Trade and Industrial Location with

TradeandIndustrialLocationwithHeterogeneousLabor∗MaryAmitiInternationalMonetaryFundandCEPRChristopherA.PissaridesLondonSchoolofEconomicsandCEPRthisversionMarch2004AbstractWeshowinthecontextofaneweconomicgeographymodelthatwhenlaborisheterogenoustradeliberalizationmayleadtoindustrialagglomer-ationandinter-regionaltrade.Laborheterogeneitygiveslocalmonopolypowertofirmsbutalsointroducesvariationsinthequalityofthejobmatch.Matchesarelikelytobebetterwhentherearemorefirmsandworkersinthelocalmarket,givingrisetoanagglomerationforcewhichcanoffsettheforcesagainst,tradecostsandtheerosionofmonopolypower.Wederiveanalyticallyarobustagglomerationequilibriumandillustrateitspropertieswithnumericalsimulations.Keywords:agglomeration,matching,spatialmismatch,inter-regionaltradeJELClassification:F12,J41,R12,R13∗Amiti,ResearchDepartment,TradeUnit,InternationalMonetaryFund,70019thStreet,WashingtonDC20431,Ph1-202-623-7767,Fax[phone],[e-mail]-sarides:LondonSchoolofEconomics,HoughtonStreet,LondonWC2A2AE,UK,tel.[phone],fax[phone],[e-mail],GeneGrossman,DiegoPuga,TonyVenablesandtheeditorandrefeeesfortheircomments.Aprevi-ousversionofthispaperwaspresentedattheCEPRWorkshop“TheEconomicGeographyofEurope:Measurement,TestingandPolicySimulations,”Villars,18/19January2002.Inthispaperwedemonstratethatwhenlaborisheterogenousandthematch-ingofskillswithjobsbelowfirst-best,theintroductionoftrademayleadtoindustrialagglomerationandinter-regionaltrade.Agglomerationtakesplacebe-causetheaveragequalityofmatchesimproveswhenfirmsinthelocalmarkethaveabiggerpoolofworkerstochoosefrom.Theforceagainstagglomerationistheexistenceoftradecosts.Atzerotradecostsregionsthatcantradealwaysspecial-ize,whereaswhentherearepositivetradecoststheymayormaynotspecialize,dependingonthevaluestakenbysomeotherparameters.Ourmodelresemblesothermodelsinnewtradetheory,exceptthatthereareheterogeneitiesintheperformanceoftasksdonebyapparentlysimilarworkers.Exanteworkersappearidenticalandnoworkerismoreproductivethananotheracrossthewholerangeofjobs.Butsomeworkersaremoreproductiveinsomejobsandotherworkersaremoreproductiveinotherjobs.Laborproductivityinthemodeldependsontechnology,trainingandtheotherfactorsofconventionalproductiontheory,butalsoonthequalityofthematchbetweenthejobandtheworker.Wepostulatethatevenwhenworkersareallocatedtothejobswheretheyaremostproductive,companiesthathavespecializedskillrequirementscanrecruitbetter-matchedworkersiftheyrecruitinlargermarkets.Weshowthatwhenthisfeatureiscombinedwithfeaturescommonlyassumedinnewtradetheory,inparticularincreasingreturns,differentiatedgoodsandtransportcosts,italonecanexplainagglomerationbyindustriesthatusespecializedskills.Anecdotalevidenceinfavorofourhypothesisiseasytofind.Togivetwoexamples,onebyanemployerandonebyanagentlookingforemployers,thegeneralmanagerofSonyUKexplainedasfollowswhyhiscompanyremainsinhigh-wageBritain:“Whatkeepsushereisthequalityofthestaffandtheresearchanddevelopmentcapacity”(FinancialTimes,January19,2002).Ontheothersideofthemarket,GavinClarkson,theownerofasoftwarecompanybutspeak-ingasamemberoftheChoktawNationofOklahoma,wasreportedasplanningtosetupatechnicaltrainingcentreinOklahomatoattractcompaniesbecause“Havingacriticalmassofpeoplewhoarehighlyskilledandresourcediswhatattractsbusinesstoanylocationorcommunity.”(AustralianFinancialReview,11April1999).ThewritersoftheFinancialTimesarticlewentfurthertoargue2that“Thesurvivalofstrugglingvolumeproducersmayormaynotprovedirectlyvitaltotheeconomyasawholebuttheirroleinprovidingskilledstaffandcom-ponentsinfrastructureforhigher-marginnichemanufacturersishardtoignore.”Thecontributionof“componentsinfrastructure”toindustrialagglomerationwasthethemeofKrugmanandVenables(1995).Ourfocusistheroleof“skilledstaff.”FormaleconometricevidenceinfavoroflaborpoolingwasprovidedbyDumaisetal.(1997).MakinguseoftheLRDmanufacturingdatabasefortheUnitedStates,theyexaminedtherelativeimportanceofMarshall’sthreereasonsforagglomerationforthelocationofmanufacturingplants;proximitytosuppliersandcustomers,laborpoolingandinformationspillovers.Theyfoundthatlaborpoolingwasbyfarthemostimportantforceforagglomeration,atleastatthemetropolitanarealevel.Newentrantstendedtolocateinareaswhereexistingfirmshadsimilarlaborrequirementstotheirown.TheDumaisetal.(1997)researchgivessupporttolaborpoolingasanag-glomerationforcebutdoesnotdifferentiatebetweendifferentreasonsthatmightmakeitimportant.Indirectevidence,however,supportsourmatchingreasons.Matchdifferentiationislikelytobemoreimportantformoreadvancedskills.Theroutinetasksthatdominateproductioninlessadvancedeconomiesdonotaffordmuchscopefordifferentiationorcreativity.Butwiththeinventionofmorecom-plicatedtasks,theroutinenatureofagriculturalandindustrialworkgiveswaytoworksituationswhichallowdifferentandvariedtypesofperformance.Ifwearecorrectinclaimingthatthisisareasonforagglomeration,thenagglomerationshouldincreasewitheconomicgrowthandshouldbemoreprevalentinindustriesthatrequiremorehigh-techlabor.Dumaisetal.(1997)findthatlaborp