Forecasting without Consequence-Evidence on the Properties of Retiring CEO

THEACCOUNTINGREVIEW

Vol.88,No.6

2013

pp.1909–1937AmericanAccountingAssociationDOI:10.2308/accr-50526

ForecastingwithoutConsequence?EvidenceonthePropertiesofRetiringCEOs’Forecasts

ofFutureEarnings

CoryA.Cassell

UniversityofArkansas

ShawnX.Huang

ArizonaStateUniversity

JuanManuelSanchez

TexasTechUniversity

ABSTRACT:WeinvestigatewhetherretiringCEOsengageinopportunisticterminal-

yearforecastingbehaviorandthecircumstancesinwhichsuchbehaviorislikelytobe

http://wendang.chazidian.comingawithin-CEOempiricaldesign,wefindthatretiring

CEOsaremorelikelytoissueforecastsoffutureearnings,andthattheyissuesuch

forecastsmorefrequentlyintheirterminalyearrelativetootheryearsduringtheirtenure

withthefirm.Further,retiringCEOs’terminal-yearforecastsoffutureearningsaremore

likelytoconveygoodnewsandaremoreoptimisticallybiasedrelativetopre-terminal

years.Opportunisticterminal-yearforecastingbehavioris:(1)morepronouncedinthe

presenceofhigherCEOequityincentivesandwhendiscretionaryexpendituresarecutin

theterminalyear,and(2)lesspronouncedinthepresenceofstrongermonitoring

mechanisms,suchashigherinstitutionalownership.Collectively,ourresultsprovide

evidenceonapotentialimplicationoftheCEOhorizonproblemthathasnotbeen

investigatedpreviously.

Keywords:forecastbias;horizonproblem;managementearningsforecasts;manage-

rialmyopia;CEOretirement.

WethankJohnHarryEvansIII(senioreditor),BeverlyR.Walther(editor),twoanonymousreferees,RichardCazier,MikeDrake,SteveKachelmeier,EdwardLi,JamesMyers,LindaMyers,RayPfeiffer,andStephanieRasmussenforcommentsthathavehelpedusimprovethequalityofthepaper.WealsothankworkshopparticipantsatTexasChristianUniversityandtheUniversityofArkansas,andparticipantsatthe2011AmericanAccountingAssociationAnnualMeetingandthe2013FARSMidyearMeetingforinsightfulcomments.WethankBenjaminAnderson,CariBurke,BenCoulter,LaurenDreher,TaylorJoo,StaceyKaden,RoySchmardebeck,TimothySeidel,andMichaelStuartforexcellentresearchassistance.Wegratefullyacknowledgethe?nancialassistanceoftheSamM.WaltonCollegeofBusinessattheUniversityofArkansas.

Editor’snote:AcceptedbyBeverlyR.Walther.

Submitted:April2011

Accepted:June2013

PublishedOnline:June2013

1909

1910Cassell,Huang,and

Sanchez

DataAvailability:Alldatausedinthestudyarepubliclyavailablefromthesourcescited

inthetext.

I.INTRODUCTION

TheorysuggeststhatChiefExecutiveOf?cers(CEOs)withshorthorizonswiththeir?rm

haveweakerincentivestoactinthebestinterestofshareholders(SmithandWatts1982).Todate,researchexaminingthe‘‘horizonproblem’’focusesonwhetherCEOsadoptmyopicinvestmentandaccountingpoliciesintheir?nalyearsinof?ce(e.g.,DechowandSloan1991;Davidsonetal.2007;Kalyta2009;Antiaetal.2010).WeextendthislineofresearchbyinvestigatingwhetherretiringCEOsaremorelikelytoengageinopportunisticforecastingbehaviorintheirterminalyearrelativetootheryearsduringtheirtenurewiththe?rm.Speci?cally,wecontrasttheproperties(issuance,frequency,news,andbias)ofearningsforecastsissuedbyretiringCEOsduringpre-terminalyears,whentheCEOwillbeinof?cewhentheassociatedearningsarerealized,withforecastsissuedbyretiringCEOsduringtheirterminalyear,whentheCEOwillnolongerbeinof?cewhentheassociatedearningsarerealized.Wealsoexaminecircumstancesinwhichopportunisticterminal-yearforecastingbehaviorislikelytobemoreorlesspronounced.

OurpredictionsarebasedonseveralincentivesthatariseorincreaseduringretiringCEOs’terminalyearwiththeir?rms.Speci?cally,relativetoCEOswhowillcontinuewiththeir?rms,retiringCEOsfacestrongincentivestoengageinopportunisticterminal-yearforecastingbehaviorinanattempttoin?atestockpricesduringtheperiodleadinguptotheirretirement.Althoughitisonlynecessaryforourhypothesesthatmanagersbelievesuchopportunisticbehaviorwillin?uencestockprices,priorworkshowsthatmanagers’opportunisticvoluntarydisclosuresin?uencestockprices(Noe1999;AboodyandKasznik2000;ChengandLo2006;Hammetal.2012).Moreover,thisworksuggeststhatmanagersuseopportunisticearningsforecaststomanipulateanalysts’(Cotteretal.2006)andinvestors’(ChengandLo2006;Hammetal.2012)perceptionsinanefforttomaximizethevalueoftheirstock-basedcompensation(AboodyandKasznik2000).BecauseSECtradingrulesrelatedtoCEOs’post-retirementsecuritytransactionsarelessstringentthanthoseineffectduringtheirtenurewiththe?rm,post-retirementtransactionscanbemadebeforetheearningsassociatedwiththeopportunisticforecastarerealizedandwithreducedregulatoryscrutiny.1

Totestourpredictions,weidentifyallCEOturnovereventsinExecuCompfrom1997through2009andperformdetailedsearchesofSEC?lings,executivebiographies,pressreleases,andrelateddisclosurestodeterminewhethertheCEOturnoverwasduetoretirement.2WerequirethateachoftheseretiringCEOshaveatleasttwoyearsofdatapriortotheirterminalyearsothatwecancomparewithin-CEOforecastingpatterns.Theseproceduresyieldasampleof862CEOretirementsfortheforecastissuance/frequencytestsand272CEOretirementsfortheforecastbiastests.

OurresultsindicatethatretiringCEOsaremorelikelytoissueforecastsoffutureearningsandthattheyissuesuchforecastsmorefrequentlyintheirterminalyearrelativetootheryearsduringtheirtenurewiththe?rm.Moreover,we?ndthatretiringCEOs’terminal-yearforecastsoffutureearningsaremorelikelytoconveygoodnewsandaremoreoptimisticallybiasedrelativeto1

2AnecdotalevidencesuggeststhatanumberofretiringCEOssellasigni?cantproportionoftheirholdingssoonaftertheretirementdate.Forexample,NicholasTaubman,formerCEOandchairmanofAdvanceAutoParts,soldabout17percentofhisholdingsforatotalofapproximately$19millionsoonafterretiringfromAdvanceAutoPartsinearly2003(Kincaid2003).Similarly,MichaelAhearn,formerCEOandchairmanofFirstSolar,soldapproximately$160millioninFirstSolarsharesshortlyafterhisretirementasCEO(Ahanotu2010).WeclassifyCEOretirementsasturnovereventswheretheCEOdoesnotsecureapositionwithanotherpubliccompanyintheyearssubsequenttohis/herdeparture.Forallturnoverevents,oursearchforsubsequentemploymentextendstotheendofourdatacollectioninlate2011.

TheAccountingReview

November2013

EvidenceonthePropertiesofRetiringCEOs’ForecastsofFutureEarnings1911pre-terminalyears.Our?ndingsthatretiringCEOsengageinopportunisticterminal-yearforecastingbehaviorrepresentapreviouslyundocumentedimplicationoftheCEO‘‘horizonproblem.’’

Next,weexaminecircumstancesinwhichtheterminal-yearforecastingbehaviorwedocumentfortheaverageretiringCEOislikelytobemoreorlesspronounced.Speci?cally,weinvestigatetheeffectsofeightfactorswehypothesizeshouldin?uencetheextentofterminal-yearopportunisticforecastingbehavior.We?ndthatopportunisticterminal-yearforecastingbehaviorismorepronouncedinthepresenceofhigherCEOequityincentivesanddiscretionaryexpenditurecutsintheterminalyear,andlesspronouncedinthepresenceofstrongermonitoringmechanisms,suchasgreaterinstitutionalownership.Otherfactorshaveasigni?canteffectonsome,butnotall,oftheforecastcharacteristicsinthepredicteddirection.

AlthoughweattributeourresultstochangesinretiringCEOs’incentivesduringtheirterminalyearwiththe?rm,ourresultscouldpotentiallybeconfoundedbyover-timetrendsinmanagementforecasts(e.g.,theeffectsofCEOtenure,seculartrends,orregulatorychanges)orbyapreferenceforconservativeaccountingbytheincomingCEO.Basedontheresultsfromabatteryofsensitivityanalyses,weconcludethatourinferencesaremostlikelynotanartifactoflongerCEOtenure,over-timetrendsinmanagementforecasts,orincomingCEOsmakingmoreconservativereportingchoices.

Ourresultsshouldbeofinteresttomarketparticipants,suchasinvestorsandanalysts,whouseinformationfrommanagementearningsforecasts.However,marketparticipants’abilitytouseourevidenceiscontingentontheirknowledgeof,orabilitytoanticipate,agivenCEO’simpendingretirement.Ourstudyshouldalsobeofinteresttostakeholders,suchasboardsofdirectorsandregulators,whoseektoimplementincentivemechanismsthatmitigateagencycon?icts.Interestingly,ourresultssuggestthatequityincentives,atoolcommonlyusedtoalignincentivesandminimizeagencycosts,canhavetheunintendedconsequenceofcreatingorexacerbatingopportunisticforecastinginaCEO’sterminalyear.3Thus,CEOand?rmcharacteristics,suchasequityincentives,mayhavecompetingeffectsonvarioushorizon-problem-inducedbehaviors.

Ourstudycontributestotwobroadresearchstreamsintheacademicaccountingand?nanceliteratures.First,ourstudycontributestoanemerginglineofresearchthatinvestigateshowindividualmanagersortheirparticularcharacteristicsin?uencecorporatepolicydecisions(BertrandandSchoar2003;Dyrengetal.2010;Geetal.2011),includingvoluntarydisclosurechoices(Bamberetal.2010;Brochetetal.2011).OurresultsthatCEOs’forecastingbehaviorbecomesmoreopportunisticwhentheirhorizonbecomesveryshortareconsistentwithpractitioners’surveyresponsesindicatingthatmanagementphilosophyisanimportantdeterminantofforecastingpolicies(NationalInvestorRelationsInstitute[NIRI]2007).Second,weextendthestreamofresearchthatinvestigatestheeffectsoftheCEOhorizonproblembyidentifyinganewimplicationoftheCEOhorizonproblem,opportunisticterminal-yearforecastingbehavior.

SectionIInextdescribespriorliteratureanddevelopsourhypotheses.InSectionIII,wediscussthedataandtheempiricalmodel.SectionIVpresentstheempiricalresultsandSectionVconcludesthepaper.

II.HYPOTHESISDEVELOPMENT

Terminal-YearForecastingBehaviorforRetiringCEOs

Ourprimaryargument,thatretiringCEOsfaceincentivestoissuedeliberatelymisleadingforecastsinanattempttoin?uencestockpricesduringtheirterminalyear,requiresthatCEOs3Forexample,Cheng(2004)providesevidencethat?rmsuseoptioncompensationtomitigatehorizon-problem-inducedreductionsinR&Dspending.

TheAccountingReview

November

2013

1912Cassell,Huang,and

Sanchezbelievethatopportunisticforecastingyieldsatleasttemporarymarketmispricingwherebystockpricesarein?uencedbybiasedforecasts.Considerableempiricalevidenceindicatesthatmanagersemploymisleadingaccountinganddisclosurepolicychoices,suggestingthattheseactivitiesyieldnontrivialbene?ts(EricksonandWang1999;Noe1999;AboodyandKasznik2000;LangandLundholm2000;Louis2004;ChengandLo2006;Brockmanetal.2008;Danieletal.2008;Gongetal.2008).4

WeexpectthatretiringCEOswillbemorelikelytosuccumbtosuchpressuresduringtheterminalyearbecausetheregulationssurroundingtheirstocktransactionsarelessstringentduringthepost-retirementperiod.Speci?cally,SECinsidertradingrulesgenerallydonotrequireretiredCEOstodisclosetheirtradesafterretirement.5Asaresult,post-retirementtradescanbemadebeforetheassociatedearningsarerealizedandwithreducedregulatoryandmarketscrutiny.Thus,relativetoCEOswhowillcontinuewiththeir?rm,retiringCEOsfacestrongerincentivestoin?atestockpricesduringtheperiodleadinguptotheirretirement.

Theprecedingdiscussionformsthebasisforourprimaryhypotheses,thatretiringCEOswillbemorelikelytoengageinopportunisticforecastingbehaviorduringtheirterminalyearwiththe?rmthaninpreviousyears.Speci?cally,wepositthat,relativetopre-terminalyears,retiringCEOswillbemorelikelytoissueforecastsoffutureearnings,andtoissuethemmorefrequently,duringtheirterminalyearandthatsuchforecastswillbemorelikelytoconveynewsthatisviewedpositivelybymarketparticipants(e.g.,investorsandanalysts)andmoreoptimisticallybiased.Weexamineforecastissuancebecauseamanagercannotissuegoodnewsorbiasedforecastsunlessaforecastisissued.Thus,retiringCEOswhohavenotpreviouslyissuedforecasts,butwouldliketoissuegoodnewsorbiasedforecastsintheirterminalyear,mustchangetheirbehavior.6Theprecedingargumentalsosupportsourinterestinterminal-yearforecastingfrequencybecauseforecastfrequencyisin?uencedbyforecastissuance.Ourprimaryhypotheses,statedinthealternativeform,areasfollows:

H1:RetiringCEOsaremorelikelytoissueforecastsoffutureearnings,andtoissuesuch

forecastsmorefrequently,intheirterminalyearrelativetootheryearsduringtheirtenurewiththe?rm.

H2:RetiringCEOs’terminal-yearforecastsoffutureearningsaremorelikelytoconveygood

newsrelativetoprevailingexpectationsthanareforecastsissuedinotheryearsduringtheirtenurewiththe?rm.

H3:RetiringCEOs’terminal-yearforecastsoffutureearningsaremoreoptimisticallybiased

relativetoforecastsissuedinotheryearsduringtheirtenurewiththe?rm.

FactorsthatAffectTerminal-YearForecastingBehavior

Wealsoinvestigateanumberoffactorsthatweexpecttoin?uencetheextentofterminal-yearopportunisticforecastingbehavior.Speci?cally,weexaminetheeffectofCEOincentivestoengage4

5

6TheoreticalworkbyBloom?eld(2002),termedTheIncompleteRevelationHypothesis,predictsanegativeassociationbetweenthecostofinformationextractionandtheextenttowhichmarketpricesre?ecttheinformation.Moreover,HirshleiferandTeoh(2003)arguethatinvestors’limitedattentioncontributestoincompletemarketresponsestopublicinformation,includinginformationconveyedbydeliberatelymisleadingdisclosures.Transactionsexecutedafteranof?cerordirectorrelinquisheshis/herinsiderstatusaresubjecttoRule16andtherelatedSEC?lingrequirementsonlyfortradesthatareexecutedwithinaperiodoflessthansixmonthsofanoppositetransaction,i.e.,anacquisitionfollowingadisposition,orviceversa,wheretheoppositetransactionwasmadewhentheof?cerordirectorheldinsiderstatus.SeeSECRule16a-2(b)formoredetails.Consistentwiththesearguments,we?ndthatapproximately18percentofsampleCEOs(152outof862)issueatleastoneforecastintheirterminalyearbutdonotissueanyforecastsduringpre-terminalyears.

TheAccountingReview

November2013

EvidenceonthePropertiesofRetiringCEOs’ForecastsofFutureEarnings1913insuchbehaviorintermsofthemagnitudeofCEOequityincentives,theincidenceofcutsindiscretionaryexpenditures,andwhethertheCEOretainsapositioninthe?rmafterretirement.7Wealsoanalyzetheeffectofthestrengthofinternalmonitoringintheformofwhetherthe?rmhirestheincomingCEOinternally,whetherthe?rmannouncestheCEO’supcomingretirementinadvance,theproportionofoutsidedirectorsontheboard,andwhethertheCEOalsochairstheboard,aswellasexternalmonitoringintermsofthepercentageofsharesheldbyinstitutionalinvestorsonsuchbehavior.Below,wedevelophypothesesforeachofthesefactorsundertwobroadcategories,CEOIncentivesandInternalandExternalMonitors.

CEOIncentives

Becausethebene?tstotheCEOofin?uencingstockpricesthroughopportunisticforecastingbehaviorareincreasingintheextentofCEOequity-basedincentivesresultingfromoptionandshareholdings,wepositthatsuchbehaviorwillbemorepronouncedamongCEOswithlargeequity-basedincentives.EquityincentivescanalsoencourageCEOstomanipulateearnings(ChengandWar?eld2005;BergstresserandPhilippon2006;Efendietal.2007).Thus,retiringCEOswithlargeequityholdingsfacestrongerincentivestoissueforecastsoffutureearningsandthisbehaviorcouldbene?ttheretiringCEOinpartbecauseSECtradingrulesrelatedtoCEOs’post-retirementsecuritytransactionsarelessstringentthanthoseineffectduringtheirtenurewiththe?rm.

SimilarargumentsapplytoretiringCEOswhoengageinrealearningsmanagementduringtheirterminalyear.ExtantresearchsuggeststhatCEOsadoptmyopicinvestmentandaccountingpolicychoicesintheyearsleadinguptotheirretirement.Consistentwiththeoreticalexpectations(SmithandWatts1982),severalempiricalstudies?ndthatCEOswithshorthorizonsreducebothR&Dspending(DechowandSloan1991;BarkerandMueller2002;Naveen2006)andcapitalexpenditures(Capex)(Xu2010).8Becausethesepolicychoicesarti?ciallyboostcurrentperiodincome,weexpectretiringCEOstoissueforecastsoffutureearningsthatimplythatthecurrentperiodresultsrepresentperformancethatwillcontinueintothefuture.Thus,weexpectthatopportunisticterminal-yearforecastingbehaviorwillbegreateramongCEOswhocutR&DandCapex.

Finally,weexpectthatopportunisticterminal-yearforecastingbehaviorwillbelesspronouncedamongCEOswhoretainapositiononthe?rm’sboardofdirectorsafterretirement.TheseCEOsaremorelikelytobeconcernedaboutthelabormarketcostsassociatedwithopportunisticforecasting,suchasreputationlossand/orthelossoftheirboardposition(Fama1980;GibbonsandMurphy1992).Inaddition,?rmsforwhichformerCEOsremainontheboardhavebetteraccountingperformanceandhigherturnover-performancesensitivityofthesuccessorCEO(Fahlenbrachetal.2011),suggestingthatCEOswhoremainasboardmemberscontinuetohaveaninterestinthe?rm.TheprecedingargumentssuggestthatCEOswhoretainapositionwithinthe?rmwillbelesslikelytoengageinopportunisticforecastingbehaviorintheirterminalyear.Ournextsetofhypotheses,statedinthealternativeform,re?ectthein?uenceofCEOincentivesonterminal-yearopportunisticforecasting:

7

8WeacknowledgethatcutsindiscretionaryexpendituresarenotCEOincentivesperse;rathertheyareexpostobservedactionsfromwhichweareinferringCEOincentives.Evidenceontheassociationbetweenthehorizonproblemandmyopicinvestmentandaccountingpolicychoicesismixed.Anumberofstudieshaveeither:(1)failedto?ndanassociationbetweenthehorizonproblemandmyopicbehavior,or(2)documentedanassociationbetweenthehorizonproblemandmyopicbehavioronlyinspeci?ccircumstances(ButlerandNewman1989;GibbonsandMurphy1992;MurphyandZimmerman1993;Pourciau1993;Wells2002;Cheng2004;Cazier2011).

TheAccountingReview

November

2013