美联储衡量金融机构的利率风险管理（英）

FinanceandEconomicsDiscussionSeries FederalReserveBoard,Washington,D.C.ISSN1936-2854(Print) ISSN2767-3898(Online) MeasuringInterestRateRiskManagementbyFinancialInstitutions CelsoBrunetti,NathanFoley-Fisher,St´ephaneVerani2023-067 Pleasecitethispaperas: Brunetti,Celso,NathanFoley-Fisher,andSt´ephaneVerani(2023).“MeasuringInter-estRateRiskManagementbyFinancialInstitutions,”FinanceandEconomicsDiscus-sionSeries2023-067.Washington:BoardofGovernorsoftheFederalReserveSystem,https://doi.org/10.17016/FEDS.2023.067. NOTE:StaffworkingpapersintheFinanceandEconomicsDiscussionSeries(FEDS)arepreliminarymaterialscirculatedtostimulatediscussionandcriticalcomment.TheanalysisandconclusionssetfortharethoseoftheauthorsanddonotindicateconcurrencebyothermembersoftheresearchstaffortheBoardofGovernors.ReferencesinpublicationstotheFinanceandEconomicsDiscussionSeries(otherthanacknowledgement)shouldbeclearedwiththeauthor(s)toprotectthetentativecharacterofthesepapers. MeasuringInterestRateRiskManagementbyFinancialInstitutions∗ CelsoBrunetti1,NathanFoley-Fisher1,andStéphaneVerani11FederalReserveBoard Firstversion:June2022;thisversion:August2023 Abstract Financialintermediariesmanagemyriadinterestrateriskexpo-sures.Weproposeanewmethodtomeasurefinancialintermediaries’residualinterestrateriskusinghigh-frequencyfinancialmarketdata.Ourmethodexploitsallavailablehigh-frequencyinformationandisvalidunderextremelyweakassumptions.ApplyingthemethodtoU.S.lifeinsurers,wefindtheirinterestrateriskmanagementstrategiesaregenerallyeffective.However,lifeinsurersaremoresensitivetochangesinlong-terminterestratesthanpropertyandcasualtyinsurers.Weshowthatthetermpremiumhelpstoexplainthedifferenceinsensitivitiesbetweenthetwotypesofinsurer. JELCodes:Keywords: G20;C58 financialinstitutions;interestrateriskmanagement;high- frequencyfinancialeconometrics;subsampling;lifeinsurers. ∗Forprovidingvaluablecomments,wewouldliketothank,withoutimplication, MarkCarey,BurcuDuygan-Bump,PeterHansen,MaxHuber,AnastasiaKartasheva,BorghanNarajabad,AndrewPatton,MattPritsker,RobertoRenò,RichRosen,OlegSokolinskiy,PavelSzerszenandparticipantsintheSocietyforEconomicMeasurementAnnualConference2023,theInternationalRiskManagementConference2023,andseminarsattheEuropeanCentralBank,St.GallenUniversity,andtheFederalReserveBoard.WearegratefultoJuliaSilbertandReneeGarrowforexceptionalresearchassistance.Theviewsinthispaperaresolelytheauthors’andshouldnotbeinterpretedasreflectingtheviewsoftheBoardofGovernorsoftheFederalReserveSystemorofanyotherpersonassociatedwiththeFederalReserveSystem. 1Introduction Financialintermediariesareexposedtointerestraterisk.Theyhavemultiplesourcesofexposurearisingfromcashflowdifferencesacrossbalancesheetcomponentsaswellascontractualorembeddedoptionswithasymmetricpayoffcharacteristics.Althoughintermediarieshaveawiderangeofassetandliabilitymanagementtoolsavailabletohedgeinterestraterisk,theydonotfullyinsulatethemselvesfromallpotentialchangesininterestratesforseveralreasons.1Financialmarketsmaybeincomplete,fullyhedgingmaybeprohibitedbyitscost,andcarryinginterestrateriskmaybeasourceofearnings.2Thus,financialintermediariescarrysomeresidualexposuretointerestraterisk,whichcouldhavesignificantconsequencesforfinancialstabilityandmacroeconomicoutcomesinbadstatesoftheworld(HolmstromandTirole,1997;BrunnermeierandSannikov,2014). Inthispaper,weproposeanewmethodtomeasurethetime-varying residualinterestrateriskexposureoffinancialintermediariesusingminute- 1Riskmanagersatfinancialinstitutionsareexpectedtomonitorandmanageinterestrateexposuresatprudentlevels,butnotfullyeliminatetherisk.Supervisorsprovidedetailedguidanceonmanagementpracticesandcoordinatetheirstandards.See,forexample,theOfficeoftheComptrolleroftheCurrency(OCC)RevisedHandbookMarch2020,theFederalDepositInsuranceCorporation(FDIC)LetteronFinancialInstitutionManagementofInterestRateRisk2010,theFederalReserveBoard(FRB)SupervisoryManualonInterestRateRisk,theNationalAssociationofInsuranceCommissioners(NAIC)Risk-BasedCapitalforInsurersModelAct,theOCC-FDIC-FRBJointPolicyStatementonInterestRateRisk1996,andtheBaselCommitteeonBankingSupervisionGuidanceonStandards2014. 2Evenanestablishedhedgingstrategymaybeexposedto“basisrisk”—thatis,it mightloseitseffectiveness. by-minutefinancialmarketdata.Wecalculatethedailyrealizedcovarianceofhigh-frequencystockreturnsforthoseintermediariesandTreasurysecurityreturns.WeconstructaconditionalcovariancebyprojectingoutaggregatestockmarketreturnsfromstockreturnsandTreasurysecurity returns.Wethenintroducerealizedgammaastheratiooftheconditional covariancetothedailyrealizedconditionalvarianceofTreasurysecurityreturns.Realizedgammaisadailyestimateofthesensitivityofanindividualfirm’sstockpricereturnstorealizedchangesininterestrates.Wecalculatereturnsatfive-minuteintervalsusingeverypossiblefive-minutegridpointinatradingday,exploitingallavailablehigh-frequencyinformationasdescribedinZhang,MyklandandAït-Sahalia(2005). Wethenproposeanewstatisticaltestofthedailyresidualinterestrateriskexposureoffinancialinterme