nancedcapital

Boards,CEOEntrenchmentandtheCostofCapitalJamesDowLondonBusinessSchoolNovember2012AbstractExistingresearchonCEOturnoverfocusesonCEOability.
Thispa-perarguesboardabilityisalsoimportant.
CorporateboardswillbereluctanttoreplaceCEOsasthismakes…nancingexpensivebysendinganegativesignalaboutboardability.
Thisdi¤ersfromexistingliterature:entrenchmentdoesnotresultfromCEOpower,noragencyproblems.
En-trenchmentismitigatedwhentherearemoreassets-in-placerelativetoinvestmentopportunities.
Thepaperalsocomparespublicandprivateequityownership.
PrivateownershipeliminatesCEOentrenchment,butasharepriceguidesinvestmentdecisions.
Finally,themodelimpliesthatboardchoiceinpubliclylisted…rmswillbeconservative.
"ThesupremeironyofbusinessmanagementisthatitisfareasierforaninadequateCEOtokeephisjobthanitisforaninad-equatesubordinate.
.
.
.
Iftheboardmakesamistakeinhiring,andperpetuatesthatmistake,sowhat"WarrenBu¤ett,1988LettertoShareholdersWhendoesaboardreplaceanunderperformingCEOAnestablishedbodyofresearchstudiesthepropositionthatCEOsareoverlysecureintheirjobs(includingShleiferandVishny(1989),BebchukandFried(2003),GuembelandCasamatta(2010)andotherpapersdiscussedbelow).
IndirectevidenceforsuchentrenchmentistobefoundinempiricalstudiessuggestingthatCEOsoftentakecapital-allocationandacquisitiondecisionsthatarenotshareholder-valuemaximizing(Rajan,ServaesandZingales(2000),LangandStulz(1994),Schoar(2002);ifshareholderscouldcontroltheircompaniese¤ectively,theywould…reCEOswhodon'tmaximizevalue.
Asaresulttakeovers,despitetheirenormoustransactionscosts,areoftenviewedasamechanismforreplacingtopmanage-ment(ShleiferandVishny(1989),GrossmanandHart(1980)).
TherearealsoIthankPatrickBolton,ElenaCarletti,PaulCoombes,FrancescaCornelli,JulianFranks,VitoGala,PieroGottardi,ChristopherHennessy,IgorMakarov,AilsaRoell,LukeTaytor,PaoloVolpinandseminarparticipantsatEUI,LBSandStockholmSchoolofEconomicsforcomments.
1empiricalstudiesthatattempttomeasureentrenchmentdirectly:Taylor(2010)estimatesthatboardsbehaveasif…ringtheCEOhasadeadweightcostof5.
9%ofthe…rms'assets.
ExistingresearchonCEOturnoverandperformancefocusesonCEOability(e.
g.
MikkelsonandPartch(1997),JenterandKanaan(2008).
Thispaperarguesthattheabilityoftheboardisalsoimportant.
Inthispaper,ImodelentrenchmentgiventhattheCEO…ringdecisionistakenbytheboard,whoremaininplaceaftertheCEOisgoneandwhocareabouthowthemarketperceivestheirdecision.
TheycarebecauseitisnotonlyCEOswhodi¤erinability.
Boardsalsodi¤er.
Therefore,thebadnewsthatisconveyedby…ringdoesn'tjustattachtothedepartedCEO.
1Thisbadnewshasimplicationsfortheabilityofthe…rmtoraiseexternal…nance.
Knowingthis,theboardwillbereluctantto…retheCEO,evenintheabsenceofanyboardagencyconicts.
Toillustrate,considertheexampleofasmallcompanywithhighgrowthpotentialthatwillneedlargeamountsofadditional…nancing.
SupposetheboarddiscoversthattheCEOisunderperforming:hemaybeneglectinghisjob,orhemightdisplaypoorjudgement–inanextremecasehecouldevenbeguiltyoffraud.
Thegutresponseisto…re.
ButthemarketwilllikelyconcludethattheinitialbadnewsthattheCEOneedstobereplacedhasnegativeimplicationsforthecompany'sfutureafterhehasleft.
Afterall,iftheboardappointedthewrongCEOonce,maybetheywilldosoagain.
The…ringdecisionisnotjustareectiononthedepartingCEO,itreectsbadlyontheboardthemselves,henceonthe…rm'sprospects.
2Ifthe…rmneedstoraiseexternalfunding,thetermsofthe…nancingwillbeworse.
Intuitively,theoutcomemaybethatthetermsaresodisadvantageousthatitisnotpossibleforthe…rmtoraisefundingfortheproject.
Inthemodel,thisisinfacttheonlypossibleoutcomeinequilibriumforasmall…rmwithnoassetsinplace.
Foralarger…rmwithexistingassets,therewillbeatrade-o¤becauseassetsinplacemitigatethelemonsproblemofissuingnewequityfollowingCEOreplacement.
Intheexistingliterature,anumberofdi¤erentexplanationshavebeengivenforCEOentrenchment.
First,CEOsmaybeabletochooseprojectstacticallysoastomakethemselvesdiculttoreplace(ShleiferandVishny(1989)).
Second,itmaybethatshareholderssimplydon'thavecontrolovercontractingwiththeCEO,asisassumedinprincipalagenttheory.
BebchukandFried(2003)arguethatahypothesisof"ManagerialPower"maybeabetterdescriptionofcorporategovernancethanprincipal-agenttheory(forexample,BebchukandKamar(2010)…ndevidencethatmanagementcanmanipulatethecorporatecharteratshareholders'expense).
Third,incertainprincipal-agentproblemsitcanbeoptimalfortheprincipaltocommittoentrenchingtheagent,and1SeealsothediscussionofHermalinandWeisbach(1998)below.
2Inthisexample,reputationattachesspeci…callytotheboard.
Thisnarrowinterpretationisnotnecessary.
Onecouldequallyinterpetthe…ringasanegativesignalabout"the…rm"-anorganizationthatallowsanincompetentCEOtobeappointedmaybedefectiveinotherways,forexample,iftheCEOisaninternalhire,itsendsabadsignalbecauseittellsusthataCEOwhoturnedouttobeincompetentwasabletothriveandbepromotedwithintheorganization.
Themainfocusofthispaperisonboards,however.
2theowner-CEOrelationshiphasparticularfeaturesthatmakethisapply(forexample,GortonandGrundy(1996),DowandRaposo(2009),GuembelandCasamatta(2010)).
Thispaperalsoarguesthatentrenchmentcanariseinequilibrium,buttheexplanationheredi¤ersbybeingspeci…ctoasituationwiththeCEO,theboard(or…rm),andthemarket.
Theexplanationdependsonthesignalgiventothemarketbythe…ringdecision,andhowthisfeedsbackintotheoptimal…ringrule.
Incontrast,GortonandGrundy(1996),DowandRaposo(2009),andGuembelandCasamatta(2010)giveexplanationsthathaveanaturalapplicationtoCEOs,butcouldapplytootheremployeeswithoutaninteractionwithmarket…nancing.
Iarguethat…ringsendsanegativesignalaboutboardability,butanalter-nativeargumentisputforwardbyHermalinandWeisbach(1998).
Theymaketheoppositeassumption:thatsomeboardsaremoreindependentandmoreaggressiveatmonitoring,inthesensethattheyhavealowercostofmonitoring.
Intheirmodel,…ringisagoodsignal.
Finally,whilemostoftheliteratureviewsentrenchmentaseitherex-postcostlyorex-antesuboptimal,Fisman,KhuranaandRhodes-Kropf(2004)ar-guethatentrenchmentisvaluablebecauseitpreventsCEOsfrombeingsub-optimally…redbymisguidedshareholders.
Animportantquestionin…nancialeconomicsis:whatkindsof…rmsshouldhaveequitythatispubliclylisted,andwhichshouldbeprivatelyownedThispapershowsthatseparationofcontroland…nancinghasacostbecauseoftheongoingneedtosendpositivesignalstothemarketsthatprovide…nance.
Adeep-pocketsownercanavoidthiscost.
PrivateownerswithsecurefundingcantakedecisionsonCEOreplacementwithoutworryingaboutthesignal.
Henceoneimplicationofthisanalysisisthatprivatelyownedand…nanced…rmswillbemorewillingto…retheirCEOs.
3Whiletheboardofapublic…rmwilloftenretainaCEOtheyknowtobeworsethanthereplacement,aprivate…rmwillsimplyreplacetheCEOwheneverthereplacementseemsbetter.
CornelliandLjunqvist(2011)documentthatwhenaprivateequity…rmappointsaCEO,theaveragetenureissigni…cantlyshorterthanforapubliccompany.
Butprivatecompaniesaren'tnecessarilysuperior:themarketmayproduceinformationthattheCEOdoesnothave.
Ianalyzethistrade-o¤todeterminewhenacompanythatisprivatelyownedand…nancedissuperiortoapubliccorporation.
Oneimplicationisthat…rmswithvaluableassetsinplaceare,otherthingsequal,lesslikelytochooseprivateequity.
Thestructureofthispaperisasfollows.
Insection(1)Isetoutthemodel,and,asabenchmark,derivethe…rst-best…ringrule.
Section(2)statesthemainresult:theboardwillnot…retheCEOunlesstheirestimateofhisabilityfallsbelowathresholdthatisstrictlylowerthanthe…rst-bestthresholdfor…ring.
Section(3)comparespublicandprivatecompanies.
3Someprivateequity…rmsmaybeconcernedthatthedecisiontakenonanindividualinvestmentwilla¤ecttheabilityofthefundtoattract…nancing,Hence,thisargumentwillnotapplytosmallPE…rmsthastdonothaveanestablishedtrackrecord.
3Section(4)concludeswithadiscussionofassumptions,resultsandempiricalimplications.
Ishowthat…rmswillbebettero¤iftheyareconservativeinmakingboardappointments:boardmemberswithextensiveexperiencemaybepreferredoverpeoplewithlittlepriorexperiencewhoseskillasboardmembersappearshigh,butisnotknownwithmuchcertainty.
Ithenshowthatthemodel'sresultsarerobustwithrespecttoacoupleofalternativeassumptions.
Alternativeboardobjectivefunctionssuchasempirebuildingorcareerconcernsyieldsimilarresults.
Thisisnotsurprising,whilethemainresultofthepaper–thatentrenchmentandanex-postsuboptimal…ringrulecanresultfromaboardthatmaximizesshareholdervalue–islessimmediate.
Similarresultsalsofollowifweallowthevalueofassetsinplacetobecorrelatedwithnewprojectpayo¤(theyareassumeduncorrelatedinthemainanalysis),althoughtheanalysisismorecomplex.
Next,Idiscussthechoicebetweendebtandequity,whilethemainpartofthepaperassumesequity…nancingonly.
FinallyIdiscussanumberofempiricalimplicationsofthemodel,inparticularontherelationshipbetweenCEOentrenchmentandinvestment.
1TheModelIassumethatCEOsareappointedand…redbyaboard,and…nanceforanewprojectisraisedfromthemarketintheformofnewequity.
Thesequenceofeventsisasfollows:First,the"old"CEOisappointed.
Second,theboardreceivesasignalabouttheCEO'sability.
ThissignalpistheprobabilitythattheprojectwillsucceedwiththeoldCEO.
piscontinuously-distributedwithE(p)=12andhassupportequalto[0;1].
Third,theboarddecideswhethertocontinuewiththeoldCEO,or…reandreplacetheCEO.
ThechanceofsuccesswiththenewCEOispositivelycorrelatedwiththeoldCEO'sability.
Iassumetheboard…resifthesignalabouttheoldCEOfallsbelowathresholdp0.
Thisincludesasspecialcasesp0=0(theCEOisnever…red)andp0=1(theCEOisalways…red).
Fourth,the…rmtriestoraisemoneyfromthemarketto…nancethecostKofinvestingintheproject.
Iftheinvestmentismade,itis…nancedbyissuingnewequityatfairvalue,giventheinformationavailabletothemarket(i.
e.
themarketforexternal…nancingiscompetitive).
Finally,thepayo¤ontheprojectisrealized.
Thisiseither0(failure)or1(success).
The…rmalsohasassetsinplacewhosepayo¤Visrealizedatthattime.
Sincetheterminalvalueofthe…rmtakestwopossiblevaluesthe…nancingcouldequivalentlybedescribedasdebt.
IassumetheassetsinplaceVareilliquidandnon-veri…ablesotheycannotbeusedto…nancetheproject(forexamplebybeingusedascollateral).
4IassumethatVisaconstant;Appendix(6.
3)containsageneralizationtothecasewhereVisarandomvariablewhosepayo¤smaybecorrelatedwiththenewproject's.
4ThisisalsoassumedinMyersandMajluf(1984).
4Chanceofsuccesswithnewmanager.
Boards'abilitiestopickmanagersvaries,sothemarketmakesinferencesabouttheirabilityfromtheirpastde-cisions(analternativeinterpretationcouldbethataboardthatpickedabadmanagerisweakandthiswilldirectlylowerexpected…rmvalue).
So,thechanceofsuccessofthenewCEOisincreasinginthechanceofsuccessoftheoldCEO.
Tobespeci…c,Iassumethatif…red,thechanceofsuccesswiththenewCEOisPr(NewCEOSucceedsjOldCEOwouldsucceed)=12Pr(NewCEOSucceedsjOldCEOwouldfail)=2:(1)forsomeparameterpKorp0orp>K21Conditionalontheboard'ssignalofCEOabilitybeingbelowaverage,theexpectedvalueofinvestingintheproject(andreplacingtheCEO)isE(pjp12asshownin(6)and(7).
Writtenmorecompactly,thisismaxf0;p(1)+2K;pKgandtheex-antepayo¤(beforelearningp)is:Emaxf0;p(1)+2K;pKg:(8)Example1Valueoftheprojectunder…rst-bestpolicywhenpisuniformlydistributed,pU[0;1]:7Ifpisuniformtheex-antepayo¤is:Z10maxf0;x(1)+2K;xKgdx=Z12K21x(1)+2Kdx+Z112xKdx=(4K3)+4(K1)28(1)(9)Condition(2)reducesto(1+)14p0for0p0);sincethisexceedstheunconditionalexpectationE(p)themarketwillbewillingto…nancetheprojectandtheequityissuancerequired,KE(pjp>p0);willbelessthanintheequilibriumwheretheCEOisalwaysretained(equation(10)).
Whenheis…red,themarket'sexpectationofthechanceofsuccessisE(2+p(1)jpp0).
Iftheboardisindi¤erentatp0between…ringandretainingtheCEO,itmustbethat…ring,whichismorecostlyintermsofdilution,iso¤setbyahigherchanceofsuccess:(2+p0(1))(1F)=p0(1R);(12)whichimpliesp00:TheCEOis…redbytheboardiftheirsignalofhisabilityfallsbelowathresholdp0;weneedtostudyhowp0isdeterminedandwhethertheprojectgets…nancediftheCEOis…red.
Iftheprojectisfunded,theninequilibriumthemarketcontributesfundingKinexchangeforafractionofequitywhoseexpectedvalueequalsthefundingraised:(V+Pr(Success))=Kwheretheprobabilityofsuccessisupdatedbythemarket,dependingonwhethertheCEOhasbeen…red.
IftheCEOisretained,themarketinfersthatp>p0;sothemarket'sestimateofthechanceofsuccessisE(pjp>p0):IftheCEOhasbeen…red,thiswillmakethemarketmorepessimisticaboutthenewprojectanditwilldemandahigherequitysharetocompensate.
IftheboardobservesasignalpthechanceofsuccesswithareplacementCEOis2+p(1);sothemarket'sestimateofthechanceofsuccessfollowinga…ringis2+E(pjpx)(13)ExE(pjpR),thechanceofsuccessfol-lowing…ring,2+p0(1);mustbecorrespondinglyhigher,2+p0(1):ThisimpliesthatthesignalabouttheoriginalCEOisbelowaverage,p0K)andstrictlylessthan1(sincefisdecreasingandf(0)0(infactasIhaveassumedthemarketiscompetitive,Y=0):Sincethe…rmiswillingtoraise…nancefortheproject,conditionalontheevent,itmustbethatX>0(otherwise,itcouldsimplystoptryingtoraisefundingconditionalonthisevent).
Therefore,wemusthaveZ=X+Y>0:(Theno-tradeliteratureassumesthattradeswon'ttakeplaceifitiscommonknowledgetheyarenotParetoimproving.
Becauseofthesimpleinformationalstructureofthismodel,ifallagentsknowthetradeisnotParetoimproving,itwillbecommonknowledgethatitisnotParetoimproving.
)11Now,in…rstbesttheCEOwouldbereplacedwheneverp0hencep0de…nedbytheintersectionf(x)=xmustalsobestrictlyincreasinginV.
Thefunctionfalsoprovidesanalgorithmfor…ndingthesolution:givenpn,de…nepn+1=f(pn).
Wecansummarizetheabovediscussioninthefollowingresult:Proposition3Assumetheboardactstomaximizethevalueofexistingequity.
TheCEOwillbe…redifhisabilitypfallsbelowathresholdp0.
Therewillbenoinvestmentfollowinga…ring.
p0isuniquelydetermined.
TheCEOwillbe…redlessoftenthanin…rstbest:p0p0)(E+p0K)(26)Inparticularifpisuniformthevalueis:V+(1p0)(1+p02K)=V+12KKV+(V+12)(K12V+rV2+(K12)2+V)(27)wherep0isgivenin(24).
The…rstbestvaluewithuniformpis,asshownin(9),(4K3)+4(K1)28(1):(28)133Publiclylisted…rmsversusprivateequity.
Theseparationbetween…nancing(ownership)andthe…ringdecision(control)hasnegativeconsequences.
Butmarketsalsohavebene…ts:inparticular,themarketmayhavetheadvantagethatitcanproduceadditionalinformationabouttheproject(Hayek(1945)).
6Sopubliclylisted…rmshavebothadvan-tagesanddisadvantagesoverprivately…nanced…rms.
Aprivate…rmdoesnothavetoworryaboutthesignalssentbyitsdecisions,butdoesnotbene…tfromtheinformationproductiongivenbythestockmarket.
Wecanexplorethistrade-o¤byadaptingthemodeldevelopedabove.
Thequestionofwhentogopublicandwhentoremain(orgo)privateseemsimportant,butthereisonlyarelativelysmallliteratureonthechoicebetweenpublicandprivate(stockmarketlisted)ownership.
PaganoandRoell(1998)consideratrade-o¤betweenliquidityandmonitoring.
Benninga,Helmantel,andSarig(2003)considerthetrade-o¤betweenthehighervaluationsofpublic…rmsasaresultofdiversi…edoutsideinvestors,andprivatebene…tsofcontrolofprivatelyowned…rms.
Theassumptionherethatthemarketprovidesasignalwhichincreases…rmvalueisrelatedtotheliteratureonhowstockpricescanimproveinvestmentdecisionsandhenceincrease…rmvalue.
Leland(1992),BootandThakor(1997),andDowandGorton(1997)areamongthetheoreticalanalysesofthisfeedbacke¤ect,whichhasbeendocumentedempiricallybyBaker,Stein,andWurgler(2003),Luo(2005),andChen,Goldstein,andJiang(2007)).
Therearedi¤erenttypesofprivatelyowned…rms.
VCfundedstart-ups,orpreviouslylisted…rmsthataretakenprivate,maybemanagedwiththeob-jectiveofraisingadditional…nancing,resellingorlistingthe…rm.
Whilesomeprivateequityfundsarewell…nanced,othersneedtosecureongoingadditionalfunding.
Thiswilla¤ecttheowners'incentives,makingthemcaremoreaboutmarketperception.
Here,Isimplyportraytheprivatelyowned…rmasbeingunconcernedwithsignalssenttopossiblefutureowners/…nanciers.
Thisisap-propriatewheretheownerhasalongtimehorizonanddoesnotneedadditionaloutside…nancingwithinthathorizon.
Icompare:Apublic…rm:asmodelledabove,butwhereanadditionalsignalisgeneratedbythemarketwhenthe…rmraisesexternal…nance,withAprivate…rm:wheretheownerprovides…nanceandtakesthe…ringdeci-sion,butdoesnotreceivetheadditionalmarketsignal.
While,inreality,marketsaggregatethesignalsofmanydispersedagents,forconvenienceImodeltheaggregatesignaldirectly.
Thiscanbeviewedastheaverageofmanyindividualsignals.
Iwillassumethatwithprobabilityq,themarketlearnstheactualoutcomebeforeinvesting.
Allotherassumptionsareunchangedfromtheprevioussection.
Followingthedecisionto…reorretaintheCEO,ifthemarketlearnsthattheprojectwillsucceed,the…rmwillissue6Otheradvantagesincludehavingastockpricetomotivatemanagers(HolmstromandTirole(1993)).
14anequityshareof=K1+V:Ifthemarketlearnstheprojectwillfail,therewillbenoinvestment.
Ifthemarketdoesnotreceiveasignal,themarketwilldecidewhethertoinvestbasedonthe…ring/retainingdecision,exactlyasinthepreviousanalysis.
Usingthesameargumentsasbefore,retainingtheCEOwillleadtoinvestmentbuttherewillbenoinvestmentfollowing…ring.
Theequityissuancewillbeasbefore,R=KV+E+p0(see(16)).
Todeterminep0;againwerequireindi¤erenceatp0:IftheCEOis…red,theboard'spayo¤isqp0(V+1K)+(1q)V:(29)Thisisbecauseifthereisasignal(probabilityq),thechanceofagoodsignalisp0:Inthatcasetheoldequitygetsallthesurplusfromtheinvestment(1K)aswellasthevalueVofassetsinplace.
Ifthereisnosignal,thereisnoinvestmentandnonewequityissuedsotheequityisworthV.
Similarly,iftheCEOisretainedtheboard'spayo¤is:qp0(V+1K)+(1q)(1R)(V+p0)(30)Equatingthepayo¤satp0;(29)and(30),wehavethesameconditionasbefore,equation(18).
Itfollowsthatp0isunchangedfromtheanalysisintheprevioussection,asgivenbyequation(19).
Thisallowsustocomputetheex-antepayo¤stothe…rmunderpublicandprivateownership.
PrivateEquity:Withprivateequity,theex-antepayo¤onthenewproject,includingthevalueofassetsinplace,is:V+Emaxf0;2+p(1)K;pKg(31)PubliclyListedEquity:Withpublicequity,theex-antepayo¤is:q(V+E(p)(1K))+(1q)(V+Z1p0xKd(x))=V+q12(1K)+(1q)Z1p0pKd(p))(32)wherep0isgivenbyequation(19).
Thisshows:Proposition5The…rmwillhavehighervalueunderprivateequitythanwithstockmarket…nancingif:Emaxf0;2+p(1)K;pKg>q12(1K)+(1q)Pr(p>p0)(E(pjp>p0)K)(33)wherep0=KVVK+E+p0:(a)Thereisathresholdvalueoftheinformativenessofthemarket'ssignal,(forgivenvaluesoftheexogenousparametersofthemodelotherthanq)suchthatifthestockmarket'sinformationalproductionisgood15(q>q0),the…rmwillbeworthmorewithstockmarket…nancingthanwithprivateequity.
Ifthemarket'sinformationispoor(qPr(p>p0)(E(pjp>p0)K):(34)(seethediscussionaccompanyingequations(25)and(26)).
Settingq=1;wehaveEmaxf0;2+p(1)K;pKgcapitalmuch.
Therefore,the…rmmaywellprefertoappointasafe,experiencedboardmemberoveralesspredictablecandidatewhoseexpectedabilityishigher.
Toillustratethisusingthemodel,Iwillshowthat…rmvalueisgreaterinwhenthemarketknowstheabilityoftheboardandwillnotmakeanyinferencesaboutboardabilitywhentheCEOisreplaced.
ThiscorrespondstothecasewheretheCEOisaniiddraw(=1):ThentheexpectedchanceofsuccesswithareplacementCEOisE(p)=12andisindependentoftheabilityofthe…redCEO.
Therearetworeasonswhy…rmvalueisgreaterinthiscasethaninthemainmodel(p0))insteadof…ring(perceivedabilityE(pjpcapitalisraisedandtheassetsinplaceareworthVL+(14+p2)andsop0isgivenby:VL+(14+p02)=(1R)(VL+(+1)p0)=VL+(+1)E+p0KVL+(+1)E+p0(VL+(+1)p0):(39)9NoticethatforgivenVLandVH,thecorrelationcantakeanyvaluebetween0and1,solongasVLisdi¤erenttoVH.
Toseethis,supposethattherealizedvalueofassetsinplaceiseV=VL+e"oreV=VH+e"(dependingonthepayo¤onthenewproject)andwheree"isindependentofthepayo¤onthenewproject.
Ife"isjustdegeneratewithe"=0;wehaveperfectcorrelationbetweentheassetsinplaceandthenewproject.
Ife"hasaverylargevariance,thecorrelationwillbeverysmall.
SinceVLandVHaresucienttoderivetheequilibrium,theequilibriumdoesnotdependonthecorrelation.
19From(39)thesolutionp0solvesp0=g(p0);(40)whereg(x)14(VL+(+1)E+x)+KVL12(2+)(1+)E+x+(1+12)VLK(+1):(41)Thereisauniquesolutionforp0on(0;1),since,asshowninAppendix(6.
3),giscontinuousanddecreasingwith0capital.
Theycandothisbywaitinguntilcapitalisraisedbefore…ringtheCEO.
Inprinciple,theywouldalsobene…tbybringingforwardcapitalraising,(sothattheydon'thavetodelayreplacingtheCEO)althoughitisnotclearthatboardstypicallyhavethepowertodothis.
Inaddition,wewouldexpectthat…rmsthat…retheCEOarelesslikelytoinvestsubsequently.
Althoughthishasnotbeenstudiedempirically,thereisonestudythatbearsontherelationshipbetweenCEOentrenchmentandinvestment.
Bhren,Cooper,andPriestley(2009)…ndthat…rmswhosemanagersarelessentrenchedinvestmore,andinvestmoreinresponsetoinvestmentopportunities.
Theyconcludethatgoodgovernancemitigatesunderinvestment,ratherthanencouragingover-investment.
JenterandKanaan(2008)documentthatCEOsarelikelytobe…redfollow-ingbadindustryperformanceaswellasfollowingbadindividualperformance.
Ifweinterpretthis…ndingasindicatedthatCEOreplacementputsrelativelytoolittleweightonindividualperformance,itissupportiveofthepredictionsofthismode(butthemodeldoesnotspeaktotheir…ndingthatindustryper-formanceseemstobegiventoomuchweight.
)WhatistheimpactofCEOreplacementonsharepriceToderiveanempir-icalpredictionfromthemodel,oneneedstorecognizethattheboard'sprivateinformationabouttheCEOwilllikelybecomepublicatsomepointastimegoesby.
Itisplausiblethatthemarketsometimes…ndsouttheabilityoftheCEO,sincebadnewscan'tbeconcealedforever.
WethereforeexpectthatwhenthemarketnolongertruststheCEO,theboardwill…re.
Sotherearetwodi¤erentkindsofCEOdismissal:theCEOcouldbe…redwhenhisabilityispubliclyknownandthedismissalwillhavenoe¤ectontheshareprice.
Ontheotherhandifthesignalisprivate,whentheCEOis…red(forverylowsignals)therewillbeanegativee¤ect.
Theaveragee¤ectonsharepricemaybesmall,evenifthereisa…rst-orderdistortioncausedbydismissalsthatare10Healsoestimatesthattheverylargest…rmshavenegativeentrenchment,whichmymodelcannotexplain.
21postponedordon'ttakeplaceatall.
Inaddition,itmaybediculttoisolateashorteventwindowoverwhichthemarketlearnsthattheCEOshouldbereplaced.
Forallthesereasons,whilethemodelpredictsCEOturnover(espe-ciallyforcedturnover)shouldhaveanegativee¤ectonshareprice,itcouldbeasmallnegativee¤ect.
Anumberofpapershavestudiedtheannouncemente¤ectsofCEOturnover,generally…ndingrathersmalle¤ectsonshareprice.
FurtadoandKaran's(1990)surveyshowsthatmoststudiesreportannounce-mente¤ectsof1%orsmaller11;forexample,Warner,WattsandWruck(1988)reportane¤ectof0.
31%(statisticallyinsigni…cant).
MorerecentstudiesthantheFurtado-KaransurveyincludeHuson,MalatestaandParrino(2004)(smallpositivee¤ect)andJenterandKanaan(2006)(negativee¤ect).
Themodelinthispaperisalsoconsistentwith…ndingsthatCEOdepartureduetoanexoge-nousevents(suddendeath,assumedindependentofeconomicvariablessuchasCEOperformanceorability)hasasmallpositivee¤ectonshareprice(Johnsonetal.
(1985),Worralletal.
(1986),HayesandSchaefer(1999)andNguyenandNielsen(2010)).
Finally,themodelinthispaperpredictsthat…rmsthataremoredependentonexternal…nancing(suchasgrowth…rms)willbemoreconservativeintheirchoiceofboardmembers.
Conservatismcouldbeproxiedbythenumberofboardmembers'otherboardmemberships.
4.
6ConclusionToconclude,thispaperhasarguedthatcorporateboardswillbereluctantto…retheCEObecauseofthefutureimpactonthe…rm'scostof…nancing,oritsabilitytoobtain…nance.
Thisentrenchmentresultdi¤ersfromexistingliteraturebecauseentrenchmentisnotaconsequenceoftheCEO'sownpower,norfromcollusionbetweenboardandCEO.
Entrenchmentismitigated,butnoteliminated,whenthe…rmhasmoreassetsinplacerelativetoinvestmentopportunities.
Thepaperalsocomparesprivateequityownershipwithpubliclylisted…rms.
PrivateownershipeliminatesthecostsofCEOentrenchment,butastockmar-ketlistingaddsto…rmvaluebecausethesharepricecanguideinvestmentdeci-sions.
Otherthingsequal,established…rmswhoseexistingassetsarevaluablecomparedtotheirinvestmentopportunitieswillchoosepubliclylistedstatus,whilemorerapidlygrowing…rmswillchooseprivateownership.
Oneimplicationofthemodelisthatpubliclylisted…rmswillbemorevalu-ableifboardmembersarechosenconservatively.
Ifboardmembersarewell-knownandpredictable,itmaybebetterfor…rmvaluethaniftheyaretalentedbutinexperienced.
A…nalremarkconcernsregulationorpublicpolicy.
Themodelshowsthat,ifboardmaximizeshareholdervalue,CEOswillbeentrenchedinthesenseofnotbeing…redevenwhentheboardlearnstheyhavelowability.
However,itdoesnotfollowthatgovernmentregulatorsoractivistinvestorsshouldbecontent11SeExhibit4ofFurtadoandKaran(1990).
22withthestatusquoandshouldrefrainfromanyintervention.
Inthemodel,itwouldbeex-anteoptimaliftheboardhadamechanismtocommittoastricter…ringrule.
Potentially,corporategovernancerulesorshareholder-inuencedchangestothecorporatechartercouldhelpachievethis.
5ReferencesBaker,M.
,J.
Stein,andJ.
Wurgler(2003)WhenDoestheMarketMatterStockPricesandtheInvestmentofEquity-DependentFirms,QuarterlyJournalofEconomics,118,969-1006.
Bebchuk,LucianandEhudKamar(2010),BundlingandEntrenchment,HarvardLawReview,123,1551-1595,Bebchuk,LucianandJesseFried(2003),ExecutiveCompensationasanAgencyProblem,JournalofEconomicPerspectives17,71-92Boot,A.
andA.
V.
Thakor(1997)FinancialSystemArchitecture,ReviewofFinancialStudies,10,693-733.
Casamatta,C.
andA.
Guembel(2010),ManagerialLegacies,EntrenchmentandStrategicInertia,JournalofFinance65,2403-2436Chen,Q.
,I.
Goldstein,andW.
Jiang(2007),PriceInformativenessandInvestmentSensitivitytoStockPrice,ReviewofFinancialStudies,20,619-650Dow,J.
andG.
Gorton(1997),StockMarketEciencyandEconomicE-ciency:IsThereaConnection,JournalofFinance,52,1087-1129.
Dow,J.
andC.
Raposo(2005),CEOCompensation,Change,andCorporateStrategy,JournalofFinance,60,2701-2727.
Fisman,R.
,R.
KhuranaandM.
Rhodes-Kropf(2004),GovernanceandCEOturnover:Dosomethingordotherightthing,workingpaper.
Furtado,E.
andV.
Karan(1990),Causes,Consequences,andShareholderWealthE¤ectsofManagementTurnover:AReviewoftheEmpiricalEvidence,FinancialManagement19,60-75Gorton,GaryandBruceGrundy(1996)Executivecompensationandtheoptimalityofmanagerialentrenchment,workingpaper.
Hayes,R.
andS.
Schaefer(1999),Howmucharedi¤erencesinmanagerialabilityworth,JournalofAccountingandEconomics27,125-148Hermalin,B.
andM.
Weisbach(1998),EndogenouslyChosenBoardsofDi-rectorsandTheirMonitoringoftheCEO,"AmericanEconomicReview88,96–118.
Holmstrom,B.
(1982)"ManagerialIncentiveProblems:ADynamicPer-spective"inEssaysinEconomicsandManagementinHonorofLarsWahlbeck(Helsinki:SwedishSchoolofEconomics)Holmstrm,B.
andJ.
Tirole(1993),MarketLiquidityandPerformanceMon-itoring,JournalofPoliticalEconomy101,678-709Jenter,D.
andF.
Kanaan(2008),CEOTurnoverandRelativePerformanceEvaluation,MITSloanWorkingPaper4594-06Jensen,M.
C.
(1986)AgencyCostsofFreeCashFlow,CorporateFinance,andTakeovers,AmericanEconomicReview,76,323-32923Johnson,W.
Bruce,RobertP.
Magee,NanduJ.
NagarajanandHenryA.
Newman(1985)AnAnalysisoftheStockPriceReactiontoSuddenExecutiveDeaths:ImplicationsfortheManagementLaborModel.
JournalofAccountingandEconomics7,151-174Kanodia,C.
,R.
Bushman,andJ.
Dickhaut(1989),EscalationErrorsandtheSunkCostE¤ect:AnExplanationBasedonReputationandInformationAsymmetries,JournalofAccountingResearch,27,59–77.
Kaplan,S.
N.
andB.
Minton(2006),HowhasCEOTurnoverChangedIn-creasinglyPerformanceSensitiveBoardsandIncreasinglyUneasyCEOs,NBERWorkingPaper12465.
LangandStulz(1994),Atestofthefreecashowhypothesis:thecaseofbidderreturns,JournalofFinancialEconomics29,315-335Leland,H.
(1992),InsiderTrading:ShoulditbeProhibited,JournalofPoliticalEconomy,100,859-887.
Luo,Y.
(2005)DoInsidersLearnfromOutsidersEvidencefromMergersandAcquisitions,JournalofFinance,60,1951-1982.
Milgrom,P.
andN.
Stokey(1982),Information,tradeandcommonknowl-edge,JournalofEconomicTheory26,17-27Mikkelson,W.
andM.
Partch(1997),TheDeclineofTakeoversandDisci-plinaryManagerialTurnover,JournalofFinancialEconomics,44,205-228.
Morris,Stephen(2001)Politicalcorrectness,JournalofPoliticalEconomy,109,231-265Myers,S.
C.
(1984)Thecapitalstructurepuzzle,JournalofFinance,39,575-592Myers,S.
C.
andN.
S.
Majluf(1984)Corporate…nancingandinvestmentdecisionswhen…rmshaveinformationthatinvestorsdonothave,JournalofFinancialEconomics13,187-221NguyenandNielsen(2010)WhatDeathcanTell:AreExecutivesPaidforTheirContributionstoFirmValue,workingpaper,JudgeBusinessSchool,UniversityofCambridge.
Pagano,M.
,F.
PanettaandL.
Zingales(1998),WhyDoCompaniesGoPublicAnEmpiricalAnalysis,JournalofFinance,53,27-64.
Pagano,R.
andA.
Roell(1998),Thechoiceofstockownershipstructure:agencycosts,monitoringandthedecisiontogopublic,QuarterlyJournalofEconomics113,187-225.
Rajan,Raghuram,HenriServaesandLuigiZingales(2000),TheCostofDiversity:TheDiversi…cationDiscountandInecientInvestment,JournalofFinance55,35-80.
Rubinstein,ArielandAsherWolinsky(1990),OntheLogicof"AgreeingtoDisagree"TypeResults,JournalofEconomicTheory51,184-193Schoar,Antoinette(2002),E¤ectsofCorporateDiversi…cationonProduc-tivity,JournalofFinance,57,2379-2403Shleifer,AndreiandRobertVishny(1986),LargeShareholdersandCorpo-rateControl,JournalofPoliticalEconomy94,461–88Taylor,Luke(2010)WhyareCEOsrarely…redEvidencefromstructuralestimation,JournalofFinance,2010,65:2051-208724Worrall,D.
,W.
Davidson,P.
Chandy,andS.
Garrison(1986),Managementturnoverthroughdeathsofkeyexecutives:E¤ectsoninvestorwealth.
AcademyofManagementJournal29,674-694.
6Appendix6.
1DetailsofderivationsDetailsforderivationofequation(19):R=KV+E+p0;(43)so1R=V+E+p0KV+E+p0(44)ThenV=V+E+p0KV+E+p0(V+p0)(45)(V+E+p0)V=(V+E+p0K)(V+p0)(46)V2+E+pV=V2+E+pVKV+p0V+p0E+pp0K(47)0=KV+p0V+p0E+pp0K(48)so:p0=KVVK+E+p(49)Detailsforderivationofequation(24):Equation(23)is:0=KV+p0(VK)+p01+p02(50)0=12p02+p0(VK+12)KV(51)so,usingtheequationforthesolutionofaquadraticequation,p0=(VK+12)r(VK+12)2+4:KV:12(52)=(VK+12)rV2+K2+142KVK+V+2KV(53)=(VK+12)rV2+K2+14K+V(54)=(VK+12)rV2+(K12)2+V(55)25notefrom(52)thattheabsolutevalueofthetermwiththesquarerootsignalwaysexceedsthe…rstterm,thereforethecorrectsolutionforp0isp0=(VK+12)+rV2+(K12)2+V(56)Wecanplotthesolution:Graphofp0withpU[0;1]andK=38:andtheshareofequitygiventothemarketisRdependingonVandsettingK=3=8:6.
2Generalizationwithrespecttoalternativeboardob-jectivesAssumethe…rmhasnootherassetsapartfromtheproject(V=0),sincethefollowingobjectivesdon'tincorporateatrade-o¤betweensendingagoodsignaltothemarketandprotectingassetsinplace.
Proposition6Iftheboard'sobjectiveistomaximizetheprobabilityofinvest-ment,thentherearemultipleequilibria:itisanequilibriumfortheCEOtobe…redforp0:SupposefurtherthatinresponsetotheCEObeing…red,themarketdoesnotinvest.
Thiscannotbeanequilibriumbecausetheboardwillbebettero¤keepingthemanager(leadingtoinvestmentandapositiveprobabilitypofpositiveequityvalue)insteadof…ring(leadingtonoinvestmentandhencezerochanceofpositiveequityvalue).
NowsupposeinsteadthatinresponsetotheCEObeing…red,themarketdoesinvest.
Thisimplies:E(pjp12sincethisincreasesthechanceofpositiveequityvalue.
ThiscontradictstheassumptionthatE(pjpp0);sotheboardwillalwayschoosetoretaintheCEO.
Whataboutifp0=0or1IntheformercasetheCEOisnever…red;inthelattertheCEOisalways…redsoBayes'ruledoesnotapplytotheout-of-equilibriumevents(retainingintheformerthecase;…ringinthelatter).
It'snaturaltoassumethatifp0=0(CEOnever…red)themarketwillnotinterpreta…ringasasignalofabetterthanaverageCEO,sothisisanequilibrium.
It'salsonaturaltoassumethatifp0=1(theCEOisalways…red)themarketwillinterpretnot…ringtheCEOasagoodsignal,sothisisnotanequilibrium.
Itisimmediatefromtheargumentintheproofthatproposition(8)couldbegeneralized:Iftheobjectiveistomaximizethemarket'sperceptionoftheboard'squalityasproxiedbyanyfunctionoftheposteriordistributionwhichstrictlyincreaseswhenthemarket'sposteriordistributionofpchangesbyshiftingupinthesenseof…rst-orderstochasticdominance,theuniqueequilibriumisfortheboardneverto…retheCEO,andthemarketalwaystoinvest.
276.
3Generalizationtothecasewhereassetsinplacearecorrelatedwiththenewprojectpayo¤Weneedtoshowgiscontinuousanddecreasingwith00;notethatthedenominatorinequation(41)isstrictlypositivesinceKV;andhisincreasing,itfollowsthatp>p0:Toseethathisincreasing,notethat@h@x=(1KV+E+x)+(K(V+E+x)2@E+x@x)(V+x)>0:(66)Tocompare…rmvaluewhenthereisnoinferenceabouttheboard(=1)tothecasewherethereislearning,notethatexpectedvaluecreatedbythe…rmis:Z1p0(pK)dp(67)inthecasewithlearningabouttheboard,andZp0(12K)dp+Z1p(pK)dp(68)withoutlearning.
Sinceinequilibrium,themarketprovidescompetitive…nancing,allthisvaluegoestothe…rm'sowners.
Asp0K;thenZ1p0(pK)dp0;(75)h(1)=V1<0;(76)h(12)=VK14+12KV14+12=12K<0:(77)h0(x)=12+K2V2x;(78)h0(0)=12+K2V<0;(79)sohasoneandonlyonepositiveroot,whichliesbetween0and12.
Byconsideringh(K)=VK12K+KK2VKK2+12=V32K2VK+12(80)wecanseethatthiscantakepositiveornegativevaluesdependingonKandV,sopmaybeeithermoreorlessthanK.
30