Centre for Economic Forecasting, London Business S

INTERNATIONALINTERESTRATELINKAGES:ATIMEVARYINGPARAMETERAPPROACHGeoffreyWilliamsCentreforEconomicForecasting,LondonBusinessSchool,SussexPlace,Regent’sPark,London.NW14SA.Tel:[phone]e-mail:[e-mail]10-99December1998IwouldliketothankGuglielmoMariaCaporale,StephenHall,BrianHenryandparticipantsatthe1999RoyalEconomicSocietyConference,UniversityofNottingham,forcommentsonthisstudy.DavidUptonprovidedexcellentresearchassistance.FinancialsupportfromESRCgrantnoR000237486,“EMUandFinancialMarkets,”isgratefullyacknowledged.Allerrorsremaintheresponsibilityoftheauthor.ABSTRACTThispaperexaminestheissueofstatecontingencyininterestratelinkagesacrosstheG7economiesbyemployingKalman(1960,1963)filtertechniquestoestimatetimevaryingparametermodelsofbilateraldependence.Ourresultsshowthatthereisconvincingevidencethatinternationallinkagesininterestratesvaryconsiderablyovertime.Thereareperiodswheninternationalinterestratemovementsareimportantindeterminingnationalinterestratesandperiodswhendomesticratesareunaffectedbyinternationaldevelopments.AlsothereisevidencethattheinfluenceoftheUSandGermanyontheotherG7economieschangesovertimeandduringparticularpolicyepisodes.Weprovideacharacterizationofthemechanismswhichmightunderlieourempiricalresults.Keywords:Interestratelinkages,Timevariance,Kalmanfilter.JELClassification:C2,E4,F3ISSN0969-6598NONTECHNICALSUMMARYItisoftenclaimedthatinternationalinterestratesshowadistincttendencytomovetogetherthroughtime.Indeedmanytheoriesofinternationaltradeadoptthisideaasbasicpremiseandempiricalstudiesoftenlookforevidenceofco-movementbothingeneraltermsandaslong-run,fundamentalprocessesdrivinginterestratebehaviour.Ifinterestratesbehavelikeassetpricesco-movementcouldbeexplainedastheresultofarbitrageamongstprofitseekingmarketparticipants.Alternatively,ifinterestratesareusedaspolicyinstrumentsco-movementmightariseduetoco-ordinationefforts,suchasthoseamongstcountriesintheEuropeanUnion(EU)orduetointerestratedominancearisingfromtheinfluenceoflargeropeneconomiesonsmalleropeneconomies.Inanyoftheabovecasesincreasingglobalizationofeconomicactivityoughttohavemadetheseeffectsmorepronouncedandasaresultweoughttobeabletoobservetheminquitegeneralframeworks.Inthispaperwetaketheviewthatwhilstassumingco-movementamongstinternationalinterestratesisintuitivelyappealing,therearecompellingreasonstoexpectthatpatternsininterestratelinkagesshouldshowsignificantvariationovertime.Wethereforetakeissuewiththeviewthatweshouldbeabletoidentifylong-runorequilibriumrelationshipsamongstinterestratesandargueinsteadthatweshouldobserveinterestco-movementthatchangesovertimeandthatmaybelikelytocoincidewithparticularpolicyepisodes.Forexample,ifinterestratesbehavelikeassetpricesthenforstableriskenvironmentswithpredictableexchangeratemovementsandfluidcapitalflowsweshouldobservestronglinksinternationally.Timevariationwouldbeexpectedtoemergeifforexample,theriskenvironmentchanges,ifthereisachangeintheexchangerateregimeorifcapitalcontrolsareimposedorremoved.Anyoreachoftheseorotherfactorswouldbeexpectedtoaffectthestrengthofthelinksbetweendomesticandinternationalinterestrates.Indeedeachandallofthesefactorshaveinfactbeenobservedinrecenteconomicactivity.Ifinterestratesarebettercharacterizedaspolicyinstrumentsthensuccessfulco-ordinationofpolicybetweencountries,inforexampleanexchangeratetargetingregime,wouldbeexpectedtostrengtheninterestratelinkages.Wewouldexpecteitherasecularincreaseinanymeasureoftheinfluenceofonerateonanotherand/orareductionintheuncertaintysurroundingsuchameasurei.e.itsvariance.Evenifpolicyco-ordinationwasnotaspecificaspiration,wemightstillexpecttimevariationinco-movementifforexampleonecountrydevelopedregionaldominanceinthesenseofso-called,“gravitymodels.”Theissueofindependenceofdomesticpolicycouldthenbeinvestigatedbyevaluatingwhethertheinfluenceofinterestmovementsinlargereconomiesonsmallereconomieshadincreased.Inthispaperwethereforeprovideastylizationoftheseideasbyinvestigatinginterestratelinkagesinatimevaryingcontext.Wearguethatifchangesintheinternationalenvironment,particularlytheincreasedglobalizationoffinancialcapital,havereallybecomeimportanttheyoughttohaveaffectedtheco-movementofinterestratesinobservableways.Also,ifpolicyco-ordinationeffortshavetrulybeeneffective,thesetoooughttohavechangedthepatternofco-movementinwaysthatcanbequantifiedinaquitegeneralframeworksuchastheonewesuggest.Ourempiricalresultsshowthatpotentiallong-runrelationshipsimpliedbystandardparityconditionsareunlikelytobeidentifiedbecauseofthepresenceoftimevariationinriskpremiaandinthecoefficientmeasuringtheinformationcontentofforeignondomesticrates.Wearguethatthistimevariationistobeexpectedaprioriandislikelytoberelatedtogeneralchangesinthefinancialenvironmentand/orspecificpolicyepisodes.ToillustratethesepointsweevaluateinterestratelinkagesusingthegeneralstatedependenceframeworkprovidedbytheKalmanfilter.Inquitegeneraltermsthereappearstobeatendencyforinterestraterelationshipstoexhibitcleartimevariationandthechangeinthestructureoftherelationshipsca