[PDF] AUDITOR SIZE AND AUDIT QUALITY Linda Elizabeth DeANGELO

View - Download AUDITOR SIZE AND AUDIT QUALITY Linda Elizabeth DeANGELO

Previous PDF

Next PDF

Journalof Accounting and Economics 3(1981)183-199.North-HollandPublishingCompany

AUDITOR SIZE AND AUDIT QUALITY

LindaElizabethDeANGELO*

UniversityofPennsylvania,Philadelphia, PA 19104,USA

Received May1981,final version receivedJuly1981

Regulatorsand smallaudit

firms allegethatauditfirm size does not affectauditqualityand thereforeshouldbeirrelevantin the selection of an auditor.Contraryto this view, thecurrent paperarguesthataudit quality is notindependentof audit firm size,even whenauditorsinitially possessidenticaltechnologicalcapabilities.Inparticular,whenincumbentauditorsearn client specificquasi-rents,auditorswith agreaternumberof clients have 'more to lose' by failing to reporta discoveredbreachin aparticularclient's records. This collateral aspect increases the auditqualitysupplied by largerauditfirms. The implications for some recentrecommendations of theAICPASpecialCommitteeon Small andMediumSized Firms are developed.1. Introduction

Largeaudit

firms areincreasinglycriticized on the basis of size aloneboth byregulatorsand by smaller firms within theaccountingprofession. 1 The allegationunderlyingthese criticisms isthatprofessionalstandardsimparta homogeneityacross different sizedauditfirms suchthatauditqualityis independentofauditfirm size.

Forexample, employing theassumptionthat

auditqualityis relativelyhomogeneousacross audit firms,ArnettandDanos (1979)arguethatsize alone'shouldnotbe a primedeterminantoffuture success' (p. 8,emphasisadded).Furthermore,they arguethat'as long as it isunfairto arbitrarilydistinguishbetween the largest eight and allother

CPAfirms' (p.

56,emphasisadded).

Because of the allegeddiscriminatoryimpact on smallerauditfirms, eighteensmall-to-mediumsizedauditfirms filedsuit in 1978 to block the divisionof the AmericanInstituteof Certified PublicAccountantsinto two practicesections.Onesection now consists of audit firms whose clients are requiredtofilereportswith the Securities and ExchangeCommission,the *Theauthorwishes to express hergratitudetoY.Barzel, H. DeAngelo, W.L. Felix, W.

Kinney,

B.Klein, E.Noreen,E.M. Rice, R.Watts,J. Zimmerman, and the referee,M.e.Jensen. lSee theReportof theSubcommitteeon Reports, Accounting, andManagementof the CommitteeonGovernmentOperations(U.S. Senate, 1977) (Metcalf Report), Arnett andDanos (1979),andtheReportof the SpecialCommitteeon Small and Medium SizedFirms(AICPA,

1980) (DerieuxCommitteeReport).

184L.E. DeAngelo,Auditorsize and quality

otherof firms whose clients are not.AccordingtoPublicAccountingReport (February

1978),thegroundsfor suit werethat'thisseeminglyinnocuous

proposalispartof acalculateddesign ...tofurtherconcentratepower in the handsof the largestaccountingfirms'. Whilethesuit waslaterdismissed, the AICPA considered the size issue sufficientlyimportanttoappointa special committee(the DerieuxCommittee)tostudyit in moredepth.

The DerieuxCommitteeReportstresses the'concern

thatsmaller firms may be replaced simplybecausethey are less wellknown,eventhoughthe smaller firms may well beprovidingas high orhigherqualityservices' (p. 5). follows: Problem:Auditorsaresometimeschosenon the basis ofarbitrary factors such as size oftheirfirm. stressingthatthe selection of a

CPAfirmshouldbebasednot onSIze,

but on the ability toprovideservice. (p. 18,emphasisadded) Inotherwords, theDerieuxCommittee'spositionisthatauditorsizeshould beirrelevantin the selection of anauditor.Justificationfor thispositionis providedby theassumptionthatauditorsize does not affect thequalityof auditservices supplied. Contraryto this view, thecurrentpaperarguesthatsize alonealters auditors'incentives suchthat,ceterisparibus,largerauditfirms supply a higher level ofauditquality.Whenaudittechnologyischaracterizedby These quasi-rents, when subject to loss from discovery of a lowerquality auditthanpromised, serve ascollateralagainstsuchopportunisticbehavior. This impliesthat,ceterisparibus,thelargertheauditorasmeasuredby numberofclients.'thelessincentivetheauditorhas tobehave opportunisticallyand the higher theperceivedqualityof theaudit." auditorsize, we firstprovideanoperationaldefinition ofauditquality (section 2). Whenqualityis costly toevaluate,self-interestedindividualswho

2This size measure isappropriatewhen client-specificquasi-rentsdonotvary acrossclientsof

a givenauditor.Whenquasi-rentsvary across clients, therelationshipbetween thequasi-rents specific to a given client and theauditor'stotalquasi-rentsbecomesimportant.Besidesaltering themagnitudeof thecollateralbondaspect of thequasi-rents,size also affects the incentives of thepartnerswithin theauditfirm. We hold theseincentivesconstantinordertoconcentrateon thecollateralaspect of client-specificquasi-rents.Thus,thecurrentanalysisisolates one previouslyunrecognized benefit ofauditorsize, sizeauditfirm. These issues are discussedfurtherinsections3 and4 of thepaper.

3Watts andZimmerman(1981) develop analternativetheory

thatpredictsthatauditorsize is asurrogatefor audit quality.Theirargumentis thatlargeauditfirmssupplyhigherquality audits because they possess acomparativeadvantageinmonitoringindividualauditorbehavior.

L.E. DeAngelo,Auditorsize and quality

185
potentiallybenefit from the exchange of quality-differentiatedauditshave incentivesto devisesurrogatesforquality.When largerauditfirms have 'moreto lose' fromsupplyingalower-than-promisedlevel ofauditquality, consumersproperlyuse size as aqualitysurrogate"(section 3). Thetheory supportsconventionalwisdom thatperceivedauditorindependenceis inverselyrelatedto thepercentageoftotalaudit feesdependentonretaining anyoneclient.Largeauditfirms are ameansof lowering thispercentage,as arealternativequalityenforcingarrangements(section 4).Implicationsfor somerecommendationsof theDerieuxCommitteewhich would subsidize smallerauditfirms at the expense of largerauditfirms and clients are developed(section 5).Conclusionsanda brief summary areprovidedin section6.

2.The qualityofauditservices

Auditservices aredemandedasmonitoringdevices because of the potentialconflicts ofinterestbetweenownersand managers as well as those amongdifferent classes of securityholders[seeWatts(1977),Wattsand

Zimmerman

(1981),andBenston(1980)forelaboration].In at least some cases,theprovisionofauditedfinancialstatementsis theleast-cost contractualresponsetoowner-managerandintra-ownerconflicts of interest, i.e., agency costs. Agency costs vary acrosspotentialclient firms andperhaps over time for a given client.

Forexample, it is well knownthatclient firms

goingpublicoften switch to BigEightauditors[CarpenterandStrawser (1971)].5Differentialagency costsacrossfirms and over time for a given firm implyaheterogeneousdemandforauditservices, i.e., differing 'levels' of auditingaredemanded. Howare different 'levels' ofauditingexchanged?In order to address this question,we define theproductionofauditsin terms of inputs andoutputs suppliedby theauditoralone,andignore client inputs." With this simplification,auditoutputcan becharacterizedasindependentverification ofmanagement-preparedfinancialdata,and consists of a statedopinion (e.g., "Ourargumentdiffers fromotherargumentsforauditorsize"found in theliterature,such as economiesof scale in theproductionofauditservices[ArnettandDanos (1979),Benston (1980), DopuchandSimunic (1979)],greaterpartnerpersonalwealth (Benston), and benefits to specializationcombinedwith the risk ofindustry-specificbusiness cycle fluctuations(Arnettand

Danos).

Thecurrentargumentisthatclient-specificquasi-rentscreate anadvantageto large -orcourse,otherreasonsbesides ademandfor'more'auditingmay alsomotivatethese switches. Forexample,BigEightfirms may possessmoreexpertise inpreparingSECdocuments,auditing largeclients, and/orgreaterindustry-specific knowledge.

6Definingauditinputsandoutputin thismannersimplifies the analysis at the cost of

bypassingsomeinterestingauditor-clientinteractionscaused by thejointnessof theproduction process.See Alchian andDemsetz(1972) for an analysis of the incentives faced byinputowners when theproductionprocessisjoint. Fora model which explicitly recognizesauditor-client interactions,seeDemskiand Swieringa (1974). 186

L.B. DeAngelo,Auditorsize andquality

anunqualifiedopinion) with anassociatedqualitydimension.The type of opinionconstant,changes in the level ofauditingareequivalenttochanges inauditquality. Thequalityof audit services is defined to be themarket-assessedjoint probabilitythata givenauditorwill both(a)discoverabreachin the client's accountingsystem, and (b)reportthebreach.

Theprobabilitythata given

auditorwill discover abreachdependson theauditor'stechnological capabilities,the auditproceduresemployedon a givenaudit,theextentof sampling,etc. Theconditionalprobabilityofreportingadiscoveredbreachis ameasureof anauditor'sindependencefrom a givenclient.Thisdefinitionof auditorindependenceis used inDeAngelo(1981) andWattsand

Zimmerman

(1981),who arguethatthe exantevalue of anauditdependson theauditor'sincentives to disclose selectively expost. Consumersincur costs ofevaluatingauditquality,i.e., of assessing the jointprobabilitythata givenauditorwillbothdiscoverandreportabreach on a given client's audit.

7First,theactualproceduresemployedon a given

auditengagementare generally notdirectlyobservedbyconsumers."Second, consumershave littleinformationabouttheincentivesinducedby the form of a givenauditor-elientcontractwhich affect theprobabilityofreportinga discoveredbreach. Forthese reasons,auditqualityevaluationcostsare likely to be significant." Whenauditquality is costly toevaluate,self-interestedindividualshave incentivesto devisealternativearrangementswhichenablequality differentiatedauditsto be exchanged.Furthermore,competitiveforcesdictate thatthearrangementchosen will be the one whichminimizesthetotalcosts of exchange (including the costs ofdifferentiatingquality).Thiscogent observationwas firstmadein amoregeneralsettingby Barzel (1977).One potentialresponseis forconsumerstodevelopsurrogatesforauditquality, i.e., to rely on someother(less costly toobserve)variablewhich is (imperfectly)correlatedwith quality. Theargumentof thecurrentpaperis thatauditorsize serves as asurrogateforauditquality.!?

7Consumers of audit services includecurrentandpotentialowners(bothshareholdersand

bondholders),managers,consumersof the firm'sproducts,employees,governmentagencies, etc.

8Certification and auditstandardsaidconsumersin assessing a lowerboundon the

probabilitya givenauditordiscovers a breach. Thesestandardsenforce aminimumlevel of auditqualityby (typically)constrainingauditorinputs.Standardizationofauditorinputsserves otherfunctions besides assuringoutputquality.First, itprovidesa defense for theauditorin litigation.Second,standardizationalso economizes oncontractingcosts, much in the same way as do'boilerplates' instandardizedcontracts.Partiesdesiring toexchangehigherthanminimum qualityaudits must rely onprivatecontracts.

9Itcannotbe the case, however, as severalauthorshaveclaimed[e.g.,Baiman(1979),Dopuch

andSimunic(1979),Kaplan(1978),Magee(1979),and Ng(1978)]thatauditqualityis unobservable,i.e., infinitely costly toevaluate.In this case, noauditswould beexchanged,i.e., themarketforauditservices would 'collapse'as inAkerlof(1970). lOSeenext page.

L.E.DeAngelo,Auditorsizeandquality187

Anotherpotentialresponsetoconsumerqualityevaluationcosts is for auditors tospecializein auniformlevel ofauditquality,bothacross clients and over time. Ifauditorssubstantiallyvary the level ofauditqualitysupplied fromperiodtoperiod,consumerswould have tore-evaluatequalityover time. Ifauditorssubstantiallyvary the level ofqualityfrom client to client, eachauditengagementwouldrequireseparateevaluationby consumers. Becausequalityevaluationis costly,consumerswillcompensateauditors whoenablethem t?avoidthese costs bymaintaininga relatively uniform qualitylevel.Auditorshaveincentivesto specialize in a uniformqualitylevel becausethey cancapturehigherfees bydoingso. However, when differential agencycostsacrossclients imply aheterogeneousdemandforauditquality, different auditorswill specialize in different (albeit uniform) quality levels. ll Whenauditorsspecialize in a givenqualitylevel, clients wishing tochange the level ofauditqualitypurchasedwill find itnecessaryto changeauditors. Adistinctionshouldbemadebetween differing levels of auditquality voluntarilyexchanged andopportunistic(unanticipated)changes inaudit quality.Adifferentlevel ofauditqualityissuppliedwhen themarket assessed jointprobabilitythatanauditorwillbothdiscover andreporta breachin aclient'srecordschanges. Significant costs of qualityevaluation provideauditorswith theopportunitytopromisea given level ofaudit qualityex anteandtoopportunisticallylower it ex post by, e.g.,reporting fewerdiscoveredbreaches thanpromised.Thisopportunityoccurs because it is costly todiscoverauditor'cheating'(theprobabilityof beingcaughtis less thanone).

Wearguebelow

thatcertainaspects ofaudittechnologyprovideauditors particularclient.However,disincentivesto'cheat'are alsoprovidedbyaudit technology, andthesedisincentivesincrease asauditorsize increases. Because largerauditorshavereducedincentives to lower auditquality opportunistically,consumersrationallyuseauditorsize as asurrogatefor auditquality.

3. Therelationshipbetweenauditorsize and audit quality

Whenaudittechnologyischaracterizedby significant client-specificstart up costs,incumbentauditorspossess costadvantagesoverpotential lOOfcourse,inorderforauditorsize to serve as asurrogateforauditquality, it must be the casethat,onaverage,larger auditfirms supply higherqualityaudits. Therelationshipbetween auditorsize andauditqualityisdevelopedin section 3. It is alsoimportantto notethatauditor sizecannotbe a would not beobservedin themarketforauditservices. Theseotherarrangementsare discussed in section 4.

11Supply sidecompetitionensuresthat,whileauditorsspecialize in a uniform level of audit

quality,auditfeesadjustso thatmarginalauditorsare indifferent to the quality 'class' to whichthey belong. 188

L.E.DeAngelo,Auditorsizeandquality

competitorsin future audits of a given client. Even when theinitialmarket forauditservices is perfectlycompetitiveinsofaras allauditorspossess identicaltechnologicalcapabilities, theseadvantagestoincumbencyimply theabsenceof perfectsubstituteauditorsinfutureperiods.

Theabsenceof

perfectsubstitutesenablesincumbentauditorsto set futureauditfeesabove theavoidablecosts ofproducingaudits, i.e.,incumbentauditorsearnclient specificquasi-rents.HTransactionscosts ofchangingauditorsalsoenable incumbents toraise futurefeeswithoutmakingitprofitableforclientsto switch. In the presence ofstart-upandtransactionscosts, therelationship between clientsandincumbentauditorsis abilateralmonopoly[see

DeAngelo

(1981)]. In abilateralmonopoly,bothpartieshaveincentivestocontinuean establishedrelationshipbecause of the absence of costlesslyavailableperfect substituteauditors(clients). Inotherwords,terminationof therelationship imposes costs onbothparties.

Ifterminated,incumbentauditorswill lose the

wealthequivalentof the client-specificquasi-rentstream.Clientswill be forced tobeartransactionscosts of switching and theduplicationofstart-up costsassociatedwithtraininga new auditor.PThetheorythereforepredicts an inverserelationshipbetween themagnitudeofstart-up/transactionscosts (and hence client-specificquasi-rents)andauditorturnover.!"Consequently, observationof the rate at which client firmschangeauditorsprovides indirectevidence on themagnitudeof thesestart-up/transactionscosts. Extantempiricalevidence isconsistentwith the existence of significant start-up/transactionscosts in the exchange ofauditservices. Inparticular,the uniformfinding ofextantstudies[BurtonandRoberts (1967),Bedingfield and Loeb

Coeand

Palmon(1979)]isthattherateat which client firmschange auditorsis low.

Forexample,BurtonandRobertsfound achangerateof

approximatelyonepercentperannumfor a sample of

Fortune500 firms for

the timeperiod

1955-1963,while Coe andPalmonfound achangerateof

approximately2percentperannumfor arandomsampleof firms listed on the COMPUSTATtape from1952-1975.Forsmallerfirms, theauditor

12A client-specificquasi-rentis the excess of a givenperiod'srevenuesovertheavoidablecosts

incurredin thatperiod,including theopportunitycost ofauditingthenext-bestalternative client.Whethera givenquasi-rentstreamconstitutesamonopolyrentdependsonwhetherthe netpresentvalue of thestreamis positive.

Fora moreextensivediscussionofthedistinction

betweenquasi-rents andmonopolyrents, see Klein,CrawfordandAlchian(1978,p.299).

13This logicapparentlyunderlies thepositionof theCommissiononAuditors'Responsibilities

(CohenCommission) thatmandatoryauditorrotationwould'considerablyincreasethe cost of auditsbecauseof thefrequentduplicationof thestart-up andlearningtimenecessaryto gain familiaritywith thecompany.. .'[CohenReport(1978,pp.108-109)].

14Similar logic can be found in Becker(1975)for the case ofemployeetrainingthatis specific

to thetrainingemployer.Thepredictionin this case isthat,ceterisparibus,firmscharacterized bylarger(smaller)amountsof specifictrainingexperience lower (higher)employeeturnover.

L.E.DeAngelo,Auditorsize andquality189

changerateisgenerallyhigher,butdoes not exceed fivepercentperannum.

Thisevidence isconsistentwith theassumption

thatclient-specific start material. toincumbentauditors hasat least twoeffects.First,auditorsbiddingfor the initialauditengagement,i.e., for thepropertyrightsto become the incumbentandcapturethequasi-rents,will 'low ball' (setaudit feesbelow totalcosts on theinitialaudit)toobtainthe client[DeAngelo(1981)]. Second,becausethesequasi-rentsare client-specific(i.e.,they have no alternativeuse tothis auditorand are notmarketabletootherauditors), incumbentauditorshavesomeincentiveto lowerqualityopportunisticallyin ordertoretaintheclientin future periods.l" Thisincentiveoccursbecauseclients can impose real costs onauditorsby termination(loss of thewealthequivalentof the client-specificquasi-rent stream).Thereforeclients canpotentiallyextractaccountingconcessions from incumbentauditorsby a crediblethreatoftermination.Inparticular, incumbentauditorshavea lessened incentive toreporta discovered breach in the client'srecords,i.e.,incumbentauditorsarenotperfectlyindependent fromclients.l"

Rationalconsumersrecognize

thatincumbentauditorsare not perfectly independentfromclients.Consequently,they lower the price they are willing to pay forsecuritiesof firmsthatretainincumbents.Inotherwords, the lowerexpected level ofindependenceis reflected inreducedclient firm value. incumbentauditors.Therefore,they have incentives to reduce this wealth impactbychoosingincumbentauditorsperceived by themarketas being moreindependent,i.e., ashavingfewer incentives to'cheat'inordertoretain thisparticularclient. Whenauditorsearnclient-specificquasi-rents,auditorswith agreater to'cheat'inorderto retain

15To the extentthatthesequasi-rentsarenotclient-specific, e.g., they aremarketabletoother

auditorsor have value inalternativeuses to thecurrentauditor,they do not provide an incentiveto'cheat'. Thequasi-rentswith which we areconcernedhere are those whose alternativeuse value is zero.

16By perfectindependence,we mean

discoveredbreachis one. Ofcourse,ifcontractingamongindividualswere costless, perfect auditorindependencecouldbeexchangedvia an exhaustively specified and perfectly enforced explicitcontract.Unfortunately,costlesscontractingalsoenablesowners and managers to eliminatethe conflicts ofinterestwhichunderliethemonitoringdemandfor (costly) auditing. Becauseauditingmodelswhichassumecostlesscontractingareinternallyinconsistentin this sense, theassumptionmaintainedthroughoutthispaperisthatexplicitcontractswhich guaranteeperfectindependenceareprohibitivelycostly tonegotiateand enforce.

190L.E. DeAngelo,Auditorsize andquality

independencewithrespectto aparticularclientbecausetheyprovidean incentiveto'cheat'inorderto retain theclientinfutureperiods. Onthe otherhand,thequasi-rentsspecific toothercurrentclientsof a givenauditor provideadisincentiveto'cheat',i.e., act as acollateralbondagainst suchopportunisticbehavior. To seethiscollateraleffect, consider anindividualauditor'sdecisionto 'cheat'insomefutureperiodwhen, due totechnological andtransactions costadvantages,theauditorearns client-specificquasi-rents.

Forthe

brandnameexpenditures)are prohibitivelycostly.We also assumethatthemagnitudeof a givenauditor's client-specificquasi-rentsdoes not varyacrossclients.'?Bothofthese assumptionsarerelaxedin section 4.

Suppose

that,atsomefuturedate, a givenincumbentauditordiscoversa breachin aparticularclient'srecords.:"The client can attempttodissuade theauditorfromreportingthe breach,perhapsbythreatsoftermination. Indeed,theclienthas a crediblethreatoftermination,shouldthe auditor by thepresentvalueof thequasi-rentsspecific to this client,which arelostif theauditor reportstruthfullyand isterminatedby the client. Acountervailingdisincentiveisprovidedby thepresentvalueof thequasi rentsspecific toothercurrentclients of theauditor.

Iftheauditor'cheats'

andiscaught,hestandsto lose someportionof thisvalue boththrough terminationbyotherclientsandthroughreducedfees fromthosethat continuetoretainhim.i?Theamounttheauditorstandsto losedependson

17Thisassumptionis areasonableone whenauditorsspecialize in auniformlevel ofaudit

qualityacrossclientsinresponseto consumer qualityevaluationcosts. Thisassumptionisnot necessary forourconclusionsto hold and, in fact, is relaxed in the followingsection.

18Nootherbreachesaresimultaneouslydiscovered by theauditor,by

assumption:The analysis wouldbecomeconsiderablymore complex if theauditorhad theincentiveto'cheat' simultaneouslyon several clients. However, the resultthata largernumberofcurrentclients lowers theprobabilityofmisrepresentationstill obtains.

19Althoughterminationisnotcostless to clients, suchterminationisrationallyassessed to be

a positiveprobabilityevent. This is the case because a discoveredbreach(which theincumbent auditorintends to disclose) increases thepotentialbenefits ofterminationto the client. Specifically,when it is costly for outsiders to discover and verify whyterminationoccurred, opportunisticclientscan'blame'theauditorandperhapsavoid or at leastmitigatethenegativequotesdbs_dbs31.pdfusesText_37

[PDF] assistant de service social - Université Rennes 2

[PDF] exemples de sujets de debat a l'oralpdf

[PDF] Le code vestimentaire - National Golf Course Owners Association

[PDF] internet et le débat démocratique - Cairn

[PDF] LE DEBAT REGLE

[PDF] Fiche méthodologique débat

[PDF] 2 • 4 • Le débat en ligne - Archives des sites des Commisions

[PDF] LE PETIT LAROUSSE ILLUSTRé 2013

[PDF] Le débat philosophique Thème : la tolérance 1 Enjeu de cet atelier

[PDF] Max et Lili » et le vivre ensemble

[PDF] La discussion à visée philosophique - mediaeduscoleducationfr

[PDF] Enseigner la morale par le débat philosophique

[PDF] Fiche doc bonheur mars 09rtf - ICEM - Pédagogie Freinet

[PDF] Classroom Debates - Northern Illinois University

[PDF] 'Dictionnaire des synonymes' de Condillac : de la synonymie - Hal