MS：灾害来临时的恐慌性抛售：债券和股票市场的证据.docx

informsMANAGEMENTSCIENCEVol.69,No.12,December2023zpp.7448-7467ISSN0025-1909(print),ISSN1526-5501(online)PanicSellingWhenDisasterStrikes:EvidenceintheBondandStockMarketsp2>J3saISIqcnp-«SlUoQSn«UoSJodJo工ZT90rtzoeqQJnZICc-Zocz-Z,二ZlKq8JoSIUJO一WulojJPUPEOWMoaThanhD.Huynh,aYingXiaaaDepartmentofBankingandFinance,MonashBusinessSchl,MonashUniversity,Victoria3145,AustraliaContaccthanh.huvnhmonash.eduz)https:/orcid.org/0000-0001-7970-1464(TDH);Ving.xiamonash.edu,皮;https:/orcid.org/0000-0003-1488-3732(YX)Keywords：physicalclimaterisknaturaldisastersoverreactionestablishment-leveldataAbstract.Thisstudyusesdisaggregatedestablishment-leveldatatoidentifyafirm,sexposuretophysicalclimateriskandexaminesinvestors*reactiontonaturaldisastersinboththeU.S.corporatebondandstockmarkets.Wefindthat,whenafirmisexposedtodisasters,investorsoverreactbydepressingthecurrentbondandstockprices,causingfuturereturnstobehigher.However,firmswithastrongenvironmentalprofileexperiencelowersellingpressureontheirbondsandstocks,althoughtheirfundamentalsweakenedfollowingdisasters.TheevidencesuggeststhatCOrPOrateinvestmentinimprovingenvironmentalprofilespaysoffwhenclimatechangeriskismaterialized.History:ThispaperwasacceptedbyCoIinMayer,SpecialSectionofManagementScienceonBusinessandClimateChange.SupplementalMaterial:Thedatafilesandonlineappendixareavailableat.2021.4018.Received:March24.2020Revised:September14,2020Accepted:January16,2021PublishedOnlineinArticlesinAdvance:August11.2021Copyright:©2021INFORMS1. IntroductionTheUnitedStatesisamongthetopthreecountrieshitbythegreatestnumberofnaturaldisastersoverthepasttwodecades(TheEconomist2017).1BlackRock(2019)predictsthatextremeclimaticeventswillcostatleast1%ofgrossdomesticproduct(GDP)ofmostmetropolitanareasintheUnitedStatesby2060-2080undera"noclimateaction*'scenario.HsiangandJina(2014)estimatethataheftycyclonecarriesthesameimpactonincomepercapitaasabankingcrisis,anditcanhaveaprolongedinfluenceontheeconomicdevelopmentofacountry.Inthisstudy,welookintothestockandbondmarketsandexaminehowinvestorsreactwhenafirmisexposedtonaturaldisasters.Giventhatstocksandbondsarethetwomajorcomponentsofafirm'scostoffinancing,examiningthemarketreactiontodisasterswilldeepenourunderstandingofthepossiblepenaltythatinvestorsimposeonexposedfirms.Thisunderstandingalsoguidesmanagersonthetypeofcorporateinvestmentsthatmighthelpmitigatethesecosts.Forpolicymakers,itisimperativetoexaminewhetherthemarketreactionisconditionaluponfirmattributes,becauseitimpliesthatpoliciespromotingefficientpricingofphysicalclimateriskmaynotbehomogenousforallfirms.Despitetheimportanceofthisfirm-levelanalysis,thesequestionshavenotbeenfullyexploredintheclimatefinanceliterature,possiblybecauseoftherestrictedavailabilityofgranulardetailsonafirmsestablishmentlocations.Ourstudycontributestothisliteraturebyusingcomprehensiveestablishment-leveldataonsubsidiaries,branches,andplantsofU.S.firmsfromtheNationalEstablishmentTime-Series(NETS)databaseandthecounty-levelinformationondisasterdamagesfromtheSpatialHaZardEventsand1.ossesDatabasefortheUnitedStates(SHE1.DUS).Afirmisdeemedtobeexposedtodisasterswhenitsestablishmentsarelocatedindisaster-struckcounties.Usingbothportfolioanalysisandmultivariateregressions,wefindapositiverelationbetweenafirmsexposuretonaturaldisastersanditsfuturestockandbondreturns.Specifically,whenafirm'sestablishmentsarelocatedincountieshitbynaturaldisasters,itsfuturemonthlystockreturnsincreaseby14.6basispoints(bps)anditsfuturemonthlybondreturnsincreaseby8.4bps.Themagnitudeiseconomicallymeaningful,becausetheseestimatesareequivalentto14%and16%ofthesamplemeansofstocksandbonds,respectively.Theseresultsarerobusttousingalternativemeasuresoffirmleveleconomicexposureandcontrollingforknownriskcharacteristics,unobservabletime-invariantdifferencesacrossfirms,andseasonalityeffects.Moreover,theseeffectsarestrongerasdisastersaremoresalient.Placebotestsofpseudo-disastersonunexposedfirmsshowaninsignificantrelation,suggestingthatourfindingsarenotspuriousresults.Wefurtherfindthat,afteradisasterstrikes,theestablishment'ssalesandthefirm,saggregatesalesdecrease.2Theseresultsserveasvalidationforouridentificationofaffectedestablishments,becauseitisp2>J3saISIqcnp-«SlUoQSn«UoSJodJo工ZT90rtzoeqQJnZICc-Zocz-Z,二ZlKq8JoSIUJO一WulojJPUPEOWMoasuccessfulincapturingdisasterimpactsonfirmfundamentals.Theyalsoindicatethatbothbehavioraltheory(investors'overreactiontodisasters;DeBondtandThaler1985)andrationalexpectation(anincreaseinriskbecauseoflossesinfirmfundamentals)couldexplainourheadlinefindings.Wethereforepresentseveralfindings,which,takentogether,leantowardthebehavioralexplanation.Acentralpredictionoftheoverreactionhypothesisisthatinvestorsoverreacttodisastersbydepressingthecurrentpricesofexposedfirms'stocksandbonds,causingthecontemporaneousrelationbetweenreturnsanddisasterstobenegative.Aspricesdropbelowfundamentalvalues,theywillbounceback.Weindeedfindthatcontemporaneousreturnsonbondsandstocksaresignificantlylowerwhenthefirmisexposedtodisasters.Asaconsequenceofthispricepressure,weobserveanincreaseinreturnsonstocksandbondsofexposedfirmsafterward.Wenextuseinsidertradingactivitiestoexaminewhetherpricechangesarepurelydrivenbythereductionoffirmfundamentals.Ifinvestorsoverreacttodisastersbyoversellingexposedrms,stocks,theninsiderscouldbebetteroffpurchasingtheirownstocks,becausestockpriceswerebelowtheirfundamentalvalues(Seyhun1990zRozeffandZaman1998).Weindeedfindthatinsiders'purchasesincrease,consistentwiththenotionthattheseinsidersbelievetheirownfirms'sharepricesareoverlydepressed.Totheextenttlatattentiontoclimatechangeinfluencesinvestors*reactiontodisasters,weexpectthatourresultsaremorepronouncedinrecentyearswheninvestors'concernsaboutclimatechangeriskheightened(Kruegeretal.2020).UsingthepublicationoftheSternReviewin2006asaneventthatarousedpublicattentiontoclimatechange(Painter2020),wefindthattheeffectofdisastersissignificantonlyafter2006.Arelatedquestioniswhyinvestorsaresurprisedbynaturaldisasters.Afterall,someregionsintheUnitedStatesaremorevulnerabletocertainprominentdisastersthanotherareas.Traditionalassetpricingtheorypositsthatphysicalclimateriskcouldbediversifiedinamarketportfolioandthus,thepricesofexposedfirms(and,inturn,theircostofcapital)shouldnotchangefollowingadisaster.Wecontendthatthisnullhypothesisisnotsupportedempirically,becausenaturaldisasterscannotbeperfectlypredicted(Rehseetal.2019)and,asclimatescientistsshow,climatechangehasfurtherexacerbatedanypredictability.3Withourdetaileddataonfirmgeography,weshowinSection3thatthereisalargedegreeofuncertaintyinthedistributionofbothprominentandnonprominentdisastersinagivenarea(e.g.zuncertaintyaboutthelikelihoodofoccurring,timing,andthepotentialdollardamages).Therandomnessofdisasterdistributionssuggeststhatgeographicdiversificationofbusinesslocationscouldnothelpafirmtoavoiddisasterscompletely,becausethereisachancethatthefirm'sestablishmentswillbehitbydisastersatsomepointintimeinthefuture.Toexaminehowfirmscanmitigatethesellingpressureontheirstocksandbondswhentheyareexposedtodisasters,weexplorewhetherinvestors*reactionisconditionalonafirmsenvironmentalprofile.Wefindthatthesellingpressureislesspronouncedamongfirmswithahighenvironmentalscore(EScore)fdespitethefactthatbothhigh-andIow-EScorefirmssufferedanequallossinestablishmentsalesafterdisasters.4Consistently,firmswithmoresustainability-orientedandlong-terminstitutionalinvestorsalsoexperiencelowersellingpressurefollowingdisasters.Theseresultsareconsistentwiththenotionthat,becauseinvestorshavenonpecuniarymotiveswheninvestinginhigh-EScorefirms(Bakeretal.2018,HartzmarkandSussman2019,Barberetal.2021),theyimposealowerpenaltyonthesefirms'stocksandbondswhenthefirmsareexposedtodisasters.Thesefindingsalsohaveimportantimplicationsforpolicymakers,whowishtoemphasizethebenefitsofcorporateinvestmentinenvironmentalprofiles:itpaysoffwhentheconsequencesofclimatechangearematerialized.Inthenextsection,weoutlineourcontributionstotheclimatefinanceliteratureanddevelopourhypotheses.Section3describesourdataandsampleselection.InSection4,wereportthemainempiricalresults,robustnessanalyses,andtestsofalternativeexplanations.WeconcludeinSection5.2. Related1.iteratureandHypotheses2.1.ABriefReviewofRelated1.iteratureandContributionsOurstudycontributestoaburgeoningclimatefinanceliteraturethatexaminestheeffectsofclimatechangerisksonfirms,andinparticular,theassetpricingimplicationsoftheserisks.Astrandofthisliteraturestudiescarbonrisk,socialcostsofcarbonemissions,andthereturnpremiumofcarbon-intensivefirms(Barnettetal.2020;BoltonandKacperczyk2020z2021;Choietal.2020;Hsuetal.2020;Ilhanetal.2021).Otherstudiesexaminetheeffectsofclimateregulatoryriskandtherelationbetweenfirmriskandenvironment,social,andgovernance(ESG)(deGreiffetal.2018,Dunnetal.2018,Hoepneretal.2020zSeltzeretal.2020).Demandsforhedgingagainstclimateriskcanalsoaffectassetprices(Engleetal.2020,HuynhandXia2021).Ourstudyisrelatedtothestrandofclimatefinanceliteraturethatexaminesphysicalclimateriskintheformofextremeclimaticevents.Hongetal.(2019)findthatinternationalmarketsunderreacttothelong-termtrendsindroughts.Aloketal.(2020)showthatmutualfundmanagerslocatednearadisasterzoneunderweightdisaster-pronestocks.Otherstudiesexaminep2>J3saISIqcnp-«SlUoQSn«UoSJodJo工ZT90rtzoeqQJnZICc-Zocz-Z,二ZlKq8JoSIUJO一WulojJPUPEOWMoatheeffectsofextremeclimaticeventsonfirmsdrawingofbankcreditlines(Brownetal.2021),insurancefirms(MassaandZhang2021)zcorporatecashholdings(DessaintandMatray2017),manufacturingfirms,performance(Hsuetal.2018),andsupplychains(BarrotandSauvagnat2016).5Althoughtheliteraturehasstudiedthepricingofclimateriskinvariousassetclasses,thereisadearthofresearchthatestimatestheeffectsofafirm'sdirectexposuretodisastersonbothstockandcorporatebondreturns,whicharethetwomajorcomponentsofthefirm,scapitalstructure.6Ourstudyaimstofillthisvoidbyusingacomprehensivedatasetonestablishment-levellocationstoidentifyafirm,sexposuretonaturaldisastersatthecountylevel.Thisidentificationisimportantbecauseafirm'smainbusinessoperationstypicallyhappenatestablishments.Becausestocksandcorporatebondsareissuedbytheexposedfirm,ifdisastersaffectthefirm,thenbothstockandbondpricescouldreacttotheseimpacts.Furthermore,bondsaremainlyheldbyinstitutions(Baietal.2019),whoareincreasinglyconcernedaboutclimaterisk(Ilhanetal.2020,Kruegeretal.2020).Totheextentthatinstitutionsdrivethemarketreactiontodisasters,weshouldexpectthattheeffectisalsopresentinbonds,therebyofferinganotherlaboratorytotestourhypothesis.Wealsocontributetotheliteraturebyshowingthatexposedfirmswithhigh-EScoreorhighsustainability-orientedinstitutioalownershipexperiencelowersellingpressure.Incontrast,thesellingpressureisgreaterforexposedfirmswithhightotalinstitutionalownership.Theseresultsdeepenourunderstandingoftheroleoffinancialinstitutionsinpricingphysicalclimaterisk.2.2. HypothesesOurstudyexaminestheeffectofdisastersonfirms?stockandbondreturns.Becausedisasterscouldabruptlyaffectafirm'sbusinessoperations,theadverseimpactofdisasterscouldcauseinvestorstocommandahigherexpectedreturnonexposedfirms(therisk-compensationhypothesis).Alternatively,investorscouldoverreacttothedisasterimpactbydepressingtheexposedfirm,sstockandbondprices.Aspricesdropbelowthefundamentalvalue,theyreboundinthefollowingmonth(theoverreactionhypothesis).Bothhypothesesleadtothefollowingprediction.Hypothesis1.Therelationbeteendisastersandone-month-aheadreturnsonstocksandbondsofexposedfirmsispositive.Thetwoexplanations(overreactionandrationalrisk-compensationhypotheses)arenotnecessarilymutuallyexclusive.Ournexthypothesesattempttoprovideindicationsastowhichexplanationislikelytoplayalargerrole.Thetrueimpactofdisasters(thelossoffundamentalvaluesbecauseofdisasters)cannotbeimmediatelyobserved.However,asCovalandStafford(2007)andotherssuggest,theexpostdeviationbetweenpricesandfundamentalvaluescanprovideuswithanindicationofoverreaction(i.e.,significantreturnreversalsafterdisasters).IfHypothesis1isdrivenbyinvestors*overreaction,thenweexpectthatthecontemporaneousreturnontheexposedfirminthemonthofdisasterswillbelower,asitreflectstheoversellingactivityofinvestors(DeBondtandThaler1985).Wethusstatethesecondhypothesis,asfollows.Hypothesis2.Ifinvestorsoverreacttodisasters,thenthecontemporaneousrelationbetzeenstock(bond)returnsanddisastersisnegative.Wealsoexaminethelong-runreturndifferentialbetweenexposedandunexposedfirms.IftherelationinHypothesis1continuesinthelongrun,thenitsuggeststhatinvestorsupdatetheirbeliefabouttheclimateriskofexposedfirms,causingthepremiumonexposedfirmstobehigherthanunexposedfirms.Suchacontinuationinreturndifferentialwouldbemoreconsistentwiththerisk-compensationhypothesis.Anotherwaytodeterminewhetherpricesaredepressedbeyondthefundamentalvalueistoexamineinsidertradingactivities.RozeffandZaman(1998)and1.ou(2014)positthat,becauseinsidershavesuperiorknowledgeabouttheirownfirms,insidertradingrevealswhethertheseinsidersbelievetheirstockpricesareundervaluedwithouttheneedtomeasurethelosstofundamentalsbecauseofdisasters.Specifically,ifstockpricesareoverlydepressed,theninsidersarebetteroffpurchasingtheirownsharesandtakeadvantageofthetemporarymisvaluation(Hongetal.2008).Ontheotherhand,ifinsidersbelievethatstockpricesjustifiablyreflectthelossoffirmfundamentalsbecauseofdisasterimpacts,thenwedonotexpecttoseeachangeintheirtradingactivity.Hypothesis3.Ifinvestorsoverreacttodisasters,theninsiderssharePUrchaSeSincreasewhentheirfirmisexposedtodisasters.Theeffectofdisastersonreturnscoulddependoninvestors*attentiontoclimatechange.Priorstudies(Kruegeretal.2020,SchlenkerandTayIor2021)suggestthatinvestorsbegantopayattentiontoclimatechangeriskssincetheearly2000s.Wethereforeexpectthattheeffectofdisasterswouldbemorepronouncedinthelateryearsofoursampleperiod.Painter(2020)suggeststhatthepublicationoftheStemReviewin2006isagoodcandidateeventtotestthisattentioneffect.TheSternReviewwasthefirstofitskindtoquantifythecoststoaddressclimatechangeanditsimpactontheglobaleconomy(Stem2007).Itattractedextensivemediacoveragearoundtheworldandarousedpublicattentiontoglobalwarming.Indeed,p2>J3saISIqcnp-«SlUoQSn«UoSJodJo工ZT90rtzoeqQJnZICc-Zocz-Z,二ZlKq8JoSIUJO一WulojJPUPEOWMoaPainter(2020)findsthatGooglesearchvolumeforclimatechangeincreasedsignificantlyafterthereleaseoftheSternReviewandthatthepricingofclimateriskinthemunicipalbondmarketissignificantonlyafter2006.Assuch,weconjecturethatinvestors*overreactiontodisasterswouldbemorepronouncedafter2006.If,however,rationalinvestorsdominatethepricingofdisasters,thentherisk-compensationhypothesiswouldpredicttheeffectofdisastersonreturnstobesignificantevenbefore2006.Hypothesis4.Theeffectofdisastersonstockandbondreturnsismorepronouncedafter2006.Takentogether,Hypotheses2-4zifsupported,suggestthattheeffectofdisastersonstockandbondreturnsisnotpurelydrivenbytheadverseimpactofdisastersonfirms'fundamentals.Investorshaveheterogenouspreferencestowardsustainability.HongandKacperczyk(2009)s