30additionaldates.Thus,wewereabletoidentify115(=85+30)uniqueinternalcontrolweaknessannouncementdates.

Forourannouncementsampleof115firms,wefindthatthemeanofthree-daycumulativeabnormalreturnsis–1.98%(untabulated),whichissignificantlynegativeandcomparabletothemeanabnormalreturnsofaround–2%documentedinBeneishetal.(2008).Whenwereplicatethemarketpricingtestsforthissampleof115firms,inuntabulatedresults,wefindthatthecoefficientonPr(ICW)PreissignificantlynegativeinModels1and4fortheannouncementsamplewhenweexaminethemarketpricingofPr(ICW)Pre,SP-ScorePreandAccrualsQualityPreoverthepre-disclosureyear(Table6results).Thecoefficientsontheothertwovariablesarenotsignificant.Theseresultsgenerallyconfirmthat,forthefirmsintheannouncementsample,stockpricesincorporateinformationaboutthelikelihoodofreportinginternalcontrolproblemsevenbeforethosefirmsdosounderSOX.Wealsofindthatnoneofthethreevariablesaresignificantintheinterimperiodfortheannouncementsample(Table7results).Finally,inthemarketpricingtestsaroundthemandateddisclosuredate(Table8results),wedonotfindanyvariableforthepre-disclosureandinterimperiodstobesignificant.OnlythecoefficientonSP-Scoreisnegativeandsignificantatthe10%levelinModel3.Overall,theresultsfromusingthissub-samplewithuniquelyidentifieddisclosuredateareverysimilartothosefromusingthefullsample.

(vii)ADRFirms

PriortoSOX,foreignfirmscross-listedintheUSweregrantedexceptionsbecauseofdistinctlawandpracticelimitationsfrominternalcontroldisclosurerequirements(Huang,2009).However,foreignfirmscross-listedintheUSmustcomplywiththe§404requirementspertainingtointernalcontroldisclosures.EventhoughforeignfirmsmighthavedistinctlydifferentcultureandorganizationalstructuresfromtheUScounterparts,theymustnowbearthisaddedcost.WhilepriorstudiesdocumentthatforeignfirmsbenefitfromUScross-listing(e.g.,Langetal.,2003;HuijgenandLubberink,2005;andCharitouetal.,2007),recentstudiesalsodocumentthatforeignlistingsonUSstockexchangeshavedeclinedforthepost-SOXperiodandthatakeyexplanationforthisdeclineislinkedtoSOX-relatedcompliancecostsforforeigncross-listedfirms(Engeletal.,2007;andPiotroskiandSrinivasan,2008).

Becausecross-listedfirmsaredifferentiallyaffectedbythemandatedinternalcontroldisclosurerequirementunderSOX,wealsoexamineourresultsforasub-sampleofforeignfirmsthatarecross-listedintheUSandtradeonUSexchangesasAmericanDepositaryReceipts(ADRs).Ourobjectiveistoanalyzewhetherourresultsaredescriptiveforasub-sampleofforeigncross-listedfirms.TheresultscanpotentiallyprovidesomeevidenceonthecostsandbenefitsofthedisclosuresoninternalcontrolsunderSOXforADRfirmsespeciallybecausemanyADRfirmsarenotrequiredtoreportontheeffectivenessofinternalcontrolsintheirhomecountries.Further,ourresultsmightserveasaninputforthemeritsofmandateddisclosuresoninternalcontrolincountrieswherethesedisclosuresarenotmandated.

󰀃C

2013BlackwellPublishingLtd

346

GHOSHANDLEE

Whenwedecomposeoursampleof672firmswithinternalcontrolproblems,wefind18ADRfirms.Becausethesamplesizeissmall,ourresultsneedtobeinterpretedwithcaution.Inuntabulatedresults,similartothefullsampleresultsfromPanelBofTable4,wefindthatthemean(median)Pr(ICW)forADRfirmsis15.3%(10.8%)whilethatforcontrolfirmsis10.9%(8.8%)andthedifferenceinPr(ICW)betweenthetwosetsoffirmsisstatisticallysignificant.Thus,ourresultssuggestthatmarketparticipantsmayhavebeenabletodiscernthatsomeADRfirmswerepronetointernalcontrolproblemsevenintheabsenceofsuchdisclosures.

WealsofindthatthecoefficientonSP-ScorePreissignificantlynegativeatthe10%levelinModels2and5forADRfirmswhenweexaminethemarketpricingofPr(ICW)Pre,SP-ScorePreandAccrualsQualityPreoverthepre-disclosureyearforADRfirms(Table6results).Thecoefficientsontheothertwovariablesarenotsignificant.Thus,forADRfirms,stockpricesincorporateinformationaboutstructuralproblemsevenbeforethosefirmsreportinternalcontrolproblemsunderSOX.NoneofthevariablesaresignificantintheinterimperiodforADRfirms(Table7results).Finally,forthemarketpricingtestsaroundthemandateddisclosuredate(Table8results),weagaindonotfindanyvariabletobesignificanteitherforthepre-disclosure,interim,ordisclosureperiods.Also,themean(median)CARaroundtheannouncementdateis0.35%(0.36%)andstatisticallyinsignificantwhichindicatesthatnonewinformationisrevealedtothemarketaroundtheinternalcontroldisclosures.Ingeneralourresultsagainsuggestthatlimitednewinformationisgeneratedaroundtheannouncementdateforforeigncross-listedfirms.

6.CONCLUSIONS

ContrarytotheclaimsthatinternalcontroldisclosuresunderSOXprovidevaluablenewinformation,priorstudiesprovideweakevidencesupportingthisconjecture.Wehypothesizethatsomeoftheinnatevalue-relevantfactorsincludingstructuralproblems,potentialinternalcontrolproblems,andpoorfinancialreportingqualitythataretypicallyassociatedwithfirms’weakinternalcontrolspre-datethemandateddisclosureperiod.TotheextentthattheinformationcontainedintheinternalcontroldisclosuresunderSOXisincorporatedinthestockpricepriortothedisclosures,marketparticipantsarelesssurprisedwhenfirmssubsequentlyreportinternalcontrolweaknessesunderSOXand,therefore,theannouncementperiodstockreturnsismuted.

Weemployafour-stepproceduretoprovideinsightsintotherelatedhypothesesbyexaminingwhether(1)thefirmattributesthatareassociatedwithinternalcontrolweaknessespre-datedisclosuresunderSOX(firmattributestest),(2)firmsreportinginternalcontrolproblemsunderSOXwerealsopronetosuchproblemsforthepre-disclosureperiod(controlproblemtest),(3)firmsreportinginternalcontrolweaknessesunderSOXhadlowerqualityoffinancialreportingforthepre-disclosureyears(reportingqualitytest),and(4)equitypricesforthepre-disclosureandinterimperiodsareconsistentwiththeexpectationthatinvestorsincorporatedinformationaboutstructuralproblems,potentialinternalcontrolweaknesses,andlowfinancialreportingqualityintosecuritypricespriortodisclosures(marketpricingtest).

Basedonasampleof672firmsthatcompliedwiththereportingrequirementsunder§302or§404,wefindevidenceconsistentwithourpredictions.Specifically,we

󰀃C

2013BlackwellPublishingLtd

INFORMATIVENESSOFINTERNALCONTROLDISCLOSURES

347

findthat:(1)thefirmattributesthatdistinguishbetweenfirmswithinternalcontrolweaknessesunderSOXandthosewithoutsuchproblemsalsodiscriminatethetwosetsoffirmsforthepre-disclosureyear;(2)firmsreportingweaknessesininternalcontrolsoverfinancialreportingunderSOXalsohadhigherprobabilitiesofhavinginternalcontrolproblemsforthepre-disclosureyearrelativetothosewithoutsuchproblems;(3)thepre-disclosurefinancialreportingqualityislowforfirmsreportinginternalcontrolweaknessesunderSOX,comparedtothosewithoutsuchproblems;and(4)muchofthevalue-relevantfactorsassociatedwithinternalcontrolproblemsarepricedoverthepre-disclosureandinterimperiods;onlysomenewinformationabouttheextentofstructuralproblemsisgeneratedaroundtheannouncementofinternalcontrolweaknesses.

Becausemuchoftheinformationcontainedintheannouncementofinternalcontrolweaknessesisincorporatedinthestockpriceduringtheperiodsleadinguptotheannouncementdate,andconsideringthattheseannouncementscontainlittlenewinformation,itisnotsurprisingthatthemarketreactiontothemandateddisclosuresoninternalcontrolisrelativelyweak.

Wealsoextendourteststoasub-sampleoffirmscross-listedintheUSandfindsimilarresults.Onepotentialinferencefromourresultsisthatthebenefitsofmandatingdisclosureoninternalcontrolsislimitedbecausemarketparticipantsareabletoassessmuchoftheinformationembeddedinthemandateddisclosuresevenwhensuchdisclosuresarenotmandated.Becausethecostsofreportingontheeffectivenessoninternalcontrolsbymanagementandauditorsmaybenon-trivial,thenetbenefitsfromsuchdisclosuresremainunclear.Ourresearchhasthepotentialtoprovideinsightsintothedebateonwhethercountriesshouldregulatethedisclosuresoninternalcontroloverfinancialreporting.

REFERENCES

Altman,E.(1968),‘FinancialRatios,DiscriminantAnalysisandthePredictionofCorporate

Bankruptcy’,JournalofFinance,Vol.23,pp.589–609.

Ashbaugh-Skaife,H.,D.CollinsandW.Kinney(2007),‘TheDiscoveryandConsequencesof

InternalControlDeficienciesPriorToSOX-MandatedAudits’,JournalofAccountingandEconomics,Vol.44,pp.166–92.

———andR.Lafond(2009),‘TheEffectofSOXInternalControlDeficienciesonFirmRisk

andCostofEquity’,JournalofAccountingResearch,Vol.47,pp.1–43.

Beneish,M.D.,M.B.BillingsandL.D.Hodder(2008),‘InternalControlWeaknessesand

InformationUncertainty’,TheAccountingReview,Vol.83,pp.665–703.

Charitou,A.,C.LoucaandS.Panayides(2007),‘Cross-Listing,BondingHypothesisand

CorporateGovernance’,JournalofBusinessFinance&Accounting,Vol.34,pp.1281–306.Chava,S.andA.Purnanandam(2010),‘CEOsversusCFOs:IncentivesandCorporatePolicies’,

JournalofFinancialEconomics,Vol.97,pp.263–78.

CommissiononAuditors’Responsibilities(TheCohenCommission)(1978),‘TheCommission

onAuditors’Responsibilities:Report,ConclusionsandRecommendations’,NewYork,NY,AICPA.

CommitteeofSponsoringOrganizationsoftheTreadwayCommission(COSO)(1992),‘Inter-nalControl-IntegratedFramework’,NewYork,NY,AICPA.

Costello,A.M.andR.Wittenberg-Moerman(2011),‘TheImpactofFinancialReporting

QualityonDebtContracting:EvidencefromInternalControlWeaknessReports’,JournalofAccountingResearch,Vol.49,pp.97–136.

Dechow,P.andI.Dichev(2002),‘TheQualityofAccrualsandEarnings’,TheAccountingReview,

Vol.77,pp.35–59.

󰀃C

2013BlackwellPublishingLtd

348

GHOSHANDLEE

———,R.SloanandA.Sweeney(1995),‘DetectingEarningsManagement’,TheAccounting

Review,Vol.70,pp.193–225.

DeFranco,G.,Y.GuanandH.Lu(2005),‘TheWealthChangeandRedistributionEffects

ofSarbanes–OxleyInternalControlDisclosures’,WorkingPaper(Toronto:UniversityofToronto).

Deloitte&Touche,Ernst&Young,KPMGandPricewaterhouseCoopers(2004),‘Internal

ControloverFinancialReporting:AnInvestorResource’,anonlineresourceavailableat:www.kpmg.no/download/202548/115886/investorresource.pdf.

Doyle,J.,W.GeandS.McVay(2007a),‘DeterminantsofWeaknessesinInternalControlover

FinancialReporting’,JournalofAccountingandEconomics,Vol.44,pp.193–223.

———,———and———(2007b),‘AccrualsQualityandInternalControloverFinancial

Reporting’,TheAccountingReview,Vol.82,pp.1141–70.

Engel,E.,R.M.HayesandX.Wang(2007),‘TheSarbanes–OxleyActandFirms’GoingPrivate

Decisions’,JournalofAccountingandEconomics,Vol.44,pp.116–23.

Ernst&Young(2005),SubmissionlettertotheSEC.FileNo.4–497,Implementationof

Sarbanes–OxleyInternalControlProvisions(Washington,DC:SEC).Availableonlineat:www.sec.gov/spotlight/soxcomp/soxcomp-turley.pdf.

Francis,J.,R.LaFond,P.OhlssonandK.Schipper(2005),‘TheMarketPricingofAccruals

Quality’,JournalofAccountingandEconomics,Vol.39,pp.295–327.

Ghosh,A.,A.MarraandD.Moon(2010),‘CorporateBoards,AuditCommittees,andEarnings

Management:Pre-andPost-SOXEvidence’,JournalofBusinessFinance&Accounting,Vol.37,pp.1145–76.

Goh,B.W.andD.Wee(2011),‘InternalControlsandConditionalConservatism’,TheAccounting

Review,Vol.86,pp.975–1005.

Gompers,P.,J.IshiiandA.Metrick(2003),‘CorporateGovernanceandEquityPrices’,Quarterly

JournalofEconomics,Vol.118,pp.107–55.

Hammersley,J.S.,L.A.MyersandC.Shakespeare(2008),‘MarketsReactionstotheDisclosure

ofInternalControlWeaknessesandtotheCharacteristicsofThoseWeaknessesunderSection404oftheSarbanes–OxleyActof2002’,ReviewofAccountingStudies,Vol.13,pp.141–65.

Hermanson,D.R.andZ.Ye(2009),‘WhyDoSomeAcceleratedFilerswithSOXSection404

MaterialWeaknessesProvideEarlyWarningunderSection302?’Auditing:AJournalofPractice&Theory,Vol.28,pp.247–71.

Hermanson,H.M.(2000),‘AnAnalysisoftheDemandforReportingonInternalControl’,

AccountingHorizons,Vol.14,pp.325–41.

Hoitash,U.,R.HoitashandJ.C.Bedard(2009),‘CorporateGovernanceandInternalControl

overFinancialReporting:AComparisonofRegulatoryRegimes’,TheAccountingReview,Vol.84,pp.839–67.

Huang,H.(2009),‘Sarbanes–OxleySection404Compliance’,ManagerialAuditingJournal,

Vol.24,pp.584–98.

Huijgen,C.andM.Lubberink(2005),‘EarningsConservatism,LitigationandContracting:The

CaseofCross-ListedFirms’,JournalofBusinessFinance&Accounting,Vol.32,pp.1275–309.Johnstone,K.,C.LiandK.Hertz(2011),‘ChangesinCorporateGovernanceAssociatedwiththe

RevelationofInternalControlMaterialWeaknessesandTheirSubsequentRemediation’,ContemporaryAccountingResearch,Vol28,pp.331–83.

Jones,J.(1991),‘EarningsManagementduringImportReliefInvestigations’,JournalofAccount-ingResearch,Vol.29,pp.193–228.

Kinney,W.,M.MaherandD.Wright(1990),‘Assertions-basedStandardsforIntegratedInternal

Control’,AccountingHorizons,Vol.4,pp.325–41.

Kinney,W.andL.McDaniel(1989),‘CharacteristicsofFirmsCorrectingPreviouslyReported

QuarterlyEarnings’,JournalofAccountingandEconomics,Vol.11,pp.71–93.

KinneyJr.,W.R.andM.L.Shepardson(2011),‘DoControlEffectivenessDisclosures

RequireSOX404(b)InternalControlAudits?ANaturalExperimentwithSmallUSPublicCompanies’,JournalofAccountingResearch,Vol.49,pp.413–48.

Krishnan,J.(2005),‘AuditCommitteeQualityandInternalControl:AnEmpiricalAnalysis’,The

AccountingReview,Vol.80,pp.649–75.

󰀃C

2013BlackwellPublishingLtd

INFORMATIVENESSOFINTERNALCONTROLDISCLOSURES

349

Lang,M.,K.LinsandD.Miller(2003),‘ADRs,AnalystsandAccuracy:DoesCross-Listingin

theUSImproveaFirm’sInformationEnvironmentandIncreaseMarketValue?’JournalofAccountingResearch,Vol.41,pp.317–45.

Naiker,V.andD.S.Sharma(2009),‘FormerAuditPartnersontheAuditCommitteeand

InternalControlDeficiencies’,TheAccountingReview,Vol.84,pp.559–87.

NationalCommissiononFraudulentFinancialReporting(TheTreadwayCommission)(1987),

‘ReportoftheNationalCommissiononFraudulentFinancialReporting’,Washington,DC,NCFFR.

Ogneva,M.,K.R.,SubramanyamandK.Raghunandan(2007),‘InternalControlWeaknessand

CostofEquity:EvidencefromSOXSection404Disclosures’,TheAccountingReview,Vol.82,pp.1255–97.

Peasnell,K.V.,P.F.PopeandS.Young(2005),‘BoardMonitoringandEarningsManagement:

DoOutsideDirectorsInfluenceAbnormalAccruals?’JournalofBusinessFinance&Account-ing,Vol.32,pp.1311–46.

Piotroski,J.D.andS.Srinivasan(2008),‘RegulationandBonding:TheSarbanes–OxleyActand

FlowofInternationalListings’,JournalofAccountingResearch,Vol.46,pp.383–425.

PublicCompanyAccountingOversightBoard(PCAOB)(2004),‘AnAuditofInternalControl

OverFinancialReportingPerformedinConjunctionwithanAuditofFinancialState-ments’,AuditingStandardNo.2,Washington,DC,PCAOB.

Schneider,A.,A.A.Gramling,D.R.HermansonandZ.Ye(2009),‘AReviewofAcademic

LiteratureonInternalControlReportingUnderSOX’,JournalofAccountingLiterature,Vol.28,pp.1–46.

SecuritiesandExchangeCommission(SEC)(2003),‘Finalrule:Management’sReportson

InternalControloverFinancialReportingandCertificationofDisclosureinExchangeActPeriodicReports’,anonlineresourceavailableat:www.sec.gov/rules/final/33--8238.htm.———(2005a),‘StaffStatementonManagement’sReportonInternalControloverFi-nancialReporting’,anonlineresourceavailableat:www.sec.gov/info/accountants/stafficreporting.htm.

———(2005b),‘RemarksbeforetheEconomicClubofMemphis.SpeechbytheSECcommis-sioner’,anonlineresourceavailableat:www.sec.gov/news/speech/spch021605psa.htm.

󰀃C

2013BlackwellPublishingLtd