FinanceandEconomicsDiscussionSeries

FederalReserveBoard,Washington,D.C.

ISSN1936-2854(Print)ISSN2767-3898(Online)

TheDollarChannelofMonetaryPolicyTransmission

RalfR.Meisenzahl,FriederikeNiepmann,TimSchmidt-Eisenlohr

2025-046

Pleasecitethispaperas:

Meisenzahl,RalfR.,FriederikeNiepmann,andTimSchmidt-Eisenlohr(2025).“TheDollarChannelofMonetaryPolicyTransmission,”FinanceandEconomicsDiscussionSeries2025-046.Washington:BoardofGovernorsoftheFederalReserveSystem,

/10.17016/FEDS.2025.046

.

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

TheDollarChannelofMonetaryPolicyTransmission*

RalfR.MeisenzahlChicagoFED

FriederikeNiepmannFederalReserveBoard

TimSchmidt-EisenlohrFederalReserveBoard

June2025

Abstract

Thispaperdocumentsanewdollarchannelthattransmitsmonetarypolicyacrossborders.Exploitinguniquefeaturesofthesyndicatedloanmarketforidentification,weshowthatchangesintheeuro-dollarexchangeratearoundECBmonetarypolicyannouncementsthatareorthogonaltosimultaneouschangesineuro-areainterestratesandstockpricesaffectU.S.leveragedloanspreads.Specifically,inresponsetodollarappreciation,investorsrequirehighercompensationforrisk,andborrowingcostsforU.S.firmsincrease.ThesefindingsimplyacausallinkbetweentheU.S.dollarandinvestors’riskappetite.

Keywords:loanpricing,monetarypolicyspillovers,dollar,institutionalinvestors,risktaking

JELClassification:F15,G15,G21,G23

*Meisenzahl:FederalReserveBankofChicago.Address:230SouthLaSalleStreet,Chicago,IL60604.Email:ralf.meisenzahl@.Niepmann:BoardofGovernorsoftheFederalReserveSystem.Address:20andCStreets,Washington,DC20551.Email:friederike.niepmann@.Schmidt-Eisenlohr:BoardofGovernorsoftheFederalReserveSystem.Address:20andCStreets,Washington,DC

20551.Email:tim.schmidt-eisenlohr@.WethankMarekJarocinskiandNiklasKroner,aswellasseminarparticipantsattheBankofIreland,CEPR“WEARE”series,ChicagoFED,FederalReserveBoard,theNewYorkFed,theUniversidadtheLosAndes,theUniversityofCambridge,theCESIfoAreaConferenceonMoney,Macro,andInternationalFinance,andWhitmanSOM,forhelpfulcommentsandsuggestions.WealsothankWenxinDuandAndreasSchrimpfformakingavailablesomedatausedinthisresearch.AnnieMcCroneprovidedoutstandingresearchassistance.TheopinionsexpressedarethoseoftheauthorsanddonotnecessarilyreflecttheviewoftheBoardofGovernors,theFederalReserveBankofChicagoorthestaffoftheFederalReserveSystem.

1

1Introduction

Theeffectsofmonetarypolicyunfoldthroughanumberofchannels.Centralbankactionsnotonlychangethemonetarypolicystancebutalsorevealinformationabouttheecon-omy(

NakamuraandSteinsson

,

2018

;

Jaroci′nskiandKaradi

,

2020

).Moreover,recentwork(

BoehmandKroner

,

2024

;

Kroencke,Schmeling,andSchrimpf

,

2021

)arguesthatmonetarypolicyactionshavelargeeffectsthatarenotreflectedintheyieldcurve.Whiletheseissueshavebeenstudiedinthedomesticcontext,littleisknownaboutwhetherchannelsinde-pendentofyieldcurvechangesplayaroleforthetransmissionofmonetarypolicyacrossborders.

Inthispaper,wedocumentanewdollarchannelofcross-bordermonetarypolicytrans-missionbystudyinghowECBmonetarypolicyspillsovertotheUnitedStates.WeshowthatdollarmovementsaroundECBmonetarypolicyannouncementsthatareorthogonaltochangesintheeuro-areayieldcurveandstockmarketpredictU.S.riskyassetprices,specif-icallyleveragedloanspreads.Inresponsetodollarappreciation,borrowingcostsforU.S.firmsincreasebecauseinvestorsinleveragedloansrequireahighercompensationforriskresultinginhigherspreads,potentiallybecauseoftheirforeigncurrencyexposures(

Bruno

andShin

,

2015

).OurfindingsimplyacausalinterpretationforthepreviouslydocumentedlinkbetweentheU.S.dollarandinvestors’riskappetite(

Avdjievetal.

,

2019

).

Westudymonetaryspilloverstothe§1.4trillionU.S.leveragedloanmarket,whichrepresentstheriskysegmentofthesyndicatedloanmarket.Leveragedloanshavevariable

2

ratesandaresyndicatedoverroughlytwo-weekperiodsduringwhichthespreadtheborrowerneedstopayisadjustedbasedonthedemandfortheloanfrominvestors.Westudyhowtheseadjustments—theso-calledspreadflexes—respondtoECBmonetarypolicyannouncements.Becauseweonlylookatchangesinthespreadwithinaloanafterinitialloantermshavebeenset,changesinthespreaddonotreflectselectioneffectsacrossborrowersbutshiftingriskpreferencesofinvestors.

Todisentangledifferentwaysinwhichmonetarypolicytransmitsacrossborders,werelatespreadflexestohigh-frequencychangesineuro-areainterestrates,theeuro-areastockpriceindex,andtheeuro-dollarexchangeratearoundECBannouncementsfrom

Altavilla

etal.

(

2019

),so-calledECBsurprises.Wefindthatasurpriseappreciationoftheeuro-dollarexchangeratebyonepercentduringthesyndicationperiodincreasesthespreadpaidonaleveragedloanbyupto22basispoints.Thisdollareffectisidentifiedwithmovementsintheeuro-dollarexchangeratethatareorthogonaltosimultaneouseuro-areainterestrateandstockpriceschanges.Tothebestofourknowledge,wearethefirsttoshowthatforeignmonetarypolicytransmitstoU.S.borrowingcoststhroughadollarchannelthatisindependentofchangesininterestratesandstockprices.

Toprovideadditionalevidenceontheeffectofdollarmovementsonleveragedloansspreadsviainvestorriskappetite,weexploreheterogeneouseffectsacrossloansandbor-rowers.First,weshowthatadollarappreciationinresponsetoECBmonetarypolicyannouncementshasastrongereffectonloanswithahigherinitialinterestrate,aproxyfor

3

theex-anteriskinessofaloan.Theresultsshowthattheriskiertheloan,thehighertheincreaseinthespreadfromdollarappreciation,consistentwithshiftingriskpreferencesofinvestors.

Next,weaddresstheconcernthatthedollarmaydirectlyaffecttheprofitabilityoffirmsthatparticipateininternationaltrade,sothatspreadsmaychangenotbecauseriskpremiagoupbutbecausetheriskinessoffirmsincreases.Toexplorethispossibility,wesplitthesampleintoborrowersfromtradedandnon-tradedindustries.Wefindthat,evenforfirmsinnon-tradedindustries,forwhomtheexchangerateshouldnotdirectlyaffectincomeandrisk,thedollarchannelisactive.

Finally,wedistinguishbetweenunderwrittenloansandrefinancing.Underwrittenloans,whichtypicallyfinanceacquisitionsorleveragedbuyouts,haveacommitted,fixedamountthatthearrangerisobligatedtodisburse.Incontrast,refinancingloanscanbewithdrawnwhenfinancialconditionstighten,introducingapotentialdownwardbias.Indeed,wefindthatourresultsarestrongerfortheunderwrittenloanssubsample.

ResultsarerobustwhencontrollingforchangesinU.S.macroeconomicandfinancialmarketvariables,usingdifferentmeasuresofriskanduncertainty,andapplyingalternativeclusteringandsamplerestrictions.Additionally,resultsdonotchangewhenweincluderesponsesofU.S.interestrates,theS&P500index,andtheeuro-dollarexchangeratetoFederalReservemonetarypolicyannouncementsthatoccurduringthesyndicationperiod.Interestingly,anappreciationofthedollarinresponsetoU.S.monetarypolicyannounce-

4

mentshasquantitativelythesameeffectonU.S.leveragedloansspreadsasdollarmovementsfromECBmonetarypolicyannouncements.ThissuggeststhatthedollarchannelnotonlytransmitsmonetarypolicyacrossbordersbutalsoamplifiestheeffectsofU.S.monetarypolicyondomesticfinancialconditions.

Altogether,ourresultsshowthatmonetarypolicytransmitsacrossbordersthroughadollarchannelthatisdistinctfromtheclassicinternationaltradechannelsthroughwhichexchangeratemovements(includingthosecausedbyforeignmonetarypolicy)affectaneconomy.Dollarappreciationlowersinvestors’riskappetiteandtightensU.S.financialconditions.

LiteratureOurpaperaddstothemonetarytransmissionliteraturethatspeakstoeffectsthatgobeyondtheyieldcurve.AninfluentialliteraturehasshownthatFedpolicyrevealsinformationaboutthestateoftheeconomy(

RomerandRomer

(

2000

),

Campbelletal.

(

2012

),

NakamuraandSteinsson

(

2018

)).Morerecentworkhasshownthatmonetarypolicy(interestrate)surprisesaffectriskyassets(seee.g.

Bauer,Bernanke,andMilstein

(

2023

)),andthateffectsofmonetarypolicycangobeyondtheyieldcurveandmaterializeinriskappetite(

BoehmandKroner

(

2024

),

Kroencke,Schmeling,andSchrimpf

(

2021

)).

1

Inad-dition,

G¨urkaynaketal.

(

2021

)findthatmonetarypolicyaffectsexchangeratesevenaftertakinginterestrateandinformationeffectsintoaccount.Weshowthatthereareeffects

ofmonetarypolicybeyondtheyieldcurvethatmatterforinternationalmonetarypolicy

1Thishasbeentermedtherisk-takingchannelofmonetarypolicy(

BorioandZhu

(

2012

)).

5

spillovers.

Wealsocontributetotheliteratureontheinternationalspilloversofmonetarypolicy.

Seminalworkby

Rey

(

2015

)and

Miranda-AgrippinoandRey

(

2020

)revealedtheimpor-tanceofU.S.monetarypolicyfortheglobalfinancialcycle.Morerecently,

Georgiadisand

Jaroci′nski

(

2023

)exploretheinternationalspilloversofunconventionalU.S.monetarypolicy.

Somogyietal.

(

2024

)documentthatU.S.monetarypolicyinterestratesurprisestransmittoothercountriesdependingontheriskinessoftheircurrency.

Jarocinski

(

2020

)studiestransatlanticspilloversofECBmonetarypolicy,focusingonspilloversfrominterestratechangesconditionalontheresponseoftheEuropeanstockmarket(informationvs.mone-tarypolicyshocks).Incontrast,weshowthatthereareECBmonetarypolicyspilloverstoborrowingcostsintheUnitedStatesthroughadollarchannelseparatefrominterestratespillovers.

Finally,thepapercontributestotheliteratureontheroleofthedollar.

Avdjievetal.

(

2019

)havearguedthatthedollarisanindicatorofrisktaking,while

BrunoandShin

(

2015

)suggestthatchangesinthedollaraffecttherisk-takingcapacityofthefinancialsector.

Jiang,Krishnamurthy,andLustig

(

2018

)findthatthedollarappreciateswiththeglobaldemandforU.Ssafeassets.

NiepmannandSchmidt-Eisenlohr

(

2023

)showthatdollarappreciationreducesthepricesofsyndicatedloansonthesecondarymarketandlowersthecreditsupplyfromU.S.banksthatfollowanoriginate-to-distributemodel.

Lilleyetal.

(

2022

)findthatafterthe2007-08globalfinancialcrisis,thedollarco-moveswithglobalrisk

6

measures.Focusingontheglobalfinancialcrisis,

StavrakevaandTang

(

2024

)documentthatU.S.monetaryeasingledtoanappreciationofthedollarthen,whichisconsistentwithflight-to-safetybehaviorrelatedtoinformationeffectsfromforwardguidance.

Obstfeldand

Zhou

(

2023

)documentthatdollarappreciationpredictseconomicdownturnsinemergingmarketsanddevelopingeconomies.OurpapershowsandquantifieshowmovementsinthedollarthatareexogenoustotheU.S.economyaffectU.S.financialconditions.

Theremainderofthepaperisorganizedasfollows.Section2providesanintroductiontothesyndicatedloanmarketandexplainsthedollarchannelofcross-bordermonetarypolicytransmission.Section3developsthemainhypothesesthatwetest.Section4presentstheempiricalspecifications,andsection5discussesthedata.Section6presentsourempiricalresults,andsection7concludes,summarizingourmaintakeaways.

2Background

Inthissection,weexplainthesetupandinstitutionaldetailsoftheleveragedloanmarketthatweexploitinouranalysisanddescribeourapproachtostudyingthechannelsthroughwhichforeignmonetarypolicyspillsovertothepriceofU.S.riskyassets.

2.1Leveragedtermloanmarketandloanpricesetting

WestudyECBmonetaryspilloversinthecontextofthe§1.4trillionU.S.syndicatedlever-agedtermloanmarket.Leveragedloansareriskyvariable-rateloansthatchargeaspread

7

overanindexrate(e.g.LibororSOFR).Syndication,theprocessoffindinginvestorsforaloanandoriginatingit,typicallytakesabouttwoweeks,duringwhichtheloanspreadcanbeadjustedbasedonthedemandfortheloanfrominvestors.

Thesyndicationprocess,illustratedinFigure

1

,startswithborrowerssolicitingbidsincludingpricingandrisk-sharingprovisionsfrombanks.Aftertheborrowerchoosesaleadarrangerbankforthesyndicationfromthesolicitedbids,initialloantermsareagreeduponbytheborrowerandtheleadarranger.Theseinitialtermsincludetheloanspread,theoriginalissuediscount,covenants,andrepaymentoptions.Importantly,initialloantermsaresetbeforetheleadarrangerassemblesthesyndicateandloandemandrealizesandreflecttheleadarranger’sinformationaboutdemandfrominvestorsatthattime.

2

Oncetheinitialloantermsareagreedupon,theleadarrangerplacestheloanwithinvestorsusingbookbuildingthatdeterminesthefinalloanspread.Bookbuildingtakesatleastoneround.Ineachround,thearrangerproposesafacilityagreementincludingallloanterms,suchasthepricingtoinvestors.If,givenproposedloanterms,thereissufficientdemand,theloanisoriginatedatthoseterms.Ifthedemandfortheloanishigherorlower,thenthereisanotherround.Basedonthedemandthatrealizedwiththelastsetofloanterms,thearrangeradjuststhepricingtermsaccordingly.Theseadjustmentsarereferredtoas“flexes.”Forinstance,ifdemandwaslow,thenthearrangermayincrease 2Competitionforthesyndicationmandatelimitsbanks’incentivestoproposeinaccuratepricingprovi-sions.Specifically,bankshavenoincentivetoproposeaspreadthatisabovetheexpectedfinalspread.If

banksofferatalkspreadthatistoohigh,thebank’sproposalwouldlikelyberejectedbytheborrower.

8

(flexup)theeffectivespread,theall-inspreadaccountingforfeesanddiscounts,inthe

nextround.

3

Theinitialloanarrangingagreementisdesignedsuchthatthefeesthebankearnsfromarrangingtheloandecreasewiththefinalloanspreadtheborrowerpays.Thiscompensationstructuregivestheleadarrangerstrongincentivestoobtainthebestpossibleloantermsfortheborrowergiventherealizeddemandfrominvestors(

Bruche,Malherbe,

andMeisenzahl

,

2020

).Thebookbuildingprocesscontinuesuntiltheofferedloantermsmeetthedemandfrominvestorsandtheloanisoriginated.Aftertheborrowerreceivesthefunds,theloanstartstradinginthesecondaryloanmarket.

Toidentifymonetarypolicyspillovers,werelatespreadflexestoECBmonetarypol-icyannouncementsthathappenduringthedescribedbookbuildingprocess.Thereby,weidentifytheeffectsofeuro-areamonetarypolicyonU.S.loanpriceswithinaloan,thatis,withoutconfoundingeffectsfromborrowerselection.Additionally,giventhattheini-tialloantermsarepreset,andtheleadarrangerhastheincentivetoobtainthebestloantermsfortheborrower,spreadflexesshouldonlybedrivenbychangesinthedemandfromoutsideinvestorsforaspecificloan.Also,becauseleveragedloansarevariablerateloans,suchthatinvestorsareautomaticallycompensatedforinterestratechangeswithoutchangestothespread,ourexerciserevealshowECBmonetarypolicytransmitstoU.S.corporate

3Weincludeadjustmentstofees,suchasdiscounts,inourcalculationoftheflexofthespreadandhencerefertothetotaladjustmentsasflexesintheeffectivespread.

Berg,Saunders,andSteffen

(

2016

)documentthatdiscountsplayasignificantroleinloanpricing.Whileloanamountscanbeflexedinsomecases,loanstakenoutforLBOpurposesgenerallydonotexhibitamountadjustments.Theabilitytoflex,therangeofflexes,andtheconsequencesofflexesforthearranger’sfeearepartoftherisk-sharinginthecontractbetweenborrowerandarranger(

Bruche,Malherbe,andMeisenzahl

,

2020

).

9

borrowingcoststhroughchangesintheriskappetiteofinvestors.

2.2Disentanglinginternationalmonetarypolicytransmissionchan-nels

Traditionally,monetarypolicyisthoughtofasoperatingthroughchangesininterestrates.Changesinpolicyratesdirectlymoveshort-terminterestrates,whilequantitativeeasingandtighteningandforwardguidancetargetthelongendoftheyieldcurve.However,recentresearchsuggeststhatmonetarypolicyaffectstheeconomythroughfactorsthatarenotreflectedintheyieldcurve(

BoehmandKroner

,

2024

;

Bauer,Bernanke,andMilstein

,

2023

).Inparticular,thesepapersarguethatthereisalotofmovementinriskyassetpricesaroundmonetarypolicyannouncementsthatcannotbeexplainedbychangesininterestrates.

Inouranalysis,weassesswhethereffectsbeyondtheyieldcurvealsoplayaroleincross-bordermonetarypolicytransmission.Wefocusonmovementsintheeuro-dollarexchangeratearoundECBmonetarypolicyannouncementsandanalyzewhethertheyaffectU.S.riskyassetprices.Toidentifythetransmissionofmonetarypolicyacrossbordersthroughthisdollarchannel,weexploitmovementsintheeuro-dollarexchangeratethatareorthogonaltoeuro-areainterestrateandstockpriceschanges.Tothatend,weobtainchangesintheyieldcurve,thestockmarket,andtheeuro-dollarexchangeratearoundECBmonetarypolicyannouncementsthathappenduringthebookbuildingprocess.Inlinewith

Altavilla

etal.

(

2019

),wecallthesehigh-frequencymovementsaroundannouncements“surprises”.

10

Controllingforinterestrateandstockmarketsurprisesiskey.First,interestratechangesdirectlyaffecttheeuro-dollarexchangerate.WhentheECBraisesinterestrates,thedollardepreciatesasinvestorsarbitrageinterestratedifferentialsbetweenEuropeandtheUnited

States.Throughthedollarchannel,dollardepreciationeasesU.S.financialconditions.At

thesametime,however,higherinterestratesintheeuroarearaiseU.S.loanspreadsasinvestorsdemandagreatercompensationforholdingU.S.assets(interestratechannel).Inotherwords,anECBinterestratehikelowersloanspreadsthroughthedollarchannelbutraisesthemthroughtheinterestratechannel.Asaresult,wecanonlyidentifytheeffectsofchangesinthedollaronloanspreadswhencontrollingforchangesininterestrates.

Second,weneedtocontrolforchangesinstockpricestoaccountforinformationeffects.

Changesinstockpricesaroundmonetarypolicyannouncements,inadditiontocapturingtheeffectsofinterestratesonfirmprofitability,reflectinformationaboutthestateoftheeconomyrevealedbythecentralbank.Informationeffectscanbesostrongthattheyoverturntheinterestrateeffect,sothathigherinterestratescanbeaccompaniedbyincreasingstockprices(

Jaroci′nskiandKaradi

,

2020

).Controllingforeuro-areastockpricesensuresthatthedollarmovementsweusetoidentifytheeffectsonleveragedloanspreadsdonotsimplycapturenewinformationabouttheeuro-areaeconomyasreflectedineuro-areastockprices.

11

2.3Distinguishingthedollarchannelfromtraditionalexchange

ratechannels

Itiswellknownthatmonetarypolicyaffectstheexchangerateandtherebygeneratesspillovereffectstoothercountriesthroughthetraditionalexpenditure-switchingchannel(

Mundell

,

1968

).Theexpenditure-switchingchannelimpliesthatanECBratehikethatde-preciatesthedollarmakesU.S.goodsrelativelycheaperandtherebyincreasesU.S.exportsanddecreasesU.S.imports.MorerecentworkhasdocumentedthatbasicallyallU.S.exportsandimportsaredenominatedinU.S.dollars(

Gopinathetal.

,

2020

),whichcontrastswiththeassumptionin

Mundell

(

1968

)ofproducer-currencypricing.

4

WhenalltradeisinvoicedinU.S.dollars,adollardepreciationshouldincreaseexportquantitiesbutshouldleaveim-portquantitiesunaffected.Allelseequal,onewouldexpectthepositiveexportseffectsofadollardepreciationtoimprovethefinancialpositionofU.S.exporters,potentiallyreducingtheirloanspreads.

Ourdollarchanneldiffersfromthisclassicchannelbecause,aswewillshow,itoperatesforbothfirmsintradedandnon-tradedsectors,whichisinconsistentwiththeviewthattheexchangerateonlyaffectstheeconomythroughinternationaltrade.Additionally,totheextentthatstandardinternationalmacromodelsdirectlylinkthedollartointerestrates,thereshouldnotbeaneffectoftheseclassictradechannelsonceinterestratesarecontrolledfor.

4Ontherelatedearlierliteratureonthedollarasavehiclecurrency,seeamongothers

Corsettietal.

(

2007

),

CookandDevereux

(

2006

),

GoldbergandTille

(

2008

)and

GoldbergandTille

(

2009

).

12

Thedollarchannelweuncoverismorecloselyrelatedtorecentworkonthebalance-sheetchannelthathasarguedthatdevaluationscanbecontractionaryinthecontextofemergingeconomiesthathavedollarliabilities(see,e.g.,

Aguiar

(

2005

)).Thisliteraturehasshownthat,totheextentthatacountryisindebtedindollars,dollarappreciationcanweighonitseconomybecauseitincreasestheburdenfromdollardebt.Thiseffectcanbestrongenoughtooverturntheexpansionaryeffectsofadevaluationarisingfromtheclassicexpenditure-switchingchannel.Ofcourse,thisforeign-currencydebtchannelcannotworkfortheUnitedStatesinexactlythesamewaybecausebothfirmsandbanksintheUnitedStatesoperateandarefundedindollars.Instead,ourdollarchannelreflectsglobalforcesthatcreatealinkbetweendollarappreciationandthepriceofriskdemandedbyinvestors.Onepaperthatmodelssucharelationshipbetweendollarappreciationandrisk-takingis

BrunoandShin

(

2015

).

3TestablePredictions

AnECBannouncementthatleadstoanappreciationofthedollaragainsttheeuroshoulddecreaserisk-takingbyinvestorsandtherebytightenU.S.creditconditions.Forexample,dollarappreciationcouldincreasetheriskininvestors’portfoliosbecauseofinvestors’cur-rencyexposures(

BrunoandShin

,

2015

).Whenthedollarappreciates,exposuresinforeigncurrencybecomeriskier,makinginvestorswithFXexposurelesswillingtotakeonaddi-tionalrisk.

13

Prediction1.DollarappreciationinresponsetoanECBmonetarypolicyannouncement

causesloanspreadsforriskyU.S.borrowerstoincrease.

Asdiscussed,ECBmonetarypolicyannouncementscanalsoaffectU.S.corporatebor-rowingspreadsthroughchangesininterestrates.HigherinterestratesintheeuroarearelativetotheUnitedStatesshould,allelseequal,raiseU.S.spreadsbecauseinvestorwillrequirehighercompensationforholdingU.S.assets.

Moreover,asexplainedinsection

2

,thereisakeyinteractionbetweenthedollarchan-nelandtheinterestratechannelbecausechangestoECBinterestratesdirectlyaffecttheeuro-dollarexchangerate.Specifically,anECBratehikedepreciatesthedollaragainsttheeuroviatheinterestrateparitycondition.Whileanincreaseineuro-areainterestratesraisesU.S.borrowingcosts,theaccompanyingdollardepreciationreducesthem.Therefore,toestimatethedollarchannel,oneneedstodirectlycontrolfortheinterestratechannel.Notdoingsogeneratesadownwardbiasinthedollarsurprisecoefficientestimate.

Prediction2.(i)Anincreaseineuro-areainterestratesinresponsetoanECBmonetarypolicyannouncementincreasesloanspreadsforriskyU.S.borrowers.

(ii)Whenregressingspreadflexesonsurprisesintheeuro-dollarexchangerate,thecoefficientestimateontheeuro-dollarexchangerateincreaseswhentheregressionalsocontrolsforsurprisesininterestrates.

Whileleveragedloansaregenerallyrisky,theirratingsrangefromBBBtoCCC+and

14

belowasFigure

7

illustrates.Becausethedollaraffectsinvestors’riskappetite,theeffectofthedollaronleveragedloanspreadsshouldbelargerforriskierloans.

Prediction3.TheeffectofdollarappreciationinresponsetoanECBmonetarypolicyannouncementonloanspreadsislargerforriskierU.S.borrowers.

Inourempiricalsection,wetestPredictions1through3.

4Specifications

Thissectionpresentsthemainspecificationsofourempiricalanalysis.

BaselineSpecificationThebaselinespecificationrelatestheeffectivespreadflexofaleveragedloantothemovementintheeuro-dollarexchangeratearoundECBmonetarypolicyannouncementsduringtheloan’ssyndicationperiod:

?EffectiveSpreadi,?t=β1DollarB+γXi+φ?Z?t+δind+δp+δl+?i,?t,(1)

where?EffectiveSpreadi,?tisthechangeintheeffectivespreadoftheloanduetoaflex

orzeroifthereisnoflex.Theindependentvariableofinterest,DollarB,isthechange

intheeuro-dollarexchangeratearoundtheECBannouncementwindowthatfallsintothesyndicationperiod.ThebaselinespecificationincludesloancontrolsXi,aswellasindustry(δind),product(δp),andleadarranger(δl)fixedeffects.Loancontrolsaretalk(initial)

15

spread,maturity,anddummyvariablesforwhetherornottheloanissponsored,rated,acov-lite