资源描述：
《11. Multinationals and Organization of the Firm.pdf》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。
Feenstra,AdvancedInternationalTradeChapter11:MultinationalsandOrganizationoftheFirmDespitethefactthatthisbookisaboutinternationaltrade,wehavesofarnotintroducedanyrolefortraders.Therehasbeenlittlescopeforfirms,letaloneeconomicorganizationmoregenerally,tohaveanysignificantinfluenceontradepatternsbetweencountries.Thiscontradictstheempiricalfactthatagooddealoftradeoccursinternallywithinfirmslocatedacrosscountries.FortheU.S.,forexample,aboutone-thirdofexportsandover40%ofimportsconsistofintra-firmtradebetweenaU.S.orforeignfirmandtheiraffiliates–seeTable11.1.Whenafirmoperatesinseveralcountriesitisamultinationalenterprise,andtheinvestmentmadeinthe1foreigncountryisreferredtoasforeigndirectinvestment(FDI).Thefirstgoalofthischapteristointroducethesefeaturesintoourearliertrademodels.Theclassictreatmentofamultinationalisthatithassomeintangibleasset(suchasknowledgeofaproductionprocess)thatitcanusetoitsadvantageinaforeignmarket.Itmustdecidewhethertosimplyexportthere,ortoinvestinthatmarketbybuildingaplantandsellingtheproduct,orwhethertoengageinajointventureorothercontractualarrangementwithaforeignfirmtoproducethegood.Wewillinitiallyfocusonthefirsttwodecisions–whethertoexporttotheforeignmarketortobuildaplantandsellthere.Thethirdcase–whethertoengageinajointventurewithaforeignfirm–willbediscussedattheendofthechapter.Noticethatthisclassicstatementoftheproblemcanequallywellworkinreverse,wherebytheacquisitionofaforeignfirmcanbringwithitsomeknowledgeofvaluetothepurchaser,thatcouldnotbe2obtainedbysimplybuyingtheproductsofthatforeignfirm.1WemightdefineFDIasacquiringsufficientassetsinaforeignfirmtoexercisesomemanagerialcontrol,thoughacquiring10%ormoreoftheassetsofaforeignenterpriseisthedefinitioncommonlyusedinpractice.2Blonigen(1997)arguesthatincreasedinflowsofJapaneseacquisitionFDIintotheU.S.during1985-1990weremotivatedbythedesiretoacquiretheknowledgeassetsofU.S.firms,combinedwiththelowvalueofthedollar.Noticethatbythisargument,FDIflowsdependonthelevelofexchangerates. 11-2Feenstra,AdvancedInternationalTradeTable11.1:U.S.ImportsandExportsthroughMultinationalCorporations,1992($Billion)TotalU.S.MerchandiseExports448.2TotalU.S.MerchandiseImports532.7ExportsthroughU.S.Multinationals:ImportsthroughU.S.Multinationals:-U.S.parenttoforeignaffiliates104.7-ForeignaffiliatestoU.S.parent92.6-U.S.parenttootherforeignfirms140.8-OtherforeignfirmstoU.S.parent107.2-OtherU.S.toforeignaffiliates15.6-ForeignaffiliatestootherU.S.16.6ExportsthroughForeignImportsthroughForeignMultinationals:Multinationals:-U.S.affiliatetoforeignparent48.8-ForeignparenttoU.S.affiliate137.8-Manufacturing,ofwhich:11.6-Manufacturing,ofwhich:37.3-WholesaleTrade,ofwhich:34.6-WholesaleTrade,ofwhich:89.2--MotorVehicles&Equipment5.2--MotorVehicles&Equipment28.7-U.S.affiliatetootherforeignfirms55.2-OtherforeignfirmstoU.S.affiliate46.7Totalintra-multinationalexports153.5Totalintra-multinationalimports230.4(percentoftotalU.S.exports)34.2%(percentoftotalU.S.imports)43.3%Sources:Mataloni(1995,Table7,p.48);U.S.DepartmentofCommerce,BureauofEconomicAnalysis(1995,TableH-25,H-27,H-31,H-33);citedinFeenstra(1999). 11-3Feenstra,AdvancedInternationalTradeWeseethatthedecisiontoengageinFDIbyamultinationalthereforeinvolvesthreeinterrelatedaspects–ownershipofanasset,locationtoproduce,andwhethertokeeptheassetinternaltothefirm–andthesecomprisetheso-called“OLIframework”ofmultinationalactivity(Dunning,1977,1981).Inthefollowingsections,weexploremodelsgivinginsightintothesevariousaspects.Webeginwiththesimplestexamplewhereafirmconsidersshuttingdownitsfactoryinonecountryandmovingittoanother,whichcanbemodeledbyamovementofphysicalcapitalacrosscountries.WhilesuchamovementofcapitaldoesnotnecessarilyoccurwithFDI(whichmaysimplyinvolveachangeinownershipwithoutanymovementofphysicalcapital),itisausefulplacetobeginouranalysis.Afterbrieflyreviewingcapitalinflowsintoaone-sectoreconomy,weusetheclassicarticlebyMundell(1957)toinvestigatecapitalmobilityinthetwo-sectorHeckscher-Ohlin(HO)model.Inthatcasewefindthatevensmalltariffshaveaverylargeimpactonthemovementofcapitalfromonecountrytotheother.Thisphenomenonof“tariff-jumping”FDIisatypicaloccurrenceinimport-substitutionregimes,butisnotrestrictedtothedevelopingcountries:manyindustrializedcountrieshaveusedtariffsorquotastoinducetheentryofforeignfirms.Thispolicyhassomeeconomicrationale,sincetheentryofforeignfirmsinsimplemodelsleadstoawelfaregainduetoincreaseddomesticwages.Thelinkbetweenforeigninvestmentandwagesisthereforeanimportanttopicforempiricalresearch.InMundell’sanalysis,domesticwagesarenotaffectedbytheforeigninvestment(dueto“factorpriceinsensitivity”intheHOmodel),butitisstillthecasethatthedeadweightlossofthetariffisreducedduetotheentryofforeignfirms.Multinationalsneednotleadtocapitalflowsbetweencountries,however,andmayinsteadjustinvolvetheownershipandlocationoffirms.Weconsiderseveralmodelsofthat 11-4Feenstra,AdvancedInternationalTradetype,distinguishingverticalfromhorizontalmultinationals.Anexampleoftheformercaseisafirmthathasheadquartersinonecountrybutdoesproductioninanothertoobtainlowerfactorpricesthere(Helpman,1984).Thelattercaseoccurswhenafirmdecidestoduplicateproductionfacilitiesandselllocallyintwoormorecountries,duetotariffsorotherbarriersbetweenthem(Markusen,1984,2002).Weidentifyconditionsunderwhichaparticulartypeofmultinationalismostlikelytoariseinequilibrium,andonceagain,findthattradepoliciescanhaveasignificantimpactonmultinationallocation.Beyondtradethatisinternaltomultinationalfirms,thereisanotherlargeportionoftradethatoccursbetweenfirmsthathavesomespecialrelationship.Anexampleiswhenafirmoutsourcessomeofitsproductionprocesstoalower-wagecountry.Inchapter4wemodeledthisasthepurchaseofintermediateinputs,butinfact,itofteninvolvesalonger-termrelationship:thefirmwantingtopurchasetheinputiscommittingitsmoney,andthesellingfirmiscommittingresourcestoproductionanddevelopmentoftheinputs.Thisrelationshipentailsrisktobothparties,andinmanycircumstancestheseriskscannotbeoffsetbylegalcontracts.Ifthecostsassociatedwiththeseincompletecontractsaretoogreat,thenthefirmwantingtopurchasetheinputswillinsteadvertically-integrateintothisactivity,becomingaverticalmultinational.Thus,aconsiderationoftransactionscostsunderincompletecontractsallowsustomodeltheinternalizationdecisionofamultinationalfirm.TherecentworkofGrossmanandHelpman(2002a,b)andothersputsthisinageneralequilibriumsetting,andinvestigatestheimplicationsofinternalizationfortrade.Transactionscostsisthefirstandforemostreasonforfirmstointegrate,butthereareothers.Forexample,jointventuresbetweenfirmsindifferentcountriescanbringwiththemknowledgeofthesupplieranddistributionnetworksineachcountry,andanyotherintangible 11-5Feenstra,AdvancedInternationalTradeassetsownedbyeitherfirm.RauchandTrindade(2000)modelthisinformationalissueasaproblemof“matching”firmsacrosscountries,wherethereissomeinherentuncertaintyintheabilitytofindagoodpartner.Aswedescribe,theefficacyofachievingmatchesdeterminestheextenttowhichmarketsareintegrated,i.e.theextenttowhichfactorpriceequalizationobtains.Thestudyoftheorganizationoffirmsanditsimplicationsforinternationaltradeisaverynewareaofresearch,andisanappropriatetopiconwhichtoconcludeourbook.CapitalFlowsinaOne-SectorEconomyWebeginouranalysiswiththesimplestcaseofacapitalinflowintoaneconomyproducingjustonegood,asinMacDougall(1960).Supposethattheproductionfunctionisgivenbyy=f(L,K),whichislinearlyhomogeneousandconcave.Normalizingtheproductpriceatunity,therentaloncapitalisr=fK(L,K),withfKK<0.DenotethedomesticcapitalstockbyK0andtheforeigninflowofcapitalbyK*.Theequilibriumrentalintheabsenceoftheforeigncapitalisshownbyr0inFigure11.1,andinthepresenceofthecapitalinflowbyr1.Theforeigncapitalearnstheamountr1K*,whichistakenoutofthecountry.TheincreaseinGDPduetothecapitalinflowis,K0+K*Dy=∫fKdK=A+B.(11.1)K0Subtractingthepaymenttoforeigncapital,r1K*=B,thenetwelfaregaintothehomecountryistheareaA.Thus,theinflowofforeigncapitalcreatesawelfaregain.Whatisthesourceofthisgain?ItisevidentfromFigure11.1thatascapitalentersitdepressestherental,sothatthepaymentstoforeigncapital(B)arelessthanthetotalareaunder 11-6Feenstra,AdvancedInternationalTraderr0CAr1r=fKBK0K0+K*KFigure11.1 11-7Feenstra,AdvancedInternationalTradethedemandcurve(A+B).Theothersideofthisisthatthecapitalinflowraisesthemarginalproductoflaborandwagesw=fL(L,K).Indeed,thetotalincreaseinthewagebillequalsthe3areaA+CinFigure11.1.AreaCistheredistributionofincomefromcapitaltolabor,whereasareaAisthenetwelfaregainanditaccruestolabor.Thesesimpleobservationshaveledtoagooddealofempiricalworkastowhetherindustriesexperiencinganinflowofforeigncapitalpayhigherwages.FortheU.S.,HowensteinandZeile(1994)andDomsandJensen(1996)findthatforeignplantspayhigherwagesthandomesticplants.BlonigenandTomlin(1999)furtherfindthatforeignplantstypicallygrowfasterthanU.S.plantsofsimilarsize,andthattheyleadtoalargerpositiveimpactonlocalwagesthandoesdomesticinvestment.CombiningU.S.datawithinformationonMexicanandVenezuelanenterprises,Aitkenetal(1996)alsofindawagedifferentialbetweendomesticandforeign-ownedenterprises,evenaftercontrollingforindustrysizeandcapital-intensity.Noticethatthesestudiesaremostoftencomparingthewagespaidbyforeignversusdomesticplants,whichisnotexactlythesameasthegeneralequilibriumwageincreaseduetoforeigninvestmentpredictedbyFigure11.1.Forexample,ifforeignfirmssimplyattractthehighest-qualityworkers,thiswouldleadtoawageincreaseoverdomesticplants,butnotnecessarilywelfaregains.Thissameproblemofself-selectionoccursinempiricalstudiesoftechnologytransferbetweenforeignandlocalfirms(throughjointventuresorotherspilloversacrossfirms).EvidenceinsupportoftechnologytransferforMexicoisfoundbyBlomstrom(1986),BlomstromandPersson(1983),andKokko(1994),andforIndonesiabyBlomstromandSjoholm(1999).Thesestudiescorrelateindustryproductivitywiththepresenceofforeign3Becausetheproductionfunctionislinearlyhomogeneous,theny0=w0L+r0K0,andy1=w1L+r1(K0+K*).ItfollowsthatDy=(A+B)=(w1-w0)L+(r1-r0)K0+r1K*,andsinceB=r1K*andC=(r0-r1)K0,thenthewagebillchangesby(w-w)L=(A+B)-(r-r)K-rK*=A+C.101001 11-8Feenstra,AdvancedInternationalTradefirms.However,thisleavesopenthequestionofwhetherthehigherproductivityisduetotheforeignplantsthemselves,orduetospillovers.ThishasbeenaddressedusingplantleveldatabyHaddadandHarrison(1993)forMorocco,andAitkenandHarrison(1999)forVenezuela.Theyfindthattheforeignplantsindeedhavehigherproductivity,butthatthereisaweaknegativecorrelationbetweentheirpresenceandproductivityofthedomesticplantsinthesameindustry.Thus,thesefindingcastsomedoubtonthehypothesisofpositivetechnologicalspilloversbetweenforeignanddomesticfirms;rather,foreignfirmsarejustmoreproductive.CapitalFlowsintheTwo-SectorHeckscher-OhlinModelWeturnnexttoafamousexampleofcapitalflowsinthetwo-sectorHOmodel,duetoMundell(1957).Supposetherearetwocountrieswithidenticaltechnologiesandhomothetictastes,initiallyengagedinfreetrade.Withcountriesproducingbothgoodsandnofactorintensityreversals,therentaloncapitalisuniquelydeterminedbythegoodsprices,whichwewriteasr(p1,p2)forbothcountries.InFigure11.2theendowmentofthetwocountriesisshowniatthepointV,whichweassumeisinsidethefactorpriceequalizationsetsotherentaloncapital4isthesameacrosscountries.Withfreetradeingoods,consumptionofthetwocountriesisiiproportionalandoccursalongthediagonalofFigure11.2,suchasthepointAD(whereDistheconsumptionvectorofeachcountry,andmultiplyingbythetechnologymatrixAconvertsthistothefactorsembodiedinconsumption).Inthecasewehaveillustrated,country1islaboriiabundant,andthemovefromVtoADinvolvescountry1importingthecapital-intensivegood.Assumethatgood1iscapitalintensive,sothat¶r(p1,p2)/¶p1>0.4Referthechapter2,Figure2.3,toreviewhowthisfigureisconstructed. 11-9Feenstra,AdvancedInternationalTrade22LO2iKViADiV1K11OLFigure11.2 11-10Feenstra,AdvancedInternationalTradeNowallowforcapitalmobilitybetweenthecountries,andsupposethatcountry1appliesaspecificimporttariffoft>0.Thisraisesthedomesticpriceofgood1inthatcountry,andsincethisgoodiscapital-intensive,bytheStolper-Samuelsontheoremitincreasesthereturntocapital,r(p1+t,p2)>r(p1,p2).Itfollowsthatcapitalwillflowfromcountry2tocountry1.IniFigure11.2,thisshiftstheeffectiveendowmentpointverticallyupwardsfrompointV.Thequestionis,wherewillthisprocessend?Thenewequilibriumallowingforcapitalmobilityrequiresthatthereturntocapitalbeequalizedacrossthecountries.Butwiththetariffdrivingawedgebetweenthepricesinthetwocountries,itisimpossibleforrentalstobeequalized:t>0alwaysimpliesthatr(p1+t,p2)>r(p1,p2).Obviously,somethinginthisequationmustchangetoachieveanewlong-runequilibrium!Toseewhatchanges,letusthinkaboutwhathappensastheendowmentsofthecountriesimoveupwardsfrompointV.Capitalisflowingtowardscountry1,whichincreasestheoutputofindustry1andreducesthatofindustry2.Sincegood1isimported,theincreasedsupplyofthatgoodlowersimports,andthereducedsupplyofgood2alsolowersexports.SoaswemoveiupwardsfrompointV,tradebetweenthecountriesisbeingreduced.Thiswillcontinueuntiltheendowmentsreachthediagonal,indicatedbythepointiV,atwhichpointthecapital/laborratiosinthetwocountriesareidentical.Thismeansthatthereisnoreasonfortrade:botheconomiesareinautarky.Whatabouttherentaloncapitalineachcountry?Sincebothcountriesareinautarky,withidenticalcapital/laborratiosandidenticalhomothetictastes,thismeansthattheequilibriumpricesarethesameacrosscountries.Itfollowstherefore,thattheequilibriumrentaloncapitalisalsothesameacrosscountries.Thetariffthatisbeingappliedbycountry1isredundant:since 11-11Feenstra,AdvancedInternationalTradetherearenoimports,thetariffnolongerdrivesawedgebetweenthepricesandrentals.SoweconcludethatthepointVisthenewequilibriumwithcapitalmobility.Productpricesandtheirentaloncapitalareunchangedfromtheiroriginalvaluesbeforethetariff(sincethisisapropertyofachangeinendowmentswithintheFPEcone).Furthermore,theentireincreaseinii5theGDPofcountry1frompointVtopointVisrepaidtotheforeigncapitallocatedthere.iDeductingthisamountfromGDP,country1endsupbackatitsoriginalconsumptionpointAD,soitswelfareisthesameasbeforethetariffwasapplied!ThisanalysisbyMundell(1957)isimportantforseveralreasons:itshowsthatatariffcanleadtoalargeenoughinflowofcapitaltoeliminatetrade;anditshowsthatthecapitalfloweliminatesthedeadweightlossduetothetariff.Bothoftheseobservationsarerealistic,andcanbeillustratedbytheVERappliedtoJapaneseexportsofautomobilesintotheU.S.,asdiscussedinchapter8.Whilethistradepolicytooktheformofaquotaratherthanatariff,itstillledalargeinflowofJapaneseautofirmsintotheU.S.–enoughtomaketheVERredundantafter1987(i.e.therewerefewerimportsfromJapanthanallowedbytheVER).Thismeansthatits6welfarecostwasalsoreduced.ThisresultsofMundell(1957)aresometimessummedupwiththephrase“tradeinfactorsisasubstitutefortradeingoods,”sincethecapitalinflowduetothetariffhastheeffectofeliminatingtrade.Thishypothesishasbeenempiricallytested.Blonigen(2002)investigatestheextenttowhichFDIandimportsaresubstitutes(whichinflowsofcapitalreducingimports)orcomplements(FDIleadingtohigherimports).Hefindsevidenceforbothhypotheses,asis5Toseethis,measureGDPasthevalueoffactorearnings.Withunchangedproductpricesandfactorprices,theii11increaseinGDPisfromVtoVisDGDP=wL+rK+rK*,sodeductingpaymenttoforeigncapitalrK*,we11obtaintheoriginalGDPofwL+rK.6Neary(1988)examinesmoresystematicallythewelfarecostsoftariffsandquotasundercapitalmobility. 11-12Feenstra,AdvancedInternationalTradealsoconfirmedbySwenson(2002)fortheU.S.andGoldbergandKlein(2000)usingdatafromLatinAmerica.TheeffectofU.S.anti-dumpingdutiesoninwardFDIisinvestigatedbyBlonigen(2000).HefindsthatthisU.S.investmentisarealisticoptiononlyforfirmswithpriormultinationalexperience.WhiletherawdatashowaparticularlyhighresponserateforJapanesefirmsshiftingtheirproductiontotheU.S.,thisismainlyexplainedbythefactthatmanyofthesefirmsarealreadymultinationals.VerticalandHorizontalMultinationalsWeturnnowfromadiscussionofcapitalflowstothemodelingofmultinationalactivity,whereafirmchoosestooperateplantsinmultiplecountries.Wedistinguishthecaseswherethemultinationalchoosestooperateitsheadquartersinonecountryanditsproductionfacilitiesinanother–theso-calledverticalmultinational–fromthecasewhereitoperatesitsheadquartersinonecountrybuthasproductionfacilitiesinmultiplecountries–thehorizontalmultinational.VerticalMultinationalsTheverticalcasewasfirstanalyzedbyHelpman(1984c)inatwo-factorframeworkwithmonopolisticcompetition.Theincentivetooperateheadquartersinonecountryandproductioninanotherarisesduetofactorpricedifferencesacrossthecountries.Asweshallsee,however,theabilityofamultinationaltospreaditsfacilitiesacrossseveralcountriesservestoamelioratethesefactorpricedifferences,andenlargestheregionoffactorpriceequalization.Supposethatsector1producesahomogeneousgood,whilesector2producesadifferentiatedgoodwithaCESutilityfunction.AvarietyofthedifferentiatedgoodrequiresaunitsskilledlaborHinitsfixedcosts(whichwethinkofas“headquarters”services),andbunitsofunskilledlaborLinitsvariablecosts.Denotingtheoutputofadifferentiatedvarietybyy2i, 11-13Feenstra,AdvancedInternationalTradethecostsofproductionareC2i(w,q,yi)=qa+wby2i,whereqandwarethewagesofskilledandunskilledlabor,respectively.Asdiscussedinchapter5,withaCESutilityfunctionthereisauniqueoutputy2i=yatwhichprofitsarezero,wherefreeentryalsodeterminesthenumberofvarietiesN2insector2.Thenthedemandforskilledlaborforeachvarietyproducedisa,andthedemandforunskilledlaborisby.Technologiesarethesameacrosscountries,andwewillfocusonthe“integratedworldequilibrium”atwhichfactorpriceequalization(FPE)isobtained.Insector1,producingthehomogeneousgood,therequirementsofunskilledandskilledlabortoproduceoneunitofoutputaredenotedbya1Landa1H.Insector2,therequirementstoproduceonevarietyofthea1La2Ldifferentiatedproductarea2Lºbyanda2Hºa.ThenwecanletA=denotethea1Ha2Hmatrixoflaborrequirementsineachindustry,whichappliesineithercountryunderFPE.LetjjjY=(y,N)'denotethevectorwhosefirstcomponentistheoutputofindustry1incountryj,12andwhosesecondcomponentisthenumberofvarietiesofthedifferentiatedproductinsector2producedincountryj.Thenfull-employmentofresourcesineachcountryrequiresthatjjjAY=V,whereVisthevectorofendowmentsincountryj,j=1,2.Letusassumeinitiallythatthefixedcostsofaandvariablecostsofbyforeachdifferentatedgoodhavetobeperformedinthesamecountry.Thentheintegratedworldequilibriumcanbeillustratedinafashionverymuchanalogoustothetwo-factorHOmodel.In121Figure11.3,theFPEsetisshownbytheparallelogram0Q0Q’0.StartingatanyendowmentjpointsuchasVwithinthisparallelogram,freetradeingoodsallowstheeconomiestoconsume 11-14Feenstra,AdvancedInternationalTrade22LO2HQ’jVjB’ADQC’1HCB11OLFigure11.3 11-15Feenstra,AdvancedInternationalTradejatpointsDthatareproportionalacrosscountries,sothatthefactorsembodiedinconsumptionj7jjADliealongthediagonalofthebox.TheslopeofthelinebetweenVandAD(inabsolutevalue)istheequilibriumratiooffactorpricesw/q.Letusassumethatindustry1,producingthehomogeneousproduct,isunskilled-laborintensive,sothata1L/a1H>a2L/a2H=by/a.Thenfromtheendowmentandconsumptionpointsjwecanreadilycomputethetradebalanceineachindustry.Thus,fromtheendowmentpointV,11country1usestheamountofunskilledlabor0Binindustry1,and0B’inindustry2.Similarly,j1fromtheconsumptionpointAD,country1usestheamountofunskilledlabor0Cinindustry1,1and0C’inindustry2.Itfollowsthatcountry1isanimporterofgood1(OBOC’).Inaddition,thereisintra-industrytradeinthedifferentiatedgoodssector.Nowletuschangeourassumptionsandsupposethatthefixedcostsofaskilledlaborandvariablecostsofbyunskilledlaborforeachdifferentatedgoodcanbebeperformedindifferentcountriesbyamultinationalfirm.Thenhowdoestheabilityto“fragment”theheadquartersandproductionactivitiesacrosscountriesaffecttheFPEset?Aquick(andentirelycorrect)answercanbeobtainedbyappealingtotheFPEsetwithmultiplegoodsdevelopedbyDixitandNorman(1980),asdiscussedinchapter3.Ranktheactivitiesbeginningwiththevariablecostsofproducingthedifferentiatedgood,whichusesonlyunskilledlabor;followedbytheproductionofthehomogeneousproductwhichusesbothfactors;followedbythefixedcosts7NotethattheconsumptionineachcountryequalsDj=[dj,(N1+N2)sj]',wherethefirstcomponentisthe122quantityconsumedincountryjofthehomogeneousgood,andthesecondcomponentistheworldvarietyofthedifferentiatedgoodtimescountryj’sshareofworldGDP. 11-16Feenstra,AdvancedInternationalTrade22R’LOV”n2VHQ’S’mV’VAD’SQ1C’HC11OLRFigure11.4 11-17Feenstra,AdvancedInternationalTradeofproducingthedifferentiatedgood,whichusesonlyskilledlabor.Thisgivesthreeratherthantwofactorrequirementsvectors,whichare(by,0),(a1L,a1H),and(0,a).Multiplyingeachofthesebythetotalworldwideamountofeachactivityintheintegratedequilibrium,andsumming121them,wewillobtaintheFPEset0RS0R’S’0inFigure11.4.ThisFPEregionisclearlylarger121thantheoriginalparallelogram0Q0Q’0.TounderstandhowFPEisachievedinthisexpandedset,supposethatthefactorendowmentsofthetwocountriesaregivenbyV’inFigure11.4.ThispointliesoutsidetheoriginalFPEset,sothatanyfirminsector1wouldchoosethecountrywithlowestwageofskilledlaborforheadquarters,andthecountrywithlowestwageofunskilledlaborforproduction.Sincecountry1isskilled-laborabundant,wecanpresumethatitwouldhavethelowerrelativewageofskilledlabor,andfirmsfromeithercountrywouldwanttoestablishheadquartersthere.Thiswillincreasethedemandforskilledlaborincountry1,tothepointwheretheentireendowmentV’canbeemployedatthesamefactorpricesasprevailingabroad.Intuitively,thisiswhatallowsFPEtoholdintheexpandedset.Becausetherearethreeproductionactivitiesandonlytwofactors,thereareactuallymanycombinationsoftheactivitiesacrosscountriesthatareconsistentwithfull-employmentandFPE.AsinHelpman(1984c),wewillfocusonthosecombinationsthatleadtoaminimumamountoftrade.WithfactorendowmentsgivenbyV’inFigure11.4,country1couldproduce1varietiesofthedifferentiatedgoodusingresourcesshownby0C’,andthendevoteitsremainingendowmentofskilledlabor,shownbyC’V’,toheadquartersservices.Theseheadquartersmserviceswouldbecombinedwithunskilledlaborincountry2,oftheamountV’V,inordertoproducevarietiesofthedifferentiatedproductinthatcountry.Additionalresourcesincountry2, 11-18Feenstra,AdvancedInternationalTrademoftheamountVQ’,wouldalsobedevotedtoproducevarietiesofthedifferentiatedproduct,2andtheremainingresourceofQ’0wouldbeusedtoproducethehomogeneousproductinsector1.Factorpriceequalizationisachievedthroughcombiningskilledlaborforheadquartersservicesincountry1withunskilledlaborforproductionincountry2.Whatisthetradepatternimpliedbythemultinationalactivity?Forfactorendowmentsin1theregion0S’Q’,aswehavebeendiscussing,thehomogeneousgoodisproducedonlyincountry2,socountry1importsthatgood.Butforthedifferentiatedgood,thetradepatternisnotsoclear.Indeed,inthecasewehaveillustratedinFigure11.4,itturnsoutthatbothcountrieshavebalancednettradeinthedifferentiatedgood.Thiscanbeseenbycomparingtheresources1usedinproductionandconsumption.Asdiscussedjustabove,country1usestheresources0C’toproducevarietiesofthedifferentiatedgood.Todetermineitsconsumption,wecantracealinefromitsendowmentpointV’toitsconsumptionoffactorsAD’(thislinehastheslopew/q).We11thendecomposethepointAD’intotheconsumptionoffactors0Cinsector1,and0C’insector12.Noticethatforthecasewehaveillustrated,theconsumptionoffactors0C’insector2isidenticaltothoseusedintheproductionofdifferentiatedvarietiesbycountry1.Itfollowsthattradeisbalancedinthatsector,andthesamewillholdforcountry2:theyhavebalancedintra-8industrytradeinthedifferentiatedgood.Howcancountry1beanimporterofthehomogeneousgood,andyethavebalancednettradeinthedifferentiatedgood?Theansweristhatthiscountryreceivesprofitsfromits8ItfollowsthatforslightlydifferentchoicesoftheendowmentpointV’,country1couldbeeitheranetexporteror1anetimporterofthedifferentiatedgoodintheregion0S’Q’. 11-19Feenstra,AdvancedInternationalTrademultinationalsengagedinproductionabroad,whichwouldshowupasinvestmentincomeinthecurrentaccount.Thus,itsbalanceoftradedeficit(duetoimportingthehomogeneousgood)isoffsetbyreceiptsfromitssubsidiariesabroad,sothatithasanoverallcurrentaccountbalance.Theamountofprofitsearnedincountry2willbeproportionaltotheunskilledlaboremployedmtherebythemultinationals,whichismeasuredbyL*ºV’V.Withbyunitsofunskilledlaborneededpervariety,thenumberofvarietiesproducedbymultinationalsincountry2isthereforeN*ºL*/by.Thevalueofheadquartersservicesprovidedbythesemultinationalsthrough9employingskilledlaborincountry1isthenaN*=aL*/by.Thisisan“invisibleexport”bycountry1,whichisusedtopayforitsimportsofthehomogeneousgood.AswemovetootherregionsoftheFPEset,thetradepatternschange.Forexample,2considerendowmentwithin0R’S’Q’,suchasV”.Onceagain,multinationalswithheadquartersincountry1willusesomeoftheunskilledlaborincountry2sothatFPEisachieved.Inthiscase,theamountofunskilledlaboremployedincountry2bymultinationalsisn1nthedistanceV”V.Wecanthinkofthevector0Vastheemploymentofresourcesbyfirmsthat1haveheadquartersincountry1,asdistinctfromthatcountry’sendowmentvector0V”.Intheabsenceofmultinationals,theemploymentofresourcesmustequalacountry’sendowment,but2nowtheycandiffer.Wecanalsoconstructthetradepatternsintheregion0R’S’Q’.Itturnsoutthecountry1canbeeitheranimporteroranexporterofthehomogeneousproductinthis10region,andalsoexportsthedifferentiatedproductaswellasheadquartersservices.9NoticethattheskilledlaborofaN*=aL*/byincountry1isequivalentlymeasuredbythedistanceC’V’.10Helpman(1984c)carefullysolvesforthetradepatterninthisregion,aswellasshowinghowendowmentsaffectthevolumeoftradeinallregionsoftheFPEset. 11-20Feenstra,AdvancedInternationalTradeSettingasidethedetailsofthetradepatterns,themostimportantmessagetotakefromFigure11.4isthatverticalmultinationalsenlargetheFPEset.ThisisadeepinsightfromHelpman(1984c),andcanbecomparedtoourresultsinchapter4,wherewediscussedtradeinintermediateinputsandwages.Wesuggestedtherethattheabilitytofragmenttheproductionprocess,andpurchasesomeunskilled-laborintermediateinputsfromanothercountry,couldleadtoanincreaseintherelativedemandandwagesofskilledlaborinbothcountries.Incomparison,wehavearguedabovethatsuchoutsourcingbymultinationalswillleadtoanenlargementoftheFPEset,andtherefore,atendencyforfactorpricestoequalizeacrosscountries.Thesetworesultsarenotnecessarilyinconsistent,butitisworththinkingabouthowtheassumptionsofHelpman(1984c)differfromthoseinchapter4.Inthemodelofchapter4withacontinuumofinputs,weshowedthatanincreaseinoutsourcingwouldoccurthroughthe“borderline”activitybeingtransferredfromthehometotheforeigncountry:thatactivitywaslessskill-intensivethananyotheractivityperformedathome,butmoreskill-intensivethananyotheractivityperformedabroad.Therefore,therelativedemandforskilledlaborroseinbothcountries,asdiditsrelativewage.Incontrast,inourdiscussionofHelpman’s(1984)modelabove,wetreatedtheproductionofthedifferentiatedgoodasusingonlyunskilledlabor,withheadquartersservicesusingonlyskilledlabor.Theproductionactivityisthereforelessskill-intensivethanthehomogeneousgoodproducedineithercountry.Thetransferofmultinationalproductionfromcountry1tocountry2raisestherelativedemandforskilledlaborincountry1,butlowersitincountry2.Thiswillraisetherelativewagesofskilledlaborincountry1,butloweritincountry2,therebymovingtheserelativewagestowardsequality,sothatFPEisachieved. 11-21Feenstra,AdvancedInternationalTradeWeseethattheassumptionsonfactor-intensitiesdifferbetweenchapter4andwhatwe11haveusedtodescribeHelpman’s(1984)model.Obviously,otherassumptionsonfactor-intensitiescanbeincorporatedintoHelpman’smodel,aswellasalternativeassumptionsonfixedcostsandtradebarriers,asweshallexploreinthenextsection.HorizontalMultinationalsInHelpman’smodel,themultinationalwillneverchoosetooperateproductionfacilitiesinbothcountries,becausewithfreetrade,thereisnobenefittodoingso.Butsupposethattherearesometradebarriers,suchasthe“iceberg”transportcostsdiscussedinchapter5.Supposefurtherthateachplanthassomespecificfixedcosts,whichmustbeborneinthesamecountryasproductioncosts.ThenwewillsaythathorizontalFDIoccursifafirmchoosestooperateproductionfacilitiesinmultiplecountries,witheachplantsellinglocallytothatmarket.HorizontalFDIrequiresthatthegainsfromsellinglocallyandtherebyavoidingtransportcostsmorethanoffsettheplant-specificfixedcosts.ThiscasehasbeenanalyzedinaseriesofpapersbeginningwithMarkusen(1984)andculminatinginthebookbyMarkusen(2002).Thesewritingscoverawholerangeofverticalandhorizontalmultinationals,usingunskilledandskilledlaborinvaryingproportions.Wewillnotattempttosummarizealltheseresults,butshalldeveloponeexampleofhowhorizontalmultinationalscanarise.Wesupposethatafirmincountryiisalreadyproducingforitshomemarket,andisdecidingwhethertoexporttocountyjorestablishaproductionfacilityandsellthere.Thisfirmhasalreadycovereditsfixedcostsofheadquartersservices,aswellasanyplant-specificfixed11Helpman(1984c)actuallyusesmoregeneralassumptiononfactorintensities,buttoillustratetheFPEregions,focusesonthecasewherethefixedcostsuseonlyskilledlabor(whichhecallsa“generalpurpose”inputH),andthevariablecostsuseonlyunskilledlabor.Thesearethesameassumptionswehaveusedabove.Inrelatedtheoreticalwork,Helpman(1985)allowsformultinationalstoproducemultipleproducts. 11-22Feenstra,AdvancedInternationalTradecostsincountryi,andthereisanadditionalfixedcostofestablishingaplantabroad.Weijsupposethereare“iceberg”transportationcostsbetweenthecountries,sothatT>1unitsoftheigoodmustbeexportedfromcountryiinorderforoneunittoarriveinj.Weletpdenotethef.o.b.pricereceivedbythefirmincountryi.Inclusiveofthetransportationcosts,thec.i.f.priceijijiabroadisp=Tp.Weborrowthedemandstructurefromourtreatmentofmonopolisticcompetitioninchapter5.WithcountryjhavingaCESutilityfunctionoverthedifferentiatedproduct,itsijdemandcforavarietysentfromcountryiis:ijijj-sjjc=(p/P)(Y/P),(11.2)jjYiscountryj’sGDPandPreferstocountryj’soverallpriceindex,definedas:1/(1-s)Pj()CNi(pij)1-s=∑i=1,(11.3)iwhereNisthenumberofproductsexportedbyeachcountryi=1,…,C.Wetreatlaborastheonlyfactorofproduction,anddonotdistinguishbyskilltype.TheiiexportingfirmincountryihastheproductionfunctionL=by,wherebisitsmarginallaborcosts.Notethatitsplant-specificfixedcostsarealreadycoveredbyvirtueofitshomesalesofijijiithisproduct.Revenuereceivedfromproducingincountryiandexportingispc=py,whereiijijoutputisrelatedtoconsumptionbyy=cT.Theprofitsfromexportingarethen:iiiiiiijijpy-wby=(p-wb)Tc.(11.4) 11-23Feenstra,AdvancedInternationalTradeSubstitutingfortheCESdemandfrom(11.3),andmaximizing(11.4)overthechoiceofthepriceijijiiip=Tp,wereadilyobtaintheoptimalpricep[1–(1/s)]=bw.Substitutingthisbackinto(11.4),weobtainthemaximizedvalueofprofits:-s1-spiyipiTijpiTijYj1piTijpiyi-wibyi===Yj,(11.5)ssPjPjsPjiijijwherethesecondequalityfollowsusingconsumptionfrom(11.2)withy=cT,andthethirdbysimplification.From(11.5),weseethathighertransportcostsreduceprofitsfromexporting,whilehigherGDPabroadincreaseprofits.Ratherthanexportingfromcountryi,themultinationalfirmcaninsteadestablishaproductionfacilityincountryj,andselllocally.Inthatcase,itwillfacetheproductionfunctionjjjL=a+by,includingtheplant-specificfixedcostspaidabroad.Forsimplicitywetreatmarginallaborcostsbasequalacrosscountries,thoughthewagesandfixedcostscandiffer.jTherearenotransportcostswhenproducingincountryjandsellinglocally,sowiththepricep,thequantitysoldis,jjj-sjjc=(p/P)(Y/P).(11.2’)jjAgain,theoptimalpriceisdeterminedasp[1–(1/s)]=bw.Usingthisand(11.2’)tocomputeprofitsearnedfromthesubsidiaryincountryj,weobtain:1-s1pjYjpjyj-wj(byj-aj)=-wjaj.(11.5’)sPjPj 11-24Feenstra,AdvancedInternationalTradeThus,inorderforthemultinationalbasedincountryitowishtoestablishasecondplantincountryj,itmustbethecasethat(11.5’)isatleastaslargeas(11.5):1-s1-s1pj1piTijYj-wjaj³Yj.(11.6)sPjsPjConversely,inordertobeahorizontalmultinational,itmustbethecasethatitisnotprofitabletoshutdownitsplantincountryiandsimplyexportfromthesingleplantincountryj.Theconditiontoensurethatis:1-s1-s1pi1pjTjiYi-wiai³Yi,(11.6’)sPisPiiwhereaisitsfixedcostsincountryi.Theleft-handsideof(11.6’)istheprofitsfromoperatingitsplantincountryi,whiletheright-handsideistheprofitsfromexportingfromcountryjtoi.Soinordertohaveahorizontalmultinationaloperatingplantsinbothcountries,both(11.6)and(11.6’)musthold.Theseconditionscanbere-expressedbymovingthefixedcoststotheright-side,multiplyingeachequationbysandmultiplyingthemtogether,inwhichcaseweobtainasinglenecessaryconditiontoensurethatthemultinationalmaintainsplantsinbothcountries:1-spipj[]1-(TijTji)1-s(YiYj)PiPj1-s1-spiTijpjTji³wiaiYj+wjajYi+s2wiaiwjaj.(11.7)PjPiThisconditionismorelikelytoholdinthefollowingregionsofparameters: 11-25Feenstra,AdvancedInternationalTradeTheorem(MarkusenandVenables,1998,2000;Markusen,2002)Horizontalmultinationalswillbemorelikelywhen:(a)transportcostsarehigher;(b)plant-specificfixedcostsarelower;(c)GDP’sarehigherormoresimilaracrosscountries.12Conditions(a)and(b)ofthistheoremfollowbyinspectionofcondition(11.7).Condition(c)requiresomeadditionalexplanation.NoticethatproductofGDP’scanbere-writtenas:ijij2ijij2i2j2YY=(Y+Y)ss=(Y+Y)[1–(s)–(s)]/2,(11.8)iiijjjijijwheres=Y/(Y+Y)ands=Y/(Y+Y)arecountyiandj’ssharesofGDP,with(s+s)=1.ij2i2j2ijItfollowsthat1=(s+s)=(s)+(s)+2ss,fromwhichthelastequalityin(11.8)isi2j2obtained.Inchapter5,wereferredto[1–(s)–(s)]asa“sizedispersionindex”ofthetwocountries,andittakesitshighestvalue(ofone-half)whenthecountriesarethesamesize.ijijDenotingthisindexbyD,wecandivide(11.8)by(Y+Y)andrewriteitas:1-sDijpipj[]1-(TijTji)1-s(Yi+Yj)2PiPj1-s1-spiTijpjTjis2wiaiwjaj³wiaisj+wjajsi+.(11.7’)PjPi(Yi+Yj)12ijjiNoticethattheleft-handsideof(11.7)isincreasinginthetransportcostsTorT,whiletheright-handsideisijdecreasing.Inaddition,therightisincreasinginthefixedcostsaora. 11-26Feenstra,AdvancedInternationalTradeijWeseethattheleftof(11.7’)islargerwhenthesumofGDP’s(Y+Y)islarger,andtherightissmaller.Inaddition,theleftislargerwhenthesizedispersionindexishigher,meaningthatthe13countriesareofsimilarsize.Undertheseconditions,(11.7’)ismorelikelytohold.MarkusenandVenables(1998)andMarkusen(2002,chap.5)demonstratetheabovetheoreminamodelwherefirmsproduceahomogeneousgoodunderCournot-Nashcompetition,whileMarkusenandVenables(2000)andMarkusen(2002,chap.6)usemonopolisticcompetitionwithaCESutilityfunction.Inbothcases,themultinationalsemployhighandlow-skilledlabor.SoinadditiontothesimilarityofGDP’sthatweobtainincondition(c),theyalsofindthathorizontalmultinationalsaremorelikelytoholdwhentherelativeendowmentsofhighandlow-skilledworkersaresimilar.Noticethatthisistheoppositeoftheresultobtainedfrom12Helpman’s(1984c)model:sinceverticalmultinationalsoccurintheregions0S’Q’and0R’S’Q’ofFigure11.4,thisrequiresthattherelativeendowmentsofskilled/unskilledworkersaresufficientlydifferent.EstimatingtheKnowledge-CapitalModelTheverticalandhorizontalmultinationalscanbecombinedinaunifiedframework,whichMarkusen(2002,chaps.7,8)referstoasthe“knowledge-capital”model.Thistermisusedbecausetheknowledgecreatedattheheadquartersisutilizedatplantsinmultiplecountries:ithasthenatureofapublicgoodwithinthefirm.Thisistheintangibleassetthatweidentifiedintheintroductionofthechapterasessentialtoanymultinationalfirm.Itisassumedthattheratioofskilled/unskilledlaborusedinheadquartersservicesexceedsthatusedinplant13Changesinthecountryshareswillalsoaffecttherightof(11.7’),butonlybyasmallamountiffixedcosts,transportcosts,andwagesaresimilaracrosscountries. 11-27Feenstra,AdvancedInternationalTradeproductionbymultinationals,whichinturnexceedsthatusedinproductionofthehomogeneouscompetitivegood.ThenMarkusen(2002,chap.8)findsthatincreasedinvestmentbymultinationalsraisestherelativewageofskilled/unskilledworkersintheskill-abundantcountry,andpossiblyintheothercountryaswell.Thereasonforthisisthattheskill-abundantcountryspecializesmoreinheadquartersservices,whiletheothercountryspecializesmoreinproductionbymultinationals,whichraisetherelativedemandforskilledlaborinbothcountries.Thisoutcomeappliestoeitherhorizontalorverticalinvestment,andisverysimilartowhatwe14obtainedfromoutsourcinginchapter4.Theknowledge-capitalmodelhasbeenestimatedbyCarr,MarkusenandMaskus(2001),andhypothesisteststodistinguishthehorizontalversusverticalmodelareperformedby15MarkusenandMaskus(1999,2002)aswellasBlonigen,DaviesandHead(2002).Thesepapersdiffersomewhatintheirprecisespecifications,andwewillcombinetheirvariousresults16inourdiscussionbelow.Thedependentvariableusedistherealsalesofforeignaffiliatesofmultinationalfirms,includingboththeirlocalandexportsales.Forhorizontalmultinationalsthelocalsalesaremostimportant,whileforverticalmultinationaltheexportsales(backtotheparentcorporation)aremostimportant.TheequationestimatedbyCarr,MarkusenandMaskus(2001)is:14Noticethattheskilled/unskilledlaborusedinplantproductintheknowledgecapitalmodelliesin-betweenthefactorintensitiesofheadquartersservicesandofthehomogeneouscompetitivegood.Thisissimilartofactorintensityofthe“borderline”activitythatwastransferredbetweencountriesintheoutsourcingmodelofchapter4,butdiffersfromtheassumptionofHelpman(1984c).15Brainard(1997)alsoestimatesahorizontalversusmultinationalsmodelinwhichfirmstrade-offthebenefitsfrombeingclosetocustomersagainstthebenefitsofconcentratingproductioninonelocation.16Intheestimationof(11.9),Blonigen,DaviesandHead(2002)measuretheskilldifferenceandGDPdifferenceasabsolutevaluesinstead.TheyarguethattheabsolutevaluesareclosertothespecificationinMaskusandMarkusen(2002)andMarkusen(2002,chap.12),whoallowforseparatecoefficientsontheskilldifferencedependingonitssign,alongwithseparatecoefficientsonitsinteractionwiththeGDPdifference. 11-28Feenstra,AdvancedInternationalTradeijij2ijRealAffiliateSales=b0+b1(GDP+GDP)+b2(GDP–GDP)+b3(Skill–Skill)ijijjjjij2+b4(GDP–GDP)(Skill–Skill)+b5(a)+b6(t)+b7(t)(Skill–Skill)iij+b8(t)+b9Distance,(11.9)wherecountryireferstotheheadquartersofamultinationalandcountryjtothelocationofitsaforeignaffiliate.Thefirstvariableontherightof(11.9)isthesumofcountryGDP’s,whichweexpecttobepositivelyrelatedtothepresenceofhorizontalmultinationals,andthereforerealaffiliatesales(b1>0).Likewise,thesecondvariableisthesquareddifferenceinrealGDP’s,whichmeasuresthesizedispersion,andwhichweexpecttobenegativelyrelatedtohorizontalmultinationals(b2<0).Whataboutforverticalmultinationals?InthemodelofHelpman(1984c),neitherthesumnorthedifferenceofGDP’sdeterminesthemagnitudeofaffiliateexportsbacktotheparentcountry;rather,itistherelativefactorendowmentsthatdeterminesthemagnitudeoftheseexports.Forexample,inFigure11.4,theaffiliateexportsareproportionaltothelinesegmentsnnV’VandV”V,whichequaltheamountofunskilledlaborusedforaffiliateproduction.ThepositionoftheendowmentpointsV’andV”relativetotheoriginalFPEsetiswhatdeterminestheseexports,notthecountryGDP’s.Thus,onetestofthehorizontalversustheverticalmodelistocomparetheexpectedcoefficientsb1>0andb2<0forthehorizontalcase,withb1=b2=0fortheverticalcase.Theestimationof(11.9)isperformedwithadatasetoninboundandoutboundU.S.foreigndirectinvestment,from1986–1994.Foreachcountrypair,eitheriorjdenotestheUnitedStates,withtheotherindexbeingthelocationofeitheraU.S.affiliate,oraforeignparentcompanywithan 11-29Feenstra,AdvancedInternationalTradeaffiliateintheU.S..Thehypothesesthatb1>0andb2<0arestronglyconfirmedintheestimation,whichsupportsthehorizontalovertheverticalmultinationalmodel.Thenexttwovariablesin(11.9)arerelatedtotheskilldifferencesbetweentheparentandihostcountry,whereSkillismeasuredbytheshareofthelaborforceincertainskilledoccupations.Inthehorizontalmodel,itisexpectedthatinvestmentandaffiliatesalesaredecliningintheendowmentdifferencesofthecountries,andthiseffectismorepronouncedwhenthecountryGDP’salsodiffer.Thus,thehypothesesb3<0andb4<0areobtainedfromthehorizontalmodel.Incontrast,withHelpman’smodelverticalinvestmentishigherwhentheendowmentsofthecountriesdiffermore,sothatb3>0ispredicted.Markusen(2002,chap.12)alsofindsinhissimulationsthatcountriesthatarebothsmallandskilled-laborabundantrelativetotheirtradingpartners,suchastheNetherlands,havethegreatestconcentrationofheadquartersforverticalmultinationals.Thus,theinteractionbetweenrelativecountrysizeandtheskillendowmentispositivelyrelatedtoverticalinvestment,sothatb4>0also.Intheestimationof(11.9),thesignpatternb3<0andb4<0isstronglyconfirmed,soonceagain,thehorizontalmodelisacceptedandtheverticalmodelisrejected.Thisoutcomeisperhapsnotsurprisinginviewofthefactthatthemajorityofforeigninvestmentsworldwidearebetweenindustrialcountries,andnotbetweendevelopedanddevelopingcountries(Markusen,1995,andLipsey,1999,2001).Theseinvestmentswouldbemoretypicalofhorizontalratherthanverticallinkages.ThistendencycanbeseenfromthedataonthestocksandflowsofglobalFDIover1980-95,asshowninTables11.2and11.3.LookingfirstattheinwardstockofFDIintheupper-halfofTable11.2,theproportionofFDIlocatedindevelopingcountriesfluctuatedbetween19%and26%over1980-95,sothe 11-30Feenstra,AdvancedInternationalTradeTable11.2:FDIStock,1980-1995InwardStock($Billion):1980198519901995TotalInwardStock481.9734.91716.92657.9DevelopedEconomies373.6538.01373.31932.7DevelopedShareas%oftotal77.5%73.2%80.1%73.9%U.S.InwardStock83.1184.6394.9564.6U.S.ShareofDevelopedStock22.2%34.3%28.8%29.2%DevelopingEconomies108.3196.8341.7693.3DevelopingShareas%oftotal22.5%26.8%19.9%26.1%ChinaInwardStock0.03.414.1129.0ChineseShareofDevelopingStock0.0%1.7%4.1%18.6%OutwardStock($Billion):1980198519901995TotalOutwardStock513.7685.61684.12730.2DevelopedEconomies507.5664.21614.62514.3DevelopedShareas%oftotal98.8%96.9%95.9%92.1%U.S.OutwardStock220.2251.0435.2705.6U.S.ShareofDevelopedStock43.4%37.8%27.0%28.1%DevelopingEconomies6.221.269.4214.5DevelopingShareas%oftotal1.2%3.1%4.1%7.9%ChinaInwardFlow0.00.12.517.3Chinese%ofDevelopingStock0.0%0.6%3.6%8.1%Source:UnitedNations(1996,AnnexTables3,4;pp.239-248);citedinFeenstra(1999). 11-31Feenstra,AdvancedInternationalTradeTable11.3:FDIFlow,1983-1995FDIInflow($Billion):1983-88*1989199019911992199319941995Totalinflow91.6200.6203.8157.8168.1207.9225.7314.9DevelopedEconomies71.8171.7169.8114.0114.0129.3132.8203.2DevelopedShareas%oftotal78.4%85.7%83.4%73.8%70.0%64.8%61.4%68.4%U.S.Inflow34.467.747.922.017.641.149.860.2U.S.ShareofDeveloped47.9%39.4%28.2%19.3%15.4%31.8%37.5%29.7%InflowDevelopingEconomies19.828.633.741.350.473.187.099.7DevelopingShareas%of21.6%14.3%16.6%26.2%30.0%35.2%38.6%31.6%totalChinaInflow1.83.43.54.411.227.533.837.5Chinese%ofDeveloping9.2%11.8%10.3%10.6%22.2%37.6%38.8%37.6%InflowFDIOutflow($Billion):1983-88*1989199019911992199319941995TotalOutflow93.7217.9240.3210.8203.1225.5230.0317.9DevelopedEconomies88.3202.3222.5201.9181.4192.4190.9270.6DevelopedShareas%oftotal94.2%92.8%92.6%95.8%89.4%85.4%83.2%85.2%U.S.Outflow14.225.727.233.539.069.045.695.5U.S.ShareofDeveloped16.1%12.7%12.2%16.6%21.5%35.9%23.9%35.3%OutflowDevelopingEconomies5.415.617.88.921.633.038.647.0DevelopingShareas%of5.8%7.2%7.4%4.2%10.6%14.6%16.8%14.8%totalChinaOutflow0.50.80.80.94.04.42.03.5Chinese%ofDeveloping8.5%5.0%4.7%10.3%18.5%13.3%5.2%7.4%Outflow*AnnualAverageSource:UnitedNations(1995,1996,AnnexTables1,2);citedinFeenstra(1999). 11-32Feenstra,AdvancedInternationalTrademajorityofFDIisindevelopedcountries.Therewasasurgeofinvestmentintothedevelopedcountriesduringthesecondhalfofthe1980s,duringwhichtimethestockofinvestmentindevelopedcountriesnearlytripledfrom$538billionto$1,373billion.Atthesametime,therewasaconsiderableincreaseinforeigninvestmentintotheU.S.,whichdoubledfrom1980to1985andagainfrom1985to1990.Since1990,thestockofinvestmentlocatedinthedevelopingcountrieshasgrownmorerapidly,whichisinlargepartduetoincreasedFDIintoChina.Thiscountryaccountsfor18.6%oftheinwardstockofdevelopingcountriesin1995,upfrom4.1%justfiveyearsearlier.ThevastmajorityofFDIenteringintodevelopedanddevelopingcountriesalikecomesfromthedevelopedcountries,asshowninthelower-halfofTable11.2.Similartothesestockfigures,themajorityofFDIflowsshowninTable11.3aretodevelopedcountries,thoughChinaisalsoalargerecipient.Forexample,in1995theUnitedStatesreceivedthemostFDI,withaninflowof$60.2billion,butChinahasthesecond-largestinflowof$37.5billion.ThesurgeinFDIflowsduringthesecondhalfofthe1980sbothcomesfromandwasdirectedtowardsthedevelopedcountries:thisflowreached$172billionin1989.Thiswasfollowedbyafallindirectinvestmentmagnitudesfrom1990to1991,witharecoverythatwasslowatfirstbuthasincreasedrecentlytoreach$203billionin1995.TheinflowofinvestmentintoChinagrewmostdramaticallyin1990-92,andagainin1994-95.Fromthisverybriefexaminationofthedata,itisapparentthatthemajorityofFDIflowsisstillbetweentheindustrialcountries,butthatanincreasingfractionisbeingdirectedtowardsthedevelopingcountries,especiallyChina.Therearetwomajorproblemswithusingsuchdatatotestmodelsofmultinationals:first,thetypesofinvestmentshavebeenchangingduringthe 11-33Feenstra,AdvancedInternationalTrade1990’s;andsecond,linkagesbetweenfirmscancomefromcontractualarrangementsacrosscountriesthatneednotbereflectedinmultinationalinvestments.Onthefirstproblem,Hanson,MataloniandSlaughter(2001)useamorerecentanddetaileddatasettoevaluatethetypesofinvestmentsmadebyU.S.multinationals.Thesedataincludedetailedinformationonthevariousrolesplayedbyforeignaffiliates:inimportinginputsfromtheirU.S.parentsforfurtherprocessing;insellinglocallywithinthehostcountries;andinexportingfromthehostcountries.Thefirstoftheserolesillustrateverticalinvestmentwhilethesecondishorizontalinvestment,andthethirdcouldbeeither.Theseauthorsfindthatbothtypesofinvestmentsareimportantinthedata,withagrowingshareofverticalinvestmentsinsomeindustriesandregions.Itisquitepossiblethatpoolingthedataacrossindustriesandcountriesasin(11.9)willhidetheunderlyingpresenceofbothactivities.Onthesecondproblem,thereisnowaythatdataonmultinationalswillreflectthevastamountofoutsourcingbetweenunaffiliatedfirms.Thisisespeciallytrueofverticalrelationships.NikemanufacturesallofitsrunningshoesinAsia,butdoesnotowntheseplants;nevertheless,thisisaverticalrelationshipverymuchinthespiritofHelpman’s(1984c)model.Soweshouldnotexpectdataonmultinationalsalonetogivethecompletepictureofhorizontalandverticalrelationshipsbetweenfirms.Estimatingthedeterminantsoftheseinadatasetthatallowsforbothisthereforeachallengeforfutureempiricalresearch.Butitisalsoachallengefortheoreticalwork:wewouldliketoexplainthedivisionbetweenactivitiesthatarekeptinternaltothemultinational,andactivitiesthatexternaltothefirm.Thiswillbediscussedintheremainderofthechapter. 11-34Feenstra,AdvancedInternationalTradeOrganizationoftheFirmBoththeverticalandhorizontalmultinationalmodelsinvolvedafirmdecidingwhethertoopenanadditionalplantinaforeigncountry,butneitherofthesemodelsaddressedthequestionofwhethertheforeignplanthadtobeownedbythemultinationalornot.Whatwouldpreventthehomefirmfromenteringintoajointventureorsomeothercontractualarrangementwithaforeignfirmratherthanestablishinganewplant?Inordertoexplainthis,weneedtoaddaricherstructuretothemodelswehaveconsideredsofar.Weshalldiscusstwoexplanationsfortheintegrationoffirms:transactionscostsunderincompletecontracts;andthe“matching”of17firmsunderincompleteinformation.TransactionsCostsAgeneralreasonfortheexistenceoffirmsprovidedbyRonaldCoase(1937)isthattherearecostsinvolvedwithmakingtransactionsonthemarkets,andwhenthesecostsbecometoohigh,theactivitiesareinsteadinternalizedwithinthefirm.ThisideawasgreatlyelaboratedbyOliverWilliamson(1975,1985),andisassociatedwiththetwinconceptsof“assetspecificity”andthe“holdupproblem.”Toexplainthese,supposethatfirmAneedssomenewinput.ItcouldaskfirmBtoproduceit,butthatwouldbeexpensiveforittodesignandtest.IffirmBmanagedtosuccessfullyproduceit,andthensellittoA,thenwhatguaranteedoesfirmBhavethatitwillbepaidenoughtocoveritscostsofdesign,letaloneitscostsofproduction?Iffirm17Therearealsootherreasonsforfirmtointegrate.Markusen(2001;2002,chaps.13-15)developsseveralprincipal-agentmodelsofthefirmsunderimperfectinformation,inwhichtheparentcompanymustgiveuprentstotheforeignpartnertoobtainitscooperation.Iftheserentsbecometoohigh,thefirmwillinsteadintegrateacrossborders.Inaddition,firmsmayintegratehorizontallyorverticallytoincreasetheirmarketpowerandprofits,asdiscussedintheindustrialorganizationliterature:seePerryandPorter,1985,FarrellandShapiro,1990,Fauli-Oller,2000,andHenrikandPersson,2001,forthehorizontalcase;Dixit,1983,MathewsonandWinter,1983,PerryandGroff,1985,Ordover,SalantandSaloner,1990,1992,HartandTirole,1990,Abiruetal,1998,andChen,2001,fortheverticalcase;aswellasthesurveybyPerry,1989,andtheworkofFeenstra,HuangandHamilton,2003,thatcombineshorizontalandverticalintegrationwithin“businessgroups”anddiscussestheimplicationsfortrade. 11-35Feenstra,AdvancedInternationalTradeAandBcouldcontractinthebeginningtospecifywhattheinputwouldcosttodevelopandproduce,thenthatwouldresolvetheproblem,butitmaybedifficulttoforeseeinthebeginningwhatthesecostswouldbe.Intheabsenceofcompletecontracts,firmBwillbeleftwithaninputwhichmayormaynotbeusefultobuyersotherthanfirmA:thisiscalled“assetspecificity.”WecanpresumethatthebettersuitedistheinputtofirmA’sneeds,thenthelessvaluableitistoalternativebuyers.Butinthatcase,firmBhasverylittlebargainingpowerinrefusingtoselltheinputtofirmA:thisisthe“holdupproblem.”Becauseoftheholdupproblem,firmBmightdecidetoproduceaninputthatisnotverywellsuitedtofirmA,butinsteadcanbesoldtootherpurchasers.Sotheincompletecontractsleadtoaninefficientoutcome.Ifthisinefficiencyistoogreat,thenfirmAmightdecidetojustdesigntheinputitself;i.e.itwillverticallyintegrateintotheproductionofthatgood,andtherebybecomeabiggerfirm.Fromthislineofreasoning,wecanseethatthedecisionofafirmtointegratedependsonthecostofmakingmarkettransactionsunderincompletecontracts,andinparticular,assetspecificityandtheholdupproblem.TheseinsightsofCoaseandWilliamsonhaveonlyrecentlybeenformalizedintomodelsofthefirm,andkeyreferencesareGrossmanandHart(1986),HartandMoore(1990,1999)andAghionandTirole(1997).Itisevenmorerecentthattheseideashavebeenincorporatedintogeneralequilibriummodelsoftrade,andreferencesincludeMcLaren(2000),GrossmanandHelpman(2002a,b),MarinandVerdier(2001),Antràs(2002)andPugaandTrefler(2002).Thesepapersarenewenoughthatwedonotattempttosummarizetheircontributionsinthisevolvingarea,buthighlyrecommendthemforfurtherreading. 11-36Feenstra,AdvancedInternationalTradeMatchingofFirmsOursecondexplanationfortheintegrationoffirmsistosharetheinformationthateachhasabouttheirownnationalmarkets.Forexample,jointventuresbetweenfirmsindifferentcountriescanbringwiththemknowledgeofthesupplieranddistributionnetworksineachcountry,andanyotherintangibleassetsownedbyeitherfirm.RauchandTrindade(2000)modelthisinformationalissueasaproblemof“matching”firmsacrosscountries,wherethereis18someinherentuncertaintyintheabilitytofindagoodpartner.Thestartingpointfortheiranalysisissimilartotheone-sectormodeldiscussedatthebeginningofthechapter,wherecapitalcombineswithlabortoproduceahomogeneousgood.Wenowsuppose,however,thattwoproducersmustformajointventureinordertosuccessfullyobtainoutput.Theseproducersmaybothbeinthesamecountry(asintheautarkyequilibrium)ormaycomefromdifferentcountries(inthetradeequilibriumwithjointventures).Letusinitiallyfocusontheautarkyequilibrium.Theamountofoutputobtainedbytwoproducersdependsontheircomplementaritywitheachother.Forconvenience,wewillarraytheproducersineachcountryaroundacircleofcircumferenceunity,asillustratedinFigure11.5.Letzijdenotetheshortestarc-distancebetweenproducersoftypeiandj,whichmeasurestheircomplementarity.Noticethatthemaximumvalueofzijis1/2.Thenweassumethattheoutputobtainedbyajointventureoftheseproducersisyij=f(Lij,zij),whereLijisthelaborused.Wedonotincludecapitalinthisproductionfunctionsincetheamountavailablefromeachproducerisequalandfixed.We18TheworkofRauchandTrindade(2000)buildsuponthatinCasellaandRauch(2002)andRauchandCasella(2003). 11-37Feenstra,AdvancedInternationalTradej1/2–k/2zijz(w*/w)SuccessfuliInternational1/2Matchesz(w*/w)1/2–k/2Figure11.5 11-38Feenstra,AdvancedInternationalTradeassumethatf(Lij,zij)isincreasingandhomogeneousofdegreeonein(Lij,zij),sothatproducersfartherawayfromeachotherobtainhigheroutput.Iflaborisavailableatthewagew,theprofitsobtainedfromthisjointventureare:maxg(w,zij)ºLf(Lij,zij)-wLij,(11.10)ijwherewehavenormalizedthepriceoftheoutputatunity.Underourassumptionthatf(Lij,zij)ishomogeneousofdegreeoneinitsarguments,theng(w,zij)willbehomogeneousofdegreeoneinzij:increasingthedistancezijbetweenproducerswillleadtoanequi-proportionateincreaseinlaborhired,andinoutputandprofits.Itfollowsthatwecanwritetotalprofitsearnedasg(w,zij)=zijp(w),wherep(w)ºg(w,1).Bytheenvelopetheorem,theamountoflaborhiredbythejointventureisLij=–zijp’(w)>0.Increasingthewagemustreducethelaborhired,sothatdLij/dw=–zijp”(w)<0,whichmeansthatp”(w)>0sopisconvex.Forconvenienceinourdiscussion,wefollowRauchandTrindadeinassumingthatp(w)hastheiso-elasticform-g19p(w)=bw,withb,g>0.Thisformwillapplyifthefunctionf(Lij,zij)isCobb-Douglas.Tosolvefortheautarkyequilibrium,weneedtomatchupproducersinasinglecountry.WeassumethatproducersareuniformlydistributedaroundthecircleinFigure11.5,andthatateachpointonthecirclethereisaunit-massofproducers:sothetotaldistributionofthemcanbethoughtofasa“unitcylinder.”Thebestmatchesareobtainedwheneachproducerhasajointventurewiththeonelocatedontheoppositesideofthecircle,sothatzij=1/2.Theprofits19Seeproblem11.4. 11-39Feenstra,AdvancedInternationalTradeearnedfromeachofthesematchesarezijp(w)=p(w)/2.Weassumethattheseareequallydividedbythetwoparticipants,soeachproducerearnsp(w)/4intheautarkyequilibrium.Theequilibriumwageisdeterminedbyequatinglabordemandandsupply.One-halfoftheproducersarematchingwiththeotherhalf(ontheoppositesideofthecircle),sothetotalnumberofmatchesis1/2.Itfollowsthathomeautarkylabordemandis:aa–zijp’(w)/2=–p’(w)/4=L,(11.11)wherethefirstequalityisobtainedbecauseproducersarematchingwiththoseontheoppositesideofthecircle(zij=1/2),andthesecondequalitysetslabordemandequaltotheendowmentLatodeterminetheautarkywagew.Letusnowintroduceaforeigncountry,withthesametechnologybuthavingahighera*laborendowmentL*>L.Underautarky,theforeignwagesatisfies–p’(w)/4=L*,anda*awouldthereforebelowerthanathome:wL,usingtheconvexityofp.Inthetradeequilibriumweallowjointventurestoformbetweenproducersacrosscountries.Weassumethatinsuchajointventure,theproducerscanemploylaborineithercountry:giventhatforeignwagesarelower,theywillchoosetolocatethere.Thiswilltendtoshiftlabordemandawayfromthelabor-scarcehometowardsthelabor-abundantforeigncountry,andthereforediminishthedifferenceintheirwages.Howfarwillthisprocessofwageequalizationproceed?Intheone-sectormodelusedatthebeginningofthechapter,fullequalizationofhomeandforeignwageswouldbeachievedthroughcapitalmobility.Thatis,equalizationoftherentalratesacrosscountriescombinedwithidenticaltechnologiesimpliesthatwagesareequalized, 11-40Feenstra,AdvancedInternationalTrade20too.Thegoalofthematchingmodelistoseehowthisresultismodifiedwhentherearesomefrictionsduetotheinformationproblemsinfindingagoodpartner.Tothisend,supposethatproducerscanenterintoajointventureacrosscountries.Ifthesejointventuresprovetobeunprofitable,thentheir“statusquo”positionistoenterintoajointventuredomestically.Presumingthattheystillmatchwithadomesticpartnerattheoppositesideofthecircle,thenthiswouldresultinstatusquopayoffsofp(w)/4andp(w*)/4foraproducerineachcountry.Noticethattheseareevaluatedattheequilibriumwageswithinternationaljointventures,whichwilldifferfromtheautarkywages.Theinternationaljointventurewillyieldprofitsofzijp(w*),wherezijisthedistancebetweenthehomeandforeignproducersandtheyemploylaborintheforeigncountry.Thetwoproducerswilldividethepayoffzijp(w*)betweenthemselvesaccordingtosomerule,thedetailsofwhichneednot21concernus.Giventhatthetotalpayofftoahomeandforeignproducerwhomatchdomesticallyis[p(w)/4+p(w*)/4],itisapparentthattheinternationaljointventureisequallyprofitableifitoccursbetweenahomeandforeignproducerwhoaredistancezapart,with,zp(w*)=[p(w)/4+p(w*)/4]Ûz(w*/w)=(1/4)[1+p(w)/p(w*)].(11.12)-gUnderourassumptionthatp(w)takesontheiso-elasticformbw,withb,g>0,wecanwritegp(w)/p(w*)=(w*/w)asanincreasingfunctionoftherelativeforeignwage,andlikewiseforz(w*/w).Wewillconfirmbelowthatforeignwagesarelessthanthoseathomeinthetrade20Seeproblem11.5.21RauchandTrindade(2000)assumethatthepayoffzijp(w*)isdividedaccordingtotheNashbargainingsolution. 11-41Feenstra,AdvancedInternationalTradeequilibriumwithjointventures,sothatz<1/2.Recallingthatthemaximumdistancebetweenanytwoproducersis1/2,thenthoseinternationalmatchesintherangezijÎ(z,1/2]willbestrictlypreferredtodomesticmatches.FollowingRauchandTrindade,letusaddsomestructuretotheinternationalmatches.Wewillassumethatwhenahomeproducerisearchesforapotentialforeignpartnerj,thedistancebetweenthemisrandomlydrawnfromauniformdistributiondefinedovertheinterval[1/2–k/2,1/2].Noticethatwhenk=1,thisintervalis[0,1/2],whichmeansthatpartnersarerandomlydistributedovertheentirecircle(recallagainthatthemaximumdistancein1/2).Ontheotherhand,whenk=0thisintervalbecomesthepoint1/2,whichisthebestmatchpossible.SotheparameterkÎ[0,1]indexestheextentofuncertaintyininternationaljointventures.ThisstructureofinternationalmatchesisillustratedinFigure11.5.Foradomesticagentlocatedatthepointlabeledi,thebestmatchwouldbeatthepointlabeled1/2ontheoppositesideofthecircle.Therangeofpossiblematchesareintheinterval[1/2–k/2,1/2]oneithersideofthepoint1/2.Ontheotherhand,thesuccessfulinternationalmatchesmustsatisfy(11.12),22whichmustbeatleastthedistancez(w*/w)awayfromthepointi.WecannowcomputethatprobabilityPthataninternationalmatchissuccessful.Thisissimplythelengthoftheintervalas[z(w*/w),1/2]ascomparedto[1/2–k/2,1/2],or,1/2-z(w*/w)1-2z(w*/w)P==.(11.13)1/2-(1/2-k/2)k22Itcanbeconfirmedthatz(w*/w)mustliewithintheinterval[1/2–k/2,1/2]inequilibrium.Ifthiswerenotthecase(i.e.ifz(w*/w)<1/2–k/2)thenallhomeproducerswouldenterintojointventureswithforeignproducers,andthedemandforhomelaborwouldbezero. 11-42Feenstra,AdvancedInternationalTradeThus,theprobabilityofasuccessfulinternationalmatchishigherwhentheratioofwagesw*/wislower(meaningthatzislower),orwhentheuncertaintykassociatedwithinternationaljointventuresisreduced.Todeterminewagesweuselabormarketequilibriuminbothcountries.Homelaborisdemandedonlybythosedomesticproducerswhoseinternationalmatcheswerenotsuccessful,orafraction(1–P)ofhomeproducers.Thehomedemandforlaboristherefore:–(1–P)p’(w)/4=L,(11.14)whichcanbecomparedto(11.11).Homelabordemandhasfallenfromautarkyduetoaproducersemployinglaborabroad,sothisresultsinafallinthedomesticwage,w0anddw*/dk<0;lim(d)dln(w/w*)/dln(L*/L)<1/(1+g),isincreasingink,anddln(w/w*)/dln(L*/L)=0.k®0Part(b)ofthistheoremisobtainedbyinspectionoftheequilibriumcondition(11.16).Ask®0,thetermmultiplyingtheintegralgoestoinfinity(meaningthatw/w*goestozero),unlessthelowerlimitofintegrationapproaches1/2sotheintegralgoestozero.Theformercaseiseasilyruledoutasanequilibrium,soask®0wemusthavez®1/2inthelowerlimitof23Noticethatz(w*/w)isdecreasingin(w/w*),whichreducesthelowerlimitofintegrationsothevalueoftheintegralontheleftof(11.16)isincreasingin(w/w*).Likewise,Pin(11.13)isincreasingin(w/w*),andthensois1/(1–P)thatappearsontheleftof(11.16). 11-44Feenstra,AdvancedInternationalTradeintegration,whichimpliesthatthewagesapproachequality.Recallthatthiswouldbetheoutcomeintheone-sectormodelusedearlierinthechapterunderperfectcapitalmobility.Therefore,thematchingmodelapproachestheoutcomewithperfectcapitalmobilityastheuncertaintyaboutinternationalpartnersiseliminated.Part(c)ofthetheoremconfirmsthatasinternationaluncertaintyincreases,theequilibriumgetsmonotonicallyfartherawayfromwageequalization:raisingkincreasesdomesticwagesandlowersforeignwages,asfewerjointventuresarepursued.Inpart(d),RauchandTrindadeinvestigatethesensitivityofrelativewagestochangesinthelaborendowments.Noticethatfrom(11.16),ifweheldzconstant(sothatPisalsoconstant),thenwereadilyobtaindln(w/w*)/dln(L*/L)=1/(1+g)>0.Relativewagesareincreasingintherelativelaborendowmentoftheothercountryforfixedz,justastheyareinautarky,becausemorelaborabroadlowersforeignwages.Incontrast,whenwagesareequalizedacrosscountriesthendln(w/w*)/dln(L*/L)=0,asoccursintheone-sectormodelunderperfectcapitalmobilityusedearlierinthechapter.Wecaninterpretthisassayingthatperfectcapitalmobilityleadstocompleteintegrationofthelabormarkets.Thenpart(d)statesthatthematchingmodelleadstogreaterintegrationofthelabormarketsthaninautarky,andmonotonicallyapproachesthecaseofperfectcapitalmobilityasuncertaintykapproacheszero.Therearemanypotentialempiricalandtheoreticalapplicationsofthismatchingmodel.Itiswellknownthatwagesdorespondtosomedegreetochangesinlocallaborsupplies(seeBernardandJensen,2000),sothatlabormarketsarenotfullyintegratedacrossregionsorcountries.Oneexplanation,accordingtothismodel,istheinformationalfrictionsinvolvedinthemovementofproducers.Thisoutcomeiscertainlynotrestrictedtotheone-sector,two-factorsetting,andwouldapplyequallywelltothetwo-sectorHOmodel.Thefindingof“factorprice 11-45Feenstra,AdvancedInternationalTradeinsensitivity”discussedintheHOmodel(seechapter1)reliesoncountriesbeingabletofreelyexportthegoodswhichuseintensivelythefactorsthataregrowinginabundance.Ifthereareinformationalfrictionsintheabilitytomatchbuyersandsellersininternationaltrade,thenfactorpriceinsensitivitywouldnolongerhold:agrowinglaborforce,forexample,wouldnecessitateincreasedexportsofthelabor-intensivegoods,butifbuyersarehardtofindthenwe24wouldexpectatermsoftradedeclineforthosegoodsandthereforefallingwages.Theideathatsellinggoodsoninternationalmarketrequiresa“match”betweenbuyersandsellersisexploredempiricallybyRauchandTrindade(2002).EarlierworkbyGould(1994)fortheU.S.andHeadandRies(1998)forCanadahadconfirmedtheimportanceofimmigrantnetworksinpromotingtradewiththeirhomecountries.RauchandTrindade(2002)expandthistoconsiderglobalbilateraltrade,andshowtheimportanceofChineseethnicnetworksinenhancingtrade.ThesenetworksaremeasuredbythelevelorfractionofthepopulationinacountrythatisofChineseethnicbackground,whichisenteredalongwithothertypicalregressorsinagravityequation.FollowingearlierworkbyRauch(1999),thetradeflowsinthegravityequationaredistinguisheddependingonwhethertheproductsarehomogeneous,withpublishedprices;differentiated,withpricesthatarenotavailable;oranin-betweencategoryof“referencepriced”products.ItisfoundthatthevariablemeasuringChinesenetworksentersmostsignificantlyinthebilateraltradeequationsfordifferentiatedgoods.Thissupportstheideathattheinformationaldifficultyisassessingthesegoodsismoreseverethanfromhomogeneousgoods,andthathavingtradersofthesameethnicbackgroundcanovercomethisdifficultytosomeextent.24Thesamelogicholdsinanylarge-countrymodelwithlabormigration.Seeproblem11.6. 11-46Feenstra,AdvancedInternationalTradeOtherempiricalresultsthatindicatetheinformationaldifficultiesininternationaltradecomefromstudiesofthelargetradersinHongKong,suchasLiandFung(SlaterandAmaha,1999).Theiractivitiesincludematchingfirmsorconsumerswhoarethebuyersofgoods,withthemostappropriateproducersofthegoods,locatedinChinaorelsewhereinAsia.Thevalueoftheseinformationalactivitiestothebuyersandsellersinvolvedwilldependonthetypeofgood(homogeneousversusdifferentiated),uncertaintyinpricesandquality,market“thickness”,andincomelevelsofthebuyers.FeenstraandHanson(2001)findthattheseregressorsaresignificantexplanationsofthemarkupschargedongoodsre-exportedthroughHongKong.Thisfurtherindicatestheimportanceofinformationinallowingforsuccessful“matches”ininternationaltrade.ConclusionsInthischapterwehavereviewedalargeliteratureontheroleofforeigndirectinvestment(FDI)andmultinationalsontrade.TheearlycontributionsofMundell(1957)andMacDougall(1960)dealtwithcapitalflowsbetweeneconomies,thoughFDIcaninsteadinvolveachangeinownershipandlocationoffirms,withoutanycapitalflows.ThelatercontributionsofHelpman(1984c)andMarkusen(1984)focusedonthedecisiontoestablishverticalorhorizontalforeignsubsidiaries.Thefactthattheparentcompanyhadinvestedinheadquartersservicesgaveitanassetthatcouldbetransferredtothesubsidiary.ItwasassumedbyHelpmanandMarkusenthatthisassetwaskeptinternaltothefirm,butofcourse,theseneednotbethecase:agreatdealofinternationaltradeoccursbetweenunaffiliatedfirms.Thiscreatesbothanempiricalandtheoreticalchallengeofexplainingcross-firmversuswithin-firmtransactions.Atthetheoreticallevel,transactionsarekeptinternaltoafirmwhenthecostofpursuingthembetweenfirmsistoogreat.Thisisthestartingpointforthetransactionscostexplanation 11-47Feenstra,AdvancedInternationalTradefortheorganizationofthefirm.ThisideahasbeenelaboratedintherecentindustrialorganizationmodelsofGrossmanandHart(1986),HartandMoore(1990,1999)andAghionandTirole(1997),andhasjustbeguntobeincorporatedintointernationaltrademodels.Thisisanewandexcitingtopic,whichmaywellreplaceincreasingreturnsandimperfectcompetition(inthe1980s),endogenousgrowth(inthefirsthalfofthe1990s)andpoliticaleconomy(inthesecondhalfofthe1990s)asthenextmajorresearchareaintrade.Oneobjectiveofthisresearchistoidentifytypesoftradingrelationshipsthatliein-betweenthetwoextremesofmarkettransactionsandverticalintegration,suchaswhenfirmsengageinajointventureorotherongoingrelationship.Wecanrefertosuchtradeasa“link”betweenthetwofirms,andthesetoflinksbetweenbuyersandsellersisreferredtoasa25“network.”Asanextensionofourstudyofmultinationals,wearenaturallyledtoadiscussionofnetworksininternationaltrade.Thesearerelationshipsthatrepresentmorethanmarkettransactions,butlessthanfullintegrationwithinafirm.Therearemanytypesofeconomicandsocialnetworksthatcanbeconsidered(Rauch,2001;RauchandCasella,2001),andwehavediscussedjustone,arisingfromthe“matching”ofagentsacrossinternationalborders.Sincefirmswillwanttobelinkedwithgoodpartnersabroad,wecanexpectthatagentswillfinditprofitabletoundertakethismatchingactivity.AmodelalongtheselinesisdevelopedbyRauchandWatson(2002),whereindividualswithexpertisechoosetobecome“internationaltradeintermediaries.”Thisalsocreatesaroleforgovernmentstosupportsuchintermediationactivities,throughtradefairs,associations,etc.Evidently,theseideasarebringingthetraderbackintointernationaltrade,whereheorsheshouldhavebeenallalong!25ThisterminologyfollowsKrantonandMinehart(2001). 11-48Feenstra,AdvancedInternationalTradeProblemsOuranalysisofcapitalflowsintheHOmodelcomparedtheequilibriumbeforeandafterthetariff,withacompleteflowofcapital,itisofinteresttoaskwhathappenstothewelfareofthecountryimposingthetariffduringthetransition.Thatis,asthecapitalinflowmovestheiicountriesfrompointVtoVinFigure11.2,howdoesthewelfareofthecapital-importingcountrychange?ThisquestionisaddressedbyBrecherandDiazAlejandro(1977).Toanswerthis,itisusefultoworkinreverse,andaskwhathappensifcapitalleavesthecountrystartingatithepointV,withthetariffonthecapital-intensivegoodappliedbycountry1.Vwillreduceproductionoftihecapital-intensiveTheoutflowofcapitalfromthepointgoodincountry1,andincreaseproductionofthelabor-intensivegood.Normally,thiswouldleadtothecapital-intensivegoodbeingimported.However,duetothetariffinthatindustry,therewillbesomerangeofcapitaloutflowthatdoesnotleadtoanyimports;instead,country1remainsinautarkybutwithachangeinrelativeprices.YoucanillustratethisinFigure11.2byiiilabelingapointE,lyingonthelinebetweenVandV,suchthatfortherangeofendowmentsiibetweenVandE,bothcountriesremaininautarkyduetothetariffappliedbycountry1.SeeFigure11.8.Nowanswerproblems11.1–11.3:ii11.1Ascapitalenterscountry1,goingfromVtoE,treatinternationalpricesasfixed.ThewelfareofarepresentativeconsumerinthatcountrycanbewrittenasV(p,I),wherep=p*+tisthedomesticrelativepriceofimportsandtisthetariff.Asinearlierchapters,weassumethattheexportableisadditivelyseparableinutility,sothatconsumptionoftheimportgoodis¶V/¶p=-d(p)and¶V/¶I=1.TotalnationalincomeequalsGDPplustariffrevenueminus 11-49Feenstra,AdvancedInternationalTrade22LO2KiViEiADiV1K11OLFigure11.8 11-50Feenstra,AdvancedInternationalTradepaymentstoforeigncapital,sothatI=w(p)L+r(p)(K+K*)+tm(p)–r(p)K*,whereK*istheforeigncapital,m(p)=d(p)–y(p,L,K+K*)areimports,andy(p,L,K+K*)isproductionoftheimportgood,whichdependsonthelaborandcapitalendowmentswiththeusualRybczynskieffects.NoticethatnationalincomesimplifiestoI=w(p)L+r(p)K+t[d(p)–y(p,L,K+K*)].ThendifferentiateconsumerutilitywithrespecttothecapitalinflowK*,andsignandinterpretthisresult,assumingthattheimportgoodiscapital-intensive.i11.2Theresultsinproblem11.1applyforinflowsofforeigncapitalbetweenpointsVandtheiipointE.AtE,country1isnowinautarky.Supposethatcapitalcontinuestoflowin:(a)Whatistheeffectofthecapitalinflowonequilibriumpricesforgoods,andontherental?(b)Thus,whatistheeffectofthecapitalinflowonconsumerwelfareincountry1?(Hint:illustratetheeffectontherentalasinFigure11.1).11.3Summarizeyourresultsfromproblems11.1and11.2bygraphingtherelationshipbetweenthecapitalinflowandwelfareoftherepresentativeconsumerincountry1.HowdoeswelfareatiiVcomparetothefreetradewelfareatV?a(1-a)11.4Supposethatf(Lij,zij)takeontheCobb-Douglasformf(Lij,zij)=Lijzij,00.11.5.Intheone-sectormodelwithperfectcapitalmobility,assumethatthetechnologiesareidenticalacrosscountriesandhomogeneousofdegreeone. 11-51Feenstra,AdvancedInternationalTrade(a)Iftherentalratesareequalizedacrosscountries,whatdoesthisimplyaboutthelabor-capitalratiosusedinproduction?(b)Therefore,howdowagescompareacrosscountries?11.6.Ratherthancapitalflows,letusinsteadconsiderinflowsoflabortoacountrythroughimmigration.Wewillcontrasttheresultsoftwomodels:(a)Ifthereisonlyonesectorintheeconomy,withfixedcapitalandfixedoutputprice,thenillustratetheeffectofimmigrationonthehomewages,andonwelfare.(b)SupposeinsteadthatweareinaRicardianmodel,asdiscussedinchapter1.Letlaborflowfromtheforeigntothehomecountry,andillustratetheimpactonthewelfareofworkersathome(notincludingthenewmigrants).ThisRicardiancaseisdiscussedingreaterdetailbyDavisandWeinstein(2002).
