Institutionaltheory,corruption,theuseofderivatives,andfirmvalue Whileacknowledgingthecontributionsoffirmdeterminants,scholarsinthefieldofinternationalbusinesshavestressedtheimportanceofi
Trang 1J of Multi Fin Manag 39 (2017) 39–59
ContentslistsavailableatScienceDirect
Managementjournalhomepage:www.elsevier.com/locate/econbase
Huong Trang Kimb, Marina Papanastassioua,∗, Quang Nguyena
b Faculty of Banking and Finance, Foreign Trade University, 91 Chua Lang street, Dong Da, Hanoi, Vietnam; and Middlesex University
©2017ElsevierB.V.Allrightsreserved
1 Introduction
TheModiglianiandMiller(1958)theoremwithperfectcapitalmarketsshowsthatriskmanagementisirrelevanttofirmvalue; hedgingwithderivativesthusdoesnotadd valuetoafirm.However,numerousstudies(e.g.,Nanceetal.,1993;Frootetal.,1993;SmithandStulz,1985;MayersandSmith,1990;MayersandSmith,1982;Bessembinder,1991)suggestthattheuseofderivativescanserveasavalue-increasingstrategyforafirmbyreducingcostsbroughtaboutbymarketimperfections.Manyempiricalstudiesinthisareahavefocusedontherelationbetweenderivativesuseandfirmcharacteristicstoinvestigatehedgingtheoryandexplainwhyfirmsusederivatives.Onlyrecentlytherehasbeenanotherstrandofresearchexploringtheimpactofderivativesuseonfirmvalue
Mostofthesestudieshaveconcentratedonunconditionaleffectsofderivativesuseonfirmvalue(e.g.,AllayannisandWeston,2001;GuayandKothari,2003;Bartrametal.,2011).Afewrecentresearchershavedevelopedteststoinvestigatevalueimplicationsofderivativesuseconditionaloncorporategovernanceoragencyproblems(e.g.,FauverandNaranjo,
Trang 2Fillingthisgapintheliterature,weexploretheuniquevalueeffectofderivativesuseforasampleof881non-financialfirmsineightEastAsiancountriesovertheperiod2003–2013withanewhand-collecteddatasetforderivativesuse.Wemakethefollowingcontributions:
First,wefocusonthevalueimplicationofderivativesuseundertheinfluenceofanenvironmentofcorruption.Extantstudiesaddressthedriversofvalueimplication,focusingonthestructuralcharacteristicsoffirm-specificresourcesandcapabilities.However,allfirmsareembeddedininstitutionalenvironments,i.e.,the“rulesofthegame”(North,1990,1994)
Akeyfactorinsuchenvironmentsiscorruption.Despitetheeffortsofgovernment,non-governmental,andmultilateralinstitutionstoreducecorruptionlevels,corruptionisawidespreadphenomenonworldwide,−whichinducesfirmstoengageinnon-marketstrategies(Dohetal.,2012)suchasbribery(Beets,2005).InEastAsiancountries,corruptionisaseriousproblem.In2013,64percentofthesecountriesscoredbelow50outof100inperceivedlevelsofpubliccorruption.1
Also,whileagreatnumberofstudiesexaminethecorrelationbetweencorruptionandeconomicgrowth,theeffectsofcorruptiononthevalueeffectofderivativesusearelittleknown
Theoretically,corruptionontheonehandcanactasa“grabbinghand”byincreasinguncertaintyandtransactioncosts(e.g.,Bardhan,1997;Quazi,2014),whichimpedesfirms’operations.Ontheotherhand,corruptionactsasa“helpinghand”
bygreasingthewheelsofcommerceandraisingeconomicgrowthinthepresenceofweaklegalandregulatoryframeworks(Bardhan,1997;Houston,2007),whichshouldimprovefirms’performance.Thesecontradictoryeffectsmayderivefromthevaryingdegreeofambiguityassociatedwithcorrupttransactionsindifferentcountries(Petrou,2014).Thus,withoutdirectlytakingintoaccountacorruptenvironment,itwouldbedifficulttodeterminewhethertheuseofderivativesisavalue-enhancingactivity.Undertakingthisresearch,weraiseimportantquestionsthathavereceivedlittleornoattention:Howdoescorruptioninfluencethevalueeffectofderivativesuse?Inlightofacorruptenvironment,doestheuseofderivativeincreasefirmvalue?
Second,overthepasttwodecades,scholarshaveexaminedandprovidedimportantinsightsintotheeffectsofderivativesuseonthevalueofnon-financialfirms.Yetimportantquestionsremain:Inwhattypeoffirmistheeffectgreater(orless),andwhatfactorsdeterminethisdifference?Inthisstudy,weshednewlightonthisgapbyexamininghowthevalueeffects
ofderivativesusemayvaryacrossforeign-ownedfirms(i.e.,foreignMNCaffiliates)anddomestic-ownedfirmsinlightof
anenvironmentofcorruption.Webreakdownthesubsetofdomestic-ownedfirmsintodomesticfirmsandmultinationalcorporations(henceforthMNCs).Althoughresearchininternationalbusiness(IB)haslongrecognizedthatbyvirtueofmultinationality,MNCshavedistinctive advantagesinbusinessoperationsvis-à-visdomestic firms(e.g.,Hymer,1976;CastellaniandZanfei,2006;AllayannisandWeston,2001),neitherIBnorfinanceresearchersprovideacomprehensiveanalysisofwhethertheuseoffinancialderivativesrewardsMNCswithhighervaluethandomesticfirms
Ontheotherhand,accordingtoCastellaniandZanfei(2006),foreignaffiliatesofanMNCarefirmswithparentcompaniesabroad,whiledomesticMNCsareeitherheadquartersornationalaffiliates.Thisimpliesthatadvantagesandcostsincurred
bydomesticMNCsandforeignMNCaffiliatesderiveprimarilyfromtheirdifferentorigins.Inparticular,inagivencountry,foreignMNCaffiliatesmightbeinabetterpositionthanlocalcounterpartsduetoownershipadvantagessuchastheirparentcompanies’expansivefinancialresources,accesstoequityandcapitalmarkets,orknowledge-basedcapabilities(Changetal.,2013;NguyenandRugman,2015).However,therealsoexistswell-supportedempiricalevidencethatforeignMNCaffiliatestendtobeatadisadvantagecomparedtotheirlocalcounterparts,astheyoftensufferfromvariouscostsofdoingbusinessabroadowingto“liabilityofforeignness”2;(e.g.,Hymer,1976;Zaheer,2002;CastellaniandZanfei,2006;HigónandAntolín,
2012).AsforeignMNCaffiliates’advantagesmayormaynotoffsetthosecosts,itremainsunknownwhether,undertheinfluenceofcorruption,foreignaffiliateswithderivativesactivitiesaremorevaluablethandomesticcounterparts.Third,weinvestigatehowthevalueimplicationofderivativesuseinacorruptenvironmentdifferedacrossdomesticfirms,domesticMNCs,andforeignMNCaffiliateswhentheyfacedexogenousshocksbroughtaboutbytheglobalfinancialcrisisof2007–2008.Thecrisiscausedsevereharmstotheworldeconomy,andincreasedvolatility,butthemagnitudeoftheeffectsofthecrisiswasdifferentacrossvariouscountriesandfirms.Althoughtherearenumerousstudiesonitseffects,littlehasbeendonetoanalyzeitsimpactsonderivativesuse.Wethuschosetheperiod2003–2013,whichprovidesanaturalexperimentoffinancialrisksandriskmanagement,toexaminetherelationshipdynamicbetweenderivativesuseandfirmvaluebefore,during,andafterthecrisis.Ourstudydoesnotmerelyinvestigateconsequencesoftheglobalfinancialcrisisforthevalueimplicationofderivativesuse,butalsoconcentratesontheroleofcorruptionlevelsinmitigatingadverseconsequences
Themainfindingsofourstudyareasfollows.ResultsofOLSestimation,aftercontrollingforendogeneityandselectionbias,consistentlyrevealthatlowcorruptionlevelsinducetheuseoffinancialderivativesandrewarddomesticfirmsanddomesticMNCswithhighervalue.Inparticular,derivativesuseisavalue-enhancingactivityfordomesticfirms
self-1 Transparency International, 2013 , Corruption Perception Index (CPI) is issued by the Transparency International This index is inversely ranked from 0
to 100, where a higher number indicates a lower level of corruption.
2 Liability of foreignness in a host country are defined as “all additional costs a firm operating in a market overseas incurs that a local firm would not incur” ( Zaheer, 1995, p 343 ) These costs are directly related to institutional distance and foreign affiliates’ weak links to the local institutional setting ( Zaheer, 2002; Bell et al., 2012; Higón and Antolín, 2012 ).
Trang 3H.T Kim et al / J of Multi Fin Manag 39 (2017) 39–59 41
anddomesticMNCs,anditincreasesthevalueofthosefirmsbetween9.87percentand11.77percent,andbetween10.78percentand12.72percentrespectivelywhentheyareinahomecountrywithlowlevelsofcorruption.3Relatedly,althoughderivativesusedoesnotaddvaluetoforeignMNCaffiliates,wefindthattheyaremorevaluableinhostcountrieswherecorruptionislesssevere.4WenoticethatthehedgingactivitiesofdomesticMNCsoutperformdomesticfirmsandforeignMNCaffiliatesintermsoffirmvalueundertheinfluenceofcorruption.Ourempiricalanalysisalsoshowsthatduringthecrisisperiod,thebenefitofderivativesusetofirmvaluedoesnotaffectallfirmtypes,andtheeffectoflowcorruptionlevels
onalleviatingthenegativeimpactsofthecrisisonderivativesuseisverymoderate.YetlowcorruptionlevelsinthehomecountryarepositivelyassociatedwiththehedgingpremiumsofdomesticfirmsanddomesticMNCsinthepost-crisisperiod
Wewilldevelopthisstudyasfollows.Section2reviewstheexistingliteratureanddevelopshypotheses.Section3cussesthesampleforthisstudy,identifiesdependentvariablesandexplanatoryvariables,andprovidessummarystatistics.Section4describesmodelspecifications.Section5presentstheempiricalresultsandananalysisoftherelationshipbetweenderivativesuseandfirmvalue.Section6summarizesthestudyanddrawsconclusions
dis-2 Theoreticalframeworkandhypotheses
2.1 Derivativesuseandfirmvalue
Hedgingtheoryarguesthatthereispotentialpositiverelationshipbetweenderivativesuseandfirmvalue.Thatlinkdependsonthedegreetowhichtheuseofderivativeseffectivelyaddressesmarketimperfectionssuchascorporatetaxes(seeSmithandStulz,1985;MayersandSmith,1990),financialdistressorbankruptcycosts(Nanceetal.,1993;Frootetal.,
1993),ortheagencycostsofdebts(MayersandSmith,1982;Bessembinder,1991).Ontheotherhand,Stulz(1996)postulatesthattheprimaryobjectiveofriskmanagementistomitigatethelikelihoodofcostlylower-tailoutcomesthatmayimposefinancingconstraintsorimpedeafirmfromcarryingoutitsinvestmentprojects.Financialderivativesarerisk-managementinstruments,soiftheuseofderivativesgeneratespositivecashflowsorvalue,thenthosederivativesaredeemedtohedgeagainstafirm’srisks,leadingtoanincreaseinfirmvalue
Although90percentofFortune500firmshavebeenusingderivativesandderivativestradingintheAsia-Pacificregionaccountsforapproximatelyone-thirdoftradingvolumeacrosstheworld,5andtheirusecontinuestoincrease,theinfluence
ofderivativesuseonfirmvaluehasnotreceivedmuchattentionuntilrecently(Bartrametal.,2011).Sofar,theliteratureregardingthevalueeffectofderivativesusepresentsmixedresultsoverall.Alargebody ofpreviousstudiesinvariablyfocusedontheunconditionalvalueeffectsofderivativesuse,andwasbroadlyconsistentwiththenotionthatderivativesuseisavalue-increasingactivity(e.g.,AllayannisandWeston,2001;ClarkandJudge,2009;Campelloetal.,2011;ChenandKing,2014),buttheestimatedmagnitudesofthehedgingpremiumvaryacrossthesestudies,rangingfromaslowas1.8percenttoashighas34percent.Someothershavefoundanegativerelationshipbetweentheuseofderivativesandfirmvalue(e.g.,NguyenandFaff,2010;Supanvanij,2011),ornoeffect(e.g.,Belghitaretal.,2013).Thus,someresearchersraisedoubtsabouttherelationshipbetweenderivativesuseandfirmvalue,andthefindingsofearlierstudies(GuayandKothari,
2003)
Recently,someresearchhasimplementedconditionalteststoscrutinizethevalueimplicationsofderivativesuseinlight
ofsomecorporateissuessuchascorporategovernanceoragencyproblems(e.g.,FauverandNaranjo,2010;Allayannisetal.,
2012);asFauverandNaranjo(2010)arguethatfirmscanhaveinternalproblemsthatmayleadtoapotentiallossinfirmvalue,thenetvalueimpactofderivativesuseisanempiricalissue
InthecontextofEastAsia,thereisonlylimitedempiricalevidenceregardingtheassociationbetweenderivativesuseandfirmvalue.Tothebestofourknowledge,nostudydirectlyinvestigatesthatrelationshipacrosscountriesinEastAsia,butafewexaminethevalueeffectsofderivativesuseacrossindustriesinonecountry.Ameer(2009),usingasampleof
40Malaysianfirms,findsthatalthoughtheuseofderivativeshasvaluerelevance,itscontributiontoafirm’svaluation
isminimal.Inaninvestigationofbothfinancialandnon-financialfirmsinIndonesia,Oktavia(2012)findsthattheuseofderivativesbynon-financialfirmscanenhanceshareholdervalue,butdoesnotaffecttheearningsofbanks
2.2 Institutionaltheory,corruption,theuseofderivatives,andfirmvalue
Whileacknowledgingthecontributionsoffirmdeterminants,scholarsinthefieldofinternationalbusinesshavestressedtheimportanceofinstitutionalvariationacrosscountriesandclaimedthatcountry-specificfactorscontributetoexplainingdifferencesintheperformanceandbehavioroffirms.Henisz(2000)andEdenandMiller(2004)assertthatdifferentialtreatmentbyhomeorhostgovernmentsorthegeneralpublicinaspecificcountrymightleadtodiscriminatoryhazards,whichmayplaceafirmatadvantageandbecomemoreopportunisticthanotherfirms.North(1990,1994),whowasamong
3 We define corruption levels based on the Corruption Perception Index (CPI), in which those countries having scores of greater than 75 are considered
to be countries with low levels of corruption, and those countries having scores of less than 75 are considered to be highly corrupt.
4 Our findings suggest that a corrupt environment may have a stronger effect on the link between derivatives use and firm value than other country-specific factors such as GDP per capita or the development of financial markets.
5 FIA 2015 FIA annual volume survey Future Industry Association.
Trang 42005)
However,governmentsandinstitutionsneverworkonlyforpublicbenefit,butalsopursueillegitimateobjectives,oftenthroughcorruption(Dohetal.,2003;Rodriguezetal.,2005).Corruptionreflectsthestateandnormsofpoliticalinstitutions;thus,corruptionobtainstodifferentdegreesindifferentcountries,andcountriesdifferremarkablyintheextenttowhichcorruptionaffectsfirms’performance.6Thetheoryhereisthatcorruptioncanhaveastrongeffectonhedgingcostsandothertransactioncosts,therebyreducingexpectedcashflows,and,ultimately,thevalueoffirmsusingderivatives.Thisisconsistentwithextantstudies(e.g.,HabibandZurawicki,2002;ShleiferandVishny,1993)demonstratingthatcorruption
isconsideredtobeamorecostlytaxonbusinessoperationsthanlegaltaxesbecausecorruptionconstrainsfirmstowastetheirresourcesonunproductiveactions(KaufmannandWei,1999).Assuch,corruptionmaydramaticallyincreasetheriskperceptionofcapitalmarketparticipantsandaffectexpectedreturnsbecauseexpectedexcessmarketreturnsvarypositivelywithmarketvolatility.Alongthisline,LeeandNg(2006)findthatacountry’slevelofcorruptionisnegativelyrelatedtoexpectedcashflows,andmayhaveasubstantiallyadverseimpactonshareholdervalue
Furthermore,althoughtodatetheexistingliteraturehasbeensilentabouttheinfluenceofcorruptiononvalueeffects
ofderivativesuse,someresearchers(e.g.VenardandHanfi,2007;Foss,2010;Petrou,2014;amongothers)observethatfirms’exposuretocorruptcountriesintheirbusinessoperationsmaytranslateintorealfinanciallosses.Moreclearly,firmsthatoperateincountrieswithhighlevelsofcorruptionarelikelytoengageincostlymarkettransactionsandless-efficienttransformationbecausesuchcountriesarelikelytohavelower-qualityinfrastructureservices,economicgrowth,andfinancialstability(Rose-Ackerman,1978,1999).Theexistingliteraturealsoshowsthatahigherlevelofcorruptionisassociatedwithhigherborrowingcosts,worsecorporategovernance,andlowerstockvaluation.Intheiranalyses,Donadelli
etal.(2014)revealthatfirmsoperatinginhighlycorruptcountriestendtohaverelativelylowreturns.Conversely,thosefirmsoperatingincountrieswithlowerlevelsofcorruptioncancapitalizeontheadvantagesgeneratedbyamorefavorableinstitutionalcontextforfirms,whichinturnhasapositiveinfluenceonfirms’performanceandprofitability(LevyandSpiller,1994;Bergaraetal.,1998).Recently,Kimetal.(2016)havefoundthatalowercorruptionlevelcorrelateswithhigherhedgingintensityinthecaseofEastAsianfirms
Buildinguponliteratureonbothderivativesuseandcorruption,wethereforehypothesizethefollowing:
Hypothesis1 Thelowerthecorruptionlevel,thehigherthelikelihoodthattheuseoffinancialderivativeswillincreasefirmvalue
Theuseofderivativesandfirmvaluefordomesticfirms,domesticMNCs,andforeignMNCaffiliates
Weconjectureintheprevioussectionthatthevalueeffectofderivativesuseislikelytobehigherforthosefirmsoperating
incountrieswithlowerlevelsofcorruption.However,derivativesuseundertheinfluenceofacorruptenvironmentdoesnothavethesamevalueeffectonalltypesoffirms.Itiswell-supportedintheinternationalbusinessliteraturethatinstitutionalenvironmentsshapefirms’behavioranddecisions,whetherthosefirmsareforeign-ownedordomestic-owned,andalsocontributetosignificantvariationsintheperformanceofdifferentfirmtypes(seeHenisz,2000;FeinbergandGupta,2009;EdenandMiller,2004).Henisz(2000)andFeinbergandGupta(2009)pointoutthatfirmsusuallyshiftresourcesandchangeinvestmentactivitiesinresponsetoinstitutionalhazards,whileEdenandMiller(2004)statethatduetoinstitutionaldistance,MNCsmightbeatadisadvantageevenwhenoperatinginthesameinstitutionalenvironments,andbearadditionalcosts
ofdoingbusinessthatlocalcounterpartsdonotencounter.Inthisstudy,weinvestigateadynamicrelationshipbetweenderivativesuseandfirmvalueoncomparisonofforeign-ownedfirms(i.e.,foreignMNCaffiliates)anddomestic-ownedfirms,thelatterofwhicharecategorizedintodomesticfirmsanddomesticMNCs
AparticularlyimportantdistinctionbetweendomesticMNCsandforeignMNCaffiliatesistheassertionthatthemer’sorganizationalroutinesandmanagementprocessarerootedinlocalinstitutionalconditions(Brutonetal.,2008),andthattheyhavewell-developedaccesstorelevantinformationandknowledgeregardingthelocalinstitutionalsettingthatencourageseffectiverunningoftheiroperations,includingknowledgeoftheeconomy,socialneedsandpreferences,norms,cultures,andlaw(Nachum,2003;Belletal.,2012);foreignMNCaffiliates,fortheirpart,areembeddedintheinstitutionalsettingsofboththeirhomeandhostcountries,sotheyoftenexperience“liabilityofforeignness”(Zaheer,2015).Ontheotherhand,inspiteofthefactthatbothdomesticMNCsanddomesticfirmsareactivelyembeddedinlocalinstitutionalsettings,thefundamentaldistinctionbetweenthemisbasedonmultinationality.Thus,wefollowHigónandAntolín(2012)
for-inarguingthatcomparisonofdomesticMNCsanddomesticfirmsrevealsamultinationalityeffect,whilecomparisonofdomesticMNCsandforeignMNCaffiliatesrevealstheeffectsofforeignness
https://www.transparency.org/what-is-corruption/
Trang 5H.T Kim et al / J of Multi Fin Manag 39 (2017) 39–59 43
First,IBliteratureoftenarguesthatMNCs,incomparisonwithdomesticfirms,possesssomefirm-specificadvantages(FSA)7 thatcompensateforthehighcostsofinternationaloperationandenablethemtocompetesuccessfully.Indeed,followingOLItheory(BuckleyandCasson,1976;Dunning,1977),IBscholarshavefoundthatMNCsshouldbeabletoexploitcostdifferentialsonaglobalscalethankstomultinationality(AllenandPantzalis,1996;Chungetal.,2010)
MultinationalitygivesdomesticMNCsanabilitytodiversifytheportfoliosoffirms,makingMNCs’cashflowslessdentondomesticmarket-levelcashflows(Krapl,2015),ultimatelydecreasingthevolatilityofacompany’scashflowsandearnings,andthusimprovingfirmvalue(Rugman,1976;KhannaandYafeh,2005).Byvirtueofmultinationality,domesticMNCsthemselvesarelikelytoreducetheprobabilityofbankruptcy(MichelandShaked,1986),overcomeinefficienciesarisingfromhighagencyandinformationcosts(KhannaandYafeh,2005),andprovideshareholderswithlowersystematicandidiosyncraticriskandsuperiorreturnopportunitiesrelativetodomesticfirms(Fatemi,1984).Furthermore,domesticMNCsalsodevelopexpertiseinhedgingrisksandaremorelikelytopossesssuperiorcapabilityforreducingexposuretomarketriskssuchasthoserelatedtoexchangeratesbyusingfinancialderivatives(e.g.,AllayannisandOfek,2001;ChoiandJiang,2009),whichleadstolowercapitalcosts.Also,MNCstransfermaterialandknowledgeresourcesamongaffiliates,therebynotonlyreducingthecostofacquisitionofthoseresourcesforaffiliatesbutalsocontributingtoahedgeagainstmarketrisks.TheseadvantagesmaycontributetoincreasingfirmvalueforMNCs
depen-Second,intermsofforeignMNCaffiliates,Kostovaetal.(2008)recognizethattheliabilityofforeignnessisthekeydriverbehindmanymisstepsandinefficiencies.ForeignMNCaffiliatesoftenbearhighercostsarisingfromtheirforeignness,such
ascostsandrisksrelatedtoexchangeratefluctuations,aswellashighagencyandinformationcosts,whicharisefromtheirlackofknowledgeaboutlocalculturesandnetworksthatcouldconnectthemwithimportantactorsinahostcountry’seconomy−ingeneral,costsarisingfromtheirweaklinkstotheinstitutionalsettingoftheirhostcountries(Hymer,1976;Zaheer,2002;Nachum,2003;CastellaniandZanfei,2006;HigónandAntolín,2012)
WefollowHigónandAntolín(2012)amongothersinassumingthatthekeydriverbehindforeignnessistheinstitutionaldistancebetweenhomeandhostcountry.Specifically,ithaslongbeenlearnedthatthephysicaldistancebetweentheheadquarterinadifferentcountryandforeignaffiliatesincreasesthegovernance,communication,andtravelcostsassociatedwithmanaginganoverseasoperation(Hymer,1976;Zaheer,2002).Intermsofhedgingactivities,suchphysicaldistancemayincreasehedgingcosts,therebyreducingthevalueeffectoffinancialderivativesuse.KhannaandPalepu(1997)alsonotethatforeignaffiliatesfacechallengesarisingfrominconsistenciesindecision-andlaw-makingbytheregulatoryinstitutionsandgovernmentsoftheirhostcountries.SuchinconsistenciesadddifficultiestotheoperationsofforeignMNCaffiliates,thuspotentiallyexposingthemtosignificantfinanciallossesandunderminingderivativeactivities.8Similarly,thedistancebetweenculturalandsocialorientationsinthehomeandhostcountriescreatesanobstaclelimitingforeignMNCaffiliates’abilitiestogainlegitimacyandotherfavorableeconomictransactionswithlocalfirmsandgovernmentauthorities(PengandLuo,2000;Luo,2001;Makinoetal.,2004)
Moreover,Kostovaetal.(2008),aswellasPetrou(2014),amongothers,arguethattheidiosyncrasiesofacorrupthostcountry’sinstitutionsandregulationscreateadditionalissues,includingariseintransactioncostsandinformationasymme-tries,whichpresentmorechallengestoforeignMNCaffiliatescomparedtodomestic-ownedfirms.Inparticular,inacorruptcountry,governmentofficialsinsearchofbribescanconstraintheperformanceofforeignMNCaffiliatesbychangingthestructureoftaxation,laws,andregulationsinordertoimpedeforeignMNCaffiliates’participationinlocalcompetitionandaccesstolocalresources(Lecraw,1984;Contractor,1990;JavorcikandWei,2009).Hymer(1976)alsoarguesthatgovern-mentssometimesalsogeneratecostsarisingfromtaxationdifferentialsorbureaucraticdelaysingovernmentapprovalsandpaperworkforforeignMNCaffiliatesthatoftenputthematadisadvantagerelativetolocalfirms.SuchunstableinstitutionalenvironmentsthreatenforeignMNCaffiliates,therebyincreasingthecostsofdoingbusiness(Makinoetal.,2004)andthecostsofimplementingderivativesactivitiesinahostcountry.Overall,itislikelythatthehighcostsofforeignnesscoulddampenthevalueeffectofderivativesuseforforeignMNCaffiliates
Basedontheabovelogic,wehypothesizethefollowing:
Hypothesis2a Inlightofacorruptenvironment,theuseoffinancialderivativesismorevaluabletodomesticMNCsthan
todomesticfirms
Hypothesis2b Undertheinfluenceofacorruptenvironment,theuseoffinancialderivativesrewardsdomesticMNCswithhighervaluethanforeignMNCaffiliates
2.3 Financialcrisis,derivativesuse,andfirmvalue
Wetakeintoaccounttheglobalfinancialcrisisintheanalysisofthevalueeffectofderivativesuseinacorruptronment.Researchershavefoundtwoseparateandsequentialeffectsofafinancialcrisisonfirms.First,acrisisleadstosuddenandmajordisruptionsinmarkets,whichresultinadeclineinfirms’investmentopportunitiesandhigherexternal
or innovation ( Rugman, 1981 ).
8 Beck and Levine (2008) note that finance can be considered a set of contracts Because derivatives are financial contracts, it is likely that legal institution influences derivatives use.
Trang 6onfirmvalueislikelytobeworseduringacrisisperiod,butitbetterintheaftermathofthecrisis
Thisargument,inaddition,derivesfromtheviewthatfirmshavedifficultyindealingwithsuddenandmajorexternalshocks(GreenwoodandHinnings,1996;RajagopalanandSpreitzer,1997),andthebenefitsofderivativesuseinreducingexposuremaynotbeobtainedduringacrisisperiod,asthecrisiscanbringaboutsignificantexchange-ratevolatilityandasharpriseinfinancialcostsandprices(SinghandYip,2000).Aswell,firmswithmorecomplicatedorganizationalstructureswillgetintogreaterdifficultieswhenamajorshockoccurs(Hannanetal.,2003;Chakrabartietal.,2007).Thus,thosedifficultiesincreasehedgingcosts,whichoffsetsthebenefitsofusingfinancialderivatives
Furthermore,wealsoexpectthatthevalueeffectofderivativesuseforfirmsincountrieswithlittlecorruptionwillbelessseverelyaffectedbythenegativeimpactoftheglobalfinancialcrisisthanwillbethecaseforthosefirmslocatedinhighlycorruptcountries.Thisisbecauselowcorruptionreducesagencycosts,resultinginlowercostsofcapital,increasedoperationalefficiency,and,eventually,betterfirmperformanceandvaluation,asagrowingnumberofempiricalstudiesoncorruptionhaveshown(e.g.,LeeandHong,2012;Petrou,2015).Takingtheaboveargumentstogetheryieldshypotheses3aand3b:
Hypothesis3a Theglobalfinancialcrisisdampensthevalueeffectofderivativesuseincorruptenvironments,butthere
isapositiverelationshipbetweenderivativesuseandfirmvalueinthepost-crisisperiod
Hypothesis3b Lowlevelsofcorruptionmitigatethenegativeimpactoftheglobalfinancialcrisisonthevalueeffectofderivativesuse
3 Sampleselectionandvariableconstruction
3.1 Datacollectionanddescriptivestatistics
Oursampleconsistsofnon-financialfirmsacrossvariousindustriesineightcountriesinEastAsia,namelyChina,HongKong,Japan,Singapore,Malaysia,Thailand,Philippines,andIndonesia.Itcomprisessolelyfirmsthathaveaccountingdatafortheperiod2003–2013ontheDatastreamdatabaseandthathaveannualreportsinEnglishforthesameyearsintheMorningstardatabaseandthefirms’websitedatabases.Weexcludefinancialfirms,whicharelikelytohavedifferentincen-tivesforusingderivativesfromnon-financialfirms.Ourfinalsamplecontains9691firm-yearobservations;itisabalancedpaneldatasetof881firms
Alldataonderivativescontractswashand-collectedfrom881firms’annualreports;weclassifyfirmsasderivativeusers
ornon-usersbasedoninformationabouttheiruseofderivatives.Afirmisclassifiedasaderivativeuserifitdisclosesthat
itusesatleastonetypeoffinancialderivativesforhedgingpurposes.Thosefirmsthatdonotdiscloseanyuseoffinancialderivativescontractsareclassifiedasnon-derivativeusers.We alsomanuallycollectednotionalnumbersofderivativecontracts,andconvertedallthosevaluesintoonecommoncurrency,USD.Inaddition,mostofouraccountingdataonexplanatoryvariablesarefromDatastream;themissingvariablesweresuppliedbytheBloombergdatabaseorinformation
inannualreports.ThesedataareprovidedannuallyandinthousandsofUSD
Inthesample,389domesticfirms,427domesticMNCs,and65foreignMNCaffiliatesareidentified.WeusedtheCorporateAffiliationsdatabasetoclassifyfirmtypes.Wedistinguishbetweentwotypesofdomesticfirms,i.e.betweenuni-nationaldomesticfirms(firmswithnooverseasinvestments)anddomesticMNCs(firmsthathaveatleastoneoverseasaffiliate).Similarly,foreignaffiliatesareincomingMNCswithaparentcompanybasedelsewhereintheworld(Pantzalisetal.,2001;CastellaniandZanfei,2006,pp.122–123and189).9Notably,mostofourdomesticMNCsareheadquarters,asweselectedthetop-listedcompaniesrankedbymarketcapitalizationontheeightsamplecountries’stockexchangesfortwoimportantreasons.Firstly,largefirmsweremorelikelytobeinvolvedininternationalbusinessactivitiesandtherebyhaveexposuretofinancialrisks.Asmanyfirmsareexpectedtohaveexposuretofinancialrisks,oursamplepotentiallyprovidesarichcross-sectionofderivativesusersandnon-users.Secondly,therewasahighlikelihoodthatlargefirmswereactivelyencouraged
toreporttheirderivativesusageintheirannualreportsduringthesampleperiod
SummarystatisticsontheuseofderivativesbythesamplefirmsisreportedinTable1.Acrossallcountries,approximately53.5percentofoursampleobservationsuseatleastonetypeoffinancialderivatives,indicatingthattheuseofderivatives
iscommonamongnon-financialfirmsinEastAsiancountries
InPanelB,themeanvalueofanyderivatives(foreigncurrencyderivativesand/orinterestratederivativesand/ormodityderivatives)isabout$339million.Whenwecalculatethemeanofthenotionalvalueofderivativecontractstototal
com-9 It is worth noting that many firms in our sample were cross-listed A disadvantage of using cross-listed firms is that they may be subject to governance
Brown, and Klapper (2003) , using a sample of East Asian firms, and Lel (2012) found no difference in the derivatives use of cross-listed firms and those that were not cross-listed.
Trang 7H.T Kim et al / J of Multi Fin Manag 39 (2017) 39–59 45
Descriptive statistics of sample firms’ derivatives use.
This table shows the number and percentage of firms that use derivatives by country, by derivative use information, and by year for all firms We present the percentage of firms using any financial derivatives (foreign currency derivatives and/or interest rate derivatives and/or commodity price derivatives) Panel
A presents the uses of derivatives based on firm-year observations by country Panel B reports the information about the use of derivatives by derivative users and non-users, and the notional value of derivatives contracts Panel C shows the trend of derivatives use over time.
Panel A: Derivatives use by country
Panel B: Firms’ derivatives use information
Panel C: Derivatives use by year
inresponsetothecrisis,whichisshownbythenumberofderivativesusersinthepost-crisisperiod−55.89percentoffirms
inoursamplecomparedto49.72percentinthepre-crisisperiod
ofequityplusthemarketvalueofequity)tothefirm’sbookvalueoftotalassets,evaluatedattheendofthefiscalyear.FauverandNaranjo(2010),Allayannisetal.(2012),andothersusethissamefirm-valuemeasurementtoconstructTobin’s
Q.WecomputeTobin’sQforatotalof9691firm-yearobservations(881firmsperyear).SimilartoAllayannisandWeston(2001),amongothers,weusethenaturallogarithmsofTobin’sQinourmultivariateteststocontrolforskewness.3.3 Independentvariables
3.3.1 Theuseofderivatives
BerkmanandBradbury(1996)statethattheidealmeasureofderivativeuseisthehedgingratioofthecontractsbeingusedtomanagerisk.Accordingtothisargument,wemeasurehedgingactivitybyderivativeuseintensity,ratherthanby
Trang 8ofafirmscaledbythatfirm’stotalassets,anditusesavalueofzerofornon-derivativeusers.Weanticipateapositiveassociationbetweentheuseofderivatives(USE)andfirmvalue(Tobin’sQ)
Intheanalysis,wedonotexaminetheeffectofspecificderivativetypesbasedonspecificrisks,namelyforeigncurrency,interestrates,andcommoditypricederivatives.Weemploythehedgingintensityofusinganyderivativetype, whichpresentsafirm’soverallderivativesuse,asmanyfirmsinoursampleusederivativestohedgemorethanonetypeofrisk.Judge(2003)showsthattherewillbeconfoundingeffectswhenfirmsthatdonothedgethespecificriskunderconsideration,buthedgeothertypeofriskexposures,areincludedinthesampleasnon-hedgers
iscalculatedastheaveragestandardizedvalueofdifferentsources,providingtheoverallextentofcorruption.Thatindexisinverselyrankedfrom0to100,whereahighernumberindicatesalowerlevelofcorruption.Anegativeassociationbetweencorruptionandfirmvalue,whichisapositiveestimatedcoefficientonCPIindex,isexpected.Wefurtherexplorethevalueeffectsofderivativesuseundertheinfluenceofacorruptenvironmentbyconstructinginteractivetermsforcorruptionandhedgingintensity
3.4 Controlvariables
Followingpriorstudies(e.g.,AllayannisandWeston,2001;FauverandNaranjo,2010;Bartrametal.,2011;Allayannis
etal.,2012),weuseseveralfirm-andcountry-specificvariablestocontrolforfactorsthathavebeenshowntoinfluencefirmvalue
3.4.1 Firmsize
Previousempiricalevidenceontheeffectoffirmsizeonfirmvalueisambiguous.AllayannisandWeston(2001),Belghitar
etal.(2013),ChenandKing(2014),andsomeothershavefoundanegativerelationshipbetweenfirmsizeandfirmvalue.YetMagee(2008)andSupanvanij(2011)reportapositivecoefficientonfirmsize,indicatingthattheeffectoffirmsizeonvalueoffirmmaybenon-linear.Therefore,weusethenaturallogarithmofthebookvalueoftotalassetsasaproxyforfirmsize,anddonotprovideanex-antepredictionregardingtheeffectoffirmsizeonfirmvalue
3.4.2 Leverage
Theliteratureoncapitalstructuresuggeststhatafirm’schoiceofcapitalstructuremayaffectitsmarketvalueincases
inwhichtherearemarketimperfections(seeHarrisandRaviv,1991;forareviewoftheliterature).Specifically,Zou(2010)arguesthatleveragealsocanprovidetaxbenefitsofdebts,therebyimprovingfirmvalue.Someotherstudiesfindthatleveragecanactasanegativesignalofinvestmentopportunitiesowingtotheagencycostsofdebt,suggestinganegativerelationshipbetweenleverageandfirmvalue(e.g.,Magee,2008;Belghitaretal.,2013).Wethereforeusetheratiooftotaldebtstototalassetsasourdefinitionofleverage,andhavenoexpectationregardingthesignoftherelationshipbetweenleverageandTobin’sQ
3.4.3 Profitability
Profitabilityissupposedtohaveapositiveeffectonthevaluationofafirmbecauseitislikelythatthemarketplacewillrewardamoreprofitablefirmwithhighervaluecomparedtoaless-profitablefirm,asAllayannisandWeston(2001)andBelghitaretal.(2013)argue.Asaresult,ifaderivativeuserismoreprofitable,itismorelikelytohaveahigherfirmvalue
Weusereturnonassets(ROA),definedastheratioofnetincometobookvalueoftotalassets,tocontrolforprofitability,andexpectapositivecoefficientonthisvariable
Trang 9H.T Kim et al / J of Multi Fin Manag 39 (2017) 39–59 47
growthisanimportantdeterminantofTobin’sQ.Inlinewithpriorstudies,weusetheratioofcapitalexpenditurestonetsalestocontrolforafirm’sinvestmentopportunities,andexpectapositiveassociationwithfirmvalue
3.4.5 Liquidity
Thefreecashflowhypothesis(Jensen,1986)statesthatfirmswithexcessfreecashflowarelikelytoinvestinnegativenetpresentvalue(NPV)projects,resultinginlowerfirmvalue.Pramborg(2004),Bartrametal.(2011)andsomeothersfindevidenceconsistentwiththatargument,yetCampaandKedia(2002)andAllayannisetal.(2012)findcontraryresults.Thus,
wehavenoconjectureonthesignofanassociationbetweenliquidityandTobin’sQ.Weusequickratio,definedascashplusshort-terminvestmentsdividedbytotalcurrentliabilities,asaproxyforliquidity
3.4.6 Accesstofinancialmarkets
Ifderivativeusershavelimitedaccesstofinancialmarkets,theirfirmvaluemaybehighbecausetherearethusfinancialconstraints,andsomanagerswilladdincentivestoundertakeonlythehighestNPVprojects,asAllayannisandWeston(2001)andJinandJorion(2006)havenoted.AsinClarkandJudge(2009)andBelghitaretal.(2013),weusedividendyield,whichismeasuredbycommondividendpersharedividedbythefiscalyear-endshareprice,toproxyforabilitytoaccessfinancialmarkets,andweexpectanegativerelationshipwithfirmvalue
3.4.7 Geographicdiversification
Theevidenceintheexistingliteratureregardingtheinfluenceofgeographicdiversification(multinationality)onfirmvalueisambiguous.Makar andHuffman(2001),and Fauverand Naranjo(2010),amongothers,provideevidencethatgeographicdiversificationispositivelyrelatedtofirmvalue,whileDenisetal.(2002),andMaramiandDubois(2013),findanegativerelationship.Nevertheless,followingAllayannisandWeston(2001),weusetheratioofforeignsalestototalsales,denotedasFORSALES,tomeasuremultinationalityinthisstudy.Weexpectthatitwillbepositivelyassociatedwithfirmvalue.WealsousethedummyvariableGEOMARKT,whichhasavalueofoneforfirmsthathavesalemarketsinforeigncountries,andzerootherwise,asanalternativemeasureforgeographicdiversification
3.4.9 Country-levelcontrolvariables
Wecontrolforcountryeffectsandacountry’stimeinvariantcharacteristicsbyusingGDPpercapitaratiotoproxyfortherelativeperformanceofthecountries,andfinancialsystemdepositstoGDP(definedasdemand,time,andsavingsdeposits
inbanksandotherfinancialinstitutionsasashareofGDP)toproxyforfinancialmarketdevelopment.ThesevariableswereobtainedfromtheWorldBank’sWorldDevelopmentIndicators.AnincreaseinGDPpercapitaandfinancialsystemdeposits
toGDPsignalsgrowthintheeconomyandtendstosignalanimprovementinproductivityandfirmvalue.Thus,apositiverelationshipbetweenfirmvalueandthesevariablesisexpected
4 Empiricalspecifications
Intestingtheabove-statedhypotheses,ourbaselinemodelscanbewrittenincondensedformsinEq.(1)asbelowLn(Tobin’sQ)ijt=˛useUSEijt+˛cCjt+!use cUSEijt∗Cjt+˛xXijt +uijti=1.n;j =1–8;t= 2003–2013 (1)Where:
-Ln(TobinsQ)ijt:Firmvalueoffirmilocatedincountryjinyeart,measuredbynaturallogarithmofbookvalueoftotalassetsminusbookvalueofequityplusmarketvalueofequitytobookvalueoftotalassets
-USEijt:Derivativesuseintensityoffirmilocatedincountryjinyeart,measuredbyanotionalamountofderivativecontractsscaledbytotalassets
-Cjt:Corruptionindexofcountryjinyeart,whichisaninverserankingofcountrycorruptionlevelsonascalefrom100(veryclean)to0(highlycorrupt)(fromTransparencyInternational)
-Xijtisavectoroffirm-andcountry-specificvariablesinyeart,includingfirmsize,leverage,ROA,capitalexpenditures,quickratio,dividendyield,foreignsalestototalsales,financialsystemdepositstoGDP,andGDPpercapita
-uijt:Errortermsclusteredbycountry
Inourinitialtests,weuseOLSestimationofEq.(1)forthesubsamplesofdomesticfirms,domesticMNCs,andforeignMNCaffiliates.Tocontrolforunobservedtime-varyingeffectsandmeasurewithin-countryandwithin-industrydifferencesintheeffectofderivativesuseandcorruptiononfirmvalue,weadoptfixedeffectsforcountry,industry,andyear.Furthermore,
Trang 10Wethenassesstherobustnessofourresultsbycarryingoutadditionalinvestigations.First,weacknowledgethatitispossiblethattheobservedrelationshipissubjecttoendogeneity.Toaddressthisconcern,weimplementtheinstrumentalvariable(IV)model.Second,oursampleconfirmspreviousresultsintheexistingliteraturethatthecharacteristicsoffirmsusingderivativesare,onaverage,quitedifferentfromthoseoffirmsthatdonot.Thesedifferencesarelikelytoleadtoselectionbiaswhenoneinvestigatesthevalueeffectsofderivativesuseundercorruptconditions.Tocontrolforthisself-selectionbias,followingBartrametal.(2011),andChenandKing(2014),weemploytheHeckmantreatmenteffectmodel.Third,wecarryoutanadditionalrobustnesstesttocheckthestabilityofthevalueeffectsofderivativesusebyusinganalternativeproxyforfirmvalue−thatis,firmmarketvalueinthousandsofUSD,calculatedbyafirm’ssharepricemultiplied
bythenumberofordinarysharesinissue
5 Empiricalresultsandanalysis
Thesecondhypothesiswewanttotestiswhethercorruptionenvironmentaffectsderivativesuseandwhetherfirmsarelikelytousederivativeswhentheyareoperatingincountrieswithlowcorruptionlevels.Inlinewithourhypothesis1,fordomesticMNCs,wefindthatcountrieswherederivativeusersarelocatedandoperatinghavelowerlevelsofcorruption(themeanofcorruptionis64.338)thancountrieswherenon-usersare(themeanofcorruptionis54.881).Thedifferencebetweenusersandnon-usersisstronglysignificant.Inthecaseofdomesticfirms,althoughwefindthatfirmsusingderivativesarelocatedincountrieswithlesscorruptionthanthosefirmsthatdonot,themeandifferencebetweenthemisnotdifferentfromzeroatanyconventionallysignificantlevel.Specifically,forforeignMNCaffiliatesweunexpectedlyfindthatderivativesuserslocatedincountrieswithslightlyhighercorruptionlevelsthannon-users.Theseresults,takentogether,suggestthatcorruptionenvironmentaffectsfirms’hedgingbehavior,butthemagnitudeofitseffectvariesacrossdifferentfirmtypes.Furthermore,weinvestigatehowdifferencesinfirm-specificresourcesandcapabilitiesbetweenderivativeusersandnon-usersmayaffectfirmvalue.Onaverage,wefindthatfirmsusingderivativesarelarger,moreprofitable,andhavehigherlevelsofexposurethanthosefirmsthatdonot,asshownbymeansoffirmsize,ROA,andforeignsalestototalsales.Theseresultsareconsistentwiththepreviousstudiessuch,asthosebyAllayannisandWeston(2001)andAllayannisetal.(2012).5.2 Multivariateanalysis
5.2.1 Valueeffectsofderivativesuseundertheinfluenceofacorruptenvironment
Table3presentstheresultsofapooledOLSestimationforoursampleofdomesticfirms,domesticMNCs,andforeignMNCaffiliates.Wefindseveralinterestingresults.Fordomesticfirms,weobservethattherearepositiveandsignificantestimatedcoefficientsonderivativesuseandcorruption(0.0418,0.154,and0.0759respectively,p<0.01).Duetoinverserankingofcorruptionlevels,theseresultssuggestthattheeffectsofderivativesuseonthevalueofdomesticfirmsvarywiththecorruptionlevelsofthehomecountries.Specifically,incountrieswithlowcorruption,11theuseofderivativesincreasesthevalueofdomesticfirmsbetween9.87percentand11.77percent,conformingtothefindingsofKimetal.(2016)thatthelowerthelevelofcorruption,thehigherthelikelihoodthatfirmswillusederivatives.Thehedgingpremiumisslightlyhigherthanthepremiumofbetween4.87percentand6.33percentreportedbyAllayannisandWeston(2001),Nain(2004),Kimetal.(2006),andMagee(2008)forsamplesofUSfirms,butbroadlysimilartothe12percentfoundbyClarkandJudge(2009)whouseasampleof412UKfirms
FordomesticMNCs,wefindresultsgenerallyconsistentwithourhypotheses2aand2b.Inparticular,alowercorruptionlevel(higherCPI)isassociatedwithhigherfirmvaluesasevidencedbyasignificantandpositivecoefficientestimateforcor-ruption(p<0.01).Wealsofindbothstatisticallyandeconomicallysignificantcoefficientsonderivativesuseandderivativesuseinteractedwithcorruption(0.0494and0.0778,respectively).TheseresultsindicatethatdomesticMNCsarelikelytousefinancialderivativesincountrieswithlowcorruptionlevels,andsuchalow-corruptionenvironmentfacilitatestheuse
11 We define corruption levels based on the Corruption Perception Index (CPI), in which those countries having scores of greater than 75 are considered
to be countries with low levels of corruption, and those countries having scores of less than 75 are considered to be highly corrupt countries.