美聯(lián)儲-超邊際借款人與貨幣政策的抵押貸款支付渠道（英）

FinanceandEconomicsDiscussionSeries

FederalReserveBoard,Washington,D.C.

ISSN1936-2854(Print)

ISSN2767-3898(Online)

InframarginalBorrowersandtheMortgagePaymentChannelofMonetaryPolicy

DanielRingo

2024-069

Pleasecitethispaperas:

Ringo,Daniel(2024).“InframarginalBorrowersandtheMortgagePaymentChannelofMonetaryPolicy,”FinanceandEconomicsDiscussionSeries2024-069.Washington:BoardofGovernorsoftheFederalReserveSystem,

/10.17016/FEDS.2024.069

.

NOTE:StafworkingpapersintheFinanceandEconomicsDiscussionSeries(FEDS)arepreliminarymaterialscirculatedtostimulatediscussionandcriticalcomment.TheanalysisandconclusionssetfortharethoseoftheauthorsanddonotindicateconcurrencebyothermembersoftheresearchstafortheBoardofGovernors.ReferencesinpublicationstotheFinanceandEconomicsDiscussionSeries(otherthanacknowledgement)shouldbeclearedwiththeauthor(s)toprotectthetentativecharacterofthesepapers.

1

InframarginalBorrowersandtheMortgagePaymentChannelofMonetaryPolicy

DanielRingo*

FederalReserveBoard

July30,2024

Abstract

Despitethewidespreaduseof?xed-ratemortgagesintheUnitedStates,Ishowthatmonetarypolicyisefectivelypassedthroughtoaggregateoutstandingmortgagedebtservice.Usingcreditbureau,lender,andservicerdataonmortgagepaymentsandoriginationsandexogenousmone-tarypolicyshocks,Iestimateamortgageratesemi-elasticityofpaymentsover10.Inframarginalborrowers—householdswhosechoicetobuyahomeorre?nancedoesnotdependonthepartic-ularmonetarypolicydecisionunderconsideration—arethemostimportantconduit,explainingoverhalfofthepass-through.Consistentlylarge且owsofinframarginalborrowingrelativetothestockofoutstandingdebtaccountforthestrengthofthischannel.Householdswithadjustable-ratemortgagesandmarginalre?nancers,thefocusofmuchoftheliteratureonmonetarypolicy’sefectonmortgageborrowers,eachexplainabout20percentofthepass-through.Ishowthemortgagepaymentchannelinducesalagintheoperationofpolicy,asthecumulativeefectsondebtservicebuildovertimeinresponsetopersistentshockstolonger-termrates.Estimatedmagnitudessuggestthatmortgagepaymentsareaprimarychannelbywhichmonetarypolicyafectsconsumption.

Keywords:Monetarypolicy;interestrates;re?nancingchannel;debtserviceJELcodes:G21;E43;E52.

*MS93,BoardofGovernorsoftheFederalReserveSystem,WashingtonD.C.20551daniel.r.ringo@.ThankstoNeilBhutta,JohnDriscoll,andseminarparticipantsattheFederalRe-serveBoardandGhentUniversityEmpiricalMacroeconomicsWorkshopforhelpfulcomments.TheanalysisandconclusionssetfortharethoseoftheauthoranddonotindicateconcurrencebyothermembersoftheresearchstafortheBoardofGovernors.

2

1Introduction

Monetarypolicycanafecthouseholds’consumptionviatheirdebtservicepayments,in-creasingtheirdisposableincomesnetofrequiredpaymentsasinterestratesdecreaseandviceversa.Residentialmortgagedebtrepresentstheoutrightmajorityofallhouseholddebtandisthelargestsourceofdebtserviceobligations,andsoshouldbeapotentchannelbywhichmonetarypolicycanin且uenceconsumerdemand.IntheUnitedStates,however,thepopularityof?xed-ratemortgagespresentsanobstacletothepass-throughofinterestratechangestopayments.Whenratesrise,debtorhouseholdswillhangontotheiroldratesandbeunafected.Whenratesfall,onlyafractionofhouseholdsforwhomre?nancingwouldbepro?tableexercisetheiroptiontodoso.Thefrictionsinherentinthisprocesshavemoti-vatedagrowingbodyofliteratureanalyzingthere?nancingchannelofmonetarypolicy.1Inaninternationalcontext,itiscommonlyarguedthattheprevalanceof?xed-ratemortgagesdampenstheefectivenessofmonetarypolicyincountriesliketheU.S.2

Inthispaper,Ishowthatthepass-throughofmonetarypolicyshockstomortgagepaymentsintheU.S.isactuallyquiterobustduetothelargevolumeofborrowingbyinframarginalborrowers.Thesearehouseholdswhosedecisiontotakeoutanewloandoesnotdependontheparticularmonetarypolicyactionunderconsideration,butwhosepaymentforthatnewloanstilldependsonthestanceofpolicy.Importantly,inframarginalborrowersincludehouseholdswhoareborrowingforthepurposesofhomepurchase,re?nancing,oranyotherends.Whileonlyasmallpercentageofhouseholdsengageinanynewmortgageborrowinginagivenyear,indollartermsannualnewborrowingrepresentsapproximatelyonequarteroftheoutstandingstockofhouseholdmortgagedebt.Thisisbecausenewloanstendtobelargerthantheunpaidprincipalbalanceofexistingloans,andexplainsthestrengthofthemortgagepaymentchannel.Historically,alargepercentageofmortgagedebthascarriedarecentlydeterminedinterestrate.

Theefectofatypicalmonetarypolicyshockonthemonthlypaymentsofaninfra-marginalborrowerissmallrelativetothatexperiencedbyamarginalre?nancer—ahousehold

1SeeAmromin,BhuttaandKeys(2020)forahelpfulsummary.

2See,e.g.,Bernanke(2007),Calza,MonacelliandStracca(2013),Rubio(2011),orKimandLim(2020).

3

thatwouldbeinducedintooroutofre?nancingtheircurrentmortgagebyasmallchangeinrates.However,inframarginalborrowersaresomuchmorenumerousthanmarginalbor-rowersthattheinframarginalchannelismorethantwiceasimportantforthepass-throughofrateshockstoaggregatepayments.I?ndthatoverthelast25years,amonetarypol-icyshockwhichcutsmortgageratesbyonly7basispointswouldreduceaggregateannualmortgagepaymentsbyapproximately$9billionin2023dollars.Inframarginalborrowerswouldaccountforabout$5billionofthisreduction,andmarginalre?nancers$2billion.Anadditional$2billionwouldgototheminorityofborrowerswithadjustable-ratemort-gages(ARMs),althoughthislatterchannelhasdiminishedinrecentyears.Marginalhomepurchaseborrowersandtheintensivemarginofborrowing(i.e.,theamountborrowedcon-ditionalongettingaloan)arenotsigni?cantchannelsofpass-throughtopaymentsinthe?rstfewyearspost-shock.

Thestanceofmonetarypolicyandthelevelofmortgageratesareendogenoustothestateoftheeconomy.Toconsistentlyestimatetheefectsofpolicydecisions,Ithereforeexploitvariationfromtheexogenous,unanticipatedcomponentsofrateresponsestoFederalOpenMarketCommittee(FOMC)announcements.Ibeginbyestimatingtheefectsofmonetarypolicyshocksonmortgageinterestrates.Itisimportantheretorecognizethatmortgageinterestratesaretiedtoyieldsonrelativelylonger-maturitydebt(asatypicalmortgagewillendureforyearsbeforeprepaying)andtheFederalReservehastoolsatitsdisposalthatafectdiferentportionsoftheyieldcurvediferently.TheFedcandirectlycontrolveryshort-termratesbysettingtheovernightfederalfundsrate.Longer-termratesdependmoreoninvestors’beliefsaboutthefuturepathofmonetarypolicydecisions,andtheFedcanattempttoin且uencethesebeliefsviaforwardguidanceandassetpurchases.IthereforeusetheseriesofmonetarypolicyshocksestimatedbySwanson(2021),whoseparatelyidenti?edshockstothefederalfundsrate,forwardguidance,andlarge-scaleassetpurchases(LSAP)

containedinFOMCannouncements.3UsingthelocalprojectionsestimatorofJord、a(2005),

3Lewis(2023)alsoestimatesshockseriesthatdisentanglethesecomponentsusingdiferentidentifyingassumptionsthanSwanson(2021).Inaddition,Lewis(2023)attemptstoisolateafourthcomponent,theFedinformationefect.InAppendixSectionA.1Ishowmyresultsarerobusttousingthesealternativeshocksseries.

4

Iestimatethataonestandarddeviationforwardguidanceshockraisesmortgageratesbyanaverageof7.4basispointsoverthethreeyearspost-shock.4Shockstothefederalfundsratedonotshowanydemonstrableefectonmortgagerates,however.TheLSAPshockisestimatedtohaveastronginitialefectonmortgagerates,butthisquicklyfades.

Next,Iestimatehowtheaveragepaymentsofmortgageborrowersareafectedbymon-etarypolicyshocksempirically.Iconstructatimeseriesoftheaveragemortgagepaymentsofalloutstandingmortgageborrowersusingnationallyrepresentativecreditbureaudatafrom1999through2023.Againusingthelocalprojectionsestimator,Iestimatethataonestandarddeviationforwardguidanceshockincreasedtheaveragepaymentsofborrowersby0.89percentafterthreeyears,orapproximately$9billionperyear.Puttingtheestimatesto-gether,eachbasispointincreaseinmortgageratesisestimatedtoincreaseannualpaymentsbyover$1billion;inotherwords,theinterestratesemi-elasticityofmortgagepaymentsisabout10.

Theestimatedefectsonmortgagedebtservicearelargeenoughtoexplainmuchoftheconsumptionresponsetomonetarypolicyshocks.Cloyne,FerreiraandSurico(2020)esti-matethatintheU.S.a25basispointgenericmonetarypolicyshockwoulddecreasedurableandnon-durableconsumptionbyapproximately0.9and0.15percent,respectively,threeyearspost-shock.Withreceentannualpersonalconsumptionexpendituresofapproximately$2trillionand$17trillioninthesecategories,thisimpliesatotalreductionofabout$45billionperyear.Myestimatessuggestthata25basispointshocktomortgagerateswouldincreaseannualmortgagepaymentsbyabout$30billionbyyearthree.Ifthemarginalpropensitytoconsumeoutofapersistentchangeindisposableincomeishigh,mostoftheefectofmonetarypolicyonconsumptioncanthereforebeexplainedbythedirectefectofthemortgagepaymentchannel.5Anymultiplierefectsfromtheshocktomortgageborrowers’

4Priorstudies,suchasHamilton(2008),Gorea,KryvtsovandKudlyak(2022),andRingo(2024),haveshowntheefectofforwardguidanceonmortgageratesintheshort-term,butthisstudyisthe?rstIamawareoftodemonstratethattheefectpersistsforyears.

5Whileestimatesofthemarginalpropensitytoconsumevaryintheliterature,mostresearchers?ndthatitishighforpermanentshockstoincome.SeeJappelliandPistaferri(2010)forareviewoftheliterature.Changesinmortgagepaymentsrepresentatransferbetweenborrowersandcreditors,sotherearepotentiallyofsettingconsumptionefects.However,mostmortgagedebtintheUS(oftenpackagedintomortgagebackedsecurities)isheldbycommercialbanks,theFederalReserve,orinstitutionalinvestors,allofwhompresumablyhavealowmarginalpropensitytoconsume.

5

consumptionwouldfurtherincreasetheimportanceofthemortgagepaymentchannel.Thestrengthofthemortgagepaymentchannelalsoprovidesanexplanationformonetarypol-icy’sefectonconsumptiondespitemicroevidencesuggestingalowintertemporalelasticityofsubstitution(see,e.g.,Bestetal.,2020).

Thetotalestimatedchangeinpaymentsincludestheefectsoninframarginalborrowers,marginalborrowers,andexistingARMs(whosemortgagorsdonotneedtoengageinanyneworiginationstobeafected).Additionally,householdsengaginginnewborrowingmayofsetsomeoftheefectofmonetarypolicyonpaymentsbymakinganintensivemarginadjustment,i.e.reducingthesizeoftheirloansinthefaceofhigherrates.Toidentifytherelativeimportanceofthesechannels,Iestimatetheirmagnitudesseparately.

Inframarginalborrowersaresimpletomodelbecause,byde?nition,theirborrowingdeci-sionsareexogenoustotherateshock.Themagnitudeoftheinframarginalborrowerchannelcanthereforebestraightforwardlyestimatedviaasimpleback-of-the-envelopecalculation.Usingtheaverageoriginationvolumeandmortgageinterestrateoverthepast25years,Icalculatethatcuttingratesby7.4basispointsforthreeyearsshouldreduceaggregatemort-gagedebtservicebyapproximately$5.2billionperannum.Theefectbuildslinearlyovertimesolongasthelowerrateismaintained,aseachmonthanewcohortofinframarginalborrowersareafectedwhilecohortsthatpreviouslygottheirloancontinuetopaythelowerrate.Comparingthissimplecalculationtothetotalestimatedchangeinpayments,itseemsthatinframarginalborrowersarereponsibleforjustoverhalfofthepass-throughofmonetarypolicy.

Inadditiontotheincreasedordecreasedpaymentsfacedbytheseinframarginalborrow-ers,arateshockcanpullinorexcludemarginalborrowers.Thenumberofsuchborrowerscanbeestimatedbytheresponsivenessoforiginationvolumestorateshocks—thedifer-encebetweenthisestimatednumberandthetotalnumberoforiginationsisthenumberofinframarginalborrowers.Marginalborrowersarevastlyoutnumberedbyinframarginalborrowersforrateshocksofatypicalhistoricalmagnitude,buttheefectontheirpaymentsper-borrowermaybemuchlarger.Thisisbecause,formarginalborrowers,thecounterfactualpaymentistheircurrentexistingpayment(withitsoldratepotentiallyverydiferentfrom

6

currentrates),notthepaymenttheywouldgetoriginatingthesameloanwithamarginallydiferentrate(asitwouldbeforinframarginalborrowers).6

Todeterminethemagnitudeofthemarginalre?nancechannel,Iestimatetheefectofthesamemonetarypolicyshockonthenumberofweeklyre?nanceoriginationsoverthe1999-2023period.Againusingalocalprojectionsestimator,I?ndtheshockdecreasesre?nancingvolumesby3.5percentoverthethreeyearspost-shock,suggestingapproximately$2billionofthetotalincreaseinpaymentscomesfrommarginalre?nancers.Forthemarginalhomepurchasechannel,Irunthesameestimatoronthenumberofweeklyhomepurchaseorigi-nations.Idonot?ndefectsonhomepurchaselendingvolumesthatareeitherstatisticallysigni?cantoreconomicallyconsequentialforaveragepaymentamounts.

TodeterminethemagnitudeoftheARMchannel,IuseamonthlypanelofservicingrecordsofindividualARMsfrom2000through2023.Thelocalprojectionsestimatorindi-catesthatthesamemonetarypolicyshockincreasesARMpaymentsbyapproximately1.2percent,equivalentto$2billioninadditionalpaymentsperyear.

Theback-of-the-envelopecalculationoftheinframarginalborrowerchannelignoresthepossibilitythathouseholdsadjustonothermargins(i.e.,loansize).Totestthisimplicitassumption,Inextestimatethemagnitudesoftheintensivemarginchannel.Todoso,Irunthelocalprojectionsestimatoronaweeklyseriesoftheaverageloansizeofneworiginations.I?ndefectsthatareclosetozeroforthe?rstfewyearsaftertheshock.Theassumptionofnointensive-marginadjustmentusedintheback-of-the-envelopecalculationoftheinframarginalchannelappearsbenign.

Together,theestimatedmagnitudesoftheinframarginalborrower,marginalre?nancer,andARMchannelsadduptobeclosetotheestimatednetefectonaggregatepayments.Theestimatesofchannel-speci?cmagnitudesuseoutcomemeasureswholelyindependentof

6Forthisreason,monetarytighteningscanincreasepaymentsviathemarginalre?nancingchannel(rel-ativetothecounterfactualofnotightening)nearlysymmetricallytothemannerinwhichmonetaryeasingslowerpayments.Ifasmalltighteningshockdissaudesaborrowerfromre?nacing,theyforegoarelativelylargereductioninpayment.Forexample,supposeaborrowerwouldchoosetore?nanceonlyifratesfalltoonefullpercentagepointbelowthatborrower’sexistingrate.A5basispointshockthatmovedratesfrom0.95percentto1.00percentbelowtheirratewouldinducetheborrowertore?nanceanddecreasetheirexistingrateby1percentagepoint.Symmetrically,a5basispointtighteningshockfrom1.00to0.95belowtheirratewoulddissuadethatsameborrowerfromre?nancing,increasingtheirrateby1percentagepointrelativetothecounterfactualofnoshock.

7

thetime-seriesmeasureofpaymentsusedtoestimatetheoverallefect.Thefactthattheresultfromthe“top-down”approachtoestimatingthemagnitudeofthemortgagepaymentchannelmatchesthatfromthe“bottom-up”approachwasnotexternallyimposed,andcon?rmsthe?ndingisrobust.Interestingly,allthreeofthesechannelsnaturallyintroducealagbetweenthemonetarypolicydecisionandthemaximumefectonpayments.Fortheinframarginalborrowersandmarginalre?nancerchannels,thisoccursbecausethechangeinpaymentsaccumulatesslowlyaseachmonthanewsetofloansisoriginatedsubjectto,orcontingenton,thepolicy-afectedrate.FortheARMchannel,alagoccursbecauseARMratesonlyadjustatperiodicintervals(oftenoneyear),orevenlessfrequentlyiftheloanisstillinaninitial?xed-rateperiod.MiltonFriedman’s“l(fā)ongandvariable”lagshaveanaturalexplanation,atleastinrecentdecades,asaconsequenceofthefeaturesofthemortgagemarketonwhichmonetarypolicyoperates.

Finally,Iturntoaconsiderationofthestatedependenceofmonetarypolicy’sefec-tivenessthroughtheinframarginalborrowerchannel.ThetotalefectofarateshockonpaymentsvariesproportionatelywiththevolumeofmortgageborrowingAmericanswillengageinoverthesubsequentthreeyears.Thisvolumehasvariedconsiderablyoverthepastquartercentury,withthree-yearpeaksofover$13trillion(inreal2023dollars)reachedduringthere?nancingboomsof2003andtheCovidpandemic,whiletheaveragethree-yearvolumewascloserto$9trillion.Three-yearvolumesneverfellbelow$6trillion,however,asless-volatilehomepurchaseborrowingsupportsthe且ooroftotalborrowingevenwhenre?nancingentersafallowperiod.These且owsarealwaysquitelargerelativetothestockofhouseholds’outstandingmortgagedebt(approximately$13trillion7),suggestingthatthe?xed-ratenatureofmostmortgagedebtonlypartiallyimpedesthepass-throughofratechangestohouseholddisposableincome.InadditiontoARMSand?xed-ratemortgagesexhibitingsimilarlagsinthepass-throughofmonetarypolicyshockstopayments,theper-borrowerefectsizesarecomparable.Duetothelarge且owsofneworiginations,Iestimatethepaymentsoftheaveragemortgageborrowerareabout75percentasresponsivetoa

forwardguidanceshockasarethoseofanARMborrower.

7Source:FederalReserveBoard(2023)

8

Mypapermakesseveralcontributions.First,thisistheonlypaperIamawareoftodemonstrateusingquasi-experimentalmethodstheexistenceofastrongmortgagepaymentchannelofmonetarypolicyintheU.S.Itisalsothe?rstIamawareofinthisliteraturetoattempttoempiricallydistinguishandquantifythemagnitudesofthevarioussub-channelsbywhichmonetarypolicyispassedthroughtomortgagepayments.My?ndingthattheinframarginalchannelissubstantiallythelargestshouldshapehowwethinkaboutthetrans-missionofmonetarypolicydecisionstohouseholdspending.Furthermore,tomyknowledgethisisthe?rststudytodemonstratethattheefectofcentralbankforwardguidanceonmortgageratespersistsforseveralyears.Thispersistenceiskeytotheoverallstrengthofthemortgagepaymentchannel,astheefectsonpaymentsthroughallofthesub-channelsIidentifygrowcumulativelythelongerratesareheldaboveorbelowtheircounterfactuallevel.

1.1RelatedLiterature

Alargebodyofliteraturehasinvestigatedthemeansbywhichmonetarypolicyafectsthemortgagemarket,andhowthesechangesinturnafectconsumption.Studiesthatestimatehowre?nancingtoalowerrateincreaseshouseholdconsumptionincludeDiMaggioetal.(2017),DiMaggio,KermaniandPalmer(2020),AbelandFuster(2021),andAgarwaletal.(2023).Otherauthorshaveinvestigatedtheheterogeneousefectsofmonetarypolicybydiferentialhouseholdpropensitiestore?nance,includingBerajaetal.(2019),Wong(2019),andCummingandHubert(2021).

Amongthese,apaperthatisparticularlyrelatedtothepresentstudyisCloyne,FerreiraandSurico(2020).TheauthorsstudyhowmonetarypolicyshocksafecttheconsumptionofhouseholdsbytheirownershipandmortgagestatusintheU.S.andtheU.K.Theyalsoestimatehowshocksafectaveragemortgagepaymentsofborrowerswithoutstandingloans,asIdointhispaper.However,Cloyne,FerreiraandSurico(2020)?ndmuchsmallerefectsofmonetarypolicyshocksonpaymentsthanIdo.They?ndamonetarypolicyshockthatcutsthefederalfundsrateby25basispointslowerspaymentsbyonly$3.60permonthfor

9

theaverageborrower.8Incontrast,Iestimatethatamere7basispointmortgageratecutinducedbyforwardguidancewouldloweraveragepaymentsbyabout$18permonthafterthreeyears.9

AsIshowinAppendixSectionA.2,thisdiferenceinour?ndingsislikelyduetoCloyne,FerreiraandSurico(2020)’suseofsurveyresponsesintheConsumerExpenditureSurvey(CEX)toconstructtheirtimeseriesofmortgagepayments.IncomparisontothecreditbureaudataIemploy,theCEXhasamuchsmaller,frequentlyrotatingsampleandreliesonindividualborrowersratherthanloanservicersaccuratelyreportingtheir?nancialdata.IshowthattheCEX-basedpaymentsseriesismuchnoisierthantheoneIconstructfromtheCCP.WhenIrunmyestimatorontheCEXdata,Idonot?ndasigni?cantefectofmonetarypolicyshocksonreportedmortgagepayments.However,anullnetefectofratesonpaymentsishardtosquarewiththemagnitudesoftheindividualchannelsIestimateinSection5,andthis?ndingismostplausiblyattributedtosmall-samplelimitationsoftheCEX.

Furthermore,myresultshelprationalizeasurprisingtensioninCloyne,FerreiraandSurico(2020)’s?ndings:despitealimitedestimateoftheefectofmonetarypolicyonmort-gagepaymentsintheU.S.,they?ndthattheconsumptionresponseofmortgageborrowerstomonetaryshocksismuchlargerthanthatofrentersoroutrightowners.Thislatterresultismoreintuitiveinlightofthelargemortgagepaymentchannelmy?ndingsdemonstrate.

Afurtherstrandoftheliteraturehasstudiedhowtheembedded-optionnatureof?xed-ratemortgagesimpliesthere?nancingchannelofmonetarypolicyisheavilystatedependent.Becauseborrowersareonlylikelytore?nancewhentheirloanis“inthemoney”(i.e.,theircurrentrateiswellabovetheprevailingmarketrate),monetarypolicyislessabletoafectborrower’spaymentswhenrateshaverisenfromlong-termlows.Bergeretal.(2021)andEichenbaum,RebeloandWong(2022)modelhowthiscomplicationcanhampertheability

8Cloyne,FerreiraandSurico(2020)reportanefectsizeof$1.20in1984dollars,andIadjustforthe300percentin且ationbetween1984and2023.

9Thisestimateiscalculatedbyapplyingmyestimateofa0.89percentpaymentresponsetoaonestandarddeviationforwardguidanceshocktoanaverageborrower’smonthlypaymentof$2,000,calculatedasde-scribedinSection2.Efectsfromtheinframarginalchannelalonewouldincreasepaymentsbyapproximately$10permonthperborrower.

10

ofratecutstostimulateconsumption.My?ndingssupporttheirmainresults,althoughIpointoutthattheoverallinframarginalborrowingchannelisnotasstatedependentasthere?nancingchannelalone.Evenwhenveryfewborrowersareinthemoneytore?nance,totalborrowingvolumesarestillsupportedbyrelativelyrate-inelastichomebuyingdemandandmonetarypolicycontinuestohavenewloanstooperateon.

Marginalcash-outre?nancingisanotherrelatedchannelbywhichmonetarypolicycouldin且uencespending,asBhuttaandKeys(2016)?ndequityextractionisquiteresponsivetointerestrates.However,Anenberg,ScharlemannandvanStraelen(2023)showthat,atleastsincetheendofthehousingboomoftheearly2000s,muchorallofthemarginalequityextractedduetomortgagerateshocksisasubstituteforotherformsofconsumerborrowing,andhaslittlenetefectonconsumption.Inadditiontore?nancing,studiesthatestimatetheefectofinterestratesandmonetarypolicyonhomepurchaseborrowingincludeBhuttaandRingo(2021)andRingo(2024)ontheextensivemargin,andDeFuscoandPaciorek(2017)ontheintensivemargin.

Inaninternationalcontext,studiesofthedebt-servicechannelofmonetarypolicyincludeHofmannandPeersman(2017)andFlod/enetal.(2021).

2Data

Dataforthisprojectcomefromseveralsources.Toconstructatimeseriesofaveragemonthlypaymentsforhouseholdswithamortgage,IusetheFederalReserveBankofNewYork/EquifaxConsumerCreditPanel(CCP).TheCCPconsistsofaquarterlypanelofcreditbureaurecordsofarandomlyselected,anonymized5percentsampleofalladultsintheU.S.withacreditrecord.Amongotherinformation,loanservicersreporttheirborrowers’scheduleddebtservicepaymentsforeachtradelinetothecreditbureaus.Overthesampleperiod1999through2023,theCCPincludesdataon2.5millionto3.6millionprimarysampleindividualswithapositivemortgagepayment

Foreachquarterfrom1999Q1through2023Q4,Itaketheaverageoftotalreportedmonthlypaymentson?rstlienmortgagesperborrower.TheCCPsamplesindividuals,not

11

householdsorloans,soloanswithcoborrowerswouldbeoverrepresentedbytakingasimpleaverageacrossobservations.Tocorrectforthisissue,Iweighteachborrowerbythefractionoftheirpaymentsforwhichtheyarethesoleborrower,suchthatthepaymentsofaborrowerwhoonlyhassoloaccountsgetsweight1andaborrowerwhoonlyhasjointaccountsgets

forwhichtheyareacoborrower,theweighttheirpaymentsgetinthequarterlyaverageis1-0.5j.Thisschemeprovidesequalweighttoeachborrowingunit,generallyahousehold.Ialsodropobservationswithreportedmonthlymortgagepaymentsinexcessof$50,000aspotentiallyrepresentingdataerrors.Usingtheweightstocorrectforthedouble-countinginherenttojointaccounts,andadjustingfortheCCP’scoverageof5percentofallindividualswithasocialsecuritynumberandacreditrecord,thedatasuggestanaverageof42millionhouseholdswithamortgageintheU.S.

ThetimeseriesofaveragemonthlypaymentsforhouseholdswithamortgageisshowninFigure1,convertedintoreal2023doll