Review of Finance-2013-Ferri-527-63

ReviewofFinance(2013)17:pp.527–563doi:10.1093/rof/rfs003

AdvanceAccesspublication:February24,2012

SayonPayVotesandCEOCompensation:EvidencefromtheUK*

FABRIZIOFERRI1andDAVIDA.MABER2

1

Downloaded from ColumbiaUniversity, and2UniversityofSouthernCalifornia

Abstract.WeexaminetheeffectofsayonpayregulationintheUK.Consistentwiththeviewthatshareholdersregardsayonpayasavalue-creatingmechanism,theregulationÕsannouncementtrig-geredapositivestockpricereactionat rmswithweakpenaltiesforpoorperformance.UK rmsrespondedtonegativesayonpayvotingoutcomesbyremovingcontroversialCEOpaypracticescriticizedasrewardsforfailure(e.g.,generousseverancecontracts)andincreasingthesensitivityofpaytopoorrealizationsofperformance.JELclassi cation:G34,G38,J33,M12

1.Introduction

In2002,theUKwasthe rstcountrytomandateanannualnon-bindingshare-holdervoteonexecutivepay(hereafterreferredtoassayonpay).Theregulations

Wethankananonymousreviewer,GennaroBernile,BrianCadman,BrianChef ns,ShijunCheng,MartinConyon,MerleEderhof,JosephGerakos,JeffreyGordon,YanivGrinstein,KevinJ.Murphy,MarcoPagano(theeditor),DhinuSrinivasan,LauraStarks,RandallThomas,anddiscussantsandparticipantsatthe2007ShareholdersandCorporateGovernanceConferenceattheSaidBusinessSchool,the2007IMOConferenceattheHarvardBusinessSchool,the2008NationalBureauofEconomicResearch(NBER)meeting,the2008InternationalAccountingSymposiumatColumbiaUniversity,the2008LondonBusinessSchoolAccountingSymposium,the6thInternationalCorpo-rateGovernanceConferenceattheUniversityofBirmingham,the2009MFAConference,the2009AAAManagementAccountingSectionMeeting,the2ndAnnualCorporateGovernanceConferenceatDrexelUniversity,theSixthAnnualNapaConferenceonFinancialMarketsResearch,the2009EFMSymposium,the2009ConferenceonTheFutureofSecuritiesRegulationattheUniversityofNotreDame,the2009FIRSConferenceandworkshopsatINSEAD,theUniversityofMaryland,andtheLondonSchoolofEconomicsfortheircomments.WeespeciallythankSidBalachandranforhishelpintheearlystageoftheproject.Thepaperbene tedfromconversationswithCarolBowie(Risk-metrics),StephenDavis(MillsteinCenterforCorporateGovernance),RichardFerlauto(AFSCME),andSarahWilson(Manifest),andfromtheexcellentresearchassistanceofCarmeloTringali.TheauthorsacknowledgeagrantfromtheMillsteinCenterforCorporateGovernanceattheYaleSchoolofManagementandresearchsupportfromColumbiaUniversity,theUniversityofSouthernCalifornia,andtheHarvardBusinessSchool.Allerrorsareourown.

ÓTheAuthors2012.PublishedbyOxfordUniversityPress[onbehalfoftheEuropeanFinanceAssociation].Allrightsreserved.ForPermissions,pleaseemail:journals.permissions@http://wendang.chazidian.com

http://wendang.chazidian.com/ at Zhongnan University of Economics and Law on October 25, 2014

528F.FERRIANDD.A.MABER

markedthe rstinstanceofanon-bindingadvisoryvoteinUKcompanylaw(Chef nsandThomas,2001)andweretoimprovethe‘‘accountability,transpar-ency,andperformancelinkageofexecutivepay’’(BairdandStowasser,2002)afteradecadeofgrowingconcernswith‘‘fatcat’’payandrewardsforfailure(e.g.,FinancialTimes,1995,1998).Sayonpaysubsequentlygarneredconsiderablein-terestfrominvestorsandpolicymakersandhasbeenadoptedinmanycountries,includingtheUSA(in2011).1

Themerits(orlackthereof)ofsayonpayhavebeenhotlydebated.Proponentsarguethatenhancedshareholdervoice,asformalizedinasayonpayvote,andreputationconcernswillhelpboardsovercomepsychologicalbarrierstonegotiat-ingwithCEOsonbehalfofshareholders,resultinginmoreef cientcompensationDownloaded from contracts(e.g.,Bebchuk,2007).Criticscounterthatsayonpayvoteswillbeig-noredatbest(sincetheyarenon-binding)and,atworst,willcausedirectorstopandertoill-informedshareholders,ultimatelyresultingintheadoptionofsubop-timalpaypractices(Kaplan,2007;Bainbridge,2008).Butempiricalevidenceisscant.ThisstudyexaminestheUKexperiencewithregardtotheeffectsofsayonpayonshareholdervalueandexecutivepaypractices.

Weperformthreesetsofanalyses.First,weexaminethemarketreactiontotheannouncementofsayonpayregulation.Accordingtothebusinesspress(e.g.,FinancialNews,2002),thesubmissionofsayonpayregulationstotheUKParliamentonJune25,2002waslargelyunanticipated,http://wendang.chazidian.comingstandardeventstudymethodologies,we ndpositiveabnormalreturnsfor rmswithexcessCEOpay(rel-ativetothelevelpredictedbyitseconomicdeterminants),especiallyininstancesinwhichexcessCEOpayiscombinedwithpoorperformance(i.e., rmsthatrecentlyprovidedlargepayoutsdespitepoorperformancerealizations).Wealso ndpositiveabnormalreturnsfor rmswithgenerousseverancecontracts,oftencriticizedasaformofrewardforfailure,thatweakenpenaltiesintheeventofpoorfutureperformance.These ndingsareconsistentwithshareholdersperceivingsayonpayasavalueen-hancingmonitoringmechanismfor rmswithweakpenaltiesforpoorperformance.Next,wetestwhetherchangestocompensationcontractsdisclosedin rmsÕremunerationreportsweretheresultofsayonpaybycomparingchangesmadebyseventy- vehighdissent(HD) rms(i.e., rmsthatexperiencedmorethan20%votingdissent)beforeandafterthe rstsayonpayvote,withchangesmadebyamatchedsampleoflowdissent rms(i.e., rmsthatexperiencedlessthan5%votingdissentbutwithotherwisesimilarcharacteristics).

OnJuly21,2010,PresidentObamasignedintolawtheDodd–FrankWallStreetReformandCon-sumerProtectionAct,oneprovisionofwhichmandatesthat,beginningin2011,publiclytraded rmsallowshareholdersanon-bindingvoteonexecutivepayatleastonceevery3years.Fordetailsabouttheadoptionofsayonpayinothercountries,seeInstitutionalShareholderServices(2007).

http://wendang.chazidian.com/ at Zhongnan University of Economics and Law on October 25, 2014

SAYONPAYVOTESANDCEOCOMPENSATION529

Ouranalysesindicatethatshareholdersviewedgenerousseverancecontractsasthemostcontroversialpaypractice.Inparticular,theyusedsayonpaytopressure rmstoreduceto12monthsnoticeperiodsthatexceeded12months(implyingseverancepaymentsgreaterthan1yearÕscompensation).Among rmswithsuchnoticeperiods,thepercentageofHD rmsthatreducedthemafterthe rstvote(80.0%)issigni cantlyhigherthanbeforethe rstvote(20.0%)andalsosignif-icantlyhigherthanamonglowdissent rmsafterthe rstvote(33.3%).Thispatternsuggeststhatreductionofnoticeperiodswasdrivennotbyageneraltrendbutratherbythesayonpayvote.ThehighrateofresponsivenessamongHD rmsisparticularlynoteworthybecausethese rmswerelikelythemostreluctanttochangecontroversialpaypracticesinthepast(otherwisetheywouldnothaveex-Downloaded from periencedHD).Interestingly,70.0%oflowdissent rms(comparedto20.0%ofHD rms)reducedtheirnoticeperiodsduringtheyearbeforethevote,suggestingthatproactivepre-votechangeshelpedtoavoidvotingdissent.Consistentwiththisconjecture,wealso ndthatHD rmsthatreducedtheirnoticeperiodswererewardedbya23.9%decreaseindissentatthesecondsayonpayvote(fromanaverageof34.3%atthe rstvote),whereasthefew rmsthatignoredthe rstvotewerepenalizedbya12.2%increaseindissentatthesecondvoteandthenallofthemreducedthenoticeperiodafterthesecondvote.Inessence,bytheendofthesecondsayonpayseason,theex-anteandex-posteffectsofsayonpaycombinedtoeliminatetheuseofnoticeperiodslongerthan1year.

Thesecondmostaffectedcompensationitemwasperformance-basedvestingconditionsinequitygrants,namely,so-calledretestingprovisions.Retestingpro-visionsprovideforreevaluatinginsubsequentyearsperformancetargetsnotachievedduringtheinitialmeasurementperiod(ratherthanallowingtheoptionstolapse)and,asaresult,areoftencriticizedforthepotentialtorewardfailure.We ndthat76.3%(5.0%)ofHD rmsand28.0%(25.0%)oflowdissent rmswithretestingprovisionsshortenedorremovedthemafter(before)the rstsayonpayvote.Thus,asfornoticeperiods,itappearsthatthesecontractualchangeswerethedirectresultofsayonpay.

Whiletheforegoinganalysisprovidesdirectevidenceoftheeffectofsayonpay,itmayunderstatethateffectbecauseitexaminesonlychangesindisclosed,andthusreadilyobservable,provisionsofcompensationcontracts.ImportantelementsofCEOpay(e.g.,meritraises,bonuses,andthesizeofequityawards)involveconsider-ablediscretion,http://wendang.chazidian.com-ingalargesampleofUK rms,we ndasigni cantincreaseinthesensitivityofCEOpaytopoorperformance,whiletherelationshipbetweenpayandothereco-nomicdeterminantsremainsunchanged.Itthusappearsthatsayonpayhas

http://wendang.chazidian.com/ at Zhongnan University of Economics and Law on October 25, 2014

530F.FERRIANDD.A.MABER

amoderatingeffectonthelevelofCEOcompensationonlyconditionaluponpoorperformance.Wefurther ndtheincreaseinthesensitivityofCEOpaytopoorper-formancetobemostpronouncedinHD rmsand rmscharacterizedashavingex-cessCEOpaybeforetheadoptionofsayonpay,twoproxiesforcontroversialCEOpaypractices.Wedonot ndasimilarincreaseinacontrolsampleofUK rmsnotsubjecttosayonpay(speci cally,UK rmstradedontheAlternativeInvestmentMarket).Takentogetherwithourevidenceofexplicitchangestocompensationprac-tices,theseadditionaltestssupportacausalinterpretationofour ndings.

Overall,ourtestssuggestthatUKinvestorsperceivedsayonpaytobeavalueenhancingmonitoringmechanismandweresuccessfulinusingsayonpayvotestopressure rmstoremovecontroversialpaypracticesandincreasethesensitivityofDownloaded from paytopoorperformance.

Ourstudycontributestotheliteratureonsayonpayintwoways.First,andmostimportant,itisthe rststudytoexaminetheimpactofsayonpayon rmsÕcom-pensationpractices,priorresearchhavingfocusedonthemarketreactiontosayonpay-relatedevents(CaiandWalkling,2011;Larckeretal.,2011).Second,oureventstudyprovidesnewevidenceoninvestorperceptionofsayonpay.Thisisimportantgiventhemixed ndingsinpriorUS-basedresearch.Forexample,whereasCaiandWalkling(2011) ndthestockpricesof rmswithexcessCEOpaytorespondpos-itivelytotheUSHouseofRepresentativesÕpassageofasayonpaybillin2007,usingslightlydifferentmethodologies,Larckeretal.(2011)reportnoabnormalreturnsaroundthesameevent(andnoornegativeabnormalreturnsaroundrelatedregulatoryevents).Thissensitivitytodifferentspeci cationsmayre ectthelowpowerofthesetting.ThepassageofasayonpaybillbyaDemocratic-controlledHousemayhavehadonlyamarginalimpactonthelikelihoodofsayonpayregulationbecausetheSenatewasRepublican-controlledandtheBushWhiteHouseopposedsayonpay(AssociatedPress,2007;NewYorkTimes,2007).TheUKgovernmentÕsdecisiontosubmitsayonpayregulationtoParliamentbecauseitrepresentsamarkedandlargelyunexpectedincreaseinthelikelihoodofsayonpayregulation,arguablyprovidesamorepowerfulsettingforaneventstudy.

Morebroadly,ourstudycontributestotheliteratureontheeffectofshareholdervoice.Earlystudiesconcludedthatshareholderproposalsaregenerallyaweakgovernancemechanism(forareviewoftheevidence,seeKarpoff,2001andRomano,2001).2However,morerecentresearchsuggeststhattheyhavebecome

2

Withrespecttocompensation,anexceptionisThomasandMartin(1999),whoreportasigni cantlylowerincreaseinpaylevelsfor rmstargetedbycompensation-relatedshareholderproposalsoverthe2-yearperiodsubsequenttotheproposalrelativetoasize/industrymatchedsample.Otherstudiesofcompensation-relatedactivismhavefocusedonmanagement-sponsoredcompensationplans,exam-iningthemarketreactiontoandshareholdersÕvotesontheseplans(ThomasandMartin,2000;MorganandPoulsen,2001;Morganetal.,2006).

http://wendang.chazidian.com/ at Zhongnan University of Economics and Law on October 25, 2014

SAYONPAYVOTESANDCEOCOMPENSATION531

increasinglyeffectivebutmostlywhentheyachievemajorityvotesupport(ThomasandCotter,2007;Ertimuretal.,2010).Forexample,Ertimuretal.(2011) ndthatbetween1997and2007boardsimplement40.3%(3.8%)ofcompensation-relatedshareholderproposalswinning(failingtowin)amajorityvote.Instarkcontrast,we ndthatasayonpaydissentvotehigherthan20%(wellbelowamajorityvote)resultsinboardsimplementing75%–80%ofshareholderrequeststoremovespe-ci cprovisions.This gureiscomparabletothesuccessrateofhedgefundactiv-ism.3Moreover,our ndingthat rmsrespondtothemerethreatofvotingdissenthighlightstheimportanceofaccountingforex-anteeffectswhenstudyingtheim-pactofshareholdervotes.

Thishighrateofresponsivenessmayre ectthefactthatsayonpayvotesareDownloaded from morelikelytogeneratereputationconcerns(relativetoshareholderproposals)be-causetheydirectlyquestiondirectorsÕchoices.Thisisaneffectivethreatbecause‘‘noinsurancepolicyformanagersordirectorscanprotectthemfromsuchrepu-tationalpenalties’’(DyckandZingales,2002).Inthissense,ourresultsarecon-sistentwiththeclaimthatpublicexpressionsofoutrage,suchassayonpayvotes,areperhapstheonlyeffectiveconstraintonCEOpaygivenboardsÕincentivestoacquiescetoCEOrequestsandthelimitedeffectivenessofothermechanisms(Beb-chukandFried,2004).Itisalsopossiblethatsayonpayvotesaremoreeffectivebecause,ratherthanforceinvestorsandboardstoagreeonaspeci cshareholderproposal,theyspurabroaderdialogaboutexecutivepay.Intheserespects,sayonpayvotesaresimilarinspirittovote-nocampaigns,whichpriorresearchhasshowntobeanincreasinglyeffectiveactivismtool(DelGuercioetal.,2008;Caietal.,2009;Ertimuretal.,2011).

Lastly,ourstudycontributestotheresearchontheeffectofmonitoringmech-anismsonexecutivepay4andcomplementsabroaderliteraturethatexamineshowexecutivepayis5affectedbyinstitutionalarrangementsthataltertheallocationofdecisionrights.Inthissense,our ndingthatsayonpayisassociatedwithsteeperpenaltiesforpoorperformanceandgreatershareholdervalueindirectlyprovidesinsightsintohow(andhowmuch)extantcompensationpracticesmaybede cient

Bravetal.(2008)andKleinandZur(2009) ndUShedgefundactiviststoachievetheirgoals(fullyorpartially)inabout60%–70%ofcases,andBechtetal.(2008) ndthesuccessrateoftheHermesUKFocusFundtorangebetween44%and100%dependingonthestatedobjective.4

Priorstudieshaveexaminedtheeffectofinstitutionalownership,boardindependence,hedgefundactivism,disclosure, nancialreporting,taxrules,andmediacoverage(HarrisandLivingstone,2002;HartzellandStarks,2003;Almazanetal.,2005;Carteretal.,2007;Bravetal.,2008;Coreetal.,2008;Go¨x,2008;ChhaochhariaandGrinstein,2009;Dikollietal.,2009;FerriandSandino,2009;Grinsteinetal.,2011;Guthrieetal.,2010).5

Researchinthisareahasexamined rmsthatgoprivatewithaprivateequitysponsor(CronqvistandFahlenbrach,2011), rmswithindependentblockholdersontheirboards(AgrawalandNasser,2010),andprivatelyheld rms(Gaoetal.,2011).

http://wendang.chazidian.com/ at Zhongnan University of Economics and Law on October 25, 2014