May28,2002HedgingForeignExchangeExposure:RiskReductionfromTransaction∗andTranslationHedgingAbstractUsingasampleofSwedishfirmsweinvestigatetheriskreducingeff,weinvestigateriskreductionfromusingdifferenthedginginstruments,’foreignexchangeexposureisincreasingwiththelevelofinherentexposure,measuredasthedifferencebetweenrevenuesandcostsdenominatedinforeigncurrency,’,wefindthattransactionexposurehedgessignificantlyreduceexposure,().Ifso,byhedgingtranslationexposure,,StockholmUniversityS-10691Stockholm,SwedenTelephone:+4686747427E-mail:bpg@∗Thispaperhasbenefitedsubstantiallyfromsuggestionsbytwoanonymousrefereesandtheeditor,,,NiclasAndrén,HosseinAsgharian,DavidBurnie,MartinHolmen,,LarsNordén,FredrikStenberg,CarstenSørensen,,the13thAnnualPACAP/FMAFinanceConference2001inSeoul,the31stFMAAnnualMeeting2001inToronto,andcommentsfromworkshopparticipantsatSchoolofBusiness,StockholmUniversity,andattheArneRydeWorkshopinEmpiricalFinance2001,LundUniversity.
1IntroductionThepurposeofthispaperistoinvestigatetherelationshipbetweenforeignexchange(FX),butyetineff(1996)arguedthatanticipatingtheconsequencesofhedgingisdiffi,hedgingactivityrisksbeingwastefultothefirm’sshareholders,’managementperceivehedgingascomplicated(seeAlkebäckandHagelin,1999).Hence,,littleeff,however,(1998).’(1998),onaverage,(1999)investigated47Swedishfirms’,usingsurveyevidenceonfirms’hedgingofforeignassets,Nydahl(1999)’(2001)analyzedthelinkbetweenFXexposureandtheuseofcurrencyderivativesforasampleofnonfinancialS&,toaddvaluehedgingmustbeaimedatminimizingtheadvertimpactfrommarketimperfections,suchasprogressivetaxrates,costsoffinancialdistress,andagencyproblems(seeSmithandStulz,1985).1
onfinancialinstrumentswithoff-balancesheetrisk,,toinvestigatewhethercurrencyderivativesusagereducedfirms’(1999)-sectionalstudyofAllayannisandOfek(2001)Guay(1999)examinedthechangeintherisklevelforfirmssurroundingeachfirm’(2000)(anditspredecessorsSFASNo105and107).HedocumentedweakassociationsbetweenderivativedisclosuresandFXexposureandsuggestedthatthiscanbeduetoinabilityincontrollingforfirms’,takentogethertheresultsofthefivestudiespresentedabovesuggestthattheuseofcurrencyderivativesmayhelptoreducefirms’’FXexposureisaffectedbydiff,,weinvestigatetheeffffectfirmsdiff,,ontheotherhand,arisesasthefinancialaccountingstatementsofforeign2Economicexposurecanbedividedinthree:theexposureofcommittedtransactions,theexposureofidentifi-ableanticipatedtransactions,andcompetitiveexposure(unidentifiableanticipatedtransactions).(seeBodnaretal,1998).Instead,,(1999)
3affi(,1999).Thereasonsforthisarethattranslationgains(losses)tendtobe(i)unrealizedandhavelittledirectimpactonfirms’cashflows,and(ii),translationexposuremanagementshouldbeinefficientinreducingfirms’,whileearlierstudiesassignallderivativesusageashedges,,givenLeland’s(1998)-sectionalregressionstoinvestigateasampleofSwedishfirms’’,().Ifso,byhedgingtranslationexposure,,(IAS21)suggeststheuseofthecurrentratemethodforself-containedforeignaffiliatesandtheuseofthetemporalmethodforintegratedforeignaffiliatesandforforeignaffiffiliate,,thesechangesalsoaff’proposaltorecommendation,,
ofFXbetas,andconcludeswithadescriptionofdeterminantsoffirms’-sectionalresultsarepresentedinthethirdsection,,Exposure,,’’inherentFXexposuresandhedgingpractices,weemploythreequestionnairesto4determineexposuresandfirms’,andcontainedquestionsconcerningtherespondents’,followedbythethirdquestionnaire,fferenceisthat,inthetwolatter,-99,and2000-01,,:(1),basedonIAS39willbeintroduced,butforthesampleperiodthiswasnotinefföringsnämnden(BFNR9)stipulatedthatfirmsshouldreportnetrevenues,investments,andemployeesforgeographicalmarkets,withconsiderablefreedomindecidingwhatthe“geographicalmarket”,deferralaccountingandmark-to-market,,wheremostfirmsreportedanetpositionandnotpercurrency,typeofexposureshedged,,accountingpracticesforinternationaloperationsandhedgingseemtosomeextenttolagthoseintheUS,atleastforthesampleperiod(seealsoHung,2001,foradiscussiononcountrydifferencesofthevaluerelevanceoffinancialstatements).4
listedattheStockholmStockExchange;(2)thefirmwasanonfinancialfirm;and(3)thefirm’(firmswithheadquarterslocatedoutsideSweden)toeliminatepotentialdifferencesbetweenfirmsthatariseduetodiff,130,and128usableresponseswereobtainedfromthefirst,second,,47percent,and49percent,.’s(1998),thefinalsampleisreducedto462observationsduetomissingdataandtheexclusionofobservationswithnoinherentFXexposure,-responsebias,(1978)andAdlerandDumas(1984)definedFXexposureastheeff(1990)proposedthatestimatesofthestockpriceexposuretoFXratechangescanbeobtainedfromthefollowingtimeseriesregressionR=α+βR+βR+ε,(1)itiMiMtFXtitFXiwhereRisthereturnonstockifromt−1tot,Risthereturnonamarketportfolio,,thereturnseriesofthetradeweightedbasketofforeignMt6currencies(theTCWindex)-relevantMt5SeeHagelin(2001)fficialtradecurrencyweightedindexprovidedbytheSwedishcentralbank,,
fficientofinteresttousisβ,firms’FXbetas,themeasureoffirms’,weareinterestedinfirms’assetFXbetas,,wefirstapplyequation(1)toestimateeachfirm’’sassetFXbetabymultiplyingthestockFXbetawiththefirm’-sectionalregressionsbelowweexclusivelyusefirms’ffinandStulz(2001)argued,includingamarketportfolio,asinequation(1),,ignoringcausality,ifthemarketiscorrelatedwiththeFXrate,wemayunderestimatefirms’(i)withouttheinclusionofamarketportfolio,and(ii)usingmarketexcessreturnsinsteadofrawreturns,implicitlyassumingamarketbetaofone,ff(2000)suggestedthatasizebasedcontrolportfolioshouldbeusedinsteadofacommonmarketindex,whichhelpstocontrolforboththemacroeconomicfactorsthatinfluencefirmvalueandthesizeeff(2000)(1)(1998),AAE7ˆˆiˆˆSpecifically,eachfirmsassetFXbetaiscalculatedasβ=β,whereβisfirmi’sassetbeta,βFXiFXiFXiFXiViistheestimatedstockbetafromequation(1),Eisthetotalmarketvalueofthefirm’sequity(atthebeginningofitheyear)andVisthetotalmarketvalueofthefirm,calculatedas(Totalbookvalueofassetsminusbookvalueiofequityplusmarketvalueofequity).,,
Guay(1999),Wong(2000),andAllayannisandOfek(2001)usedmonthlyobservationswhileNydahl(1999),bi-weekly,andmonthlyoverlappingreturns,usingdailyobservations,andapplytheNewey-West(1987),earlierstudies,suchasJorion(1990)andAmihud(1993),relyingontheassumptionthatcapitalmarketsreactinstantaneouslytoexchangeratechangeshaddifficultiesinidentifyingsignificantcontemporaneousFXexposurecoeffi(1994)showedthatlaggedchangesinexchangeratesdemonstratedasignificanteff,andfindsupportforlaggedeff,,(1998),Nydahl(1999),Guay(1999),Wong(2000),andAllayannisandOfek(2001).InTable1,PanelA,,fromapossible617,dueto(i)missingdata,whichincludetoofewdatapointsforbetaestimations,andincompletesurveyresponses(73observationsdeleted);and(ii)thefirm’ssurveyresponseindicatednorevenuesandnocostsdenominatedinforeigncurrencyforaparticularyear(82observationsdeleted).[InsertTable1here]About13(24)percentoftheFXbetasaresignificantatthefive(ten)percentlevel,,theTCWindexmaycapturetheaverageFXexposureofallfirms,butformanyindividualfirms,,manyfirmsmayhavelowinherentFXexposureinthefirstplace,andthird,evenforfirmswithlargeinherentFX7
exposures,theusageofoperationaland/,9PanelB,presentstheabsolutevalueoftheassetFXbetas,wherethesampleisdividedintothreeequallysizedgroupsbasedonfirms’bookvalueoftotalassets,whichismotivatedbythepotentialdiff,namelythatthesmallerthefirm,(.,1998),wherelargerfirms,duetoeconomiesofscale,,itmayreflectthatlargerfirmsaremorelikelytobemultinationalcorporations(MNCs)withproductionandsalesinmanycurrencies,whichcanreduceFXexposure,,aremorelikelytobeexportersorimporters,’,therespondentswereaskedwhetherornottheyhedgedFXexposurewithderivativesandifso,iftheirhedgingconcernedtranslationand/,theywereaskedwhethertheyusedforeigndenominateddebtforhedgingpurposes,and,inthesecondandthirdquestionnaire,ifthisconcernedtranslationand/,wewereabletodefinevariablesthatreflecteachfirm’’sFXexposureissignificantlyrelatedtoproxiesforforeigninvolvement(seeHeandNg,1998;Nydahl,1999;Wong,2000;andAllayannisandOfek,2001).Lackofbetterproxieshasledmoststudiessofartouseone-sidedmeasures,suchasexport9TheabsolutevalueoftheFXbetascapturesthemagnitudeofexposure,
ratio(,1998,andAllayannisandOfek,2001).Analternativemeasureisthenetexposure,suggestedbyMarston(2001).Thismeasure,anetoflongandshortpositionsinforeigncurrency,supposedlycapturesexporters’aswellasimporters’,wecancreateaproxyforfirms’,denotedNE,isdefinedasthedifferencebetweenthepercentageofrevenues(FR)andthepercentageofcosts(FC),asmeasuredbytheabsolutevalueoffirms’assetFXbetas,10weusetheabsolutevalueofthismeasureforouranalysis,denotedabs(NE).Insteadofthismeasure,themeasureproposedbyMarston(2001),NE,couldbeused,whichwouldenableustomodelthesignoftheexposure(),,(notreported)whereweusedonlysignificantFXbetas,,(NE),focusestheanalysisonthemagnitudeoftheFXexposure,whichisthemaininterestofthestudy,,totalassets,FR,FC,andabs(NE),,,,largerfirmshavelowerabs(NE)thansmallerfirms,whichmaypartlyexplainthelowerFXbetasfor10Formally,thismeasureisdefinedasabsNE=abs(FR−FC),whereFRisthepercentageofrevenuesdenominatedinforeigncurrency,,
,thelowerFXbetasoflargerfirmscanpotentiallybeexplainedbyotherfirmcharacteristics,suchasdegreeofmultinationality,,ourproxyforfirmsizeisthelogarithmoftheaveragebookvalueoftotalassetsforeachfirmyearobservation,,firmsmayhedgewithdiff,firmsmaychoosetohedgediff,toexploretheimpactonfirms’FXexposurefromhedgingwithfinancialinstruments,,H,fferenttypesofFXexposure;,,denotedTRTL,,TR,issettooneforfirmsthathedgetransactionexposureonly,andthethird,TL,,-posurethatfirmshedgedwithcurrencyderivatives,-sectionalregressions(notreported).,foroursample,(medianvalue90percent),andthosethathedgedtranslationexposurehedgedonaverage66percent(medianvalue75percent).10
,weclassifyhedgersintofourdistinctgroups,(CDFD),(CD)issettooneifthefirmusecurrencyderivativesbutnoforeigndenominateddebttohedge,andthethirddummy(FD).[InsertTable3here]InTable3,PanelA,-thirds(63percent)ofthefirmyearobservationsareforfirmsthathedgedtheirFXexposureusingfinancialinstruments(H).Almostalllargefirmshedge,,,itismostcommontohedgetransactionexposureaswellastranslationexposure,,,weinvestigatetheimpactfromhedgingdifferenttypesofFXexposures,ffectfrominherentFXexposureandfinancialhedgingonfirms’11
FXbetas,,toinvestigatetheeffectonfirms’FXbetasfrominherentFXexposure,werunthefollowingcross-sectionalregression¯¯410APP¯¯jkˆβ=α+αabs(NE)+δYEAR+γIND+ε,(2)¯¯01jiFXiiikij=1k=1AˆwhereβistheestimatedFXbetaforfirmi’sassets;abs(NE)istheabsolutevalueoftheFXiijnetoffirmi’sproportionofrevenuesandexpendituresdenominatedinforeigncurrency;YEARik13isayeardummyforyearjandfirmi;INDisanindustrydummyforindustrykandfirmi;iandε[InsertTable4here]Table4,model(a),presentstheresultfromregressionequation(2).Thecoefficientforabs(NE)ispositiveandsignificantonaonepercentlevel,whichisconsistentwithMarston(2001)thatalargerdiff(b)inTable4isanaugmentedversionofregressionequation(2)(i)morelikelytobeMNCs,andthereforehavenaturalhedgesbuilt-intotheirbusiness,and(ii)thattheyaremorelikelytohedge,,weaddvariablesrepresentingfinancialhedgestomodel(b),whicharereportedinTable4,models(c)-(e).Inmodel(c),itisshownthatbyaddingthedummyforhedgers,H,tothemodeltheFXexposureisasbeforeincreasinginabs(NE),thehedgingdummyissignificantandnegative,suggestingthatfirmsthatusefinancialinstrumentstohedge13WeaddyeardummiesandindustrydummiesinordertoaccountfordifferencesinabsoluteFXbetalevelsbetweenyearsandindustries,
,and,arguably,(d)presentsevidenceregardingtheriskreducingeffectfromdifffficientsforallthreedummyvariables,(1999),(1997),whoarguedthatifthetranslationFXrateisacloseproxyforaweightedaverageoffutureFXratesatthetimecashflowsoccur,,byhedgingtranslationexposure,,however,,,model(e)presentsevidenceregardingtheriskreducingeffectfromdifferenttypesofinstruments:firmsthatusecurrencyderivativesandforeigndenominateddebt(CDFD),firmsthatuseforeigndenominateddebtonly(FD),aswellasfirmsthatusecurrencyderivativesonly(CD).FirmsinallclasseshavesignificantlylowerFXbetaestimatesatanyconventionalsignificancelevelthannon-hedgers,suggestingthatfirmsarereducingexposureeff,butyetineffff
Consequently,ifmanagementfailstoreducetotalriskbyhedging,,asuccessfulhedgingprogrammayincreaseshareholdervaluebyreducingcostsrelatedtodiff,asmeasuredbyfirms’foreignex-changebetas,,,andalsobecauselargerfirmsmayhedgemoreeffi,wefindevidencethatfinancialhedgesareeffectiveinreducingfirms’-tivesand/-causetranslationexposureandtransactionexposuretendtoaffectfirmsdiffff().Ifso,byhedgingtranslationexposure,-terestingextensionstothisstudywouldbetoincludevariablesthatreflectfirms’abilitytoalterpricingpolicyinresponsetoFXchanges(seeBodnaretal.,1999,andAllayannisandIhrig,2001)aswellastheusageofoperationalhedging(seeMarshall,2000).ReferencesAdler,,“ExposuretoCurrencyrisk:DefinitionandMeasurement,”Fi-nancialManagement13(1984),,,“Exposureandmarkups,”ReviewofFinancialStudies14(2001),,,“ExchangeRateExposure,HedgingandtheUseofForeignCurrencyDerivatives,”JournalofInternationalMoneyandFinance20(2001),
Alkebäck,,“DerivativeUsagebyNonfinancialFirmsinSwedenwithanInternationalComparison,”JournalofInternationalFinancialManagementandAccounting10(1999),,Y.,“EvidenceonExchangeRatesandValuationofEquityShares,”inExchangeRatesandCorporatePerformance,,eds.(IrwinProfessionalPublishing,1993).Bartov,,“FirmValuation,EarningsExpectations,andtheExchange-rateExposureEffect,”JournalofFinance49(1994),,.,,,“-financialFirms,”FinancialManagement27(1998),,.,,,“Pass-throughandExposure,”Workingpaper(WeissCenterforInternationalFinance,1999).Bodnar,,“EstimatingExchangeRateExposures:Some‘Weighty’Issues,”Workingpaper(NBER,2000).Butler,.,MultinationalFinance,2nded.(South-WesternCollegePublishing,1999).Copeland,,“WhyDerivativesDon’tReduceFXRisk,”McKinseyQuar-terly(,1996),,B.,“TheTheoryoftheTradingFirmRevisited,”JournalofFinance33(1978),ffin,,“InternationalCompetitionandExchangeRateShocks:ACross-countryIndustryAnalysisofStockReturns,”ReviewofFinancialStudies14(2001),-241Guay,.,“TheImpactofDerivativesonFirmRisk:AnEmpiricalExaminationofNewDerivativeUsers,”JournalofAccountingandEconomics26(1999),,N.,“WhyFirmsHedgewithCurrencyDerivatives:AnExaminationofTransactionandTranslationExposure,”AppliedFinancialEconomics11(2001),,,“TheForeignExchangeExposureofJapaneseMultinationalCorpora-tions,”JournalofFinance53(1998),,M.,“AccountingStandardsandValueRelevanceofFinancialStatements:AnInter-nationalAnalysis,”JournalofAccountingandEconomics30(2001),,P.,“,”JournalofBusiness63(1990),,.,“AgencyCosts,RiskManagement,andCapitalStructure,”JournalofFinance53(1998),,.,“ForeignExchangeRiskManagementinUK,USAandAsiaPacificMultina-tionalCompanies,”JournalofMultinationalFinancialManagement10(2000),,.,“Theeffectsofindustrystructureoneconomicexposure,”JournalofInter-nationalMoneyandFinance20(2001),,,“ASimple,PositiveSemi-definiteHeteroscedasticityandAutocor-relationConsistentCovarianceMatrix,”Econometrica,55(1987),-70815
Nydahl,S.,“ExchangeRateExposure,ForeignInvolvementandCurrencyHedgingofFirms:SomeSwedishEvidence,”EuropeanFinancialManagement5(1999),,,ManagingintheTurbulentWorldEconomy:CorporatePerformanceandRiskExposure(JohnWileyandSons,Inc.,1997)Wong,.,“TheAssociationBetweenSFAS119DerivativesDisclosuresandtheFor-eignExchangeRiskExposureofManufacturingFirms,”JournalofAccountingResearch38(Autumn,2000),
’srevenuesandcostsaredenominatedinforeigncurrencies?Revenues...%Costs...%
(FX)=itα+βRiMt+βR+ε,’sstock,RMtisthereturnonthemarketindex,RFXisthetreturnontheorthogonalizedtradeweightedcurrencyindex,andε,thenumberandpercentageofstockFXˆbetas,β,thataresignificantonafivepercentlevelofsignificanceFXi(α=),andonatenpercentlevelofsignificance(α=),weeklyobservations,-West(1987),descriptivestatisticsfortheabsoluteassetFX¯¯A¯¯ˆbetas,β,arepresented,wheretheobservationsaredividedinto¯¯FXithreeseparategroupsbasedonthebookvalueoftotalassets(theaverageofbeginning-of-yearandend-of-yearvalue)foreachfirmyearobservation.ˆPanelA:NumberofSignificantβFXiα=α=(%)significant(%)1997796()12()19989014()22()19999216()29()200010317()30()2001989()17()Total46262()110()¯¯A¯¯ˆPanelB:DescriptiveStatistics,β¯¯
;FCistheshareofcoststhatisdenominatedinforeigncurrency;andabs(NE)istheabsolutevalueofnetexposure,wherenetexposureisdefinedas,(FR−FC),thediff(NE)
,-sizedgroups,basedonbookvalueoftotalassets(theaverageofbeginning-of-yearandend-of-yearvalue).PanelAcon-tainsthenumberandpercentage(inparenthesis)ofobservationsthatareforfirmsthathedge(H),andthenumbersandpercentagesofobserva-tionsthatareforfirmsthathedgetransactionexposureandtranslationexposure(TRTL),thathedgetransactionexposureonly(TR),andthathedgetranslationexposureonly(TL).-berandpercentage(inparenthesis)ofobservationsthatareforfirmsthathedge(H),andthenumbersandpercentagesofobservationsthatareforfirmsthatusecurrencyderivativesandforeigndenominateddebttohedge(CDFD),thatusecurrencyderivativeonly(CD),andthatuseforeigndenominateddebtonly(FD).(1998-2001)LargeMidSmallAllobsTotal128128127383H121(95)89(70)32(25)242(63)TRTL65(51)20(16)3(2)88(23)TR49(38)60(47)29(23)138(36)TL7(5)9(7)0(0)16(4)(1997-2001)LargeMidSmallAllobsTotal154154154462H145(94)115(75)47(31)307(66)CDFD93(60)48(31)3(2)144(31)CD35(23)42(27)41(27)118(26)FD16(10)24(16)3(2)43(9)20
-SectionalRegressionsThistablereportstheresultsofmultipleregressionanalysesonfactorsaff,whichisthestockFXbetaestimatedwithweeklyoverlappingstockreturnsusingdailydataandcorrectedfor¯¯A¯¯ˆ,model(c)isβ=α+αabs(NE)+αSIZE+αH+¯¯01i2i3iFXi4j4kΣδYEAR+ΣγIND+ε,=1jk=1iikiindependentvariablescapturinginherentexposureareabs(NE),theabsolutevalueofthedifferencebetweentheshareofrevenuesandcostsdenominatedinforeigncurrencies;andSIZE,;H,adummysettooneifafirmreportedthatitusedderivativesand/orforeigndenominateddebt;CDFD,adummysettooneifafirmusedderivativesandforeigndenominateddebt;CD,adummysettooneifafirmusedderivativesbutnotforeigndenominateddebt;FD,adummysettooneifafirmusedforeigndenominateddebtbutnotderivatives;TRTL,adummysettooneifafirmhedgedtransactionexposureandtranslationexposure;TR,adummysettooneifafirmhedgedtransactionexposureonly;andTL,,YEAR,andindustrydummies,IND,-values,usingNewey-West(1987)robuststandarderrors,=1(a)(b)(c)(d)(e)()()()()()abs(NE).()()()()()SIZE−−−−()()()()H307−()TRTL88−()TR138−()TL16−()CDFD144−()CD118−()FD43−()
