TheAssociationbetweenIndividualAuditPartners9RiskPreferencesandthe

CompositionoftheirClientPortfolios

EliAmir

LondonBusinessSchool

Regent'sPark,LondonNW14SA,UK

Eamir@london.cdu

Juha-PekkaKalkinki*

UniversityofOulu,DepartmentofAccountingandFinance

P.O.Box4600,FIN-90014UniversityofOulu,Finland

Juha-Pekka.Kallunki@oulu.fi

HenrikNilsson

StockholmSchoolofEconomics,DepartmentofAccounting

P.O.Box6501,SE-11383Stockholm,Sweden

Henrik.Nilsson@hhs.se

October2011

*Correspondingauthor.WethanktheseminarparticipantsalBocconi(Italy),theUniversityof

Cyprus,IEBusinessSchool(Spain),theNorwegianBusinessSchool,BI(Norway),the

UniversityofNewSouthWales(Australia),theUniversityofAuckland(NewZealand),the

UniversityofOulu(Finland)andVictoriaUniversity(Wellington,NewZealand)foruseful

comments.EliAmirisgratefultotheLondonBusinessSchoolforresearchsupport.Wealso

thankEuroclearSweden,FinansinspektionenandtheSwedishtaxagencyforproviding[he

requisitedata.Anypossibleerrorsareourown.Thisstudyhasbeenevaluatedandapprovedby

TheRegionalEthicalReviewBoardinUmea,Sweden(DNR08:074O).

TheAssociationbetweenIndividualAuditPartners9RiskPreferencesandthe

CompositionoftheirClientPortfolios

Abstract

Werelaxtheassumptionoftenmadeinempiricalauditingresearchthatauditorsare

homogenousindividualswithsimilarpersonalcharacteristics.Basedontheexistingtheoriesin

economicsandbehavioralsciences,wcexplorewhetherauditpartners1attitudetowardsrisk,as

measuredbytheirpersonalcriminalconvictions,arcreflectedinthecompositionoftheirclient

portfolios.AnalyzingauniquedatasetofSwedishauditpartners'criminalconvictions,wefind

thattheclientsofauditpartnerswithcriminalconvictionsarecharacterizedbygreaterfinancial,

governanceandreportingriskthanthoseofauditpartnerswithoutcriminalconvictions.Also,

auditpartnerswithcriminalconvictionschargehigherauditfeescompensatingthemfor

acceptingriskierclients,aftercontrollingforotherauditriskvariables.

JELClassification:M41,M42,G30,K42

Keywords:AuditPartners,CriminalConvictions,ClientPortfolios,AuditRisk,Accounting

Conservatism

1

OntheAssociationbebveenIndividualAuditPartners9RiskPreferencesand

theCompositionoftheirClientPortfolios

1.Introduction

Researchshowsthatauditors'portfoliomanagementdecisionsareaffectedbytheclient's

perceivedrisk,andthatauditorsassessclientriskonthebasisofclient-specificcharacteristics,

suchasfinancialcondition,thelevelofcorporategovernanceandthequalityoffinancial

reporting.Withfewexceptions,thislineofresearchtypicallyassumesthatauditorsare

homogenousintheirattitudestowardsclientrisk.Todate,littleisknownabouthowindividual

auditpartners'personalriskpreferencesaffectclientportfoliodecisions,althoughitisthe

partnerswhomakethesedecisionswithintheauditfirm.Thelackofempiricalarchivalresearch

intheareaisduetodatalimitations,becauseinmanycountries,includingtheUnitedStatesand

theUnitedKingdom,individualauditorsarenotrequiredtosigntheirnamesontheaudit

reports.However,sucharequirementiscurrentlyinpracticeinsomecountries,whichfacilitates

bringingtheanalysisdowntothelevelofindividualauditpartnersinthesecountries,i

ThisstudyutilizesaproprietarydatasetfromSwedentoexplorewhetherauditpartners'

personalriskpreferences,asmeasuredbytheirpossiblepriorcriminalconvictions,arereflected

intheirclientportfoliorisk.2

1See,forinstance,ChinandChi(2008)andChenetal.(2010).ThePublicCompanyAccounting

OversightBoard(PCAOB)hasissuedaconceptreleasesolicitingcommentsonwhetheritshouldrequire

engagementpartnerstosignauditreportstoincreasetransparencyandenhancepartners'accountability

(PCAOB2009).

Thecriminalconvictionsdiscussedherearcnotrelatedtotheauditors'work.Ouranalysisdocsnot

implythatcriminallyconvictedauditorsnecessarilyexhibitillegalorevenfaultyprofessionaljudgment

intheirauditwork.Instead,wearguethatcriminalconvictionsindicateagreaterpropensityforrisk-

taking,ingeneralandinparticulartheauditwork.

2

Ourempiricalresultssupportthepredictionthatauditpartnerswithcriminalconvictions

engageinriskierauditsthanthosewithoutcriminalconvictions.Inparticular,wefindthat

clientsofconvictedauditpartnershaveagreaterfinancialrisk,asreflectedintheirfinancial

ratios,andagreatergovernancerisk,asreflectedintheirboardcomposition,thanclientsof

auditpartnerswithoutcriminalconvictions.Moreover,clientsofauditpartnerswithcriminal

convictionshaveagreaterfinancialreportingrisk;thesefirmsreportlessconservatively.We

alsofindsupportforthepredictionthatauditfeesinauditsperformedbyconvictedauditorsare,

onaverage,higherthaninthoseperformedbyauditorswithoutcriminalconvictions,to

compensatetheseauditorsfortakingadditionalrisk.Finally,ourresultsshowthatauditpartners

withcriminalconvictionsaremorelikelytobemalesemployedinNon-Big-5auditfirms.This

findingsuggeststhatBig-5auditfirmsarelesstolerantofrisk-takingauditorsandhaveastricter

screeningpolicyfortheiremployees.3

Wcproceedasfollows:Section2providesareviewoftheliteratureanddevelopsour

predictions.Section3discussesthesampleanddata,andSection4presentsourempirical

analysisandresults.Section5discussesourconclusionsandpolicyimplications.

2.LiteratureReviewandTestablePredictions

2.1.CriminalConvictionsasanIndicatorofRisk-TakingBehavior

Becker(1968)andEhrlich(1973)arguethatindividualsengageincriminalactsifthe

expectedgainfromthatactisgreaterthantheexpectedcosts;theseindividualsarcwillingto

■'WeinterviewedseveralauditpartnersinleadingSwedishfirms.Thesepartnersconfirmedthatthelarge

auditfirmsarelesswillingtotakehigh-riskclientstoavoidlossofreputation.Thesefirmsalsocharge

higherfeesforauditservicesofferedtotheseclientstocompensatethemselvesforadditionalriskand

evenavoidtakingtheseclients.Consequently,smallerauditfirmshavemorehigh-riskclientsthatpay

higherauditfees.Auditpartnersofthelargeauditfirmsalsopointedoutthattheyhavedismissed

auditorsexhibitingexcessiverisk-taking,andtheseauditorsarcoftenhiredbysmallerfirms.These

inteniewsshouldbeviewedasanecdotalevidencesupportingcurpredictions.

4

taketheriskofbeingcaughtandconvictedincourtbecausetheexpectedbenefitscompensate

themfortakingsuchrisks.Hence,adecisiontoengageincriminalactivitiescanbeseenas

rationalbehaviorunderuncertainty.AsEhrlich(1973)pointsout,animportantimplicationof

thesemodelsisthatarisk-neutralindividualwillspendmoretimeonillegalactivitiesrelativeto

arisk-avoider,andarisk-seekerwillspendmoretimeonsuchactivitiesrelativetoboth.Hence,

criminalconvictionsreflectanindividual'srisk-seekingattributeregardlessofthetypeor

seriousnessofthecrime.

Unliketheeconomicapproachtocrime,behavioralresearchlinkscrimetopersonal

attributes,andinparticulartooverconfidence;individualswhoengageincriminalactivitiestend

tobeoveroptimisticandunderestimatetheprobabilityofnegativeoutcomes(seeAideetal.

(2006),Garoupa(2003),PalmerandHollin(2004),andWalters(2009)).Furthermore,Sandroni

andSquintani(2004)findthatoverconfidenceisamajordeterminantoftrafficaccidents,and

McKenna(1993)showsthattheillusionofcontrol,relatedtooverconfidence,characterizes

riskydrivers.OverconfidenceimpliesiiTationalratherthanrationalrisk-takingbehavior;

however,aspointedoutbyCarr-HillandStern(1979),theeconomicandbehavioralapproaches

tocrimeshouldbeseenascomplementaryratherthanconflicting.

Auditors,likeotherindividuals,possessdifferentpersonalcharacteristics,including

differentpreferencesregardingriskandtendencytocriminalbehavior.Severalexperimental

studieshaveexaminedthediversityinauditors'personalcharacteristicsrelatedtotheir

propensitytotakerisk,andhowthesecharacteristicsaffecttheauditworkunderdifferent

circumstances.Forexample,Schatzbcrgctal.(2005)findapositivelinkbetweenauditors'

levelsofmoralreasoningandauditmisreporting.Theyalsofindthatastheeconomicpenalties

increase,auditormisreportingandfeesdecreaseandthemoralreasoningeffectdiminishes.

5

CohenandTrompeter(1998)findthatthemoreaggressiveauditpartnersoflargeauditfirmsare

themorelikelytheyaretorecommendacceptingtheclient'srelativelyaggressiveaccounting

choicesinordertoretaintheclientintheincreasinglycompetitivemarketsofauditservices.

Theyalsofindthatauditorsarcmorewillingtobidfbrexistingthanpotentialclients,afinding

thatisconsistentwiththeprospecttheoryofKahnemanandTversky(1979)-individualsare

risk-aversewithrespecttogainsandrisk-seekingwithrespecttolosses.Fanner(1993)

examinesriskattitudesofindividualauditorsinlargeauditfirmsandfindsthattendenciesfbr

bothriskaversionandriskpreferenceoccuramongauditors.Krishnan(2005)arguesthathigh-

qualityauditorsconstraintheopportunisticfinancialreportingofthefirm.

Priorstudieshavealsoexaminedbehavioralbiases,suchasoverconfidence,among

auditors.Messierelal.(2008)findthatauditpartnersexhibitsignificantoverconfidenceinthe

abilityoftheirsubordinatestodetecterrors.KennedyandPeecher(1997)findthatauditorsare

overconfidentregardingboththeirownandtheirsubordinatcs,technicalknowledge;this

overconfidenceincreaseswiththeknowledgegapbetweensupervisorsandsubordinates.

Whilemanystudieshavelookedatauditfirmleveldata,recentstudieshavefocusedmore

onoffice-levelanalysis.Forexample,ReynoldsandFrancis(2000)examineclientaccrualsand

auditorgoingconcernreportsattheofficelevelofanalysisandfindevidencesuggestingthat

auditorstreatedlargerclientsmoreconservatively.Whileaudit-firm-levelandoffice-level

analyseshavegreatlycontributedtoourunderstandingofthecausesandconsequencesofaudit

risk,qualityandpricing,significantinsightscanbeobtainedbybringingtheanalysesdownto

thelevelofindividualauditors.Auditingrequiresexpertiseandskillsaswellaspersonal

judgmentinassessingtheoverallauditrisk,theappropriatescopeoftheauditandthe

6

accountingchoicesmadebyclientfirms.Therefore,auditors'personalattributeshavea

significanteffectontheirdecisionsregardingtheauditwork.

Theliteraturediscussedabovesuggeststhatauditorsdifferfromeachotherintheir

personalriskpreferences.Wearguethatauditorswhohavebeenconvictedofcrimeshavea

higherpropensityforriskintheirpersonalaswellastheirprofessionalwork.Inparticular,

economictheoryofcrimeimpliesthatauditorswithcriminalconvictionsmaybemorewilling

totakerisksiftheexpectedgainfromtakingsuchriskisgreaterthantheexpectedcosts.For

instance,theymayprefertoengageinauditingmoreriskycompaniesortoallowthese

companiestoadoptlessconservativeaccountingiftheycanchargehigherauditfees(aformal

modelispresentedintheAppendix).Inaddition,behavioralstudiesimplythatauditors'

criminalconvictionsmayreflectover-confidence,whichleadstomoreaggressiverisk-seeking

behavior.

2.2.TestablePredictions

Priorexperimentalstudieshaveshownthatauditorsevaluateclient-relatedriskstoavoid

lossesfromauditengagement.Forinstance,Johnstone(2000)arguesthatauditpartnersevaluate

client-relatedrisks,andusethoseevaluationstodeterminewhethertheauditfirmwillsuflera

lossontheauditengagementduetopotentiallitigation.Otherstudiesshowthatauditorsassess

clicnt-spccificriskbyusingtheclient'sfinancialcondition(PrattandSticc,1994),levelof

corporategovernance(CohenandHanno,2000),andqualityoffinancialreporting(Cohenand

Trompeter,1998).

Theauditfirmlevelresultsonwhetherauditorsvarybytheirtolerancetowardsclientrisk

aresomewhatcontradictory.JonesandRaghunandan(1998)arguethatlargeauditfirmsavoid

7

riskbecausetheyhavemoretolosefromauditfailure.Bycontrast,FrancisandKrishnan(2002)

arguethatlargeauditfirmsaccepthigh-riskclientsbecausetheycandiversifythatriskacross

theirclientportfolio.Therefore,thequestionofhowauditpartners'personalriskpreferences

affectclientportfoliodecisionsisstillanopenone.

Wearguethatauditpartnerswithpriorcriminalconvictionsaremorewillingtoaccept

riskierclientsthanauditorswithoutcriminalconvictions.Consequently,weexpectclientsof

auditpartnerswithcriminalconvictionstobecharacterizedbyhigherfinancialrisk,weaker

governanceandmoreaggressivefinancialreporting,thanclientsofauditpartnerswithout

criminalconvictions.

SimunicandStein(1990)focusonfinancialriskandfindthatauditor'sriskincreaseswith

thedegreeofclients'financialrisk.Hence,anauditpartnerwithapriorcriminalconviction,

whoismorewillingtoengageinriskierauditengagement,ismorelikelytohaveclientswith

greaterfinancialrisk.Thisleadstothefollowingprediction:

Prediction1:Clientsofcriminallyconvictedauditpartnershaveagreaterfinancialriskthan

thoseofnon-convictedauditpartners.

Weakercorporategovernanceincreasestheauditriskbecauseweakergovernanceis

associatedwithhigherlikelihoodoffraud,misappropriaticnsoffunds,andopportunistic

accountingdecisions,whichcouldalsoleadtohigherlitigationrisk.Anauditpartnerfacing

clientswithweakergovernanceisrequiredtotakethatintoaccountinplanningandexecuting

theaudit.CohenandHanno(2000)reportthatthegovernancestructureoftheclientfirmaffects

auditors'decisiontoaccepttheclient.Duetotheirgreatertoleranceofrisk,auditpartnerswith

8

criminalconvictionsaremorelikelytotolerateweakergovernanceoftheirclientfirms.This

leadstothesecondprediction:

Prediction2:Clientsofauditpartnerswithpriorcriminalconvictionshaveweakergovernance

thanclientsofauditpartnerswithoutcriminalconvictions.

Thelevelofconditionalaccountingconservatism,oftenmeasuredastheasymmetric

recognitionofgoodandbadeconomicnewsinearnings,isregardedasafundamental

characteristicofhighqualityfinancialreporting.Also,asWatts(2003)andSkinner(1997)

argue,theriskoflitigationcompelsauditorstorequireconservativeaccountingbecause

damagesareassessedwhenfinancialstatementsoverstatenetassetsandprofitabilityratherthan

understatethem.Qiang(2007)findsthatlitigationriskispositivelyrelatedtoconservatism.

Basuetal.(2001)findthattheclientsofBig-8auditorsshowagreaterdegreeofconservatism

thanthoseofNon-Big-8auditors.Khrisnan(2005)comparestheasymmetrictimelinessof

earningsofArthurAndersen'sHouston-basedclientswithacontrolgroupandfindsevidencein

supportoftheviewthatriskyclientsshowalowerdegreeofearningsconservatism.

Weexpectauditpartnerswithcriminalconvictionstobemoretolerantofaggressive

accountingtreatment,resultinginlessconservativeaccounting.Thisisbecauseauditpartners

withcriminalconvictionsarewillingtotakeagreaterriskwhenselectingclients,comparedto

auditpartnerswithoutcriminalconvictions.Thisleadstothethirdprediction:

Prediction3:Clientsofauditpartnerswithcriminalconvictionsreportlessconservativelythan

thoseofauditpartnerswithoutcriminalconvictions.

9

Ourfirstthreepredictionsdealwithdifferentaspectsofauditrisk-financial,governance

andreportingrisks.Ourfinalpredictionaddressesthecompensationfortakingadditionalaudit

risk.Priorstudiesonauditpricing(forinstance,Sinmnic,1980)havefoundthatauditfees

increasewithfirmrisk.Also,consistentwiththeviewthatauditfeessummarizetheauditor's

unobservableprivateinformationaboutclientfirms*risk,Hribaretal.(2010)findthathighaudit

feespredictaccountingfraud,restatements,andSECcommentletters,aftercontrollingforother

measuresofaccountingquality.

Iffirmsauditedbyauditpartnerswithcriminalconvictionsareindeedriskierthanthose

auditedbyauditpartnerswithoutconvictions,theformerfirmsshouldpayhigherauditfeesthan

thelattertocompensatetheauditpartnerforassuminggreaterrisk.Inparticular,auditpartners

facesignificantpersonalreputationandauditfirm-levellitigationrisksincasesofaudit

failures.4Beingawareofthesegreaterauditrisksinherentintheirclients,auditpartnersrequire

compensationforbearingthatriskintermsofhigherauditfees.Moreover,auditpartnersmay

increasethescopeoftheauditinanattempttomanagetheperceivedhighclientrisk,increasing

auditfees.Inequilibrium,firmswithhigherriskareauditedbythoseauditorswhoarewillingto

taketheriskforahigherauditfee.Wethereforepredictthatfirmsauditedbyauditpartnerswith

criminalconvictionspayhigherauditfees,aftercontrollingforfirm-risk.

Prediction4:Finnsauditedbyauditpartnerswithcriminalconvictionspayhigherauditfees

thanthoseauditedbyauditorswithoutcriminalconvictions,aftercontrollingforfirm-risk.

Personalreputationriskislinkedtotheprobabilityoflosingexistingclientsandnotobtainingnew

ones;totheprobabilityofbeingsanctionedbyhisauditfirm,andtheprobabilityofbeingsanctionedby

thestateorregulator.Auditfirmlevellitigationriskislinkedtotheprobabilityofbeingpenalizedfor

auditfailuresorbeingsued.

10

3.InstitutionalBackground,SampleandDataSources

3.1.AuditProfessioninSweden

TwotypesofauditorsinSwedenareallowedtoauditlistedfirms:approvedauditorsand

authorizedauditors.Tobecomeanapprovedauditor,thecandidatemustobtainabachelor's

degree,practiceforatleastthreeyears,andpassanexaminationinprofessionalcompetence.To

becomeanauthorizedauditor,theapprovedauditormustobtainamaster'sdegree,complete

fiveyearsofpractice,andpassanexaminationforauthorizedauditors.Theauditcertificationis

validforfiveyears;afterwhichtheauditormustreapplytothesupervisoryboardforlicense

renewal.5AlllimitedliabilitycompaniesinSwedenmustbeauditedonanannualbasis.

Auditinglistedfirmsoftenrequiresseveralauditorsduetothecomplexityoftheaudit.Hence,

auditfirmsformteamsofauditorsledbyauditpartners.Theseteamleadersareresponsiblefor

keyclient-specificdecisions,includingthescopeandpricingoftheauditengagement.

Therefore,leadingauditors'personalattributes,asmeasuredbytheircriminalconvictions,are

likelytoplayanimportantroleinaudit-relateddecisions.

TheauditmarketinSwedenislargelycontrolledbyPWC,Ernst&Young,KPMGand

Deloiuc,whichemploy56%of(heauthorizedauditors,and29%oftheapprovedauditorsin

Sweden.6Theirmarketshareisabout40%,anditisincreasingwithclientsize.Auditorsand

registeredpublicaccountingfirmsaresubjecttoindependentqualitycontroleverysixyears,and

anauditorwhoauditsatleastonepubliccompanyshouldbeevaluatedeverythreeyears.Also,

mandatorypartnerrotationwasnotpracticedinSwedenduringthesampleperiod.

'AccordingtotheSwedishMinistryofJustice,therewere2,321approvedauditorsand1,787authorized

auditors2008(S.se/sb/d/584).

6OursampleperiodstillincludesArthurAndersen,andhence,wehavefivebigfirmsinoursample.

11

Giventhenatureoftheirwork,auditorsareexpectedtoexhibithighpersonalintegrityduetothe

fiduciaryresponsibilityentrustedtothem.Normally,convictedindividualsinwesterncountries,

includingSweden,cannotcompletethetrainingandobtainthelicenseallowingthemtobecomecertified

publicaccountants.However,inmanywesterncountries,includingSweden,individualsarcbarredfrom

becomingcertifiedauditorsonlyiftheyhavebeenconvictedofviolentoreconomic-relatedcrimes.Also,

itispossiblethatauditorsareconvictedofcrimesafterreceivingtheirlicenseandthesecrime

convictionsrarelytriggerthewithdrawalofalicense.Inparticular,theSupervisoryBoardforPublic

Accountants(SBPA)inSwedenexaminesthecriminalrecordsofapplicantseveryfiveyearswhen

certificationisrenewed.TheSupervisoryBoardalsocheckswhetherauditorshavehadanydisputeswith

thetaxauthorities,thattheyarenotbankruptandthattheydonothavelegalguardianship.However,

criminalrecordchecksarelimitedtoeconomic-relatedcrimesandcrimesrelatedtotheauditprofession.

AuditingstandardsissuedinSwedenarebasedonInternationalAuditingStandards,but

someadditionsandchangeshavebeenmadetocertainstandardstomakethemconsistentwith

Swedishlaw.Tomaintainauditorindependence,SwedishandEuropeanlawsrequirethat

auditorshaveneitherafinancialinterestnoranyclosepersonalrelationshipwiththeclient.

Auditfailuresmayresultinlitigationbyclientsanddisciplinarysanctionsbytheprofession.As

inmostotherEuropeancountries,thelitigationriskisrelativelylowinSweden.TheSBPA

issuesdisciplinarysanctionsagainstauditorsincertaincases,althoughthesecasesarenot

common.7

3.2.SampleandDataSources

OursampleincludesalllistedcompaniesinSwedenmonitoredbytheSwedishsecurities

regulator(Finansinspektionen)duringtheperiod1999-2007.Theidentitiesofauditorsinall

7Accordingtothe2010-2011GlobalCompetitivenessReportoftheWorldEconomicForum,thequality

ofSweden'sauditingandreporingstandardsisranked2ndintheworld(aheadoftheUnitedS(a:es).See:

:〃/issues/global-comDetiiiveness.

12

listedSwedishcompanieswerealsoobtainedfromFinansinspektionen.Dataonauditors'

criminalconvictionsandsuspectedcriminalactionsaretakenfromBra(TheSwedishNational

CouncilforCrimePrevention).Ourdatasetcontainsinformationonthecriminalactivityofall

Swedishcitizenssince1974.Specifically,itcontainsinformationaboutindividualswhohave

beenfoundguiltybyacourtoflaworreceivedsummarypunishmentsbyprosecutors.8This

databaseismorecomprehensivethantheofficialcrimerecords,becauseitcontainsallcriminal

convictionsinSwedensince1974,regardlessofthetypeofcrimeorwhethertheseconvictions

havebeenexpungedfromtheofficialcrimerecordsavailableonSwedishcitizens.The

informationcontainedinthedatabasewascollectedfromallSwedishcourtsandprosecution

authorities.Foreachauditorregistered,thisdatasetincludesdetailsofthecrimeandthe

punishment(thelengthofunconditionalprisonsentences,suspendedsentencesandmonetary

fines)andthedetailsofthecrime(foreachcrimeanexactreferencetothelaworlawsviolated

isgiven).Thedatasetdocsnot,however,containinformationonminoroffenceslikespeeding,

parkingandviolationsoflocalbylawsforwhichthepunishmentisanon-the-spotfine.Hence,

thedatabasedoesnotcontaininformationaboutminorcrimes.

Wemeasurecriminalbehaviorintermsofcriminalconvictionsandhavingbeenunder

investigationforsuspectedcriminalactions.Whilecriminalconvictionsareproofofprior

criminalactivity,focusingonlyonactualconvictionscouldpotentiallycauseaselectionbias.

Thisisbecausetheburdenofproofisheavierinmoreseriouscrimes,asindicatedbyKorsell

(2001).Consequently,seriouscrimesarelikelytobeunderrepresentedinthedatasetofactual

'Bra(bra.se)aimstoreducecrimeinSwedenbyprovidinganddisseminatingdataoncrimeand

crimeprevention.BraalsoproducesSweden'sofficialcrimestatistics,evaluatesrefbnns,conducts

researchandprovidessupporttolocalcrimepreventionagencies.Also,acriminalinvestigationdoesnot

alwaysleadtoprosecutionandtrial.Ifthesuspectconfessesto!hecrimeanditisclearwhatthe

punishmentwillbe,theprosecutormaypronounceaso-calledorderofsummarypunishment(Source:

SwedishProsecutionAuthority,aklagare.se).

13

criminalconvictions.Thisselectionbiascouldbereducedbyincludingdataonindividuals

havingbeenunderinvestigationforseriouscrimes.Ourdatasetonsuspectedcriminalactions

containsinformationonallSwedishcitizenswhohavebeenunderinvestigationforserious

crimes.Suspicionofacrimeinthisstudymeansthatapoliceinvestigationhadbeenlaunched

buttheprosecutoreventuallydecidednottopursuethecaseincourt.Thedatabaseismaintained

bytheNationalPoliceBoardandismainlyusedbythepouce,thetaxauthorities,thecustoms

andthecoastguardstocoordinatepreliminaryinvestigationsofindividualsinordertoprevent,

discoverandinvestigatecrimes.Finally,accountingandmarketdataforSwedishlistedfirms

wereobtainedfromThomson'sDatastream.

Table1presentsalistoftheconvictionsbylawsviolatedandbyauditfirmsize(Big-5

versusNon-Big-5).NotethatwhiletheproportionofconvictionsishigheramongNon-Big-5

firms(18.5%)thaninBig-5firms(10%),themostseriouscrimesinthesamplewerecommitted

byBig-5auditors.Table1clearlyshowsthatoursampleisnotdominated