IMFWorkingPapersdescriberesearchinprogressbytheauthor(s)andarepublishedtoelicitcommentsandtoencouragedebate.TheviewsexpressedinIMFWorkingPapersarethoseoftheauthor(s)anddonotnecessarilyrepresenttheviewsoftheIMF,itsExecutiveBoard,orIMFmanagement.NATn?2024InternationalMonetaryFundWP/24/242IMFWorkingPaperResearchDepartmentBilateralTradeinServicesandExchangeRates:EvidenceofDominantCurrencyPricingPreparedbyNanLiandSergiiMeleshchukAuthorizedfordistributionbyPetiaTopalovaNovember2024IMFWorkingPapersdescriberesearchinprogressbytheauthor(s)andarepublishedtoelicitcommentsandtoencouragedebate.TheviewsexpressedinIMFWorkingPapersarethoseoftheauthor(s)anddonotnecessarilyrepresenttheviewsoftheIMF,itsExecutiveBoard,orIMFmanagement.ABSTRACT:Thispaperestimates,forthefirsttime,theexchangerateelasticityofbilateraltradeinservices,providingindirectevidenceofbothproducercurrencypricinganddominantcurrencypricinginservicestrade.Wedevelopedanoveldatasetofbilateraltradeflowsinservices,coveringtwelvebroadservicesectorsacross245countriesfrom1985to2022.Wefindthat,similartomanufacturingtrade,thevalueofservicestradeismorecloselyassociatedwithUSdollarexchangeratesthanwithbilateralexchangerates,althoughthisrelationshipvariesbyservicecategory.Zeroinginontourism,whereproxiesfortradevolume(suchastouristarrivalsandhotelstays)areavailable,wefindthatbilateralexchangeratesplayalargerroleontourismvolumecomparedtothedollarexchangerates.Inaddition,inthecontextofglobalsupplychain,wefindthatdownstreamdollarexchangeratemovements,ratherthandownstreambilateralexchangerates,affectthedemandforserviceimportsviaforwardlinkages.Keywords:tradeinservices;exchangerates;dominantcurrencypricing;producercurrencypricing;globalvaluechainAuthor’sE-MailAddress:nli@;smeleshchuk@INTERNATIONALMONETARYFUNDBilateralTradeinServicesandExchangeRates:EvidenceofDominantCurrencyPricingPapageorgiou,AndresRodriguez-Clare,PetiaTopaQiuyanYin.WealsothankInexperienceswithcompilIncomeCountriesprogramoftheUKsFo11IntroductionInrecentyears,servicesingforcebehindthenextwaveofglobalization(WTO,2019;Baldwin,2022a,b).Accountingforincreasinglycrucialrolesinglobalvaluechains(GVCs).Yet,despitetheburgeoningimportancetionsafectservicestrade.inantcurrencypricing”(DCP),oferingnewinsightsintotheroleofexchangeratesinglobaltradedynamics.Researchshowsthatthedominantcurrency(e.g.,theU.S.dollar)exchangeratedrivesglobalgoodstradeintheshortrun(Gopinath,2015;Gopinathetal.,202pricesarestickyinthedominantc(GopinathandRigobon,2008;GoldbergandTille,2008,2016;Amitietal.,2022;Bozetal.,2022).nershaveonlyalimitedefectontradevolumesandprices(Bozetal.,2019).centration,andexhibitgreaterpricestickiness(FrancoisandHoekman,2010;KlenowandMalin,2010).The“proximityburden”facedbycertainservices(e.g.tourism),wheredirectinterac-exchangeratesandtrade(Hill,1977).Thesediferencesmayafectthesensitivityofexporters’tingacrosscompetitors—factorstoexchangerate丑uctuations(Gopinathetal.,2010;Amitietal.,2022;Mukhin,2022).2Figure1:GlobalTradeinGoodsandSerNotes:This?gureshowsthe(normalized)timeseriesofglobalexportsinservicesandingoods.Thenominal1985valuesofservicesexportsandofgoodsexportsareeachnormalizedto1.Source:TheWorldBank’sWorldDevelopmentIndicatorDatabase.trade.includingWTO-UNCTAD-ITCTradeinCommercialServicesDatabase,UNComtradeandEuro-weightedefectiveexchangerates(SpilimbergoandVamvakidis,2003;MayerandSteingress,2bilateralanddollarexchangerates.FollowingtheempiricalstrategyofGopinathetal.(2020),wepricestickinessinservices(BilsandKlenow,2004;NakamuraandSteinsson,2008;Klenowandincludesvariouslaggedvariablesalongwithtimeanddirectedcountry-pa1Forexample,KlenowandMalin(2010)?ndsthatservicesproducerstaketwiceaslongtochangepricescomparedtonondurablegoodsproducersandthreetimesaslongcomparedtodurablesgoodsproducers.BilsandKlenow(2004)attributethislowerfrequencyofpricechangesinservicesre丑ectstothelowervolatilityofconsumerdemand.Inthecaseofgoodstrade,GoldbergandTille(2008)andGopinathandRigobon(2008)presentmicro-levelevidencethatpricesoftenremainstickyinUSDforuptotwoyears.2Identifyingexogenousexchangerateshocks,aswellasexogenousdemandshocksandrelevantrelativeprices,isalmostimpossible.Therefore,ourresultsshouldnotbeinterpretedasstructuralestimatesbutasthe?rstanalysis3Ifpricesarestickyintheproducer’scurrency(PCP),importpricesinthedestinationcurrencyofimportquantitieswithrespecttoimportpricesdiferssigni?cantlyfrom1,importvaluesintheconstant.Incontrast,underlocalcurrencypricing(LCP),importpricesareunafectedbybilateralexchangerates(zeropass-through),resultinginnochanexporter’scurrencyorUSdollars.Speci?cally,ifcountryj’scurrencydepreciatesby10%againstcountryi’scurrency,thevalueofservicesimportedfromibyj(measuredinj’scurrency)byabout2%intheshortrun.3TheefectoftheUSDexchangerateistwicethatofthebilateralstatisticallyinsigni?cant,whilethecumulativeelasticityoftheUSDexchangerateincreaseThis?ndingalignswithandgeneralizestherecentsingle-countrystudybyAmadoretal.(2024),oftheUSDollarasavehiIfthesensitivityofservicesdemandtopricesisthesameirrerency,therelativemagnitudesofbilateralandUSdollarexchangerateelasrelativeextenttowhichPCPandDCPareadopted.Comparingtheseresultswiththosefrommanufacturingtrade,we?ndthatthedominanceofDCPrelativetoPCPislesspronouncedinservicestradethaninmanufacturingtrade.Whiletheratioofdollarexchangerateelasticitytobilateralelasticityisabout?vebasedonmanufacturingtrade,itisonlytwoforservicestrade.rates.Intheshortterm,theUSDexchangerateseemstobemoreimportantthanthebilat exchangerateforfoursectors:Transport,Financial,Telecommunications,andIntellectualProperty followingtheprevailingempiricalstrategyintheliteraturetoassesshowservicestradeisrelatedtoexchangeratemovements.Nevertheless,werefertothereduced-formestimatesofexchangeratecoe伍cientsasexchangerateelasticitiesofservicestrade.3Thissuggeststhatpriceincreasesoutweighreductionsintradedquantities,similartothe?ndingsofGopinathetal.(2020),whichestimatethatpricepass-throughhasalargerefectthanquantityelasticityinmanufacturingtrade.4Services.Incontrast,OtherbusinessservicesaremuchafectedbytheUSDexchangerate.ThereisnoevidenceofeitherPCPorDCPforConstruction,whichsuggeststhatLCPdominatesinthismarkupelasticitiestoshocks.?ndevidenceofbothPCPandDCP—althoughtourismrespondsmoretomovementsirate,whiletheefectofbilateralexchangeratesbecomesstronger.Thisimpliesthatforcountriesspecializingintourism,thetraditionalMundell-Flemingmechanismofboostingexportsthrocurrencydepreciationisstillefective.restoftheworld.UnderDCP,valueofimportstotheUnitedStatesdenominatedinUSdollarsdonotreacttobilateralexchangerateswhereasUSexportsdosigni?cant)responsesofoverallservicesexportsfromthU.S.,consistentwithDCP.Acrosssectors,howeverimportsofFinancial,ITandOtherbusinessimportanceofproducercurrencyprparticularlywithinglobalvaluechains(GVCs)(seeWB,2020).Inthesecases,exchangeratebetweentheinitialimporter5and(ii)theUSDexchangerateofdownstrre-exportedbyimporterj.dollar,thevalueofservicesimportedbyjfromiincreases,particularlywhenthoseservicesarethevalueforimportedservices.ThisanalysiscontributestotheexistingliteratureonhowGVCparticipationafectstheworkingsofexchangerates,whichhinmanufacturinggoods(e.g.Amitietal.,2014;DeSoyresetal.,2021;Adleretal.,2023).InadditiontocontributingtothegrowingliteratureonDCP(GoldbergandTille,2008,2016;Chung,2016;Lyonnetetal.,2022;Chenetal.,2022),ourworkwhichuntilnowhasprimarilyfocusedonmercofexchangerateshockstopricesandquantities(e.g.,AtkesonandBurstein,2008;Bermanetal.,2012;Amitietal.,2014;Devereuxetal.,2015;AuerandSchoenle,2016;Amitietal.,2022).Thispaperisthe?rsttodevelopandstudyagloballyrepresentativedatspecializationbycountries,thisstudybearssigni?cantimplicationsfomonetarypolicy,optimalexchangeratepolicies,andeforttocontainlargeexternalimbalance6basicpatternsofglobalseSection5explorestheefectsofforwardGVCparticipation.WeconcludeinSection6withabrief2HowDoesTradeinServicesDiferfromTradeinGoods?BeforeproceedingtoGopinathetal.,2010;Amitietal.,2022;Mukhin,2022)hasestablishedthata?rm’scurrencyThedesiredpass-throughdepends,inturn,theintensityofimportedintermediateinputs.Exporterstendtoinvoicetheirsalesinacurrencytheelasticityofdemandinthatmarketvarieswiththeirmarketshare.Therefore,thesensitivitytheshock.Loweringthedestinationcurrencypriceincreasesthecostoflowermarkups.Ifstrategiccomplementaritiesinpandgeneralcharacteristics.Inwhatfoll7andtradeinmanufacturinggoods,diferencesthatcouldresultinavarietyofinvoicingcurrencyFirst,incontrasttomanufacturingtrade—whereexportersareoftenalsoimporters(Bernardetal.,2009;KuglerandVerhoogen,2009;ManovaandZhang,2009)—servicesgenerallyemployaoutputformanufacturingproductionis2ThecontrastinthesharesectortranslatesintolowersensitivityofmFigure2:DistributionofInputSharesintheProductionofServicesandManufacturedGoodsKerneldensityestimatesImportedKerneldensityestimatesKerneldensityestimates0246LaborShareKerneldensityestimates0246 GoodsServicesNotes:This?guregraphs,foryear2014andacross56countries,kerneldensityestimatesoftheshareofimportedinputs(leftpanel)andtheshareoflaborinputs(rightpanel)inservicesindustries(redplots)andinmanufacturingindustries(blueplots).Source:WorldInput-OutputDatabase(2016).“proximityburden”(Hill,1977).A?rm’schoiceofcurrency,then,dependsonitsrivals’pricingWhenexporterscompetemainlywithotherlocalprovidersinthecustomer’slocatitopriceinlocalcleadstolocalcurrencypricing.Yetiftradeoccursattheexporter’slocationandiftheexporter8servesalsothelocalconsumers(aswithTourism),thenthman,2010;Miroudotetal.,2013;HoekmanandShepherd,2021).Allthesefactorsgiverise,insomeimpliesthatsomeserviceexportersmayactivelyadjusttheirmarkupsbychoosingLCPorDCP(ifothercompetitorsadoptDCP)soastoensurethestabilityoftheirmarketshares(Amitietal.,2014,2022).Finally,someservicestraderelieslessthanothersonexternal?nance.Forexample,Borchertsector.Duygan-Bumpetal.(2015)establishthatfewservicesectorsdethattheycanhardlyserveascollateral.Theservicestrade’sindependencefromextandStein,2021).3DataDescriptionidenti?edintheBalanceofPayment(BoP)Statisticsincludesonlyservicesprovidedbymodes1,4Thusmode1serviceisprovidedbyresourceslocatedintheexportingcountryanddeliveredtotheimportingcountry;oneexampleistechnicalhelpprovidedbyacustomerservicerepresentativeinIndiaoverphonetoaUSconsumerwhosecomputerhasbeeninfectedbyavirus.Mode2,consumptionabroad,applieswhenapersonfromtheimportingcountrytravelstotheexportingcountrytoconsumetheserviceaswhenaclientfromSingaporecomestohisUSlawyer’so伍ce.Undermode3,theexporterhasacommercialpresenceintheimportingcountry;bythisde?nition,a?rm’sforeigna伍liatesalestohost-countryconsumersarecountedastheexportofservices.Mode4,whichinvolvesthepresenceofnaturalpersons,isinthatsensetheoppositeofmode2.Soinmode4,anaturalpersonfromtheexportingcountrytravelstotheimportingcountrytoprovideaservice;examplesincludea?nancialconsultantvisitingherclientinaforeigncountryand(say)BillyJoelgivingaconcertintheformerSovietUnion.9thea伍liatesofamultinationalenterpris3.1DataConstructionandSourcesBilateralServicesTradeDatabase.WedevelopedaBilateralServicesTradeDatabase(BiTS),consolidates,imputesandreconcilesthelatesto伍ciaUNCTAD-ITCAnnualTradeinCommercialServicesDatabaseservingastheprimarysource,supplementedbydatafromUNCOMTRADEandhistoricalrecordsfromEurostat.5DescriptionsofthedataconstructionareavailableintheAppendix(formoredetailsseeLietal.,forthcoming).isbasedontheExtendedBalanceofPaymentsServicesClassi?cation(EBOPS2010),whichalignswiththe6theditionoftheIMFBalanceofPaymentsandInternationalInvestmearliestyearsarelimited.lack?nancialincentivestocloselymonitorservicesimports.Consequently,servicesbanks,tendstobemorereliablethanimportdata.Forthisreason,theBiTSprimarilyuses?llinmissingexpothehierarchyofthevariouscommercialsBPM6,CommercialservicescompriseGoods-relatedservices,Transport,Travel,andOthercom- 5Itiscriticalthatouranalysisreliesono伍cialdataratherthanestimateddata.Otherdataonbilateralservices丑ows,suchasthoseprovidedbyWIODandWTO-OECDBaTIS,includevaluesestimatedusinggravity-basedmodels;theyareunsuitableforourpurposesbecausetheywouldarti?ciallyimplytheabsenceofanyrelationshipbetweenhealth,education,orotherpersonalreasonstotravel).Theinto:Construction;Insuranceandpensionservices;Financialservices;Chargesforuseofintellec-tualproperty(e.g.,Royaltiesandlservices);Otherbusinessservices;andPersonal,cultural,andrecreationalservices.Ouranalysishamperviablestatisticalanalysis.Table1:CommercialServicesClassi?cations,BPM61-digit2-digit(description)Goods-relatedservicesSA:ManufacturingservicesonphysicalinputSAy:GoodsforprocessingabroadSAz:GoodsforprocessinginthereportingeconomySB:Maintenanceandrepairservicesn.i.e. SC:TransportSC1:SeatransportSC2:AirtransportSC3:OthermodesoftransportSC4:PostalandcourierservicesSD:TravelSDA:BusinessSDB:PersonalOthercommercialservices(OCS)SE:ConstructionSE1:ConstructionabroadSE2:ConstructioninthereportingeconomySF:InsuranceandpensionservicesSF1:DirectinsuranceSF2:Re-insuranceSF3:AuxiliaryinsuranceservicesSF4:PensionandstandardizedguaranteedservicesSG:FinancialservicesSG1:Explicitlychargedandother?nancialservicesSG2:Financialintermediationservicesindirectlymeasured(FISIM)SH:Chargesforuseofintellectualpropertyn.i.e.SH1:FranchisefeesandtrademarklicensingfeesSH2:LicensesfortheuseofoutcomesofresearchanddevelopmentSH3:Licensestoreproduceand/ordistributecomputersoftwareSH4:Licensestoreproduceand/ordistributeaudio-visualmaterialsSI:Telecom,computer,andinformationservicesSJ:OtherbusinessservicesandrelatedproductsSI1:TelecommunicationsservicesSI2:ComputerservicesSI3:InformationservicesSJ1:ResearchanddevelopmentservicesSJ2:ProfessionalandmanagementconsultingservicesSJ3:Technical,trade-related,andotherbusinessservicesSK:Personal,cultural,andrecreationalservicesSK1:Audio-visualandrelatedservicesSK2:Otherpersonal,cultural,andrecreationalservicesSL:Governmentgoodsandservicesn.i.eSource:TheInternationalMonetaryFund’sBPM6.USTradeinServices.Tocomplementourcross-counUSBureauofEconomicAnalysis(BEA).TherearetworeasonsforexaminingUSevidenceinparticular.First,thedatacouldbeofhigherqualitybecausetheBEAhasmadetremendousUnitedStatesprovideanMultilateralServicesTrade.WeusedataonmultilateralservicestradeforsummarizingglobalDevelopmentIndicators(WDI)andtheIMF’sTradeinServicesDatabase.AlthoughWDIencom-thedisadvantageofWDIdataisthattheyarcialservices,Travel,andTransport.Thatis,theyincludenobreakdownofOthercommercialservices—incontrasttoIMFdata,Tourism.Separatedataonpricesandquantitiesarenotavailableforbilateralservicestrades.establishments—bytourists’respectivecountriesoforigin—fromEurostaEurostatdata,whichcoverstheperiod1991–2018,relyonreportsfrom:(a)the28EUcountriesplusIceland,Ireland,Norway,Turkey,andSwitzerlandAustralia,Brazil,Canada,China,Japan,Korea,Russia,SouthAfrica,Ukraine,andtheUnitedStates.3.2BasicPatternsofInternationalTradeinServicesBeforedescribingandanalyzingthedata,weexaminethebreadthoftheBiTS.Table2givestheshareofworldexports(asrecordedinWDI)thatiscoveredbytheBiTS’sthreemaincategories.sector.Thetable’slastcolumnshowstheshareofworldservicesexportsthatiscoveredbyBiTSinanyspeci?ccategory,averagedacross2005-2097%ofworldtransport,and97%oftravelaswellasnearlyalltotalservicestrade.Table2:WorldwideServiceExportsGlobalexportsGrowthBiTScoverageTotalservicesTransportTravel6.0%Othercommercialservic4,554.49.0%Notes:Column[1],globalserviceexportvalueinbillionsofUSdollars(asrecordedintheWDIfor2022);column[2],averageannualgrowthrateofexportvalueovertheperiod1995–2022;column[3],averageshareofworldtradecoveredbytheBiTSduring2005-2022.Kingdom,andtheUnitedStates.EveninChina,whichistraditionallyviewedasamanufacturingexporter,morethanathiditionallyassociatedwithservicestradeincreasinglyascomponentswithinGVCs(HeuserandMattoo,2017).Asaresult,manyservicestradabilityofservicesiswiIntermsofindustries,theincreaseinservicesexportshservicesthatcanbedeliveredatadistance(e.g.,ITservices,?nancialintermediation,R&Dandprofessionalservices,entertainmentservices).Inthemid-1980s,thesectorsofTransport,Travel,andOCSeachaccountedforaboutathirdoOCSincreasedtoalmost70%in2019,whilethesharesofsuandTraveldeclinedovertime.ThFinancialservices,ITservicesandChargesforuseofintellectualproperty.Overall,thetop5servicesindustries(Transport,Travel,Finance,IT,andOtherbusinessservices)nowaccountforFigure3:CompositionChangesofServicesTradebyIndustryNotes:DatasourcesaretheWorldBank’sWDIdatabaseforthechartonthelefthandside,andIMFInternationalTradeinServicesdatabasefortheright.ItturnsoutthatthevalueoftradeishighlyconcentratedinarelativelysmallshareoftradingFigure4:ServicesExportingbyIncomeGroupNote:This?gureplotsthevalue,intrillionsofUSD,ofservicesexportsbyhigh-income(gray),middle-income(blue),andlow-income(orange)countrygroupsfrom1985to2022.Source:TheWorldBank’sWorldDevelopmentIndicatorsdatabase.Figure5:RatioofManufacturingExportstoServiceExports,AEsvs.EMDEsNote:This?gureplotsthetimeseriesoftheGDP-weightedaverageofthemanufacturing/servicesexportsratioforadvancedeconomies(AEs,dashedblueline)andforemergingmarketanddevelopingeconomies(EMDEs,solidorangeline).Source:TheWorldBank’sWorldDevelopmentIndicatorsdatabase.(seeforexample,HavemanandHummels,2004;Helpmanetal.,2008).4ServicesTradeandExchangeRates:EmpiricalEvidence4.1ConceptualFrameworkwork.TomotivatetheempiricalLetqijdenotethe(log)quantityofimportsbycountryjfromcountryi,letpij-pjbethe(log)(log)realexpendituresintheimportingcounexpenditures)inthedestinationcountry:△qij;t=-σij(△pij;t-△pj;t)+△yjt;(1)w.r.t.importprices,althoughitismorecommonlygivenastructuralinterpretation.Forexample,thisdemandfunctioncanarisefromaconstantelasticityofsufunctioninwhichbuyersregarddiferentvarietiesasimperfectsubstitutes.6Thisisknownastheelasticity;however,modelsofimperfectcompetition(e.g.,Melitz-typsimilarformulations.AsinGopinathetal.(2020),?rmsfacethreecurrencychoiceswhenpricproducercurrencypricing(PCP),localcurrencypricing(LCP),anddominantcurrencypricing(DCP).Supposeacertainfractionθjofimportsareinvoicedinproduceri’scurrency(PCP),afractionθinlocalimporterj’scurrency(LCP),andafractionθinthedominantcurrency—heritbedollaroreuro(DCP).Assumethatθj+θ+θ=1.Thenthf 6Thiselasticityofdemandcanbeeitherconstant(inthecaseofCESdemand;cf.AtkesonandBurstein,2008)ortimevarying(aswiththehomotheticdemandaggregatorofKimball,1995).ofdiferentinvoicingcurrenciesexpressedintheimportingcountry’scurrency:inithatareimportedbyjandinvoicedincurrencyk.InCalvo’s(1983)“stickyprices”setup,△p,t=(1-δ)(p-,t-p,t-1)forp-,tthe(log)resetpriceincurrencyk.Thein丑ationofservicesincountryjthataHereeij,tisthe(log)bilateralnominalexchangeratebetweeniandj(unitsofimporterj’scurrenperunitofexporteri’scurrency);e$j,tisthe(log)dollarexchangerateofj(unitsofj’scurrencyexpressthechangeinlogvalueofservicesimcurrency)asfollows:△vij,t=△pij,t+△qij,t=(1-σij)θj△eij,t+(1-σij)θ△e$j,t-σij△pj,t+△yj,t.(4)Forcountriesotherthamandyj,tthenthefollowingstatem?Producercurrencypricingoccursupto1-σij.inexchangerates;also,thevalueofi’simportsfromj(ini’scurrency)isindependentof7GopinathandRigobon(2008)documentthatgoodspricesarerigidinthecurrencyinwhichtheyarereportedlypriced(withamediandurationofelevenmonths).changes,scaledby1-σij,intheimportercurrency–USDexchangerate.Aslongasσij1—thatis,providedthevaluationefectdoesnotcompletelyofsetthevolumethedemandelasticityisunknown,wecannottestfortheexistenceofLCP(e.g.whenσijiscloseto1).4.2EmpiricalStrategyGopinathetal.,2020):LLherevij,tisthevalueofservicesimportedbyjfromiattinimporterj’scurrency,andτ=0,1,2andtheexporter’sproducerpriceindex(PPI)(tocapturemarginalcostshifters),alongwiththeirtwolags.Wealsocontrolfordirecteddyadic?xedefectsλijandtime?xedefectsδt.Theformertime,institutions)thathavebeenshowntomatterforservicestrade(e.g.,Head,MayerandRies,2009;EatonandKortum,2018)andpossiblyforitschangesandofservicestrade(e.g.geopfbetweenbilateral(dollar)exchangeratesandthevalueimportedbycountryjinj’scurrencyiscoe伍cientsforthecontemporaneousandlaggedexchangeratesΣ=0ατandΣ=0βτ.Standardfrequencyofthisanalysis)aremarginalcostsshiftersshouldnotbeinterpretedasprovidingestimatesforstructuralparameters.Rather,theyprovide4.3ExchangeRateElasticitiesofAggregateServicesTradeFirst,weapplytheregressionspeci?cationEquation(5)tototalservicetrade;theestimatesareTable3:ExchangeRateElasticitiesofOverallServicesTrade(1)(2)(3)(4)△eij,t△e$j,t0.436***(0.028)0.223***0.391*** (0.039)(0.078)0.453***(0.057)0.176*** (0.056)0.539*** (0.097)0.264***(0.046)0.0190.222*(0.057)(0.133)0.601***(0.087) (0.102)0.593*** (0.141)R2DyadicFEsYesYesYesYesTimeFEsYesYesYesYesDyadsNotes:ThecountrysampleexcludesU.S.bothasanexporterandanimporter.Thedependent