IMF-货币政策对利率的传导：来自30个欧洲国家的风格化事实（英）-2024.1.docx

INTERNATIONA1.MONETARYFUNDMonetaryPolicyPass-ThroughtoInterestRates:StylizedFactsfrom30EuropeanCountriesRobertBeyer,RuoChen,Claire1.i,FlorianMisch5EzgiO.Ozturk5and1.evRatnovskiWP/24/9IMFWorkingPapersdescriberesearchinprogressbytheauthor(三)andarepublishedtoelicitcommentsandtoencouragedebate.TheviewsexpressedinIMFWorkingPapersarethoseoftheauthor(三)anddonotnecessarilyrepresenttheviewsoftheIMF,itsExecutiveBoard,orIMFmanagement.WoRK-NGPAPER2024JANWP/24/9©2024InternationalMonetaryFundIMFWorkingPaperEuropeanDepartmentMonetaryPolicyPass-ThroughtoInterestRates:StylizedFactsfrom30EuropeanCountriesPreparedbyRobertBeyer,RuoChen,Claire1.i,FlorianMisch*,EzgiO.Ozturk,and1.evRatnovski*AuthorizedfordistributionbyHelgeBergerJanuary2024IMFWorkingPapersdescriberesearchinprogressbytheauthor(三)andarepublishedtoelicitcommentsandtoencouragedebate.TheviewsexpressedinIMFWorkingPapersarethoseoftheauthor(三)anddonotnecessarilyrepresenttheviewsoftheIMF,itsExecutiveBoard,orIMFmanagement.ABSTRACT:Theextenttowhichchangesinmonetarypolicyratesleadtochangesinloananddepositratesforhouseholdsandfirms,referredtoas,pass-through,isanimportantingredientofmonetarypolicytransmissiontooutputandprices.Usingdataonsevendifferentbankinterestratesin30Europeancountries,differentapproaches,andthefullsampleaswellasasubsampleofeuroareacountries,weshowthata)thepass-throughinthepost-pandemichikingcyclehasbeenheterogenousacrosscountriesandtypesofinterestrates;b)thepass-throughhasgenerallybeenweakerandslower,exceptforratesofnon-financialcorporationloansandtimedepositsineuroareacountries;c)differencesinpass-throughovertimeandacrosscountriesformostdepositratesarecorrelatedwithfinancialsectorconcentration,liquidity,andloanopportunities,andd)theeffectsofpass-throughtooutstandingmortgageratesonmonetarytransmissiononpricesandoutputareheterogenousacrosscountries.RECOMMENDEDCITATION:Beyer,R.,Chen,R.,1.i,C.,Misch,F.,Ozturk,E.,andRatnovski,1.(2024).MonetaryPolicyPass-ThroughtoInterestRates:StylizedFactsfrom30EuropeanCountries.IMFWorkingPaperWP/24/9JE1.ClassificationNumbers:E43;E52Keywords:MonetaryPolicyTransmission;MonetaryPolicyPass-ThroughAuthor'sE-MailAddress:rbeyer(g>imf.org;rchenimf.org;yli13imf.org;fmischimf.org;eozturkimf.org;lratnovskiimf.orgCorrespondingauthor.TheauthorswouldliketothankOyaCelasun,1.uisBrandaoMarques,VincenzoGuzzo,SebastianWeber,PelinBerkmen1MahirBinici,OzgeEmeksiz1EzequielCabezon.SalvatoreDeH1Erba1FazurinJamaludin,RuiMano1VinaNguyen,YuShiandECBstaffforusefulcomments.KaylaQin,AgnesaZalezakovaandWeiZhaoprovidedusefulassistance.WORKINGPAPERSMonetaryPolicyPass-ThroughtoInterestRates:StylizedFactsfrom30EuropeanCountriesPreparedbyRobertBeyer,RuoChen,Claire1.i,FlorianMischCorrespondingauthor.,EzgiO.Ozturk,and1.evRatnovskiTheauthorswouldliketothankOyaCelasun,1.uisBrandaoMarques,VincenzoGuzzo,SebastianWeber,PelinBerkmen,MahirBinici1OzgeEmeksiz,EzequielCabezon,SalvatoreDelErba,FazurinJamaludin1RuiMano,VinaNguyen,YuShiandECBstaffforusefulcomments.KaylaQin.AgnesaZalezakovaandWeiZhaoprovidedusefulassistance.Contents1. Introduction52. TheRoleofPolicyPass-ThroughforMonetaryPolicyTransmission63. Data84. StylizedFacts84.1 Inthepost-pandemichikingcycle,pass-throughhasbeenheterogeneousacrosssectors,loananddeposittypes,andcountries84.2 Thepass-throughhasbeenweakerthistime,excepttoNFCloanrates104.3 Thepass-throughhasbeenslowerthistime,excepttoNFCloanrates134.4 1.owerbankingcompetition,amplehouseholdandNFCdeposits,andhighbankingliquidityaresometimesassociatedwithlowerdepositratepass-through174.5 Pass-throughtoratesofexistingmortgagesweakenedovertimeastheshareoffixedratemortgagesincreased,withheterogeneouseffectsacrosscountries215. Conclusions22References23Annex I. DataDescription26Annex II. AdditionalChartsandResults301. IntroductionThepost-pandemicsurgeininflationhastriggeredthemostaggressivemonetarypolicytighteningindecades.Thishas,inturn,generatedrenewedinterestintheaggregateeffectsofmonetarypolicy.Thispaperanalyzesthepass-throughofpolicyratesinEuropeaneconomiestolendinganddepositratesofcreditinstitutions,referredtoasbankinterestrates,therebyprovidinginsightsintoanimportant(butnottheonly)aspectofmonetarypolicytransmission.Theimpactofmonetarypolicy-inducedchangesinloananddepositinterestratesforhouseholdsandnon-financialcorporations(NFCs)isoneofmanychannelsthroughwhichmonetarypolicyaffectsfinancialconditions,andtherebyoutputandprices.Bankinterestratescanaffectagents'investment,consumptionandsavingdecisions,debtservicecostsandsavingsincome,aswellasassetvalues,allofwhichaffectrealeconomicoutcomes.Theextentofpass-throughofmonetarypolicyratechangestobankinterestratesthereforeaffectsmonetarypolicytransmissionandeffectiveness.Inturn,pass-throughtobankinterestratescanbeinfluencedbyfinancialsectorconcentration,liquidity,andtheavailabilityofdepositsrelativetoprofitablelendingopportunities.Thesecharacteristicscanaffectincentivesforbankstocompetefordepositsandpotentiallyunderminepass-throughinhikingcycles.Methodologically,analyzingpass-throughhasadvantages:theeffectsofmonetarypolicyonbankloananddepositratesarearguablymoredirect,oftenwithshorterlagsandfewerconfoundingfactorscomparedtomonetarypolicytransmissiononoutputandprices.Focusingonmonetarypolicypass-throughinEuropeancountriesduringthepost-pandemichikingcycle,thispapermakestwokeycontributionstothepreviousliterature.First,wecompiledataonsevendifferentinterestratesfor30Europeancountriesbothwithineuroareaandnon-euroareacountrieswithindependentmonetarypolicy(definedashavingnoofficialpegandnotbeingfullyeuroized).ThisdatasetcontainsinterestratesforkeytypesofnewloansanddepositsofbothhouseholdsandNFCs.Second,wepresentarangeofearlyevidenceonstylizedfactsonmonetarypass-throughduringthepost-pandemichikingcycleusingamixofdescriptivestatisticsandregressionanalyses,bothforthefullsampleofcountriesandasubsetofeuroareacountries(EA-12countries).Inparticular,wedocumentthat:a)thepass-throughinthepost-pandemichikingcyclehasbeenheterogenousacrosscountriesandtypesofinterestrates;b)thepass-throughhasbeenweakerandslowerrelativetopasthikingcycles,exceptforratesofnewloanstoNFCsandNFCtimedeposits(andcontrollingforthepeculiaritiesofthecurrentcycleincludingpossibleanticipationeffectsoflowinitialratesdoesnotalterthesefindingqualitatively);c)theweakeningofpass-throughtodepositratesisinpartassociatedwithhigherfinancialsectorconcentration,moreampledepositsrelativetoloanopportunities,andliquidity;andd)usingdataonoutstandingmortgagesasanexample,theeffectsofpass-throughonaggregatemonetarypolicytransmissionareheterogenousacrosscountries.Takingaccountofthepeculiaritiesofthepost-pandemichikingcycleincludingpossibleanticipationeffectsandverylowinitialratesdoesnotchangethesestylizedfacts.Ourpapercontributestotheempiricalliteraturestudyingmonetarypolicypass-throughtobankinterestrates,includinginEurope.Onestrandofthisliteraturefocusesontheeuroareaandmostlycoversthe2000s.Forexample,SorensenandWerner(2006),BernhoferandvanTreeck(2013),SanderandKleimeier(2006)documenthowpass-throughdiffersacrosscountries,interestratetypes,andintheshortvs.longrun(seeforexampledeBondt12005,forasurvey),andSaborowskiandWeber(2013)examinethedeterminantsofpass-throughtotheaggregatebanklendingrateinalargepanelofcountries.OtherpapersfocusonEuropeancountriesoutsideoftheeurozone.Forexample,GregorandMelecky(2018)focusonpass-throughtolendingratesinCzechRepublic,Egertetal.(2007)studytheinterestratepass-throughinfivecentralandeasternEuropeancountries,Beckeretal.(2012)analyzethepass-throughtomortgageratesintheUK,andHansenandWelz(2011)analyzethepass-throughtolendinganddepositratesinSweden.Ameta-datastudyoninterestratepass-through,basedonmorethanathousandestimatesfrom50studies,findsthattheaveragepass-throughfrommonetarypolicyratestobanklendingratesisaround0.8(GregorandMelecky,2021).Ourpaperisalsorelatedtoanotherstrandoftheliteraturewhichexploitsheterogeneityinthepass-throughacrossbankswithincountries.Forinstance,Wangetal.(2022)usedataonU.S.banksandshowthatbankmarketpoweraffectspass-through,andthattheeffectsdependonthelevelofthepolicyrate.Kho(2023)usescountry-andbank-leveldatafrom13euroareacountriesfrombeforethepost-pandemichikingcycleandshowsthatthedegreeofbankingsectorconcentrationhasasymmetriceffectsonpass-throughtoovernight(ON)depositratesduringtighteningandlooseningcycles.1.dpez-Quiles(2021)estimatespass-throughtomoneymarketinterestratesusingtransaction-leveldataofthelargestbanksintheeuroarea.Importantly,ourpaperovercomesacriticallimitationoftheexistingliteraturebycoveringthepost-pandemichikingcycleandusesdataonindividualcountriesorcoveringonlyasubsetofeuroareacountries.Hence,ourpaperisrelatedtorecentstudiesanalyzingthepass-throughduringthepost-pandemichikingcycle,whichallreportfindingsthatareconsistentwithours.Forexample,fortheeuroarea,ByrneandFoster(2023)findthatpass-throughisgenerallyweakernowrelativetopreviouscycles,MesserandNiepmann(2023)findthatpass-throughtodepositrateshasbeenmoresluggishthaninthepast,1.ane(2023aand2023d)reportsthatthepass-throughtoNFCloanshasbeenstrongandthatloanratesstartedtorisebeforetheECBhikeditspolicyratesinJuly2022,and1.ane(2023c)findsweakerpass-throughtoO/Ndepositratesthantotimedepositratesinthepost-pandemichikingcycle.ForotherEuropeaneconomies,CabezonandKovachevska(2024)findsthatpass-throughinNorthMacedoniaisweakercomparedtoregionalpeersandtheeuroarea,andBinici(2024)attributesrelativelyweakpass-throughinMaltainthepost-pandemictighteningcycletoabundantliquidityinthebankingsystem.Thepaperisorganizedasfollows:Section2discussesconceptuallytheroleofpolicypass-throughinoverallmonetarypolicytransmission.Section3describesthedata.Section4presentsstylizedfactsonmonetarypolicypass-through.Section5concludes.2. TheRoleofPolicyPass-ThroughforMonetaryPolicyTransmissionMonetarypolicyusuallyusestheshort-termmoneymarketinterestrateasitsoperationaltarget.Themoneymarketrateistransmittedviafinancialmarketstodepositandlendingratesofcreditinstitutions('bankinterestrates'),andthroughthemtothebroadereconomy.Theabilityofthecentralbanktoachieveitsultimateobjectivesprice,financial,andoverallmacroeconomicstabilityhingesinpartonthetransmissionofthepolicyratetootherinterestrates.Inturn,bankinterestratesaffectthemacroeconomythroughseveralrelatedchannels:1. Interestratechannel:Changesininterestratesalterfirminvestmentbehaviorandhouseholdconsumptionandsavingchoices.Higherrealinterestratesincreasethehurdle(break-even)ratesforinvestment,therebyreducingit,andtheymakesavingmoreattractive,whichinturnreducesconsumption.Thisisthestandardneo-classicalchannelofmonetarypolicy(GoodfriendandKing,1997).2. Cashflowchannel:Changesininterestratesaffectagents*incomeandcostsrelatingtoexistingassetsandliabilitiesthatcarryvariableinterestrates.Inotherwords,tightermonetarypolicycandirectlyincreaseinterestincomeanddebtSerViCecosts,butthecashfloweffectsdependonwhetherhouseholdsandfirmsarenetborrowersorsaversandtheirexposuretovariableratedebtanddeposits.Withdifferentpropensitiestoconsumeorinvestacrosshouseholdsandfirms,thischannelcanaffectaggregatemacroeconomicoutcomes(see,e.g.,Auclert,2019).Theno-financialsectorinagiveneconomymaypredominantlyconsistofeitherlosersorwinners,sothattheaggregateeffectsdonotnetoutevenifthemarginalpropensitiestoinvestandconsumeareidentical.Inthatcase,theaggregateconsumptionandinvestmenteffectscouldbeoffsetbychangesinthevolumeofloansthroughchangesincreditconditions.1.owpass-throughtodepositratesmaymatterinthiscontextgiventhatmanyhouseholdsinEuropearenetsavers.3. Balancesheetchannel:Higherinterestratesreducethevalueofassetsusedascollateral,therebytighteningagents'financialconstraints.Forfirms,thisamplifiestheneoclassicalimpactofinterestratesoninvestment,whileforhouseholdsthiscanaltertheconsumptionofdurablegoods(KiyotakiandMoore,1997).Additionally,wealtheffects(lowerassetvaluationsduetohigherinterestrates)reducethehouseholds'propensitytoconsume(1.ettauand1.udvigson12004).4. Bankingchannel:Higherinterestratestendtobeassociatedwithhigherbanknetinterestratemargins,therebyincreasingbankprofits(Borioetal.,2017).Interestratesmayalsoaffectbankfundingconditions,ashigherdepositratesmayincreasedepositsupply.Higherbankprofitsandampledepositsmayincreasebanklendingcapacity(KashyapandStein,2000;Drechsleretal.,2017).Seealso1.ane(2023a)foradiscussionofthischannel.Obviously,thereareothermonetarypolicytransmissionchannelswhichdonotdirectlydependoninterestratepass-through;see,forexample,Ceviketal.(2023),1.ane(2023a),andMishkin(1996)formoredetailsontheinterestratechannelsdiscussedhereandacomprehensivelistofmonetarytransmissionchannelsmorebroadly.Table1:Effectsofmonetarypolicypass-throughtobankinterestratesontotherealeconomyChannelsthatdirectionallyweakentheeffectsOfmonetarypolicytighteningshowninorangeChanneIsofMPtransmissionMechanismofchannelduringMPtighteningEffectsofhigherpass-throughduringMPtighteningEffectsofhigherpass-throughtoloanratesEffectsofhigherpass-throughtodepositratesInterestratechannelHigherhurdleratefornewinvestment,andhighersavingsremuneration1.essinvestmentMoresaving÷1.essconsumptionCashflowchannelHigherinterestincomeanddebtservicescostforexistingexposures1.owercashflow÷1.essconsumptionandinvestmentHighercashflow"÷moreconsumptionandinvestmentBalancesheetchannel1.owervalueofcollateralTighternon-pricecreditconditions÷1.essinvestmentBankingchannel1.owerbanks'networthTighterbankfundingconditions÷1.esslendingHighercostofbankfundingMoresupplyofdeposits->MorelendingOnbalance,ahigherpass-throughofmonetarypolicytointerestratestendstostrengthenmonetarypolicytransmissionthrougheachofthesechannels(Table1).However,ahigherpass-throughtodepositratescancreateambiguouseffectsinsomecases.Forinstance,higherpass-throughinahikingcyclereducesprivateinvestmentandconsumptionthroughahigherhurdleratefornewinvestmentandsavingsremuneration,butthiseffectisweakenedbyhighercashflowsfromtheexistingstockofsavings.3. DataWecompileadatasetonloananddepositinterestratesofcreditinstitutionsin30Europeaneconomiesusingdatafromcentralbanksandstatisticalagencies,referredtoasbankinterestrates.Ourdataincludessevendifferenttypesofinterestrates:interestratesonovernightdepositsandtimedeposits,bothforhouseholdsandNFCs,consumerloans,mortgages,andloanstoNFCs.Theinterestratesgenerallyrefertonewdepositsandloans,butwealsocollectdataoninterestratesofoutstandingmortgages.Insometimeseries,thereisthedistinctionbetweenpurenewbusinessandnewbusinesscombinedwithrenegotiations.Wegenerallyusepurebanksratesfornewbusiness,butforNFCloanratesinCzechRepublic,Cyprus,Estonia,FinlandandRomaniacountries,weuseinformationfrombothseriestocompiledataserieswithfewergaps.Ourdataisatmonthlyfrequencyandconstitutesanunbalancedpanel,spanning30Europeancountriesthatareeitherpartoftheeuroareaorhaveindependentmonetarypolicy(inthesensethattheydonothavehardpegsorarecompletelyeuroized).Twothirdsofcountrieshavedataonallseventypesofinterestrates(AnnexTableA2).Someseriesrea