基爾世界經(jīng)濟研究所-外籍工人如何促進出口（英文）-2017.2-41正式版

KIEL

WORKINGPAPER

HowForeign-born

WorkersFoster

Exports

No.2071February2017

LéaMarchal,ClémentNedoncelle

KielInstitutefortheWorldEconomy

ISSN2195–7525

KIELWORKINGPAPER NO.2071|FEBRUARY2017

ABSTRACT

HOWFOREIGN-BORNWORKERSFOSTEREXPORTS

LéaMarchalandClémentNedoncelle

Weinvestigatetheexport-enhancingeffectofforeignworkersatthefirmlevel.Wefirstdevelopatheoreticalframeworkofheterogeneousfirms,assumingthatforeignworkersallowforproductivitygainsandconveyvaluableinformationonforeignmarkets.Weillustratethatforeignworkersfosterexportsattheextensiveandtheintensivemargins.Thiseffectcanbedecomposedinageneraleffect–towhichanyforeignworkercontributes–andadestination-specificeffect–towhichonlyforeignworkerswhowerebornintheexportdestinationcontribute.WetestthesetheoreticalpredictionsusingFrenchfirm-leveldataoverthe1997-2008periodandapropensityscorematchingmethodtoaddressendogeneityconcerns.Wefindthatforeign-bornworkers,andespeciallyskilledindividuals,fosterexportsatbothmargins.Onaverage,afirmemployingforeign-bornworkersexports30%moreinvaluethanacontrolfirm.Wefindevidencethatthisincreaseisspreadoveralldestinations,suggestingthattheeffectofforeign-bornworkersgoesbeyondadestination-specificinformationalchannel.

Keywords:Foreign-bornworkers,Exports,Firms,Heterogeneity,Productivity

JELclassification:F14,F22,F16

LéaMarchal

KielInstitutefortheWorldEconomyKiellinie66,D-24105Kiel,Germany

Email:lea.marchal@ifw-kiel.de

ClémentNedoncelle

UniversityofLille

LEM-CNRSUMR9221

CitéScientifique,batSH2

59655Villeneuved’Ascq,France

Email:clement.nedoncelle@

Theresponsibilityforthecontentsofthispublicationrestswiththeauthor,nottheInstitute.Sinceworkingpapersareofapreliminarynature,itmaybeusefultocontacttheauthorofaworkingpaperaboutresultsorcaveatsbeforereferringto,orquoting,apaper.Anycommentsshouldbesentdirectlytotheauthor.

2

Introduction

Anextensiveliteratureinvestigatesthefirm-leveldeterminantsoftradeperformanceonforeignmarkets.Empiricalregularitiessuggestthattradeoutcomesaremainlydeterminedby(i)theidiosyncraticfirmcharacteristicsand(ii)thecapacityofthefirmtoovercomelargecountry-specifictradecosts.Successfulexportersaremoreproductiveandlargerintermsofemployment,capital-intensityandfinancialcapabilitiesascomparedtonon-successfulexportersandnon-exportingfirms(Bernardetal.,2012).Thiswithin-industryselectionofexportingfirmsthroughproductivityhasbeenrationalizedbytheseminalmodelofMelitz(2003).Ontheotherhand,informationalbarrierswhichareusuallyapproximatedbythegeographicandculturaldistancebetweencountries,areknowntodeterexportoutcomes(DisdierandHead,2008).

Foreignworkersimpactbothaforementionedtradedeterminants.First,somepapersshowthatemployingforeignworkersgeneratesaproductivity-enhancingtaskspecializationwithinthefirm(PeriandSparber,2009).Thisliteratureechoesanotherstrandofresultssupportingtheproductivity-enhancinge?ectofculturalandethnicdiversityamongskilledworkers(Goldinetal.,2011;Traxetal.,2015).Sofar,onlyalimitedattempthasbeendonetobridgethisliteratureinvestigatingtheproductivitye?ectofimmigrants,andthetrade-migrationliterature.Tothebestofourknowledge,Mitaritonnaetal.(2016)arethefirsttoexplicitthelinkbetweenforeignworkers,productivityandexports.

Second,anumberofstudiesprovideevidencethatimmigrantsconveyvaluableinformationontheirorigincountrieswhichdecreasesadvaloremandfixedcostsfacedbyexporters.Thus,theyfostertradebetweentheiroriginandhostcountriesattheextensiveandtheintensivemargins.Theseresultshavemainlybeensupportedbymacro-levelstudies(Gould,1994;Rauch,2001;ParsonsandWinters,2014).Atthefirmlevel,theimpactofthefirm’sworkforceonitsexportoutcomeshasattractedlittleattention.Hiller(2013)showsthatfirmsshouldemployforeignworkersinordertoaccesstheknowledgeembeddedintheforeignpopulation.BothHiller(2013)andHatzigeorgiouandLodefalk(2016)findthatanincreaseinforeignemploymentatthefirmlevelisassociatedwithanincreaseinexportstotheimmigrantorigincountry.

Inthepresentpaper,weinvestigatethedi?erentchannelsthroughwhichforeignworkersimpactfirm-levelexportsatbothmargins.Tothisend,wedevelopatheoreticalframeworkwithheterogeneousfirmsinmonopolisticcompetitionrestinguponthemodelofMelitz(2003).Weassumethatforeignworkersallowtheirfirmtobemoreproductive,andconveyvaluableinformationtotheiremployeronforeignmarkets,particularlyontheirorigincountry.Ourmodelpredictsthatforeignworkersfosterexportsatbothmarginstowardanydestinationcountry.Thise?ectcanbedecomposedinagenerale?ect–towhichanyforeignworkercontributes–andadestination-specifice?ect–towhichonlyforeignworkerswhowerebornintheexportdestinationcontribute.Doingso,weshowthatexportcuto?saredestination-andfirm-specificanddependonthefirm’sexogenousproductivityanditsemploymentofforeignworkers.

WetestthesepredictionsusingadatasetonFrenchmanufacturingfirmsoverthe1997-2008period.Weidentifyforeign-bornworkersinacomprehensivematchedemployer-employeedataset(fromfirms’annualemployeedeclarations)thatwecombinewithtradedataatthe

2

firm-destinationlevel(fromtheFrenchcustoms)andbalancesheetdata(fromtheFrenchtaxauthority).Awareofthereversecausalitybias–foreignemploymentispotentiallydrivenbyfirm-levelexportperformance–wetakefulladvantageofthefirm-leveldatabyestimatingthetrade-inducede?ectofforeign-bornworkersusingapropensityscorematching(PSM)approach.Weestimatethee?ectofmultipletreatments–allrelatedtoforeignemployment–onexportoutcomes.Inotherwords,weestimatethedi?erenceinexportoutcomescomingfromdi?erencesinforeignemployment.Wefindthatbothmarginspositivelyreacttotheemploymentofforeign-bornworkers.Onaverage,afirmemployingforeign-bornworkersexports30%moreinvaluethanacontrolfirm.Thispro-tradee?ectisnotrestrictedtoskilledworkers:unskilledforeignemploymentisalsoassociatedtoapro-tradee?ect,inparticularfortheintensivemargin.Then,ourresultsshowthatforeignemploymentmattersnotonlyforexportstowardimmigrants’origincountriesbuttowardanyexportdestination,whichdirectlyrelatestoapro-tradee?ectofforeignworkersattheintensivemargin.Thisresultcanbeattributedtoaproductivity-enhancinge?ectand/oraninformationale?ectofforeignworkers,assupportbytheory.Ourresultsarerobusttoalternativematchingprocedures,alternativesub-samplesandalternativetreatmentsthatallinformusabouthowforeignworkersfavorexports.

Thecontributionsofthepaperarethefollowing.First,weproposeatheoreticalmodelofheterogeneousfirmsrationalizingtheexport-enhancinge?ectofforeignworkers.Tothebestofourknowledge,theonlyattempttoprovideatheoreticalframeworktoshowhowmigrantsfostertradehasbeenmadebyFelbermayrandToubal(2012).Wedepartfromthisarticlebyprovidingaheterogeneous-firmapproachtothisresearchquestionandbyfocusingontheexportside.Moreover,werationalizethee?ectofforeignemploymentontheintensivemarginoftrade,whilemostpapersfocusontheextensivemargin.

Second,wedepartfromexistingempiricalstudiesonthetrade-migrationnexusbyproposingafirm-levelanalysisandanalternativeestimationstrategytoinsulateourresultsfromreversecausality.Whilemostpaperstendtoovercomeendogeneityissuesusingregional-levelvariablestoinstrumentthefirm’sforeignemployment,weestimatethetrade-inducede?ectofemployingforeignworkersthankstoaPSMapproach.Thismethodallowsustorelyonfirm-levelinforma-tion,insteadofaggregatemigrationdata.Ontopofthat,ourstrategyallowsustoshowthat,withinlocallabormarkets,foreignemploymentfavorsexports.Ourresultsthuscomplementexistingstudiesinvestigatingthetrade-migrationnexusatthelocallevel.

Finally,weprovideadditionalinsightstotheblack-boxofexportsuccessandfailure.Despiterecentcontributions,whatmakesanexportersuccessfulisstillnotfullyrationalized.Ouranalysisprovidesevidenceofanadditionaldeterminantoffirms’exportsuccess.

Therestofthepaperisorganizedasfollows.Inthenextsection,wepresenttherelatedliterature.InSection3,wepresenttheFrenchfirm-leveldataandasetofstylizedfacts.InSection4,wedevelopatheoreticalframeworkrationalizingthee?ectofforeignworkersonexports,thatweestimateinSection5.Thelattersectiondisplaysafirstsetofresultssupportingthepro-tradee?ectofforeign-bornworkersderivedfromaPSMapproach.Wepursueour

3

analysisinSection6byprovidingfurtherevidencethatforeignworkersfavorexportstoanydestination,andnotonlytotheirorigincountry.Section7concludes.

Exportsandimmigration

Thepreferenceandinformationalchannels

Alargesetofpapersprovideaggregateevidenceonthepro-tradee?ectofimmigrants.TheseminalpapersofGould(1994)andsubsequentworksurveyedbyRauch(2001)andParsonsandWinters(2014)highlighttwochannels.First,immigrantsconveyinformationandpromotetrustbetweentheirhomeandhostcountries.Theirsocialcapitalreducestransactioncosts,whichfostersbilateraltrade(bothexportsandimports).Second,immigrantsexpandbilateralimportsduetotheirpreferencesforthegoodsproducedintheirhomecountry.

FelbermayrandToubal(2012)attempttoquantifythetwoaforementionedchannels.UsingstocksofforeignindividualsinOECDcountriesin2000,theyfindthatthepro-tradee?ectofimmigrantsonbilateraltradeaccountsfor37%ofthetotale?ect,theremaining63%beingattributedtothepreferencechannel.

Moststudiessuggestthatimmigrantsexertagreatestpro-tradee?ectondi?erentiatedgoodsforwhichthepricefailstotransmitrelevantinformation.Theliteraturealsosuggestsalargerpro-tradee?ectofskilledandvoluntarymigrants,withrespecttoloweducatedandforcedmi-grants.Finally,asdocumentedbyHatzigeorgiouandLodefalk(2014),existingstudiesadoptinginstrumentalvariabletechniquesshowthatthecausalrelationrunsfromimmigrationtotrade.

Alimitednumberofstudiesusefirm-leveldatatoanalyzehowimmigrantsimpactexports.Hiller(2013)showsthatimmigrantsmayreducebothadvaloremandfixedexportcostsbyrelaxinginformationalbarriersthankstotheirsuperiorknowledgeofforeign-markets.Nonethe-less,inordertoaccesstheknowledgeembeddedintheforeignpopulationoftheircountry,firmsshouldindeedemployforeignworkers.Usingemployer-employeedataonDanishmanufacturingfirmsover1995-2005,shefindsthatforeignemploymentincreasestheexportedquantitiesandimpactsthecompositionofexports,whilethelocalpresenceofforeignershasalimitedimpactonexports.Tohighlightcausality,sheinstrumentsforeignemploymentbytheaveragenumberofimmigrantsfromagivenoriginemployedinotherfirmsinthesameindustry,orinthesameregionofthefirm.Yet,aspointedoutbyParsonsandWinters(2014),sheassumesthatregionalimmigrantstocksareexogenoustothefirm,whilearguingthatforeignemploymentiscorrelatedwiththelocalpresenceofforeigners,whichslightlyweakensherargument.

Similarly,Parrottaetal.(2016)useDanishemployer-employeedataoverthe1995-2007period,toinvestigatethecausale?ectofanincreaseinlaborforceethnicdiversityonexportoutcomesatbothmargins.Theymeasurediversityusingdi?erencesinspokenlanguagesacrossworkers.Theyfindthatmorediversefirmsperformbetteronforeignmarketsalongallextensivemarginmeasures.Thesefirmshaveahigher"relationalcapital"whichtranslateintoanincreasedabilitytoinitiate,manage,andexpandinternationalbusiness.Tocontrolforendogeneity,theyuseashift-shareinstrumentàlaCard(2001)toidentifysupply-drivendiversityfromexogenous

4

changesinthelocallaborsupplyinthe1990’s.Doingso,theyassumeacorrelationbetweenpastandcontemporaneousimmigration,andthatindividualfirmsdonotimpactlocaleconomicoutcomes.Yet,thelatterargumentcanbechallengedincaseoflargefirmsandsmallsectorsintheDanisheconomy.

Then,HatzigeorgiouandLodefalk(2016)useSwedishemployer-employeedataover1998-2007andfindthatforeign-bornworkers(inparticularskilledandrecentlyarrived)increaseexportsatbothtrademargins,allthemoreforsmallfirms.Toovercomeendogeneitybias,theyuseaGMMestimatorandinstrumentthefirm-levelemploymentofforeign-bornworkerswiththeaverageforeignemploymentinotherSwedishfirmsandtheaverageforeignemploymentinthefirm’sindustry.Theirinstrumentisvalidundertheassumptionthatfirmemploymentstrategiesareexogenoustopriortraderelationships.Toconvincethereader,theypresentaSwedishbusinesssurveywhichindicatesthathiringofforeignworkersmaynotnecessarilybeendogenoustotrade.

Theoretically,thee?ectofimmigrationonexportshasbeenrationalizedbyPeriandRequena-Silvente(2010)usingthemodelofChaney(2008).Theassumethatimmigrantslowerfixedexportcosts.Thus,lessproductivefirmsthatwerebelowtheproductivitythresholdtoexport,becomeabletoentertheexportmarketwhenemployingimmigrants.Theyconcludethatthetrade-enhancinge?ectofimmigrantsshouldtakeplaceattheextensivemargin,andcorrobo-ratethispredictionusingdataonSpanishprovincesoverthe1995-2008period.However,theirtheoreticalmodelisnotfullyinlinewithempiricalevidence,assomestudiesalsofindane?ectontheintensivemargin,supportingtheideathatimmigrantsdonotonlyreducefixedexportcosts.Tofullyrationalizetheexport-enhancinge?ectofimmigrants,onbothtrademargins,onealsoneedstoassumethatimmigrantseitherloweradvaloremexportcosts(throughtheinformationalchannel)orincreasefirmproductivity.

Theproductivitychannel

TheindirectimpactofforeignemploymentonfirmproductivityhasbeenputforwardbyPeriandSparber(2009).Intheirseminalpaper,theyshowthatnativesandimmigrantsareimperfectsubstitutes.Theemploymentofimmigrantsthusgeneratesataskspecializationwithinfirms,andgeneratedgainspreventnatives’wagestodecreaseduetoimmigration.

Recently,Mitaritonnaetal.(2016)explicitlyanalyzethelinkbetweenimmigrationandproductivitygains.UsingFrenchfirm-leveldataoverthe1995-2005period,theyfindthatanincreaseinthelocalsupplyofimmigrantsincreasestotalfactorproductivityoffirmslocatedinthatarea.Thise?ectisfoundtobestrongerforfirmswithinitiallylowproductivityandsmallsize.Inaddition,theauthorsfindthatthispositiveproductivitye?ectwasassociatedwithlargerexports.Toaddresspotentialendogeneity,theyinstrumentthelocalsupplyofimmigrantsbyashift-shareinstrumentbasedonthespatialdistributionofimmigrantsin1990.Thismethodrestupontheideathatpastandcontemporaneousimmigrationsharesarehighlycorrelatedduetonetworke?ects.Theauthorsexcludethepossibilitythatthespatialdistributionofimmigrants

5

in1990couldbecorrelatedwithchangesinlabordemandduetofirms’productivityshocks,arguingthatlocaleconomicoutcomesarenotimpactedbyindividualfirms.

Aemergingstrandoftheliteratureshowsthatculturaldiversityeventuallyincreasesfirmproductivity,forinstance,byenhancinginnovationandproblemsolving(Goldinetal.,2011).Traxetal.(2015)usedataonasampleofGermanfirmsover1999-2008toanalyzethelinkbetweenculturaldiversityandfirmproductivity.Theyfindthattheshareofforeignemploy-menthasnoimpactonproductivity,whilethediversificationoftheworkforcewithrespecttonationalitiesincreasestotalfactorproductivity.Inaddition,regionaldiversificationsignificantlyimpactsfirmproductivity.Toinsulatetheirresultsfromreversecausality,theauthorsuseaGMMapproachandconstructashift-shareinstrumentusingregionalemploymentsharesbynationalitiesin1987andnationwideemploymentgrowthratesforeachnationality.Then,thisinstrumentisinteractedwiththefirm’sshareofforeignemploymentwithrespecttotheregionalforeignemployment.

Nonetheless,somepapersshowthatethnicdiversitycancreatelinguisticandculturalfric-tions,whichcancreatecommunicationproblemsandweakensocialtiesbetweenworkers.UsingDanishemployer-employeedataover1995-2005,Parrottaetal.(2014)findevidencethatafirm’sworkforcediversityinethnicitynegativelyimpactsitstotalfactorproductivity.Theytacklereversecausalitybyconstructingashift-shareinstrumentwherethefirm’sdiversityisinstrumentedusingthelocaldiversityofthelaborsupply.Thelatterisconstructedusingthe1990’scompositionandcontemporaneouspopulationstocks.Anumberofpapersinmanage-mentsciencesalsopresentmixedevidenceregardingtheadvantagesofmulticulturalismatthefirmlevel(Loth,2009;GoodallandRoberts,2003).

Dataandstylizedfacts

3.1 Data

WemergethreedatasetsprovidingusinformationonFrenchfirmsoverthe1997-2008period,usingauniqueadministrativeFrenchfirmidentifier(theSIRENnumber).

Administrativeemployer-employeedata

First,weusethefirms’annualemployeedeclarations(DéclarationsAnnuellesdesDonnéesSo-ciales,DADS)containingexhaustiveinformationontheemploymentoffirmssettledontheFrenchmetropolitanterritoryfrom1997to2008.Thisadministrativedatabaseismadeofcom-pulsoryreportsprovidedbyeachemployeronthegrossearningofhisemployees.Allwage-payingindividualsandlegalentitiesestablishedinFrancearerequiredtofilepayrolldeclarations;onlyindividualsemployingcivilservantsareexcludedfromfilingsuchdeclarations.Thedatasetisthusmadeofinformationatthefirm-employee-yearlevel.

Moreespecially,wehaveinformationaboutthegeographicalzoneofbirthofeachworker.TheetrangvariableallowsustoknowwhetheranemployeewasborninFranceorinaforeign

6

country.1Weassumethatwhentheetrangvariableisleftempty,theworkerwasborninFrance.Thedatasethoweverdoesnotcontaininformationabouttheexactcountryofbirthofforeignworkers.Intherestofthepaper,weconsiderforeignworkersasforeign-bornworkers.Notethatwecannotidentifynaturalizedindividualswhothusareconsideredasforeigners.Wealsohaveinformationonthesocio-professionalcategoryofeachworker.Wecombinethisinformationwithaclassificationofcategoriesintowhite-andblue-collaroccupationstoidentifyskilledandunskilledworkers(Bombardinietal.,2015).

Weaggregatethisdatasetatthefirm-levelandcount,foreachfirm,thenumberofnativeandforeignworkersforeachskillcategory.Afterremovingobviousoutliersandextremevalues,DADSdataareinlinewithmacro-levelevidence.Forinstance,in2006intheIle-de-Franceregion,13.6%ofworkersareforeign-born,whilethepartial2006censusestimatesthatimmi-grantsrepresent12.9%oftheworking-agepopulation.Thefinalsampleismadeof21,157,647observationsatthefirm-yearlevel,thatcorrespondstoanaverageof2,000,000firmsperyear.Inthissample,foreign-bornworkersrepresent7.49%ofallworkers,whichisclosetotheestimatesproposedbyBrückeretal.(2013).

Theadvantageofusingfirm-leveldatainourcaseistwofold.First,werelyonfirm-leveldatatosolelyfocusontheimmigrantworkingpopulation.Converselytocensusdata,ourdatasetexhaustivelycoverstheemploymentofforeignersinFrance.Thisdatasetisthusparticularlyappropriateforaconsistentidentificationofthepro-tradee?ectofforeignworkersonexports,withinthefirmboundaries.Additionally,itallowsustoidentifyfirm-levelmechanismsthatcannotbecapturedusingaggregatedata,inwhichthecausaldirectionmaybeflawed.Second,ourdatasetallowsustoestimatethepropertradee?ectoftheemployedforeignersbythefirm,whileexistingstudyusingregionalimmigrationdatatendtoestimatethee?ectofimmigrationontheaveragelocalfirmperformance,alsoaccountingforexternalitiesarisingfromtheproximityofimmigrants.

Customstradedata

Wethenusefirm-leveltradedatafromtheFrenchcustomsoverthe1997-2008period.Thisdatabasereportsthevolume(intons)andthevalue(inEuros)ofexportsforeachCN8product(EuropeanUnionCombinedNomenclatureat8digits)anddestination,foreachfirmlocatedontheFrenchmetropolitanterritory.Someshipmentsareexcludedfromthisdatacollection.InsidetheEU,firmsarerequiredtoreporttheirshipmentsbyproductanddestinationcountryonlyiftheirannualexportvalueexceedsthethresholdof150,000Euros.ForexportsoutsidetheEU,allflowsarerecordedunlesstheirvalueissmallerthan1,000Eurosoroneton.Yet,thesethresholdseliminateaverysmallshareofthetotalexports.Fromthisdataset,weonlykeepmerchandiseshipments,excludingagriculturalandservicesexports.

1TheDADSdataallowustodistinguishtwocategoriesofforeignworkers:thoseborninsidetheEuropeanUnion,andothers.Intheremainingofthepaper,wedonotexploitthatinformationbecausethegroupofforeignworkersbornoutsidetheEUistobroad.Tousethisgroup,onewouldneedtoassumethataworkersborninSwitzerlandeventuallyhasthesamepro-tradeimpactthanaworkerbornonanothercontinent.

7

Thedatasetconsistsof26,186,006observationsatthefirm-year-destination-productlevel,thatweaggregateinto7,110,894observationsatthefirm-year-destinationlevelandinto1,381,500observationsatthefirm-yearlevel.OncecombinedwiththeDADSdata,weobtainadatasetof21,157,647firm-yearobservationsoverthe1997-2008period,inwhich1,043,790areexporters(representing98%oftotalFrenchexports)and20,113,857arenon-exportingfirms.

Balance-sheetdata

Wecompletethepictureusingabalance-sheetdatasetconstructedfromreportsofFrenchfirmstothetaxadministrationoverthe1997-2008period(BénéficesRéelsNormaux,BRN).Thisdatasetcontainsinformationonthevalueadded,totalsales,capitalstock,debtstructureandothervariablesatthefirmlevel.Importantly,thisdatasetiscomposedofbothsmallandlargefirms,sincenothresholdappliesonthenumberofemployeesforreportingtothetaxadministration.

Thedatasetcontainsbetween550,000and650,000firmsperyear(around50%ofthetotalnumberofFrenchfirms).Intotal,thedatasetismadeof5,850,838firm-yearobservationsofwhich5,425,621canbemergedintothesampleof21,157,647firm-yearobservations.Dependingontheyear,thesefirmsrepresentbetween90%and95%ofFrenchexportscontainedinthecustomsdata.

3.2 Stylizedfacts

DescriptivestatisticsarereportedinTable1.Westartbypresentinganumberoffirmcharac-teristics:theprofit,therevenue,thetotalassets,thecapitalintensitymeasuredbytheassetsperemployee,theageandtheapparentlaborproductivitymeasuredbythevalueaddedperemployee.Then,wereportanumberofstatisticsontheexportactivities.Notethatthepartic-ipationdummytakesthevalueofoneifthefirmisanexporteratagiventime,zerootherwise.Weinferthatfirmsexportinaverage2.67thousandsofEurosoverthestudiedperiod.

Lookingatfirms’employment,weobservethatnon-exportersandexportersemployabout96%ofnativesovertheirtotalworkforce.Notethatabout76.29%offirmsdonotemployanyforeignworkerinthesample.Wealsoseethatmostforeign-bornworkersholdlow-skilledjobs.Onaverage,exporterstendtohiremoreskilledforeignworkersthannon-exporters.

Wenowfocusonthecharacteristicsoffirmsthatemployforeign-bornworkers,ascomparedtofirmsthatdonot.Weplotthedistributionsoftwofirm-levelcharacteristicsthataregener-allyassociatedtotradeoutcomesforthesetwogroupsoffirms.Figure1presentstheKerneldistributionsoftheassets(inlog)forbothgroups.Thetwodistributionsareverycloseandhavesimilarshapes.Thedistributionoffirmsemployingnoforeignworkerisslightlyontheleftwithrespecttotheseconddistribution.Itsuggeststhatfirmshiringnoforeignworkermaybesmaller.

ThesameconclusionisreachedwhenlookingatthedistributionsofthecapitalintensityshowninFigure2.Inthiscase,thetwodistributionsarehardlydistinguishable.Althoughnotreportedinthepaper,weobtainsimilarpictureswhenplottingthedebtstructureandthe

8

Table1:Summarystatistics

Obs.

Mean

Std.Dev.

Min.

Max.

FIRMCHARACTERISTICS

Profit(inthousandsofEuros)

5,425,621

0.210

16.722

-14,710

8,099

Revenue(inthousandsofEuros)

5,425,621

6.014

597

0

697,523

Assets(inthousandsofEuros)

5,425,621

13.860

1,298

0

1,266,449

Capitalintensity

5,425,621

83.210

2,323

0

1,265,299

Age(sincecreation)

4,787,889

16.050

13.220

0

88

Apparentlaborproductivity

5,425,621

60.910

6,688

0

1.35E+07

TRADE

Exportedvalue(inthousandsofEuros)

1,043,790

3,218

71,816

0

1.51E+07

Exportedquantity

1,043,790

1,762

43,028

0

8,766,293

Exportdestinations

1,043,790

5.960

10.170

1

174

Exportedproducts

1,043,790

22.400

102.195

1

10,194

Participationdummy

4,603,472

0.049

0.217

0

1

EMPLOYMENTPERTYPEOFFIRM

PanelA:Exporters

ShareofFrenchworkers

748,160

0.958

0.145

0

1

Shareofforeign-bornworkers

748,160

0.041

0.145

0

1

PanelB:Non-exportingfirms

ShareofFrenchworkers

4,603,472

0.961

0.136

0

1

Shareofforeign-bornworkers

4,603,472

0.038

0.018

0

1

SKILLEDEMPLOYMENTPERTYPEOFFIRM

PanelA:Exporters

ShareofFrenchskilledworkers

748,160

0.165

0.311

0

1

Shareofforeign-bornskilledworkers

748,160

0.008

0.071

0

1

PanelB:Non-exportingfirms

ShareofFrenchskilledworkers

4,603,472

0.077

0.213

0

1

Shareofforeign-bornskilledworkers

4,603,472

0.004

0.045

0

1

9

employmentofthefirms.Thereby,thetwogroupsoffirmsdonotseemtobeverydi?erentwhenwelookatva