Corporate governance, external market discipline and firm productivity

2020-02-10 来源：好走旅游网

JournalofCorporateFinance17(2011)403–417

ContentslistsavailableatScienceDirect

JournalofCorporateFinance

journalhomepage:www.elsevier.com/locate/jcorpfin

Corporategovernance,externalmarketdisciplineandﬁrmproductivity

GloriaY.Tiana,GarryTwiteb,⁎

abSchoolofBankingandFinance,AustralianSchoolofBusiness,UniversityofNewSouthWales,Sydney,NSW2052,AustraliaSchoolofFinance,ActuarialStudiesandAppliedStatistics,AustralianNationalUniversity,Canberra,ACT0200,Australia

articleinfoabstract

UsingasampleofAustraliancompaniesoverthe2000–2005period,weexaminetheimpactofinternalcorporategovernanceonfirm'stotalfactorproductivity,takingintoaccounttheinteractionbetweeninternalgovernanceandexternalmarketdiscipline.Ourempiricalfindingspointtoasubstitutioneffectbetweenproductmarketcompetitivenessandfirm-levelcorporategovernance.Overall,internalcorporategovernancemechanisms–moreefficientboardsandgreaterCEOstock-basedcompensation–areeffectiveinstrumentsforimprovingfirmproductivity.However,internalgovernanceislesseffectivewhenafirmfacesahighlycompetitiveproductmarket.Wefindonlyweakempiricalsupportforanassociationbetweenfirm'sownershipstructureandproductivity,andnosupportforanassociationbetweenindustrytakeoverintensityandfirmproductivity.

Articlehistory:

Received26July2009

Receivedinrevisedform23November2010Accepted13December2010

Availableonline16December2010JELclassiﬁcation:G30

Keywords:

Firmproductivity

CorporategovernanceExternalmarketdiscipline

“Productivityisn'teverything,butinthelongrunitisalmosteverything.”–PaulKrugman,1990,TheAgeofDiminishedExpectations

1.Introduction

Doesaﬁrm'schoiceofinternalcorporategovernancemeasuresinﬂuenceﬁrmperformance?Therehasbeenagrowingempiricalliteratureonthelinkbetweencorporategovernanceandﬁrmperformance.Thispaperbuildsonthisliteraturebyconsideringthedifferentialeffectofinternalcorporategovernanceonﬁrmperformance,takingintoaccounttheinﬂuenceofexternalmarketdisciplinefacedbytheﬁrm(intermsofproductmarketcompetitivenessandanactivemarketforcorporatecontrol).Incontrasttotheexistingliterature,wemeasureﬁrmperformanceusingtotalfactorproductivity(TFPhenceforth).1Productivity,deﬁnedastheresidualproductionoutputbeyondandabovethecontributionofinputcosts,isarguablyabettermeasureoftheﬁrm'srealeconomicperformance.Giventhatpriorliteraturehasestablishedthatgoodinternalgovernanceisrelatedtobetterﬁnancialproﬁtabilityandhighermarketvaluation,2onewouldexpectthatgoodgovernancealsohasapositiveimpactontheunderlyingdeterminantofﬁrmvalue,namely,productivity.Theprimarypurposeofthisstudyistoﬁllinthisgapinourknowledge,toaddresswhethergoodinternalgovernanceindeedhelpsimproveﬁrmproductivity.

⁎Correspondingauthor.SchoolofFinanceandAppliedStatistics,AustralianNationalUniversity,CanberraACT0200,Australia.Tel.:+61261259190;fax:+61261250087.

E-mailaddresses:y.tian@unsw.edu.au(G.Y.Tian),garry.twite@anu.edu.au(G.Twite).1TheexceptionsareHillandSnell(1989),PaliaandLichtenberg(1999),andBarthetal.(2005)whoutilizedtotalfactorproductivityinconsideringtheinﬂuenceofownershipstructureonﬁrmperformance.Totalfactorproductivityhasbeenusedelsewheretoconsidertheimpactofleverage(KimandMaksimovic,1990)andbankruptcy(MaksimovicandPhillips,1996)onﬁrmperformance.2Forexample,JensenandMurphy(1990),HartzellandStarks(2003)andCoreetal.(1999).0929-1199/$–seefrontmatter©2010ElsevierB.V.Allrightsreserved.doi:10.1016/j.jcorpﬁn.2010.12.004

404G.Y.Tian,G.Twite/JournalofCorporateFinance17(2011)403–417

Further,wedrawuponanotherstrandofliteraturethatdocumentsalinkbetweenindustry-widecompetitivenessandﬁrm-levelinternalcorporategovernanceinstruments,suchasmanagerialincentiveschemes(AggarwalandSamwick,1999),businesscombinationlaws(GiroudandMueller,2010)andanti-takeoverprovisions(Chouetal.,2008).Weexplicitlyexaminewhetherthedegreeofexternalmarketdisciplineaffectsthelinkbetweeninternalgovernanceandﬁrmproductivity.

Australia,liketheUS,isacommonlawcountrywithawelldevelopedcapitalmarket,albeitsmaller.TheconventionalwisdomisthustoclassifyAustraliaashavingan“outsidersystem”,similartothatevidencedintheUS.However,Australia'scorporategovernanceismorerepresentativeofan“insidersystem”(DignmaandGalanis,2004).Firstly,shareholdingsinAustraliaaremoreconcentrated,withmanycorporationshavinglargeblockholders(LaPortaetal.,1999;LambaandStapledon,2003).Inoursample,theaverageﬁrmhasanownershipconcentrationratioof35%(measuredascommonstocksheldbyinsiders).Secondly,executivecompensationinAustraliaislesstiedtostockperformance.Inoursample,theCEOofanaverageﬁrmreceivesonly6%oftotalpayinsharesorstockoptionswhile73%remunerationintheformsofﬁxedsalaryorcashbonus.Thirdly,AustraliancorporateboardshavealongtraditionofseparatingtherolesofchairmanandtheCEO.Furthermore,giventhattheAustralianequitymarketisrelativelysmallandcorporateownershipisconcentrated,itisnotalwaysfeasible(orsensible)tohavecompleteboardindependence.Inoursample,theaverageAustraliancorporateboardhas5to6directors,withmajoritybeingnon-executivedirectors.WhilstAustraliancompaniesfaceproductmarketcompetitionthatiscomparabletotheirUScounterparts,takeovers(inparticularhostilebids)arelesscommonintheAustralianmarket.3Insum,theseinstitutionaldifferencessuggestthatAustraliarepresentsasanideallaboratorytoexaminetheinﬂuenceofinternalcorporategovernanceandexternalmarketdisciplineoncorporateproductivity.

Intheﬁrstpartofourempiricalanalyses,wedeﬁneandcalculateTFPforasampleofAustraliancompaniesovertheperiodof2000to2005.Inthesecondpart,weexaminehowTFPisimpactedbyvariouselementsofinternalcorporategovernance(boardcharacteristics,ownershipstructure,andCEOcompensation),differentproxiesofexternalmarketdiscipline(productmarketcompetitionandtakeoverthreat)andtheirinteractions.

Ourmainﬁndingisthatinternalcorporategovernancematters,buttheextentdependsontheexternalmarketenvironment.Overallinternalcorporategovernanceisaneffectiveinstrumenttoimproveﬁrmproductivity.FirmswithmoreefﬁcientboardsandhigherCEOstock-basedcompensationhavehigherproductivitythanotherwise.However,internalgovernanceislesseffectivewhenaﬁrmfacesacompetitiveproductmarket.Thereisastrongsubstitutioneffectbetweeninternalgovernanceandexternalcompetition.Further,weﬁndonlyweaksupportforarelationshipbetweenownershipstructureandproductivity.Ourresultsarerobusttovariousmodelspeciﬁcationsandcontrollingforanumberofﬁrm-levelfactorsandﬁxedeffects.4Insummary,ourﬁndingsareconsistentwiththeargument,datingbacktoatleastAlchian(1950)andStigler(1958),forasubstitutioneffectbetweeninternalcorporategovernanceandexternalmarketdiscipline,measuredasthelevelofindustry-widecompetition.However,weﬁndnoassociationbetweenthelevelofindustry-leveltakeoverintensityandtheeffectivenessofinternalgovernance.

Thisstudyaddstotheexistingliteratureinseveralways.Firstly,beyondownershipstructurethathasbeenexaminedinexistingstudiesconcerningtherelationshipbetweencorporategovernanceandﬁrmproductivity,weexaminetherolesplayedbyboardsandmanagerialincentivecontracts.Ourﬁndingsindicatethatboardstructureisofparticularimportance.Secondly,wetest,explicitly,theimportanceofexternalmarketdisciplineandhowitinteractswithinternalgovernanceindeterminingﬁrmproductivity.Weprovideempiricalevidenceinsupportoftheargumentthatproductmarketcompetitionservesasasubstituteforinternalgovernance.Wecontrolforunobservableﬁrmheterogeneity,thereforeensuringtherobustnessofourresults.Finally,thisstudyistheﬁrsttointegratecorporategovernanceandproductivityliteratureinthecontextoftheinstitutionaldifferencesintheAustralianenvironment,complementingpriorstudiesthatfocusprimarilyontheU.S.corporateenvironment.

Therestofthepaperisorganizedasfollows.Section2reviewsrelevantliteratureandsummarizeskeytestablehypotheses.Section3describesthedataandpresentsdescriptivestatistics.Section4outlinesourempiricalframework,presentsthemainﬁndingsanddiscussesrobustnesschecks.WeconcludewithSection5.2.Relatedliteratureandtestablehypotheses

Ourpaperisrelatedtoseveralstrandsofexistingliterature—measuringﬁrmperformancebasedonproductivity,theroleofinternalcorporategovernanceinenhancingﬁrmperformance,andhowexternalmarketdisciplineimpactstherelationshipbetweentheﬁrm'sinternalgovernanceandﬁrmperformance.

Productivityrepresentsimprovedefﬁciencyinaﬁrm'seconomicperformance,aftertakingintoaccountthecontributionofinputcosts.Ithaslongbeenrecognizedineconomicsthattotalfactorproductivityisasigniﬁcantcomponentofeconomicgrowthattheﬁrmlevel.5However,isﬁrmproductivityrelatedtoﬁrmvalue?PaliaandLichtenberg(1999)ﬁndapositivecorrelationbetweentotalfactorproductivityandﬁrmvalue(measuredasTobin'sQ).Whilstgoodinternalgovernancehasbeenfoundtoenhanceﬁrmvalue,itremainstoconsidertheimpactofinternalgovernanceandexternalmarketdisciplineonﬁrmproductivity.

Duringtheperiodof2000to2005,theUSmarketwasfarmoreactivethattheAustralianmarket,withthenumberoftenderoffersintheUSmarketrepresenting167%ofthetotalnumberoflistedﬁrms,whereasisAustraliathenumberoftenderoffersrepresentedonly74%ofthetotalnumberoflistedﬁrms.4Ananalysisofthelinkbetweenﬁrmproductivityandcorporategovernancedoesraiseendogeneityconcerns.Inparticular,biasesintroducedbytheeffectofunobservedﬁrmcharacteristicsarearealpossibility.Whileestimatingﬁxedeffectspanelregressionspotentiallyaddressesthisissue,oneshouldinterprettheseresultswithsomecaution.5Forasummaryoftheliteraturelinkingeconomicgrowthandtotalfactorproductivitybothattheﬁrmandeconomywidelevel,seeHulten(2000).

3G.Y.Tian,G.Twite/JournalofCorporateFinance17(2011)403–417405

2.1.Relationshipbetweeninternalcorporategovernanceandﬁrmperformance

Ingeneral,empiricalstudiesmeasurethelevelofinternalgovernancealongfourdimensions:boardcharacteristics,ownershipstructure,managerialcompensationandshareholderrights.

Researchersineconomics,managementandﬁnancehavelongbeenanalyzingtheimportanceofownershipstructure.Forinstance,ShleiferandVishny(1986)arguethatblockholdershaveboththeabilityandincentivetomonitorcorporatemanagers,thusreducingagencycostsandenhancingﬁrmperformance.Morcketal.(1988)andMcConnellandServaes(1990),amongothers,identifyacurvilinearrelationshipbetweenTobin'sQandthefractionofcommonsharesownedbycorporateinsiders.Firmperformancemayalsobeaffectedbyvariouscharacteristicsoftheboardofdirectors.Alongthisline,manystudiesﬁndthatﬁrmperformanceisnegativelyrelatedtoboardsizeandthatnon-executivedirectorsalsopositivelyaffectﬁrmvaluation,whileotherspresentcontradictingevidence.Forinstance,Jensen(1993)arguesthatsmallerboards,beingabletoexercisecontrolmoreeffectively,shouldhaveapositiveimpactonﬁrmvalue.Yermack(1996)andConyonandPeck(1998)havepresentedevidencethatﬁrmswithsmallerboardshavehigherTobin'sQ.However,Colesetal.(2008)challengethesearguments,andﬁndthatcomplexﬁrmshavelargerboardswithmoreoutsidedirectors,andtheperformanceofcomplexﬁrmsindeedincreaseinboardsize.Borokhovichetal.(1996)ﬁndthatboardindependencecanaddﬁrmvalue,whileBhagatandBolton(2008)reportnegativerelationbetweenﬁrmperformanceandproportionofoutsidedirectors.Athirdaspectofinternalgovernancehastodowithmanagerialcompensation.InordertoalignaCEO'sownintereststothoseofshareholders,theoptimalcontractshouldhavetheCEO'scompensationpackagelinkedtoﬁrmperformance.Stock-basedcompensationsuchasrestrictedsharesandstockoptionsarepaidtotheCEO(andothertopmanagers)toimproveﬁrmperformance,potentiallythroughenhancedproductionefﬁciency(e.g.Coreetal.,1999;andMurphy,1999).Lastbutcertainlynottheleast,agrowingnumberofstudieshaveexaminedvariouscomposite-indexmeasuresofgovernance.Forinstance,Gompersetal.(2003)constructedtheGIMindexbasedon24anti-takeoverprovisions,whileBebchuketal.(2009)createdanentrenchmentindexbasedononly6oftheseprovisions.Bothstudieshavedocumentedapositiverelationbetweeninternalgovernance(asproxiedbytherespectiveindex)andﬁrmperformance.

Whenstudyingtheimportanceofinternalcorporategovernancetoﬁrmperformance,theexistingliteraturehasfocusedprimarilyonﬁnancialmeasuressuchasaccountingratesofreturnandstockmarketvaluation.Literatureonhowcorporategovernanceinﬂuencesﬁrmproductivityismuchlessextensive.HillandSnell(1989),examining122Fortune500ﬁrms,proposethatownershipstructureinﬂuencesproductivitybothdirectlyandindirectlythroughdiversiﬁcation,R&Dandcapitalintensity.PaliaandLichtenberg(1999)ﬁndthatmanagerialownershipchangesarepositivelyrelatedtochangesinproductivity,basedonasampleofU.S.manufacturingﬁrms.UtilizingapanelofNorwegianﬁrms,Barthetal.(2005)documentanegativerelationshipbetweenfamilyownershipandﬁrmproductivity,inparticularwhenafamily-ownedﬁrmismanagedbyafamilymember.Ontheotherhand,KokeandRenneboog(2005)ﬁndthat,inpoorlyperformingUKﬁrms,thepresenceofadominantoutsideshareholderhasasigniﬁcantpositiveimpactonﬁrmproductivity.Martikainenetal.(2009)ﬁndthat,forasampleofS&P500ﬁrms,thosecontrolledbyfamiliesaremoreproductive.Theyattributethisdifferencetothehigherproductiveefﬁciencyoffamilyﬁrms.

Withrespecttothepotentialimpactonproductivity,verylittleattentionhasbeengiventowhetherotheraspectsofinternalcorporategovernancebeyondownershipstructure–inparticularboardcharacteristicsandmanagerialcompensation6–alsoinﬂuenceﬁrmproductivity.Giventhisgapinourknowledge,ananalysisofvariouselementsofgovernancefromthepointofviewofproductivityshouldprovideadditionalinsightsonthebeneﬁtsofinternalcorporategovernance.2.2.Roleofexternalmarketdiscipline

Ithaslongbeenarguedthatproductmarketcompetitionhelpsimproveﬁrmperformance,andservesasapowerfultooltoresolveconﬂictsbetweenshareholdersandmanagers(e.g.Alchian,1950;Stigler,1958).Hart(1983),ShleiferandVishny(1997)andAllenandGale(2000)allpositthatmoreproductmarketcompetitioncouldbeaneffectivemechanismtoreducemanagerialslackorinefﬁciency.7Insupportoftheseconjectures,Nickell(1996)ﬁndsthattheﬁrm'smarketpowerisrelatedtolowerproductivity,withgreaterindustry-levelcompetitionassociatedwithhighergrowthinﬁrm-levelproductivity.AggarwalandSamwick(1999)ﬁndthatmorecompetitiveindustriesarecharacterizedbyweakerpay-for-performanceincentives.Chouetal.(2008)documentthatﬁrmswithweakergovernancestructurestendtobeincompetitiveindustriesortheyhavelowmarketpower.Further,governancehasasigniﬁcantimpactonﬁrm'sstockmarketreturnsonlywhenproductmarketcompetitionisweak.GiroudandMueller(2009)ﬁndthattheimpactofgoodinternalgovernance,asmeasuredbytheGIM-index,onﬁrmperformanceissmallandinsigniﬁcantincompetitiveindustries,whilelargeandsigniﬁcantinnon-competitiveindustries.GiroudandMueller(2010)investigatehowﬁrmsincompetitiveandnon-competitiveindustriesperformdifferentlyafterthepassageofbusinesscombinationlaws,whichreducesthetakeoverthreatandinturnweakensﬁrminternalgovernance.Theirpaperdocumentsthat,onlyinnon-competitiveindustries,doessuchweakenedinternalgovernancecoincideswithsigniﬁcantlyworsenedﬁrm-leveloperatingperformance.

Analternativemeasureoftheexternalmarketdisciplineiswhetheraﬁrmfacesanactivemarketforcorporatecontrol.However,thedirectionoftheassociationbetweentheleveloftakeoverthreatandﬁrmperformanceisunclear.WhilstMasulisetal.(2007)

AnexceptiontothisisBulanetal.(2007),whoreportapositiverelationshipbetweenﬁrmproductivityandCEOpay-performancesensitivityfromstockoptions.7Notallagreewithpostulateofapositiverelationshipbetweenproductmarketcompetitionandﬁrmproductivity.Analternativetheoreticalviewisinfavorofanegativerelationship.Forinstance,inmorecompetitiveenvironment,ﬁrmsfacegreatermarketuncertaintyandmorevolatilecashﬂows,sotheyarelesswillingtoinvest,whichinturnharmsﬁrmproductivity(e.g.Levinetal.,1985).Nickell(1996)providesabalancedviewonthisdebate.

6406G.Y.Tian,G.Twite/JournalofCorporateFinance17(2011)403–417

Table1

Summarystatistics—productivityestimates.Thistablepresentssummarystatisticsforourproductivitymeasures.PanelApresentsthemean,median,standarddeviation,minimum,maximum,andnumberofobservationsforeachvariable—output,labourandcapitalinputcosts,totalfactorproductivity(TFP)estimatedwithsectorﬁxed-effect,andlabourproductivityisthelogarithmoutputperemployee.PanelBpresentstheoveralltime-constantinputfactorloadingsintheestimationofTFP,togetherwithtime-constantfactorloadingsforfourindustrysectors—mining(energy),manufacturing,servicesandtrade.VariablesareasdeﬁnedinAppendixA.

PanelA:output,inputcostsandTFPVariablename

Output($mil)Labour($mil)Capital($mil)

Totalfactorproductivity(TFP)Labourproductivity

PanelB:TFPinputcostfactorloadingsInputfactorloadings

Overall

Labour($mil)Capital($mil)

0.2810.719

Energy0.2620.738

Manuf.0.2690.731

Services0.4480.552

Trade0.3510.649

OBS23062306230623061362

MEAN511.7383.87353.942.489.79

MEDIAN267.5332.56137.752.339.67

STDDEV2450.95367.662247.32

2.996.85

MIN0.020.000.01−12.20

0.12

MAX37917.763665.1032246.17

14.9225.25

arguethathighertakeoverthreatshouldberelatedtobetterﬁrm'sﬁnancialperformance,anumberofotherpapers(e.g.ShleiferandVishny,1986;Stein,1988)predictthatlowerlikelihoodoftakeoverthreatincreasesﬁrm'sproductivity.Thus,theusefulnessofmarketforcorporatecontrolinpromotingbetterperformanceisanempiricalmatter.

Insupportofapositiverelationshipbetweentheleveloftakeoverthreatandﬁrm'sproductivity,LichtenbergandSiegel(1987)ﬁndthattheleastproductiveplantsaremostlikelytothetargetofatakeoverandthattakeoverstendtobeassociatedwithincreasedproductivity.Inasimilarvein,BertrandandMullainathan(2003)investigatehowﬁrmproductivitychangesafterthepassageofbusinesscombinationlaws,whichreducesthelikelihoodofatakeoverthreat.Theydocumentthatbusinesscombinationlawscoincidewithlowertotalfactorproductivity.

Buildingonexistingtheoriesandempiricalstudiesdiscussedabove,wetestthefollowingspeciﬁchypothesesinthisstudy.H1.Betterinternalcorporategovernancereducesagencyproblemsandincentivizesmanagerstoimproveﬁrmproductivity.H2.Externalmarketdisciplineactsasasubstituteforinternalcorporategovernance.Therefore,thepositivelinkbetweengovernanceandﬁrmproductivityisweakerinmoredisciplinedindustries(morecompetitiveproductmarket,ormoreactivetakeovermarket)thanotherwise.3.Dataanddescriptivestatistics

Ourdatacomefromvarioussources.Westartwithalistof1080companiesinConnect4Boardroomdatabaseovertheperiodof2000to2005.Usingthislist,weobtaindetailedexecutiveanddirectorstatusandcompensationdata,whichwethenusetogenerateﬁrm-levelcorporategovernancemeasuresconcerningCEOcompensationandboardcharacteristics.OurprimaryownershipdatasourceisWorldScope,fromwhichweobtainthetime-varyingownershipconcentrationratio.FinancialdataaresourcedfrombothBvD'sORBISdatabaseandAspectHuntleydatabase.Weexcludeﬁrmsintheﬁnancialindustry,duetotheirdifferentreportingstructures,andﬁrmswithlessthan2yearsofconsecutiveﬁnancialdata.Ourﬁnalsampleincludes2306ﬁrm-yearobservationscovering543differentAustraliancompaniesovertheperiodof2000–2005.3.1.Measuringtotalfactorproductivity(TFP)

Weestimateaﬁrm'sTFPinspiritsimilartothatinotherexistingstudies(suchasPaliaandLichtenberg,1999;Bulanetal.,2007).TFPistypicallyusedintheliteratureasameasureoftechnologyadvanceorenhancedefﬁciencyinproduction.WeuseaversionofCobb–DouglasproductionfunctiontoderiveTFP,intwosteps.First,weestimatethetime-invariantinputfactorloadingsinapanelsetting,assumingthatfactorloadingsarethesameforallﬁrmswithinaparticularindustrysector.WethencalculateTFPforeachﬁrmineachyearastheresidualbetweenactualoutputandtheestimatedoutputusingthetime-constantfactorloadings.Theestimationprocedureisasfollows:

LβK

Yit=Ait×Lβit−1×Kit−1

→yit=ait+βLlit−1+βKkit−1+εitsubjectto:βL+βK=1󰀁󰀃

ˆˆait=yit−βLlit−1+βKkit−1

ð1Þ

ð2Þ

G.Y.Tian,G.Twite/JournalofCorporateFinance17(2011)403–417407

whereYistheoutput,ListhecostsoflabourandKisthecostofphysicalcapitalinput.Lower-caselettersrepresentlogarithmofthesevariables,andat,logarithmA(technologyorefﬁciency),istermedasTFP.Outputismeasuredusingoperatingrevenue,labourinputusingcostsofemployees,andcapitalinputusingnetproperty,plantandequipment(PP&E).8ThesevariablesaresourcedfromORBISdatabase.Becausedifferentindustrystructuresmightaffecttherelativeimportanceoftheinputvariables,wealsoincludeaﬁxedeffecttoaccountforsector-levelheterogeneityintheﬁrststep.Theindustrysectorsweconsideraremanufacturing,services,utilitiesandmining(energy),andtradeandtransportation.Finally,wetrimpotentialoutliersat1%and99%.

Table1reportsdescriptivestatisticsofﬁrmoutput,inputcosts,andtheTFP.Weobservethat,onaverage,capitalaccountsformorethantwo-thirdsinputcostsinproduction.Theaverageoperatingrevenueofoursampleﬁrmsis$512million,averagelabourcostsof$8millionandcapitalcostsof$354million.Overallinputfactorloadingsforlabourandcapitalare0.281and0.719respectively,andtheaverage(median)TFPestimatedwithsector-ﬁxed-effectis2.48(2.33).Asexpected,therearesomedifferencesacrosssectors:forinstance,theimportanceofcapitalismoreobviousformanufacturersandenergycompanies,whilelabouriscriticalforﬁrmsinserviceﬁrms.TheseestimatesaresomewhatdifferentthanthoseofPaliaandLichtenberg(1999),whoestimatedproductionfunctionsforasampleofU.S.manufacturerbetween1983and1992,alsousinglabourandcapitalinputs.93.2.Internalcorporategovernanceandexternalmarketdiscipline

Wemeasurecorporategovernancealongoneofthreedimensions:boardcharacteristics,ownershipstructure,orCEOcompensation.10Onthedimensionofboardcharacteristics,ourprimarymeasureisNEDS,whichistheproportionofnon-executivedirectorssittingonaboard.Onthedimensionofownershipstructure,ourprimarymeasureisownershipconcentrationratio(COhenceforth),thepercentagecommonequityheldbyinsidersasagroup,asreportedinWorldscopedatabase.OnthedimensionofCEOcompensation,ourprimaryvariableisCEO_STKPAY,thelogarithmtransformationofdollarvalueCEOstock-basedcompensation.Wepositthatinsiderownershipprovidesincentivesforexecutivestooperatetheircompaniesmoreefﬁciently,thatmoreindependentboardisbetteratmonitoringtopmanagers,andthatstock-basedcompensation(inparticularstockoptions)incentivizestheCEOtoimproveﬁrmproductionefﬁciency.Insummary,weexpectapositiverelationshipbetweenTFPandmeasuresofgoodinternalcorporategovernance.

Externalmarketdisciplineismeasuredalongoneoftwodimensions:productmarketcompetitiveness(COMPETITIONhenceforth),ormarketforcorporatecontrol(MKTCONTROLhenceforth).Toassessthecompetitivenessofproductmarket,weﬁrstconstructtheHerﬁndahl-Hirschmanindexusingﬁrmsalesorassets(HHIhenceforth),basedon4-digitGICScodes.11AlowerHHIindicatesstrongerproductmarketcompetition,orlessconcentrationofmarketpowerheldbylargeplayerswithintheindustry.WethenrankallindustriesbasedontheirHHIscores,andsetthebinaryindicatorCOMPETITIONtooneifaﬁrmoperatesinanindustrywhoseHHIisinbottomtercileinaparticularyear,andzeroforthoseﬁrmsinthetoptercile.Toassesswhethermarketforcorporatecontrolimposescrediblethreattotheexistingmanagers,weﬁrstconstructindustry-widetakeoverratio(M&Ahenceforth),deﬁnedasthenumberofbothcompleteandincompletemergersandhostiletakeoversdividedbythenumberofactiveﬁrmsineachindustry.12AhigherM&Aimpliesmoreactivemarketforcorporatecontrolandagreatertakeoverthreat.ThebinaryindicatorMKTCONTROListhussettobeoneifaﬁrmoperatesinanindustrywhoseM&Ainaparticularyearisinthetoptercile,andzeroforthoseﬁrmsthatareinthebottomtercile.13Wepositthattougherexternalmarketenvironment(i.e.morecompetitiveproductmarket,orgreatertakeoverthreat)wouldforceaﬁrmtooperatemoreefﬁciently,leadingtobetterproductionefﬁciency.WethusexpectapositiverelationshipbetweenTFPandindicatorsofexternalmarketdiscipline.Inaddition,ifexternalmarketservesasasubstituteforefﬁcientinternalcorporategovernance,weexpecttoﬁndinternalgovernancehavinglessimpactonTFPinthepresenceofgreatermarketdiscipline.

Intheempiricalregressionmodelstofollow,wealsoconsiderasetofﬁrm-levelcontrolvariablesthatmayinﬂuenceﬁrmproductivity.Inparticular,weconsider:ROA,ﬁnancialproﬁtabilitymeasuredasreturnonassets;SIZE,logarithmofﬁrm'stotalassetsplusone;AGE,logarithmofnumberofyearssinceﬁrm'sincorporation;GROWTH,themarket-to-bookratioofcommonequity;and

ThisdeﬁnitionofTFPisconsistentwiththelabourproductivityliterature.Forrobustnesschecks,wealsotestalternativespeciﬁcationsofTFP.Forinstance,wealsoconsiderthecostofmaterialsasanadditionalinput.The3-factorTFPmeasurehasacorrelationof0.828withthe2-factorTFPreportedinthepaper,andourregressionresultsremainqualitativelythesameusingthisalternativemeasure.9InPaliaandLichtenberg(1999),Table2,inputshareofone-yearlaggedlabourisonaverage0.697andtheinputshareofone-yearlaggedcapitalisaround0.119.Thereareseveralfactorsthatcontributetothedifferencebetweentheirandourestimates:(i)mostimportantly,thetwostudiesutilizedifferenttimeperiods/countrysettings,(ii)givendatalimitation,wemeasurelabourinputbyitsdollarcostratherthanthenumberofemployees,whichisthevariableusedinthepriorstudy;and(iii)weseparatetheestimationofinputfactorloadingsfromananalysisofwhataffectsTFP.Inparticular,inour1st-stepinputfactorloadingequation,weassumetime-invariantestimatesinsteadofcontrollingforyearﬁxedeffects.However,weincludeyearﬁxedeffectsinthe2nd-stepregressionanalysiswhenweinvestigatetheseparateimpactsofcorporategovernanceandproductmarketcompetitiononTFP.10Therearecurrentlynoequivalents,inAustralia,ofGompersetal.(2003)anti-takeoverindexortheBebchuketal.(2009)entrenchmentindex.11GiroudandMueller(2010)deﬁnedHHIbasedon3-digitSICcodes.BecauseAustraliaisarelativelysmalleconomywithbigplayersinsomeindustries,suchdetailedsegmentation–165differentindustriesat3-digitSIClevel–isnotfeasibleforoursample.Excludingtheﬁnancialsector,wehave20(9)industrygroupsbasedon4-digt(2-digt)GICScodes.Atabledetailingthe4-digitGICSindustryclassiﬁcationandassociatedHHIscoresispresentedinAppendixB.12Incompletetakeoverscouldalsorepresentrealthreattoincumbentmanagement.Wealsoconsideredonlycompletedmergersandacquisitions,andtheresultsarequalitativelythesame.Inaddition,duetothelowfrequencyofM&Aeventsthroughoutoursampleperiod,thetakeoverratiosaredeﬁnedat2-digitGICSindustrylevel.13Effectively,wereduceoursamplesizebyaboutonethird.Inrobustnesstestsnottabulatedforbrevity,wealsoredeﬁnethesetwoindicators(i)oneifaﬁrmoperatesinanindustrywhoseHHIscoreorM&Aratioisinbottomtercileinaparticularyear,andzerootherwise;and(ii)basedonmedianHHIscoresandmedianM&Aratios,thusincludingallsampleﬁrms.Theresultsarequalitativelythesameasthosereportedinthepaper.

8408G.Y.Tian,G.Twite/JournalofCorporateFinance17(2011)403–417

Table2

Summarystatistics—keyvariables.Thistablepresentssummarystatisticsofkeyvariablesforthesample,whichincludes543Australiancompaniesovertheperiodof2000–2005.PanelApresentsthemean,median,standarddeviation,minimum,maximum,andnumberofobservationsforeachvariable.PanelBpresentsthePearsoncorrelationsforeachpairofvariables.Thekeyvariablesinclude:Productivity—TFP;Internalcorporategovernance—OWN_PCT,BODSIZE,NEDS,CEO_STKPAY,andCEO_INCENT;Externalmarketdiscipline—HHI_sales,HHI_asset,andM&A;Firmlevelcontrolvariablesare:ROA,SIZE,AGE,GROWTH,andLEVERAGE.VariablesareasdeﬁnedinAppendixA.PanelA:descriptivestatisticsVariablenameTFPROASIZEAGE

GROWTHLEVERAGEOWN_PCTBODSIZENEDS

CEO_STKPAYCEO_INCENTHHI_salesHHI_assetM&A

PanelB:correlationmatrix

ROA

TFPROASIZEAGE

GROWTHLEVERAGE

BODsize

TFP

BODsizeNEDSOWN_pctCEO_STKPAYCEO_INCENT

0.090

0.300

SIZE0.3540.449

AGE0.1550.2240.351

GROWTH−0.100−0.132−0.156−0.096

LEVERAGE0.3040.2180.4430.1400.0871.000

HHI_sales0.1020.1510.0930.0360.0210.010

M&A0.0480.038−0.0100.056−0.061−0.025

OBS23062306230618621909230616452306230616941694214221422142

MEAN2.48−4.3711.0331.052.2738.3733.495.750.663.056.100.190.181.29

MEDIAN2.331.8510.7422.001.3640.7733.315.000.670.010.000.140.120.52

STDDEV2.9922.402.3324.543.3324.4023.542.530.215.0715.020.150.131.57

MIN−12.20−91.58

0.003.000.000.250.001.000.000.000.000.050.050.00

MAX14.9242.8819.40126.0021.5699.7499.8618.001.0015.5998.960.750.686.15

NEDS0.0670.304

OWN_pct0.046−0.062−0.129

CEO_STKPAY0.1090.1620.116−0.045

CEO_INCENT0.0450.060−0.016−0.0150.733

LEVERAGE,theratiooftotalliabilitiestototalassets.Thesecontrolvariablesarewinsorizedat1%and99%tominimizeimpactofoutliers.AsinNickelletal.(1997)andKokeandRenneboog(2005),weuselaggedvaluesofcontrolvariablestoreducereversecausality.

Table2providessummarystatisticsofthenumericalmeasuresofTFP,internalcorporategovernanceandﬁrm-levelcontrols.14ItalsoreportsthecorrelationmatrixofTFPandcontemporaneousvariables.Weobservethatﬁrmswithhigherproductivityarelargerﬁrms,morelikelytohavebetterﬁnancialperformance,haveexistedforalongerperiod,andusemoredebtﬁnancing.Wealsoﬁndthatproductivityisnegativelycorrelatedwithgrowthopportunitiesasperceivedbytheequityinvestors.Withouttakingintoaccountotherfactorsthatinﬂuenceproductivity,TFPisnegativelycorrelatedwiththedegreeofproductmarketcompetition,butispositively,albeitmuchweaker,withthelevelofactivityinthemarketforcorporatecontrol.Intermsofinternalgovernancevariables,productivityispositivelycorrelatedwithmorenon-executivedirectors,largerboardsandmoreCEOstock-basedcompensation.Inaddition,ownershipconcentrationratioisnegativelycorrelatedwithboardvariables.

Withregardtoﬁrmcharacteristics,weobservethat,thoughTFPispositiveoverthesampleperiod,ﬁnancialperformanceismuchweaker(i.e.meanROAisnegative);thatmarketvalueofequityisonaveragemorethantwicethatoftheﬁrm'sbookvalueofequity;andthatanaverageﬁrmwouldborrowabout38%ofitstotalassets.

4.Internalgovernance,marketdisciplineandﬁrmproductivity

Ourmaininterestsarewhethergoodinternalcorporategovernanceisassociatedwithhigherﬁrmproductivity,andwhetherexternalmarketdisciplineaffectsthisassociation.

14DeﬁnitionsofvariablesarepresentedinAppendixA.

G.Y.Tian,G.Twite/JournalofCorporateFinance17(2011)403–417409

Table3

Differenceinmeans.Thistablepresentsthemeanofthekeyvariables:Productivity—TFP;Internalcorporategovernance—OWN_PCT,BODSIZE,NEDS,CEO_STKPAY,CEO_INCENT;Externalmarketdiscipline—HHI_sales,HHI_assets;ﬁrmlevelcontrolvariablesare:ROA,SIZE,AGE,GROWTH,andLEVERAGE.VariablesareasdeﬁnedinAppendixA.Thefullsampleisseparatedintotwosub-samples,competitiveandnon-competitiveindustries,basedonbinaryindicatorvariableCOMPETITION,whichtakesavalueofoneforcompetitiveindustries.Theindicatorequalsoneifaﬁrm'sHHI_salesscoreisinthebottomtercileinaparticularyear,zerootherwise.Wetestforthedifferencesinmeansbetweenthetwosub-samplesusingat-test.Differencesinmeansandtheirassociatedt-valuesarereported,with*,**and***indicatingstatisticalsigniﬁcanceat10%,5%and1%respectively.

TFP

Non-competitiveCompetitive

DifferenceinmeansT-stat

2.7342.1950.5394.33⁎⁎⁎

NEDS0.6840.6440.0403.99⁎⁎⁎ROA

Non-competitiveCompetitive

DifferenceinmeansT-stat

−3.844−3.475−0.369−0.35

BODSIZE6.0385.5460.4924.05⁎⁎⁎

SIZE11.39610.9440.4524.07⁎⁎⁎

OWN_PCT36.81438.274−1.460−1.02

CEO_STKPAY3.6083.2410.3661.62AGE34.32030.0094.2213.55⁎⁎⁎

CEO_INCENT0.0670.0640.0030.30

GROWTH2.5982.5600.0380.24

HHI_sales0.2980.0940.20445.43⁎⁎⁎

HHI_asset0.2580.1030.15533.31⁎⁎⁎LEVERAGE40.24439.1581.0870.94

4.1.Fixed-effectregressionmodels

Wenowspecifythesourcesofﬁrmproductivitybyexamininginternalgovernance,externalmarketdisciplineandtheirinteractioninamultipleregressionframework.SimilartoHimmelbergetal.(1999)andPhametal.(2008),weacknowledgethatordinaryleastsquaresmodelcannotsufﬁcientlyaddresstheissueofendogeneity.Wethusconsiderthefollowingﬁxed-effectmodel:

ai;t=αi;t+γ1ðGOVÞi;t−1+γ2ðMKTÞi;t−1+γ3ðGOVÞi;t−1×ðMKTÞi;t−1+γ4ðControlsÞi;t−1+βt+λi+εi;t

ð3Þ

whereatisTFPwithsectorﬁxedeffects;tstandsforadeterministicgrowthcomponent(i.e.timeimpact),andistandsforﬁrmﬁxedeffect,whichcapturesunobservedﬁrmheterogeneitythatmayberelatedtobothinternalgovernanceandﬁrmproductivity.15Weuseagenericterm,GOV,torepresentinternalcorporategovernance,takingoneofthreevalues:ownershipconcentrationratio(CO),theproportionofnon-executivedirectorssittingontheboard(NEDS)orthelogarithmofCEOpaythatisstock-based(CEO_STKPAY).Likewise,weuseanothergenericterm,MKT,torepresentthetoughnessofexternalmarketdiscipline,asmeasuredbyHHI_salesscoresorM&Aratio.Wealsoconsidertheinteractionbetweeninternalgovernanceandthestateofexternalmarketenvironment,inwhichMKTnowtakesoneofthetwobinaryvalues:productmarketcompetition(COMPETITION),ormarketforcorporatecontrol(MKTCONTROL).Therefore,regressioncoefﬁcientγ1capturesthebase-lineimpactofinternalgovernanceonTFP;coefﬁcientγ3captureshowthisimpactchanges,whentheexternalmarketistough(i.e.MKTequalsone);andcoefﬁcientγ2measuresthedirectimpactoftheexternalmarkettoughnessontheaveragelevelofTFP.4.2.Empiricalﬁndings

Webeginbysplittingthesampleﬁrmsintotwogroups–thoseinnon-competitiveindustriesandthoseincompetitiveindustries–wherecompetitivenessisdeﬁnedusingtheindicatorvariableCOMPETITION.Table3presentsthemeansofeachofthenumericalmeasuresofTFP,internalcorporategovernanceandﬁrm-levelcontrols.Weobservethatﬁrmsinnon-competitiveindustries(i.e.greaterconcentrationofmarketpower)havehigheraverageproductivity.Theseﬁrmsarelargerandhavesurvivedlonger.Inaddition,theircorporateboardsarebigger,withmorenon-executivedirectors;thismaybeareﬂectionoftheobservationthatlargerﬁrmsaremorecomplextomanage(Colesetal.,2008).

Doesinternalgovernancemattertoﬁrmproductivity?Table4presentsﬁxed-effectregressionresultswhenweconsideronlytheimpactofinternalcorporategovernanceonTFP,forthefullsampleandalsoseparatelyforcompetitiveandnon-competitiveindustries.16Whilethecontinuousgovernancevariablesareallinsigniﬁcantatthe5%levelforthefullsample,separatesub-

Zhou(2001)pointsoutthatﬁxedeffectsmodelsuchasHimmelbergetal.(1999)mightunderestimatetheassociationbetweenmanagerialownershipandﬁrmperformance.16Aswenoted,ananalysisofthelinkbetweenﬁrmproductivityandcorporategovernancedoesraiseendogeneityconcerns.Inparticular,biasesintroducedbytheeffectofunobservedﬁrmcharacteristicsarearealpossibility.Forexample,lowlevelsofinsiderownershipmaytheoptimalinlargeﬁrmsthathavelowerintangibleassets,morestablecashﬂowsandwherethedemandformonitoringislow.Wecanfurtherarguethattheseﬁrmsarebetterabletoaccessexternalﬁnancingforpromisinginvestments,therebyimprovingproductivity.Whilewehavesoughttoaddresstheissueofunobservedheterogeneityacrossﬁrmsbyestimatingﬁxedeffectspanelregressions,thisisanincompletesolution,whichassumesstationaryrelationshipswiththeunobservablevariables.Thus,oneshouldinterprettheseresultswithsomedegreeofcaution.

15410G.Y.Tian,G.Twite/JournalofCorporateFinance17(2011)403–417

Table4

InternalCorporateGovernanceandFirmProductivity.Thistablepresentsregressionresultsofproductivity(TFP)onmeasuresofinternalcorporategovernance(OWN_PCT,NEDS,andCEO_STKPAY)andﬁrmlevelcontrolvariables(ROA,SIZE,AGE,GROWTH,andLEVERAGE).VariablesareasdeﬁnedinAppendixA.Resultsarepresentedforthefullsampleandseparatelyforﬁrmsinnon-competitiveandcompetitiveindustries,basedonbinaryindicatorvariableCOMPETITION,whichtakesavalueofoneforcompetitiveindustries.Theindicatorequalsoneifaﬁrm'sHHI_salesscoreisinthebottomtercileinaparticularyear,zerootherwise.Inallmodels,weincludeﬁrmﬁxedeffectswithstandarderrorsrobusttoclusteringbyindividualﬁrmsovertime.Regressioncoefﬁcientsandtheirassociatedt-valuesarereported,with*,**and***indicatingstatisticalsigniﬁcanceat10%,5%and1%respectively.

Non-competitive

GOV=NEDSGOV=OWN_PCTGOV=CEO_STKPAYROASIZEAGEGROWTHLEVERAGEAdjustedR2OBS

0.0010.16−0.083−1.86⁎0.0092.64⁎⁎⁎−0.043−1.75⁎0.0071.86⁎0.754553

0.0041.44−0.019−0.560.0113.57⁎⁎⁎−0.036−1.89⁎0.0041.470.699682

0.0041.45−0.028−0.950.0083.49⁎⁎⁎−0.049−2.17⁎⁎0.0072.25⁎⁎0.7651235

0.0051.10−0.247−4.97⁎⁎⁎0.0123.35⁎⁎⁎−0.059−2.460.0061.240.738553

0.0041.27−0.037−0.870.0982.77⁎⁎⁎−0.033−1.550.0051.310.786682

0.0051.76⁎−0.114−3.37⁎⁎⁎0.0114.08⁎⁎⁎−0.046−1.88⁎0.0071.80⁎0.7741235

1.0262.56⁎⁎

Competitive0.2490.79

Fullsample

0.4941.88⁎

−0.006−1.80⁎

−0.001−0.41

−0.001−0.24

0.0234.02⁎⁎⁎0.0010.20−0.011−0.240.0144.17⁎⁎⁎−0.046−1.86⁎0.0061.540.760553

−0.002−0.300.0051.43−0.065−1.65⁎0.0103.02⁎⁎⁎−0.032−1.500.0051.400.805682

0.0030.810.0041.38−0.062−1.94⁎0.0135.23⁎⁎⁎−0.052−2.19⁎⁎0.0071.94⁎0.7751235

Non-competitive

Competitive

Fullsample

Non-competitive

Competitive

Fullsample

sampleregressionsrevealquitedifferentresults.Whilegovernancedoesnotmatterwithinthecompetitiveindustries,withinthenon-competitiveindustries,bothNEDSandCEO_STKPAYhavepositiveandsigniﬁcantcoefﬁcientsat5%level,indicatingthatmoreindependentboardsandhigherCEOstock-basedcompensationareassociatedwithbetterﬁrmproductivity.

Intermsoftheeconomicsigniﬁcanceofkeyvariables,thecoefﬁcientof1.026onNEDSreportedinnon-competitiveindustriesindicatesthataone-standarddeviationincreaseinNEDSisassociatedwitha0.216(8.7percentage)increaseinTFP.Whilethe

Table5

Internalgovernance,marketcompetitivenessandﬁrmproductivity.Thistablepresentsregressionresultsofproductivity(TFP)onmeasuresofinternalcorporategovernance(OWN_PCT,NEDS,andCEO_STKPAY),productmarketcompetitiveness(HHI_sales,COMPETITION),theirinteractiontermsandﬁrmlevelcontrolvariables(ROA,SIZE,AGE,GROWTH,andLEVERAGE).VariablesareasdeﬁnedinAppendixA.InModelA,productmarketcompetitivenessismeasuredusingcontinuousHHI_salesscores.InModelsB,CandD,competitivenessismeasuredusingbinaryindicatorCOMPETITION.Theindicatorequalsoneifaﬁrm'sHHI_salesscoreisinthebottomtercileinaparticularyear,zerootherwise.Inallmodels,weincludeﬁrmﬁxedeffectswithstandarderrorsrobusttoclusteringbyindividualﬁrmsovertime.Regressioncoefﬁcientsandtheirassociatedt-valuesarereported,with*,**and***indicatingstatisticalsigniﬁcanceat10%,5%and1%respectively.

ModelA

GOV=NEDSGOV=OWN_PCTGOV=CEO_STKPAYHHI_salesCOMPETITIONGOV*COMPETITIONROASIZEAGEGROWTHLEVERAGEAdjustedR2OBS

0.0041.40−0.001−0.050.0104.34⁎⁎⁎−0.042−1.77⁎0.0083.27⁎⁎⁎0.7761235

0.8602.05⁎⁎

0.6311.74⁎−0.877−2.02⁎⁎0.0041.61−0.042−1.64⁎0.0104.90⁎⁎⁎−0.038−1.94⁎0.0041.590.7711235

−0.067−0.230.0041.010.0051.80⁎−0.074−2.48⁎⁎0.0114.85⁎⁎⁎−0.041−1.90⁎0.0051.82⁎0.7741235

0.1300.54−0.014−1.82⁎0.0041.49−0.046−1.560.0135.86⁎⁎⁎−0.037−1.71⁎0.0041.460.7811235

ModelB

0.7922.52⁎⁎

−0.006−1.80⁎

0.0122.40⁎⁎

ModelC

ModelD

coefﬁcientof0.023onCEO_STKPAYreportedinnon-competitiveindustriesindicatesthataone-standarddeviationincreasein

G.Y.Tian,G.Twite/JournalofCorporateFinance17(2011)403–417411

Table6

Internalgovernance,takeoverthreatandﬁrmproductivity.Thistablepresentsregressionresultsofproductivity(TFP)onmeasuresofinternalcorporategovernance(OWN_PCT,NEDS,andCEO_STKPAY),marketforcorporatecontrol(M&AandMKTCONTROL),theirinteractiontermsandﬁrmlevelcontrolvariables(ROA,SIZE,AGE,GROWTH,andLEVERAGE).VariablesareasdeﬁnedinAppendixA.InModelA,marketforcorporatecontrolismeasuredusingcontinuousM&A.InModelsB,CandD,takeoverthreatismeasuredusingbinaryindicatorMKTCONTROL.TheindicatorequalsoneifaﬁrmoperatesinanindustrywhoseM&Ainaparticularyearisinthetoptercile,andzerootherwise.Inallmodels,weincludeﬁrmﬁxedeffectswithstandarderrorsrobusttoclusteringbyindividualﬁrmsovertime.Regressioncoefﬁcientsandtheirassociatedt-valuesarereported,with*,**and***indicatingstatisticalsigniﬁcanceat10%,5%and1%respectively.

ModelA

GOV=NEDSGOV=OWN_PCTGOV=CEO_STKPAYM&AMKTCONTROGOV*MKTCONTROLROASIZEAGEGROWTHLEVERAGEAdjustedR2OBS

0.0041.320.0090.300.0104.41⁎⁎⁎−0.043−2.62⁎⁎⁎0.003.40⁎⁎⁎0.7751235−2.418−1.35

0.1230.39−0.154−0.360.0041.61−0.038−1.470.0104.89⁎⁎⁎−0.038−2.70⁎⁎⁎0.0031.520.7651235

−0.024−0.110.0000.090.0051.81⁎−0.150−4.98⁎⁎⁎0.0114.86⁎⁎⁎−0.040−2.69⁎⁎⁎0.0051.84⁎0.7751235

−0.137−0.870.0201.090.0041.58−0.064−2.22⁎⁎0.0135.86⁎⁎⁎−0.039−2.64⁎⁎⁎0.0041.530.7761235

ModelB

0.4171.38

−0.004−1.21

−0.003−0.24

ModelC

ModelD

CEO_STKPAYisassociatedwitha0.117(4.7percentage)increaseinTFP.Concentratedinsiderownership,OWN_PCT,isstatisticallynegativelysigniﬁcant,albeitonlymarginally,indicatingthathigherinsiderownershipisassociatedwithpoorerﬁrmproductivity.Thecoefﬁcientof-0.006onOWN_PCTindicatesthataone-standarddeviationincreaseinOWN_PCTisassociatedwitha0.142(5.7percentage)decreaseinTFP.

Focusingonﬁrmcharacteristics,weﬁndthatTFPispositivelyrelatedtoﬁrmageand,toalesserextent,theuseofdebtﬁnancing.Somewhatsurprisingly,weﬁndthatbothﬁrmsizeandgrowthopportunities(asmeasuredbythemarket-to-bookratio)haveanegativeassociationwithﬁrmproductivity,althoughregressioncoefﬁcientsarestatisticallysigniﬁcantat5%onlyinsomemodelspeciﬁcations.17InsimpleOLSmodelswithoutﬁxedeffects(nottabulatedforbrevity),productivityisalsopositivelyrelatedtoaﬁrm'spastﬁnancialproﬁtability,laggedROA,butthisimpactbecomesinsigniﬁcantoncewecontrolforunobservedﬁrmeffects.

Sofar,ourresultshavesuggestedthatinternalcorporategovernanceplaysadifferentroleincompetitiveandnon-competitiveindustries:goodgovernancepromotesbetterproductionefﬁciencywhentheﬁrmfacesarelaxedexternalenvironment;however,itmatterslittlewhentheﬁrmoperatesinamoredisciplinedmarket.Thispointstoapotentialsubstitutioneffectbetweenthetoughnessofexternalmarketandtheeffectivenessofinternalcorporategovernance.

Doesproductmarketcompetitivenesssubstituteforinternalgovernance?Table5summarizesregressionresultsofourmainﬁxed-effectmodel,Eq.(3),withfourmodelspeciﬁcations.ModelAconsidersdirectimpactofexternalmarketdiscipline,usingcontinuousHHIscores,withouttakingintoaccountinternalcorporategovernance.InmodelsB,CandD,weregressTFPonmeasuresofinternalgovernance,thestateofexternalmarketcompetitiveness,COMPETITION,andtheirinteractionterms.TheindicatorvariableCOMPETITIONisusedtofacilitateinterpretation.Inallmodels,weincludeﬁrmﬁxedeffectswithstandarderrorsrobusttoclusteringbyindividualﬁrmsovertime.

Withrespecttoexternalmarketdiscipline,byitself,ModelAresultsshowthatTFPincreasesintheHHIscores,indicatinganegativeassociationbetweenﬁrmproductivityandthedegreeofexternalmarketcompetition.ThisisconsistentwithTable3inthat,onaverage,ﬁrmsincompetitiveindustrieshavelowerproductivity.18Withrespecttoeconomicsigniﬁcance,thecoefﬁcient

17ThisevidencecontrastswithPaliaandLichtenberg(1999)whoﬁndapositiveassociationbetweenthemarket-to-bookratioandTFP.OnepotentialexplanationforthisnegativecoefﬁcientonGROWTHisthatTFPisameasureofcurrentproductivityreﬂectingbothcurrentandpastinvestmentdecisions.Growthopportunities,ontheotherhand,involvefutureinvestmentschoices.Totheextentthatcurrentandfutureinvestmentdecisionsaresubstitutes,onewouldanticipateanegativecorrelationbetweengrowthopportunitiesandﬁrmproductivity.Togetherwithdifferencesinthesampleperiodsandvariablesdeﬁnitions,thisexplanationmayexplainthedifferenceinresults.18Whiletheobservednegativeassociationbetweenﬁrmproductivityandcompetitionisinconsistentwithexpectations(Nickell,1996),possibleexplanationsforthisresultmaylieinthecompositionoursampleofAustraliaﬁrms.Forexample,oursampleisdominatedbylarge,lowproﬁtabilityﬁrmsincompetitiveindustries,whereresourcesmaybedirectedtowardsdeferringentryratherthanimprovingproductivity(Nickell,1996),ortheselaggardﬁrmsarediscouragedfrominnovatingbecauseoflowpost-innovationrents(Aghionetal.,2005).Alternatively,ﬁrmsincompetitiveindustriesmayfacegreatermarketuncertaintyandmorevolatilecashﬂows,sotheyarelesswillingtoinvest,whichinturnharmsﬁrmproductivity(Nickell,1996).

412G.Y.Tian,G.Twite/JournalofCorporateFinance17(2011)403–417

Table7

Robustness—alternativemeasuresofTFP.Thistablepresentsregressionresultsofproductivity(TFP)onmeasuresofinternalcorporategovernance(OWN_PCT,NEDS,CEO_STKPAY),productmarketcompetitiveness(COMPETITION),theirinteractiontermsandﬁrmlevelcontrolvariables(ROA,SIZE,AGE,GROWTH,andLEVERAGE).VariablesareasdeﬁnedinAppendixA.ModelsBandDareasdeﬁnedinTable5.TFPismeasuredusingalternativedeﬁnitions.WeestimateTFPusingbothvalueaddedandsalesastheoutput.Wealsoconsiderathirdinputvariable,dollaramountofmaterialscost.Forbrevity,wedonotreportthecoefﬁcientsonﬁrm-levelcontrolvariables.Inallmodels,weincludeﬁrmﬁxedeffectswithstandarderrorsrobusttoclusteringbyindividualﬁrmsovertime.Regressioncoefﬁcientsandtheirassociatedt-valuesarereported,with*,**and***indicatingstatisticalsigniﬁcanceat10%,5%and1%respectively.

Base-lineTFP

TFPbasedonsales

TFPbasedonvalueaddedModelB0.4080.64

0.0250.880.2990.69−0.038−1.020.554778

−0.002−0.060.2670.67−0.005−0.120.452577ModelD

3-factorTFP

ModelB

GOV=NEDSGOV=CEO_STKPAYCOMPETITIONGOV*COMPETITIONAdjustedR2OBS

0.6311.74⁎−0.877−2.02⁎⁎0.7711235

0.7922.52⁎⁎

ModelDModelB0.3890.61

ModelDModelB

1.0893.06⁎⁎⁎

ModelD

0.0122.40⁎⁎0.1300.54−0.014−1.82⁎0.7811235

0.8851.33−0.837−1.98⁎0.5647780.9341.57−0.634−1.500.4655770.6031.46−1.328−2.70⁎⁎⁎0.7931235

0.0060.380.2150.75−0.010−0.480.7981235

of0.860onHHI_salesreportedinModelAindicatesthataone-standarddeviationincreaseinHHI_salesisassociatedwitha0.129(5.2percentage)increaseinTFP.

ThisaverageimpactofexternalcompetitiononTFPbecomesinsigniﬁcantonceweaddinternalgovernanceandtheinteractionbetweengovernanceandcompetition.InModelB(non-executivedirectors)andModelD(CEOstock-basedcompensation),goodgovernanceisstatisticallypositivelyrelatedtobetterﬁrmproductivity,whiletheinteractiontermisnegativeandsigniﬁcant.Intermsoftheeconomicsigniﬁcanceofkeyvariables,thecoefﬁcientof0.792onNEDSreportedinModelBindicatesthataone-standarddeviationincreaseinNEDSisassociatedwitha0.166(6.7percentage)increaseinTFP.Whilethecoefﬁcientof0.012onCEO_STKPAYreportedinModelDindicatesthataone-standarddeviationincreaseinCEO_STKPAYisassociatedwitha0.061(2.5percentage)increaseinTFP.Withrespecttotheinteractionterms(i.e.additional

Table8

Robustness—alternativemeasuresofinternalgovernance.Thistablepresentsregressionresultsofproductivity(TFP)onmeasuresofinternalcorporategovernance(BODSIZEandCEO_INCENT),productmarketcompetitiveness(COMPETITION),theirinteractiontermsandﬁrmlevelcontrolvariables(ROA,SIZE,AGE,GROWTH,andLEVERAGE).VariablesareasdeﬁnedinAppendixA.Inallmodels,weincludeﬁrmﬁxedeffectswithstandarderrorsrobusttoclusteringbyindividualﬁrmsovertime.Regressioncoefﬁcientsandtheirassociatedt-valuesarereported,with*,**and***indicatingstatisticalsigniﬁcanceat10%,5%and1%respectively.

ModelB

GOV=BODSIZEGOV=CEO_INCENTCOMPETITIONGOV*COMPETITIONROASIZEAGEGROWTHLEVERAGEAdjustedR2OBS

0.4580.31−0.014−0.510.0041.64⁎−0.055−1.91⁎0.0104.79⁎⁎⁎−0.038−2.71⁎⁎⁎0.0031.490.7651235

0.0381.51

0.0610.130.1910.78−0.381−0.570.0041.75⁎−0.082−2.81⁎⁎⁎0.0124.97⁎⁎⁎−0.037−2.53⁎⁎0.0041.67⁎0.6801235ModelD

G.Y.Tian,G.Twite/JournalofCorporateFinance17(2011)403–417413

Table9

Quietlifevs.empirebuilding.Thistabledistinguishesbetweenthe“quietlife”and“empirebuilding”hypotheses.Inaddressingthe“quietlife”hypothesis,weconsiderLabour/Assets,theratiooflabourcostovertotalassetandR&D/Assets,theratioofR&Dinvestmentovertotalasset.While,inassessing“empirebuilding”hypotheses,weconsiderCapEx/Assets,theratioofcapitalexpenditureovertotalassetandFCF/Assets,theratiooffreecashﬂowovertotalasset.PanelApresentsthemeansofeachofthenumericalmeasuresofTFP,quietlifeandempirebuilding,forsub-samplesofcompetitiveandnon-competitiveindustries,wherecompetitivenessisdeﬁnedusingtheindicatorvariableCOMPETITION.Wetestforthedifferencesinmeansbetweenthetwosub-samplesusingat-test.PanelBpresentstheresultsoftheﬁxedeffectregressionofthemeasuresof“quietlife”(Labour/Assets,R&D/Assets)and“empirebuilding”(CapEx/Assets,FCF/Assets)oninternalcorporategovernance(NEDS),productmarketcompetitiveness(COMPETITION)andtheirinteractionterms.VariablesareasdeﬁnedinAppendixA.Inallmodels,weincludeﬁrmﬁxedeffectswithstandarderrorsrobusttoclusteringbyindividualﬁrmsovertime.InPanelB,regressioncoefﬁcientsandtheirassociatedt-valuesarereported.*,**and***indicatingstatisticalsigniﬁcanceat10%,5%and1%respectively.PanelA:differenceinmeans

TFP

Non-competitive(D=0)Competitive(D=1)DifferenceinmeansT-stat

PanelB:ﬁx-effectregression

Labour/assets

GOV=NEDSCOMPETITIONGOV*COMPETITIONAdjustedR2OBS

−0.007−0.250.1420.880.0490.260.4191235

R&D/assets−0.052−1.38−0.002−0.360.0051.000.345758

CapEx/assets0.0371.12−0.041−1.070.0210.450.5031235

FCF/assets−0.008−0.22−0.041−1.04−0.003−0.060.2801235

2.7342.1950.5394.33⁎⁎⁎

Labour/assets0.1570.282−0.125−3.39⁎⁎⁎

R&D/assets0.0040.005−0.001−0.48

CapEx/assets0.2070.2000.0070.75

FCF/assets0.0910.111−0.020−2.39⁎⁎

impactofgovernanceforﬁrmsincompetitiveindustries),thecoefﬁcientof−0.877ontheinteractionofNEDSandCOMPETITIONreportedinModelBindicatesthataone-standarddeviationincreaseintheinteractiontermisassociatedwitha0.184(7.4percentage)decreaseinTFP.Whilethecoefﬁcientof−0.014ontheinteractionofCEO_STKPAYandCOMPETITIONreportedinModelDindicatesthataone-standarddeviationincreaseintheinteractiontermisassociatedwitha0.071(2.9percentage)decreaseinTFP.Thus,externalcompetitionreducestheneedforandbeneﬁtofinternalgovernanceinpromotingproductionefﬁciency.

InModelC,ownershipconcentrationisnegativelyrelatedtobetterﬁrmproductivity,whiletheinteractiontermispositivebutinsigniﬁcant.Thenegative,albeitweak,relationshipbetweenconcentratedinsiderownershipandﬁrmproductivityisconsistentwithpriorevidenceontherelationshipbetweenblock-holdingsandﬁrm'sﬁnancialperformance(Phametal.,2008)andfamilyownershipandtotalfactorproductivity(Barthetal.,2005).Theregressioncoefﬁcientof−0.006onOWN_PCTindicatesthataone-standarddeviationincreaseinOWN_PCTisassociatedwitha0.141(5.7percentage)decreaseinTFP.

Overall,theeconomicsigniﬁcanceoftheresultspresentedinTable5isverysimilartotheeconomicsigniﬁcanceoftheresultspresentedinTable4andtheregressionresultsconcerningcontrolvariablesarealsosimilartothosereportedinTable4.

Hence,ourresultssuggestthat,overall,internalcorporategovernanceisaneffectiveinstrumenttoimproveﬁrm

productivity.However,internalgovernanceislesseffectivewhenaﬁrmfacesacompetitiveproductmarket.Thereisastrongsubstitutioneffectbetweeninternalgovernanceandexternalcompetition.

Giventhatactivemarketforcorporatecontrolmayserveasanalternativewayofimposingtougherexternaldiscipline,werepeataboveanalysesusingindustry-leveltakeoveractivitiesinAustralia.Table6isstructuredinthesamefashionasTable5,replacingHHI_salesscoreswiththecontinuousmeasureoftakeoverintensitywithineachindustry,M&A,andthestateindicatorofexternalmarketdisciplinewiththeindicatorvariableMKTCONTROL.Iftakeoverthreatrepresentsaneffectivemechanismtodisciplinemanagers,wewouldexpecttoﬁndsimilarresultsasinTable5.

Doesmarketforcorporatecontrolmattertoﬁrmproductivity?ResultsfromTable6suggestthat,inAustralia,industry-leveltakeoveractivitiesdonothavedirectimpactonTFP.Inaddition,whenthesampleisseparatedusingindicatorvariableMKTCONTROL,neitherinternalgovernancenoritsinteractionwiththeindicatorisstatisticallysigniﬁcant.Regressioncoefﬁcientsofﬁrm-levelcontrolsaresimilartothosepresentedinTable5;however,thesigniﬁcanceofgrowthopportunitieshasincreased.

Insummary,weﬁndthatinternalcorporategovernanceaffectsﬁrmproductivitydifferentlyincompetitiveandnon-competitiveindustries.Incompetitiveindustries,marketcompetitionreducestheopportunityformanagerialslackorinefﬁciency,andthusweakenstheeffectivenessofinternalgovernance.Ontheotherhand,innon-competitiveindustries,internalcorporate

414G.Y.Tian,G.Twite/JournalofCorporateFinance17(2011)403–417

governance–measuredeitherbyboardindependenceorCEOstock-basedcompensation–ismoreusefulandhasapositiveinﬂuenceonTFP.TheseﬁndingsareconsistentwithUSstudiesusingﬁnancialperformancemeasures,suchasHart(1983),Chouetal.(2008),andGiroudandMueller(2009a).

IncontrasttoUSevidenceregardingtheeffectivenessofmarketforcorporatecontrol,wedocumentthatinAustraliatakeoverthreatdoesnotinﬂuenceﬁrmproductivityeitherdirectlyorindirectlyviathelinkbetweeninternalcorporategovernanceandproductivity.Thisisnotentirelysurprising.IntheUS-type“outsidersystem”,marketforcorporatecontrolcanserveasaneffectivemethodofmanagerialdiscipline.Australia,ontheotherhand,ismoreofan“insidersystem”.Dignam(2005)ﬁndthatovertheperiodof1992–2001,therewereroughly400takeoversinAustraliabutonly7%ofthemweresuccessfulhostilebids.Inoursample,theaverageindustry-leveltakeoverthreatmeasureislessthan2%.HencetheAustralianmarketischaracterizedbyaverysmallnumberofhostiletakeovers,whichinturnsuggeststhattheeffectivenessoftakeovermarketdisciplinarymechanism,whichprevailedintheU.S.,ismostlikelyabsentinAustralia.4.3.Robustnesschecks

Wemeasureﬁrmperformanceusingtotalfactorproductivity,TFP.OurﬁrstrobustnesscheckconcernsalternativedeﬁnitionsofTFP.Forinstance,studiesonproductivitycommonlyusenumberofemployees,whileourdeﬁnitionofTFPconsidersdollaramountofbothlabourandcapitalinputs.Ourchoiceisduetodatalimitations,aswewilllosemorethanathirdofoursampleifweadoptthisalternativemeasure.Next,weconsiderathirdinputvariable,dollaramountofmaterialscost,inestimatingEqs.(1)and(2).RegressionresultsarequalitativelythesameasthosepresentedinTables4–6.19WealsotestTFPformulasusingvalueaddedandsalesastheoutput,respectively.RegressionresultsconcerninggovernanceandmarketcompetitionbecomelesssigniﬁcantforthesetwoalternativeTFPmeasures,butthecoefﬁcientshavesignsthatareconsistentwithourmainﬁndings.20RelevantrobustnesstestresultsaresummarizedinTable7.Forbrevity,wedonotreportthecoefﬁcientsonﬁrm-levelcontrols,whichareconsistentwiththosereportedinTable5.

Inthemainanalysespresentedearlier,wehavechosen,oneachdimensionofinternalgovernance,onespeciﬁcmeasure.Asasecondrobustnesscheck,weconsideralternativeinternalcorporategovernancemetrics.AsanalternativetoNEDS,weconsiderBODSIZE,deﬁnedasthelogarithmofnumberofdirectorssittingontheﬁrm'sboard.Wepositthatsmallerboardsaremoreefﬁcientandthusassociatedwithhigherproductivity.21AsanalternativetoCEO_STKPAY,weconsiderCEO_INCENT,percentageofCEOtotalcompensationthatisstock-based.

Table8presentsregressionresultsusingthesealternativegovernancemetrics,withexternaldisciplinebeingmeasuredbyproductmarketcompetitiveness.However,boardsize,percentageofstock-basedcompensationandtheirinteractionswiththeproductmarketcompetivenessindicatorareallstatisticallyinsigniﬁcant.

Resultsofthefollowingrobustnesstestsarenottabulatedforbrevity.TheoverallmedianHHI(sales)inoursampleis0.195,andthesescoresarebasedon4-digitGICSindustryclassiﬁcation.Whenwere-calculatethesescoresusing2-digitGICS,theoverallmedianHHIdropsto0.189.TheAnti-trustDivisionoftheJusticeDepartmentintheU.S.usesHerﬁndahlindextomonitormarketconcentrationandtoavoidpotentialcollusion,andanindexvaluebeyond0.180isconsideredanticompetition.Toensurethatourresultsarerobusttodifferentdeﬁnitionsofcompetitiveness,weresetthebinaryindicatorCOMPETITIONtobeoneif(a)aﬁrmoperatesinanindustrywhoseHHIscoreisinbottomtercileinaparticularyear,(b)HHI_salessitsbelowthemedianvalue,insteadofinthebottomtercile,inaparticularyear,or(c)HHI_salesislessthanorequalto0.180.RegressionresultsareingeneralslightlyweakerbutqualitativelyconsistentwiththemainﬁndingsreportedinTable5.Inaddition,wealsorepeatallregressionsusingmarketpowerconcentrationmeasurebasedonassetsinsteadofsales.Ourkeymessageremains:thereisasubstitutioneffectbetweenproductmarketcompetitionandcorporategovernance.

Totieourﬁndingstopriorstudiesconcerningﬁrm'sﬁnancialperformance,weconsiderROAandTobin'sQasalternativedependentvariables.22Theresultsfromtheseadditionalrobustnesstestsareconsistentwithourkeyresult,inthatincompetitiveindustries,externalmarketdisciplinereducestheneedandeffectivenessofﬁrm-levelcorporategovernance.WithrespecttoROA,COMPETITIONhasapositiveandmarginallysigniﬁcantimpactat10%level;ourmeasuresofinternalcorporategovernancearegenerallypositivebutinsigniﬁcant,whiletheinteractiontermissigniﬁcantlynegativewheninternalgovernanceismeasuredonthedimensionofboardindependence.ResultsforTobin'sQ,proxiedasmarketvalueofequityplus

Inapreviousversion,TFPwasindeeddeﬁnedusingallthreeinputfactors.WethankananonymousrefereeforcorrectlypointingoutthatmoststudiesonTFPconsideronlytwofactors.20Firmproductivitydeﬁnedusingthesealternativemeasuresarehighlycorrelatedwithourbase-lineTFPmeasure.PearsoncorrelationcoefﬁcientsofthesealternativemeasureswithTFPareallsigniﬁcant,withintherangeof0.730to0.860.21Anotherboardvariable,duality,whichhasbeenusedextensivelyintheexisting(U.S.)literatureandexpectedtohavenegativeimpactonﬁrmperformance,isnotsuitableforourstudy.AustraliancompanieshavealongtraditionofseparatingCEOsandboardchairmen,thusmakesthedualitymeasureextremelyskewed—morethan98%ofoursampleﬁrmshaveboardchairmenthatarenotcompanyCEOs.22WhenregressingROA,weconsidermodelsbothwithandwithoutlaggedROA.Likewise,withTobin'sQ,weconsidermodelsbothandwithoutthelaggedmarket-to-bookequityratio.ThecorrelationbetweenTFPandROAis0.300,andbetweenTFPandTobin'sQis−0.099.ThenegativecorrelationbetweenTFPandTobin'sQisincontrasttotheresultsreportedbyPaliaandLichtenberg(1999)whoﬁndapositiveassociationbetweenTFPandTobin'sQ.Seefootnote12foronepotentialexplanationforthisdifferenceinresults.

19G.Y.Tian,G.Twite/JournalofCorporateFinance17(2011)403–417415

totaldebtovertotalassets,arelessconclusive.Intermsofsignsoftheregressioncoefﬁcients,corporategovernanceispositivelyassociatedwithQ,whileCOMPETITIONandtheinteractiontermsarebothnegativelyassociatedwithQ.Theinteractiontermisonlymarginallysigniﬁcanceat10%levelwheninternalgovernanceismeasuredintermsofboardindependence.

Concerningtheimportanceofownershipstructure,priorresearchhasdocumentedacurvilinearrelationshipbetweenﬁnancialperformance(e.g.Tobin'sQ)andshareownershipofcorporateinsiders(e.g.Morcketal.,1988).Asalastrobustnesstest,weincludethesquaredinsiderownershipconcentrationratioinourregressionmodelsrelatedtoTables5–6.Wedonotdetectthesigniﬁcantnon-linearrelationshipbetweenTFPandinsiderequityownershipdocumentedinU.S.studies.

Lastbutcertainlynotleast,weattempttoaddressthequestionofwhichhypothesis–quietlifeorempirebuilding–ismoreconsistentwithourﬁndings.Unfortunately,duringoursampleperiodof2000–2005,therearenosigniﬁcantexogenoussourcesofvariationsofinternalcorporategovernanceinAustralia.InthespiritofGiroudandMueller(2010),weconsidertwoproxiesforeachhypothesisandthenregresseachoninternalgovernance,externalmarketcompetitivenessandtheirinteractionterms.Inaddressingthe“quietlife”hypothesis,weconsiderLabour/Assets,theratiooflabourcostovertotalassetsandR&D/Assets,theratioofR&Dinvestmentovertotalassets.While,inassessing“empirebuilding”hypotheses,weconsiderCapEx/Assets,theratioofcapitalexpenditureovertotalassetsandFCF/Assets,theratiooffreecashﬂowovertotalassets.

PanelA,Table9presentsthemeansofeachofthenumericalmeasuresofTFP,quietlifeandempirebuilding,forsub-samplesofcompetitiveandnon-competitiveindustries,wherecompetitivenessisdeﬁnedusingtheindicatorvariableCOMPETITION.PanelB,Table9presentstheresultsoftheﬁxedeffectregressionofthemeasuresof“quietlife”(Labour/Assets,R&D/Assets)and“empirebuilding”(CapEx/Assets,FCF/Assets)oninternalcorporategovernance,productmarketcompetitivenessandtheirinteractionterms.

Whilstthereissomeindicationthatﬁrmsinnon-competitiveindustriesonaverageenjoylowerlabourcostsandpossesslowerfreecashﬂows,neitherinternalgovernance(asmeasuredinboardindependence)norexternaldiscipline(asmeasuredbyproductmarketcompetitiveness)haveastatisticallysigniﬁcantimpactonanyoneoftheseproxies.Hence,weareunabletodisentanglethetwohypotheses.

5.Conclusionanddiscussion

Theintegrationofinternalcorporategovernance,externalmarketdisciplineandﬁrmproductivityhaslargelybeenoverlookedinthecurrentdebateoncorporategovernance.Priorempiricalstudies,focusinglargelyonmetricsofﬁnancialperformance,havefoundthatbetterinternalgovernanceisassociatedwithenhancedﬁrmperformance.Existingliteraturealsopointsoutthatexternalproductmarketcompetitioncouldserveasasubstituteofinternalgovernance,inthatcompetitionmitigatesagencyproblems.

UtilizingasampleofAustraliancompaniesovertheperiodof2000–2005,wetesthowcorporategovernanceinﬂuencesanothermetricofﬁrmperformance,namely,productivity.WeﬁrstestimatetheinputsharesoflabourandcapitalcostsusingaCobb–Douglasproductionfunction,andattributetheresidualbetweenactualeconomicoutputandestimatedoutputasthemeasureofproductionefﬁciency,ortotalfactorproductivity(TFP).Next,weregressTFPonmeasuresofinternalcorporategovernance,externalmarketdisciplineandtheirinteractionterms,takingintoaccountﬁxedeffects.

Ourmainﬁndingssuggestthatinternalcorporategovernancecouldaffectﬁrmproductivity,howeverdifferentlyincompetitiveversusnon-competitiveindustries.Overall,bettergovernance–morenon-executivedirectors,smallerboardsandhigherCEOstock-basedcompensation–hasapositiveimpactonproductivity.However,whenaﬁrmoperatesinacompetitiveindustry,thetougherproductmarketcompetitionreducesagencyproblemsand,inturn,weakenstheeffectivenessofinternalcorporategovernance.Corporategovernancematterslittletoﬁrmproductivityduetothissubstitutioneffect.

Weﬁndonlyweakevidencesuggestingthattheinsiderownershipconcentrationratioservesasaneffectivegovernancechannel.Inaddition,unlikeproductmarketcompetitiveness,marketforcorporatecontrolasmeasuredbythelevelofindustry-widetakeoverthreatisfoundtohavenosigniﬁcantimpactonTFP;nordoesitservetosubstitutetheeffectivenessofinternalcorporategovernanceinpromotingbetterproductivity.

Acknowledgements

WegratefullyacknowledgetheﬁnancialsupportunderaFaultyResearchGrantfromAustralianSchoolofBusiness(PS15941),UniversityofNewSouthWales.WethankJeffryNetter(theEditor),ananonymousreferee,JosephFan,BruceGrundy,MarkHumphery,NingGong,TomSmith,SheridanTitmanandparticipantsatthe2010AsianFinanceAssociationConferenceandtheUNSWcorporategovernancePhDseminarfortheirinsightfulsuggestions.WeareparticularlyindebtedtoRonMasulisforhishelpfulcommentsandsuggestionsthroughouttheevolutionofthispaper.RockyHeandPingYuhaveprovidedexcellentresearchassistance.Allremainingerrorsareoursalone.

416G.Y.Tian,G.Twite/JournalofCorporateFinance17(2011)403–417

AppendixA.Variabledeﬁnition

VariableTFP

OutputLabourCapitalGOV

OWN_pct

Variablelabel

TotalfactorproductivityOutput

LabourinputCapitalinput

Variabledeﬁnition

OperatingrevenueCostofemployees

Costofcapitalcalculatedbasedonnetproperty,plantandequipment(PP&E)

DatasourceORBISORBISORBIS

Internalcorporategovernance

OwnershipconcentrationratioPercentagecommonequityheldbyinsidersasagroup,

asreportedinWorldScope

NEDSNon-executivedirectorsPercentageofnon-executivedirectorsontheﬁrm'sboardBODsizeCorporateboardsizeTotalnumberofdirectorssittingontheboard

CEO_incentiveCEOstock-basedremunerationPercentageofCEO'stotalcompensationthatisstock-basedCEO_stockCEOstock-basedremunerationLogarithmofdollarvalueCEO'sstock-basedcompensationplusoneMKTStateofexternalmarketdisciplineHHI_salesHHIindex,salesHerﬁndahl–HirschmanIndex(4-digitGICS)basedonﬁrm'ssalesHHI_assetHHIindex,assetHerﬁndahl–HirschmanIndex(4-digitGICS)basedonﬁrm'stotalassetsCompetitionCompetitionindicatorIndicatorvariableforcompetitiveproductmarket.Itequalsone

ifaﬁrmoperatesinalow-HHIindustry,zerootherwise

M&ATakeoverratioNumberoftotaltakeovertransactions(successful&failed)

dividedbynumberoftotalactiveﬁrmswithinanindustry

MktControlTakeoveractivityindicatorIndicatorvariableforactivemarketforcorporatecontrol.Itequals

oneifaﬁrmoperatesinahigh-M&Aindustry,zerootherwise

CONTROLFirm-levelcontrolvariablesROAReturnonassetsReturnonassets,afterdepreciationandtaxesSizeAgeGrowthLeverage

FirmsizeFirmage

GrowthopportunitiesFinancialleverage

Logarithmofﬁrm'stotalassetsplusone

LogarithmofnumberofyearssinceincorporationMarket-to-bookratioofcommonequityRatiooftotalliabilitiestototalassets

WORLDSCOPECONNECT4,

CONNECT4,CONNECT4,CONNECT4,

owncalculationowncalculationowncalculationowncalculation

DATANALYSIS,owncalculationDATANALYSIS,owncalculationOwncalculation

DATANALYSIS,owncalculationOwncalculation

ORBIS&DATANALYSIS,calculation

ownownownownown

AppendixB.HHIat4-digitGICSindustrylevel

4-digitGICS101015102010202020302510252025302540255030103020303035103520401040204030404045104520453050105510

Industrygroup

EnergyMaterialsCapitalgoods

Commercialservices&suppliersTransportation

Automobiles&componentsConsumerdurables&apparelConsumerservicesMediaRetailing

Food&staplesretailingFood,beverage&tobacco

Household&personalproductsHealthcareequipment&servicesPharmaceuticals&biotechBanks

DiversiﬁedﬁnancialsInsuranceRealestateITsoftware

Communicationsequipment&servicesOfﬁceelectronics&othersTelecommunicationservicesUtilities

2-digtGICS101520202025252525253030303535404040404545455055

Sector

EnergyMaterialsIndustrialsIndustrialsIndustrials

ConsumerdiscretionaryConsumerdiscretionaryConsumerdiscretionaryConsumerdiscretionaryConsumerdiscretionaryConsumerstaplesConsumerstaplesConsumerstaplesHealthcareHealthcareFinancialsFinancialsFinancialsFinancials

InformationtechnologyInformationtechnologyInformationtechnologyTelecommunicationUtilities

HHI_sales0.2440.0950.1210.1160.2090.1860.1090.0800.4480.0590.2280.6860.7210.1970.404n.a.n.a.n.a.n.a.0.0720.2000.3150.3520.210

HHI_asset0.1130.0920.1390.1820.1560.2170.1480.0820.4870.1010.1590.6050.6230.1630.258n.a.n.a.n.a.n.a.0.0930.1610.2590.3430.116

G.Y.Tian,G.Twite/JournalofCorporateFinance17(2011)403–417417

References

Aggarwal,R.,Samwick,A.,1999.Executivecompensation,strategiccompetition,andrelativeperformanceevaluation:theoryandevidence.J.Finance.54,

1999–2043.

Aghion,P.,Bloom,N.,Blundell,R.,Grifﬁth,R.,Howitt,P.,2005.Competitionandinnovation:aninverted-Urelationship.Q.J.Economics.120,701–728.Alchian,A.,1950.Uncertainty,evolutionandeconomictheory.J.Polit.Economy.58,211–221.

Allen,F.,Gale,D.,2000.Corporategovernanceandcompetition.In:Vives,X.(Ed.),CorporateGovernance:TheoreticalandEmpiricalPerspectives.Cambridge

UniversityPress,Cambridge,pp.23–94.

Barth,E.,Gulbrandsen,T.,Schone,P.,2005.Familyownershipandproductivity:theroleofowner–manager.J.Corp.Finance.11,107–127.Bebchuk,L.A.,Cohen,A.,Ferrell,A.,2009.Whatmattersincorporategovernance?Rev.Financ.Studies.22,783–827.

Bertrand,M.,Mullainathan,S.,2003.Enjoyingthequietlife?Corporategovernanceandmanagerialpreferences.J.Polit.Economy.111,1043–1075.Bhagat,S.,Bolton,B.J.,2008.Corporategovernanceandﬁrmperformance.J.Corp.Finance.14,257–273.

Borokhovich,K.A.,Parrino,R.,Trapani,T.,1996.OutsidedirectorsandCEOselection.J.Financ.Quant.Anal.31,337–355.Bulan,L.,Sanyal,P.,Yan,Z.,2007.CEOIncentivesandFirmProductivity,Workingpaper.BrandeisUniversity.

Chou,W.,Ng,L.,Wang,Q.,2008.ProductMarketCompetitionandCorporateGovernance,Workingpaper.GeorgiaInstituteofTechnology.Coles,J.L.,Daniel,N.D.,Naveen,L.,2008.Boards:doesonesizeﬁtall?J.Financ.Economics.87,329–356.

Conyon,M.,Peck,S.,1998.Boardsizeandcorporateperformance:evidencefromEuropeancountries.Eur.J.Finance.4,291–304.

Core,J.E.,Holthausen,R.W.,Larcker,D.F.,1999.Corporategovernance,chiefexecutiveofﬁcercompensation,andﬁrmperformance.J.Financ.Economics.51,

371–406.

Dignam,A.,2005.Thetakeoverpanel,themarketefﬁciencyprincipleandthemarketforcorporatecontrol—anempiricalstudy.CompanySecuritiesLawJournal.

23,58–64.

Dignma,A.,Galanis,M.,2004.Australiainside-out:thecorporategovernancesystemoftheAustralianlistedmarket.MelbourneUniv.LawReview.28,623–653.Giroud,X.,Mueller,H.M.,2009.CorporateGovernance,ProductMarketCompetition,andEquityPrices.CEPRDiscussionPaperNo.6974.Giroud,X.,Mueller,H.M.,2010.Doescorporategovernancematterincompetitiveindustries?J.Financ.Economics.95,312–331.Gompers,P.,Ishii,J.,Metrick,A.,2003.Corporategovernanceandequityprices.Q.J.Economics.118,107–155.Hart,O.D.,1983.Themarketmechanismasanincentivescheme.BellJ.Economics.14,366–382.

Hartzell,J.C.,Starks,L.,2003.Institutionalinvestorsandexecutivecompensation.J.Finance.58l,2351–2374.

Hill,C.W.,Snell,S.A.,1989.Effectsofownershipstructureandcontroloncorporateproductivity.Acad.Manage.J.32,25–46.

Himmelberg,C.P.,Hubbard,R.G.,Palia,D.,1999.Understandingthedeterminantsofmanagerialownershipandthelinkbetweenownershipandperformance.

J.Financ.Economics.53,353–384.

Hulten,C.R.,2000.TotalFactorProductivity:aShortBiography.NBERWorkingPaper7471.

Jensen,M.C.,1993.Themodernindustrialrevolution,exit,andthefailureofinternalcontrolsystems.J.Finance.48,831–880.Jensen,M.C.,Murphy,K.J.,1990.Performancepayandtop-managementincentives.J.Polit.Economy.98,225–264.Kim,M.,Maksimovic,V.,1990.Debtandinputmisallocation.J.Finance.45,795–816.

Koke,J.,Renneboog,L.,2005.Docorporatecontrolandproductmarketcompetitionleadtostrongerproductivitygrowth?Evidencefrommarket-orientedand

blockholder-basedgovernanceregimes.J.LawEconomics.48,475–516.

Krugman,P.,1990.TheAgeofDiminishedExpectations:theUSEconomicPoliciesinthe1990s.MITPress,Massachusetts.LaPorta,R.,Lopez-de-Silanes,F.,Shleifer,A.,1999.Corporateownershiparoundtheworld.J.Finance.54,471–517.

Lamba,A.,Stapledon,G.,2003.TheDeterminantsofCorporateOwnershipStructure:AustralianEvidence,Workingpaper.UniversityofMelbourne.

Levin,R.,Cohen,W.,Mowery,D.,1985.R&DAppropriability,opportunityandmarketstructure:newevidenceonsomeSchumpeterianhypotheses.Am.Econ.Rev.

Proceedings.75,20–24.

Lichtenberg,F.,Siegel,D.,1987.Productivityandchangesinownershipofmanufacturingplants.In:Lichtenberg,F.,Siegel,D.,Jorgenson,D.,Mansﬁeld,E.(Eds.),

SpecialIssueonMicroeconomics.BrookingInstitution,Washington,pp.643–683.

Maksimovic,V.,Phillips,G.,1996.EfﬁciencyofBankruptFirmsandIndustryConditions:TheoryandEvidence.CESDiscussionPaper100248.

Martikainen,M.,Nikkinen,J.,Vahamaa,S.,2009.Productionfunctionsandproductivityoffamilyﬁrms:evidencefromtheS&P500.Q.Rev.Econ.Finance49,

295–307.

Masulis,R.W.,Wang,C.,Xie,F.,2007.Corporategovernanceandacquirerreturns.J.Finance.62,1851–1889.

McConnell,J.,Servaes,H.,1990.Additionalevidenceonequityownershipandcorporatevalue.J.Financ.Economics.27,595–612.

Morck,R.K.,Shleifer,A.,Vishny,R.W.,1988.Managementownershipandmarketvaluation:anempiricalanalysis.J.Financ.Economics.20,293–315.Murphy,K.J.,1999.Executivecompensation.In:Ashenfelter,O.,Layard,R.(Eds.),HandbookofLaborEconomics.ElsevierScienceB.V,pp.2485–2563.Nickell,S.,1996.Competitionandcorporateperformance.J.Polit.Economy.104,724–746.

Nickell,S.,Nicolitsas,D.,Dryden,N.,1997.Whatmakesﬁrmsperformwell?Eur.Econ.Review.41,783–796.

Palia,D.,Lichtenberg,F.,1999.Managerialownershipandﬁrmperformance:are-examinationusingproductivitymeasurement.J.Corp.Finance5,323–339.Pham,P.K.,Suchard,J.A.,Zein,J.,2008.CorporateGovernanceAndAlternativePerformanceMeasures:EvidencefromAustralianFirms,Workingpaper.University

ofNewSouthWales.

Shleifer,A.,Vishny,R.W.,1986.Largeshareholdersandcorporatecontrol.J.Polit.Economy.94,461–488.Shleifer,A.,Vishny,R.W.,1997.Asurveyofcorporategovernance.J.Finance.52,737–783.Stigler,G.,1958.Theeconomicsofscale.J.LawEconomics.1,54–71.

Yermack,D.,1996.Highermarketvaluationofacompanywithasmallboardofdirectors.J.Financ.Economics.40,185–211.

Zhou,X.,2001.Understandingthedeterminantsofmanagerialownershipandthelinkbetweenownershipandperformance:comment.J.Financ.Economics.62,

559–571.

因篇幅问题不能全部显示，请点此查看更多更全内容

查看全文

全部栏目

Corporate governance, external market discipline and firm productivity