Doeslumpinessmatterinanopeneconomy?Studyinginternationaleconomicswithr - PDF document

Doeslumpinessmatterinanopeneconomy?StudyinginternationaleconomicswithregionaldataPeterDebaereDepartmentofEconomics,UniversityofTexas,Austin,TX78712,USAReceived30August2002;receivedinrevisedform25June2003;accepted28July2003Thispaperaddressestheempiricalquestionwhethertheregionaldistributionofproductionfactorswithincountriesiseversouneventhatittriggersspecializationofproductionthatmakesregionsproducedifferentsetsofgoodsatdifferentfactorprices.Duetothedifferentwelfareeffectsoftradepolicyinacountrywithregionalspecialization,thisisanimportantquestion.Inaddition,itisaquestionaboutthelegitimacyoftreatingacountryasarelativelyhomogenousunit.Inansweringthesequestions,Iimplementtheconceptof‘‘lumpiness’’asintroducedbyCourantandDeardorff[J.Polit.Econ.100(1992)198].IfindthatlumpinessoranunevenregionaldistributionofproductionfactorsthatinducesintranationalspecializationanddifferentregionalfactorpricesisnotanissueforJapan,theUnitedKingdomandmaybenotevenforIndia.2003ElsevierB.V.Allrightsreserved.Keywords:Heckscher–Ohlin;Multiconeproduction;Factorpriceequalization;Regionalintegration;EconomicJELclassification:F11;F15;R12Inrecentyears,economistshaverediscoveredeconomicgeography.Duetothisrenewedinterest,economicinteractionsofregionshaveincreasinglybeenintegratedinfieldsotherthanregionaleconomics.Thelatterhasbeenfertileforinternationaleconomics.AsadvocatedbyKrugman(1991),regionsinacountryhavebecometestinggroundsforinternationaltradetheories.Thefocusonregionshaschallengedwhatisoftenabasicpresumptionofinternationaleconomics:thatacountryisafairlyhomogeneous,0022-1996/$-seefrontmatter2003ElsevierB.V.Allrightsreserved.doi:10.1016/j.jinteco.2003.07.005 *Tel.:+1-512-475-8537;fax:+1-512-471-3510.E-mailaddress:debaere@eco.utexas.edu(P.Debaere).www.elsevier.com/locate/econbaseJournalofInternationalEconomics64(2004)485–501 undifferentiatedunit.Ithasalsochallengedthedistinctionbetweenregionalandinterna-tionaleconomics.ThechallengethatIinvestigatedoesnotcomefromthe‘‘new’’geographythatisassociatedwiththeworkofKrugmanandothers,butratherfromthestandardperfectcompetitionanalysis.InaHeckscher–Ohlinmodel(HO),internationaltradeisdeter-minedbycountries’factorendowments:capital-abundantcountriesexportcapital-inten-siveproductsandlabor-abundantcountriesexportlabor-intensivegoods.Sincecountriesareintegratedeconomies,theirregionaldistributionofproductionisbyassumptionirrelevantfortheirtradepattern.CourantandDeardorff(1992)challengethisview.Theystudyhowunevenregionaldistributionsofendowmentscanbeanindependentcauseforinternationaltrade,makingregionsthepreferredunitofanalysisininternationaleconom-icsandacountry’sinternationaltrade‘‘aparticularaggregationofinterregional(trade’’.Intheirmodel,verydifferentregionalendowmentsinducespecializationbetweentheexistingregions;regionsproducedifferentgoodsandtheirfactorsarepaiddifferentrewards.Moreover,anationaltradepolicycanhavedifferentwelfareeffectsacrossregionsanditmayimpactonfactorreturnsinwaysdifferentfromthestandardpredictionsofStolperandSamuelsonatthenationallevel(seeMelvin,1985foradiscussion).CourantandDeardorffcallacountrythatischaracterizedbysuchanunevendistributionofresourcesalumpycountry.IinvestigateforJapan,theUnitedKingdomandIndiawhetherthedistributionofeconomicactivity/resourcesacrosstheirregionsissouneventhatitinducesspecializationandunequalfactorprices.Itaketheagglomerationofresourcesasgivenandwonderwhatitimpliesforproduction(andhencetrade).Weknowfromempiricalstudiesthattherearewagedifferenceswithinacountry.Thesewagedifferencesare,however,notprimafacieevidenceofintranationalspecialization.Inparticular,sincetherearemanyexplanationsforregionaldifferences,Iinvestigatewhetherspecializationinducedbylumpinessisoneofthem.Myfindingssupportthenotionthatintranationalproductionfactorsdonotvaryenoughtoinducespecializationacrossregionsandtotriggerdifferentfactorprices.Iarguethatspecializationduetolumpinessisnotaregionalphenomenonandspecialization KrugmanandVenables(1995)bestsummarizethealternativeperspectiveof‘‘new’’geographyinaprogrammaticarticle.Theyexplainagglomerationofeconomicactivitywithincreasingreturnsandtransportationcostsand‘‘talkaboutinternationaltradewithoutmentioningcountries.’’Alltradeflowsbetweenanytwopointsinspacematter,notjustthosecrossingarbitrarylines,calledborders.Internationaltradeisaspecialcaseofregionaltrade.Whetherdifferencesinendowmentsgeneratespecializationacrosscountriesisacentralquestionininternationaleconomics.Leamer(1996)raisedtheissueinthetradeandwagesdebatesincetheeffectoftrade(changingprices)onwagescriticallydependsonwhethercountriesproducethesamegoodsornot.Schott(2003)Leamer(1987)DebaereandDemiroglu(2003)implementingDeardorff(1994)provideevidencethatitisimpossiblethatallcountriesproducethesamegoodsatthesamefactorprices.(Countryendowmentsdonotlieinonediversificationcone.)ThisiscorroboratedbyEvenettandKeller(2002).Notealsothattheideaofinternationalspecializationofproduction(andnon-FPE)wasimplicitlypresentinOhlin’swritingsfromtheverybeginningonwards(seeFlametal.,2002)Thepresentpapergoesonestepfurtherandaskswhetherregionalendowmentdifferencesinducespecializationinanopeneconomy.Hanson(2000)forasurveyoftheliteratureontheregionalvariationinwages.SeeMachin(1996)regionalwageinequalityintheUnitedKingdom.P.Debaere/JournalofInternationalEconomics64(2004)485–501 becauseofendowmentdifferencesoccursprimarilyattheinternationallevelwheremoresubstantialfactorendowmentdifferencesbetweencountriesexist.Myanalysisbuildsonstudiesthatillustratehowtheorieswithperfectcompetitionandendowment-driventradestillhavearoletoplaynexttoeconomicgeography.MoroneyandWalker(1966)provideanearlytestofHOacrosstheregionsoftheUnitedStates.Inmorerecentyears,Davisetal.(1997)predictJapaneseregionalproductioninaHeckscher–Ohlin–Vanek(HOV)setting,ahigher-dimensionalversionoftheHOmodel.HansonandSlaughter(1999)studythesectoralreallocationacrossUSstatesinaHOVframework.Inaddition,Smith(1997)andKim(1995,1999)theexplanatorypowerofHOwithotherreasonsforagglomerationsuchasinternalandexternalscaleeconomies.FairlystrongempiricalsupportfortheHO(V)modelisfoundinallcases.1.Acriterionfor‘lumpiness’Consideramodelwithperfectcompetition,manygoods,manycountriesandmanyregionspercountry.Therearetwofactors,labor,,andland,,whicharenotmobilebetweencountriesorregions.Withinaregion,thereislabormobilitybetweentheland-andthelabor-intensivesector.Inaddition,theconstant-return-to-scaletechnologyandthepricesthatconsumersandproducersfacearethesameeverywhere.Countriesandregionsareopeneconomiesthatcanfreelytradewithouttransportationcosts.Assumefornowthattradeequalizesfactorreturnsintheentireworld.Tradethusreplicatestheintegratedworldeconomy(IWE)—ahypotheticalworldinwhichallfactorsareperfectlymobile.Thereisfullemployment.Fig.1illustratesacriticalcriterionforlumpiness.Thesizeoftheboxisdeterminedbyonecountry’slandandlaborsupply.Thetwosolidvectorsrepresentthefactorinputsthatsatisfythefactordemandsattheworldinsectors1,themostcapital-intensivesectorintheworld,and3,theleastcapital-intensivesector.Forsimplicity,IassumethereareonlytworegionsinthecountrythatIstudy.characterizestheendowmentsofthetworegions.(’scoordinatesaremeasuredforoneregionandfromfortheother.)Asthedashedlinesindicate,onecaneasilyemploytheregions’resourcesinsectors1and3.(Measuredfrom,thecoordinatesofindicatethelandandlaborusedinsectors1and3ofoneregion;measuredfrom,thecoordinatesofindicatethesamefortheotherregion.)Therefore,regionalendowmentsdonotinducespecializationwithinthesolidparallelogram;outsidethatarea,theydo.IntheempiricalimplementationIrelyonthe‘‘lenscondition’’thatisslightlystricterthanFig.1aandthathasbeenstudiedbyDeardorff(1994)DebaereandDemiroglu(2003)Fig.1bdepictsthelensconditionforourparticularcountrythatispartofaworldthatreplicatestheIWE.Fortwoproductionfactors,theconditionstatesthatregionshaveequalfactorpricesandproducethesamegoodsifandonlyiftheregionalendowmentlens(in Asonechoosesthecountrysmallenough,onecanruleoutthecaseinwhichnosuchcombinationofthetwosectorswiththemostextremefactorintensitiesexistsbecausethecountrycontainsmoreresourcesthanthefactoruseofthetwosectorscombined.P.Debaere/JournalofInternationalEconomics64(2004)485–501 dashedlines)liesinsidethegoodslens(insolidlines).NotethatDemirogluandYun(1999)showthatthis‘‘lenscondition’’isonlyanecessaryconditionwhentherearemorethantwofactors.Oneconstructstheregionallensasfollows.Ranktheendowmentsforthetwo Fig.1.(a)WorldIWEconditionforlumpiness;(b)lensconditionsatisfied;(c)lensconditionviolated. DebaereandDemiroglu(1998)showthattheconditioncanholdforjustanygroupofcountries(regions)withFPE.P.Debaere/JournalofInternationalEconomics64(2004)485–501 regionsbyland/laborratioandconcatenatethecorrespondingvectorsctorsTr,Lr],inincreasinganddecreasingorderoftheirland/laborratiostartingfromtheorigin.(Theheightoftheboxequalsourcountry’sland,withthewidthasitslabor.)ForanycountrythatexperiencesFPE,thegoodslensisdrawninasimilarway.Assumethreesectorsandsumtheirlandandlaborinputacrosstheregionstoobtainthetotalfactoruseineach,i.e..Next,rankagainandconcatenatethecorrespondingfactorinputvectorsvectorsTi,Li],fromtheoriginbyincreasinganddecreasingland/laborratio.Fig.1b,theregionallensliesinsidethegoodslens,whichpointstoFPEandnospecialization.Lumpinessdoesnotmatterhere.Withmorefactors,thegoodslensenvelopstheregionallensforanytwoofitsfactors.ThelensconditionisbestillustratedwithaviolationasinFig.1cinwhichtheendowmentlensdoesnotentirelylieinsidethegoodslens.Oneregionhastoomuchland;itcouldneveremployallitsresourcesforagivensetoffactorprices.Consequently,regionscannotproducethesamesetofproductsatthesamefactorprices;theyendupproducingdifferentgoods.TheconditioninFig.1bisslightlystricterthanFig.1asinceitdependsonthefactorinputsofallsectors—notjustthemostextremeones.IftheconditioninFig.1bisnotviolated,thenthelumpinessconditionofFig.1aisautomaticallysatisfied.Tostudylumpinessempirically,IcheckconditionFig.1bwithdatafromJapan,theUKandIndia.Fig.1bhastheadvantagethatitdoesnotrequireknowledgeabouttheentireintegratedworldeconomy.Sofar,IhavedefinedtheIWEforallcountriesoftheworld.Thisisbynomeansnecessary.OnecanhaveFPEamongagroupofcountriesthatisfarsmallerthantheentireworld.Byfocusingonthelensconditionforonecountryonly,Idonothavetotakeastandontheentiresetofcountries,ifany,amongwhichthereisFPE.Ionlyhavetorelyontheproductionpatternofonecountryasitisfoundinthedatatocheckwhetheritisconsistentwithnospecialization.Aspresentedhere,specializationofproductionhingesonregionalfactorimmobility.Indeed,mobilefactorscanundoregionaldifferencesinfactorreturnsorinrelativefactorabundance.However,CourantandDeardorff(1993)showinaversionofthemodelwithamenitiesthatspecializationofproductionandunequalfactorpricescancoexistwithinterregionalfactormobility.Inacountrywithamenitiesinvariousregions,consumersmay,forexample,bewillingtoincuralowerwageinordertobenefitfromamenities,andhencetheymaydecidenottomovetohigher-wageareas.Inotherwords,assumingregionalfactorimmobilityisaconvenient,yetnotanessentialsimplificationofreality.2.ThecaseofJapanInthissection,Icomparethedistributionofregionalendowmentswiththedistributionofsectoralfactoruse.IdrawthelensconditionofFig.1basdescribedintheprevious Assoonasacountryisbiggerinsize,sothatmorethantwosectorsareneededtoemployitsresources,onehastoexactlyknowthesizeofthesectorsintheintegratedworldeconomy(IWE)tobeabletojudgewhetherthereisaviolationornot.CourantandDeardorff(1993)alsoconsiderproductionamenities,suchas,forexample,regionalvariationsinclimate,andtheyshowhowproductionamenitiescanreconcilefactormobilitywithlumpinessontheproductionside.P.Debaere/JournalofInternationalEconomics64(2004)485–501 sectionwithactualdata.Istudywhetherthelensofregionalendowmentsindeedliesinsidethegoodslenswiththeactualfactorinputs.Toconstructthegoodsandtheregionalendowmentlenses,considerforexamplehigh-andlow-skilledlabor.Forthegoodslens,oneneedsnationaldataonthetotalamountofhigh-()andlow-skilledlabor()usedineachsectorinJapan.ThisnumberbydefinitionamountstothesumofallfactorsemployedinthatsectoracrossthevariousregionsasinEq.(1).Fortheregionallens,onealsoneedstheregionalendowmentsthatbydefinitionamounttothesumoftheregions’high-andlow-skilledlaborthatisusedinallNext,oneranksfactor-inputandendowmentdatainincreasinganddecreasingorderoftheirhigh-skilled/low-skilledratios.Toobtainthegoodslens,oneconcatenatesthecorrespondingsectoralfactorinputvectorsinbothorderings,startingfromorigin;thevectorsadduptothetotalhigh-andlow-skilledendowmentofJapan,whichisthesizeoftheendowmentbox.Idrawalensforhigh-andlow-skilledlabor,oneforcapitalandlow-skilledlaborandoneforcapitalandhigh-skilledlabor.AviolationinanyoftheselensesimpliesspecializationnoFPE;noviolationsuggestslumpinessdoesnotmatter.Fortheempiricalanalysis,IrelyonDavisetal.’sdata.Theyhaveendowmentandproductiondatafor10JapaneseregionsthatarebasedonJapan’s47prefectures.BasedontheEmploymentStatusSurvey,Davisetal.providefactorendowmentsforhigh-andlow-skilledlaborfor10regions.TheregionalcapitalstocksareconstructedwiththeperpetualinventorymethodwithinvestmentflowsfromtheAnnualReportonPrefecturalAccounts.ThepricedeflatorsareformtheAnnualReportonNationalandthedepreciationrateis0.133.Asforthesectoralfactoruse,capitalisconstructedinthesamewayastheregionalstocksusinginvestmentflowsfor30sectors.TheinvestmentflowsarefoundintheAnnualReportofCorporationSurveyandtheCensusofManufacturing.ThesectoralemploymentfigurescomefromtheEmploymentStatusSurveyTable1reportswhatfractionofthetotalJapaneseendow-mentsisusedinthevarioussectorsandwhatfractionoftheseendowmentsisallocatedineachofthe10regions. The10JapaneseregionsareHokkaido,Tohoku,Kanto,Hokuriku,Tokai,Kinki,Chugoku,Shikoku,Kyushuand,finally,Okinawa.Inthepresenceofnontradedgoods,onecouldarguethatthelensesshouldbedrawnwithdatafromonlythetradedgoodsectors(seeHelpmanandKrugman,1985).Inmycase,drawingthelensesforallsectorsoronlyformanufacturingplusagriculture(toproxyfortradedgoods)doesnotaffecttheresults.P.Debaere/JournalofInternationalEconomics64(2004)485–501 Fig.2showsthelenswithhigh-andlow-skilledlabor.Astheendowmentsarenormalizedwiththetotalfactorsupplies,thesidesoftheendowmentboxesaddupto1.Thereismorevariationinthefactorinputsthanintheregionalfactorsupplies.Thelens Table1SkilledlaborUnskilledlaborCapital(A)Japan:theregionalendowmentsTotalendowmentofJapan=1(fractionoftotal)Hokkaido0.0340.0410.047Tohoku0.0560.0700.076Kanot0.3970.3600.343Hokuriku0.0390.0430.046Tokai0.1230.1300.127Kinki0.1800.1680.157Chugoku0.0580.0610.066Shigoku0.0250.0310.032Kyushu0.0810.0900.097Okinawa0.0050.0060.007(B)Japan:thesectoralfactoruseTotalendowmentofJapan=1(fractionoftotal)Ag/forestry/fishery0.0010.1070.091Mining0.0010.0020.004Const0.0730.0690.036Food0.0200.0320.021Textile0.0050.0170.006Apparel0.0030.0170.002Lumber0.0020.0070.003Furniture0.0020.0060.002Pulp,paper0.0060.0090.009Printing0.0240.0150.008Chemicals0.0290.0160.026Petrol/coalproducts0.0020.0010.007Rubber0.0030.0060.004Leather/leatherproducts0.0030.0020.000Stone,clay,glass0.0080.0150.013Iron/steel0.0070.0130.028Non-fe0.0060.0060.008Fabricatedmetal0.0140.0220.011Machinery0.0320.0310.022Electricalmachinery0.0660.0620.044Transportequip0.0250.0330.037Precisioninstru0.0090.0090.005Othermanufacturing0.0120.0160.011Transportation/communication0.0440.1070.053Electricity/gas/water0.0090.0090.072Wholesale/retail0.2510.1520.075Finance/insurance0.1040.0450.016Realestate0.0110.0050.018Otherservices0.2190.1540.096Publicadministration0.0090.0150.274Davisetal.(1997)P.Debaere/JournalofInternationalEconomics64(2004)485–501 conditionisnotviolated.Also,forcapital-skilledlaborandcapital-unskilledlabor(notreported)istheregionallenswellinsidethegoodslens.Theextenttowhichregionalendowmentsdiffercannotbeasourceofregionalspecialization.ThissupportsDavisetal.’sclaimthatlumpinessshouldnotmatterforJapaneseproduction.3.ThecaseoftheUnitedKingdomInthissection,IextendtheempiricalanalysisoflumpinesstotheUnitedKingdom.Iusedataofthe11regionsintowhichtheCentralStatisticsOfficedividestheUnitedKingdom.TheUnitedKingdomhas,likeJapan,afairlypronouncedpatternofregionalconcentrationanddispersionofitseconomicactivity.ThefairlydenselypopulatedstringofregionsthatrunsfromtheSouthEastacrosstheWestMidlandsthroughtheNorthWestcoversalittlelessthan20%oftheentireareaoftheUnitedKingdom.Still,in1988,thisareaaccountedforabout54%oftheavailablejobsintheUnitedKingdom.Notsurprisingly,onasubregionallevel,thepictureisevenstarker.Outofthe65subregionalunitsthattheCentralStatisticsOfficedistinguishesfortheUnitedKingdom,the15mostdenselypopulatedsubregionsaccountforabout7%oftheareaoftheentireUK.Intermsofemployment,however,itfillsabout48%ofthejobsintheUK.InordertodrawthegoodslensandtheregionallensfortheUnitedKingdom,Icollectendowmentandsector-levelfactorusedatafor1990.Theproductionfactorsarehigh-andlow-skilledlabor.Ialsoconstructregionalandsectoralcapitalstocks.Table2reportstheregionalfactor Fig.2.Japan,high-skilledlabor(HSL)versuslow-skilledlabor(LSL). TheregionsareforEngland,YorkshireandHumberside,theNorth,theEastMidlands,EastAnglia,theSouthEast,theSouthWest,theWestMidlandsandtheNorthWestplusWales,ScotlandandNorthernIreland.Forthedistributionofhigh-andlowskilledlaboracrossthedifferentregions,Ihavetorelyonthe1991P.Debaere/JournalofInternationalEconomics64(2004)485–501 Table2SkilledlaborUnskilledlaborCapital(A)UnitedKingdom:theregionalendowmentsTotalendowmentoftheUK=1(fractionaloftotal)North0.0460.0570.061YorkshireandHumberside0.0750.0900.086EastMidlands0.0650.0720.065EastAnglia0.0340.0370.037SouthEast0.3610.2840.296SouthWest0.0810.0840.077WestMidlands0.0820.0950.084NorthWest0.1050.1120.106Wales0.0430.0530.056Scotland0.0850.0900.103NorthernIreland0.0220.0270.028(B)UK:thesectoralfactoruseTotalUKendowment=1(fractionaloftotal)Agriculture,forestry,fishing0.0190.0290.021Mining0.0110.0160.018Food,drinksandtobacco0.0370.0590.046Textile,leather,footwear0.0130.0210.012Wood,corkandfurniture0.0070.0110.007Paper,printingandpublishing0.0320.0270.029Basicchemicals0.0240.0240.025Pharmaceuticals0.0070.0060.006Petroleumrefineriesandproducts0.0060.0070.022Rubberplasticproducts0.0110.0120.011Stone,clayandglass0.0090.0110.012Ferrousmetals0.0120.0140.015Non-ferrousmetals0.0040.0060.004Fabricatedmetalproducts0.0120.0130.012Othernon-electricalmachinery0.0240.0260.021Computersandofficeequipment0.0110.0100.008Electricalequipment0.0100.0110.009Electronicequipment0.0200.0200.016Shipbuilding0.0020.0030.002Othertransportequipment0.0010.0020.002Motorvehicles0.0160.0200.017Aircraft0.0110.0100.008Instruments0.0040.0040.004Othermanufactures0.0030.0040.003Electricity,gas0.0200.0220.061Construction0.0800.1270.072Wholesaleandretail0.0910.1060.108Hotelsandrestaurants0.0160.0170.016Transportandstorage0.0440.0430.046Communications0.0160.0150.017Financeinsurance0.0930.0740.125Realestateandbusinessservices0.0960.0720.120Community,socialservices0.2380.1590.106RegionalTrends,OECD,owncalculations.P.Debaere/JournalofInternationalEconomics64(2004)485–501 suppliesandthefactorinputsofthevarioussectors.BotharereportedassharesofthetotalfactorendowmentsoftheUK.Ifirstdescribethedataforthegoodslens.FromanOECDworkingpaper,,ItaketheemploymentfiguresforallthesectorsoftheUnitedKingdomforbothhigh-andlow-skilledlabor.High-skilledlaborconsistsofhigh-skilledwhite-collarworkerswhichcomprisethefollowingcategoriesoftheIndustryStandardClassificationofOccupationsissuedbytheInternationalLaborOrganization:10(legislators,seniorofficialsandmanagers),20(professionals)and30(techniciansandassociateprofessio-nals).Iconsiderallotherworkers(theOECDdistinguisheslow-skilledwhite-collarworkersandhigh-andlow-skilledblue-collarworkers)ofthelow-skilltype.TheOECDprovideslabordatafor42sectors.Iregroupthelabordataintothe33-sectorclassificationoftheOECDInput–OutputTables(1995)tomakethemconsistentwiththeinvestmentdatathatareneededfortheconstructionofsector-levelcapitalstocks.IrelyontheconcordanceoftheOECDInput–OutputTables(1995)forthisreclassification.ThesectoralinvestmentflowswithwhichIconstructthesectoralcapitalstocksoftheUnitedKingdomaredrawnfromtheOECDNationalAccountDetailedTables.Theinvestmentflowsfrom1976to1990aretransformedintostockswiththeperpetualinventorymethod.TheinvestmentdeflatorisalsofromtheOECDandthedepreciationrateis0.133.Toobtaindirectplusindirectfactorinputs,IcombinesectoralfactorinputdatawiththeUK’sInput–outputtablefor1990,alsofoundintheOECDInput–OutputTables(1995).Inordertobeabletodrawtheregionallens,endowmentdataareneededforthe11regions.TobeconsistentwiththefactorinputdatathatIdescribedabove,IsumtheOECDhigh-andlow-skilledlabordataofthepreviousparagraphacrosssectorsinordertoobtaincountrywideendowments.Ithenapplytotheseendowmentfigurestheregionaldistribu-tionofhigh-andlow-skilledlaborthatistakenfromtheRegionalTrendsoftheCentralStatisticalOffice.TheRegionalTrendsprovidesforthe11regionsoftheUnitedKingdomthetotalnumberofpeopleofage16yearsandolderthatareactiveineachregion.Inaddition,itbreaksdownthesenumbersperregionaccordingtooccupation(seeTable10.9).ThesefiguresallowmetoidentifywhatfractionofthetotalUK-employedlaborforceishigh-skilledorlow-skilledineachregion.ThesefractionsarethenmultipliedwiththeOECDendowmentsmentionedabove.EstimatesofregionalcapitalstocksfortheUnitedKingdomareobtainedalongsimilarlines.TheaggregatecapitalstockoftheUKisjustthesumofthecapitalstocksthatareusedinthevariousindustriesdiscussedinthepreviousparagraph.IderivetheregionaldistributionofcapitalfromTheRegionalTrends.Moreinparticular,IconstructregionalcapitalstockswiththeprivateandgovernmentgrossfixedcapitalformationfiguresfromRegionalTrendsforthesame15years,1976–1990.Irelyupontheperpetualinventorymethodandusethesamedeflatorsandthesamedepreciationratehere.TheregionaldistributionofthethusobtainedcapitalstocksisthenappliedtothetotalUKcapitalstockthatisbasedonOECDinvestmentfigures.Fig.3presentsfortheUnitedKingdomthelensforlow-skilledlaborvs.capital.Asbefore,theendowmentsarenormalizedwiththetotalfactorsupplies,sothatthesidesof Abroaderdefinitionofhigh-skilledlaborincludinglow-skilledwhite-collarworkersgeneratesqualitativelysimilarresults.P.Debaere/JournalofInternationalEconomics64(2004)485–501 theendowmentboxaddupto1.Thehigh-skilledvs.capitalandthehigh-vs.low-skilledlaborlens(notreported)aresimilarandconfirmfortheUKtheresultsthatwereobtainedforJapan.Theregionalendowmentlenslieswellinsidethegoodslens.4.AssessmentofresultsThereissignificantvariationineconomicactivityacrosstheregionsofacountry.Myanalysisshows,however,thatthereisnotenoughregionalvariationinthefactorsuppliesintheUnitedKingdomandJapantoinduceregionalspecializationandtomakedifferentregionsproducedifferentgoodsatdifferentfactorprices.Myfindingsaccordwellwiththeempiricaltradeliteratureoninternationalspecialization.DebaereandDemiroglu(2003)Schott(2003)showthatspecializationisespeciallyaninternationalphenomenon.Inparticular,thesubstantialdifferenceinendowments,especiallybetweendevelopinganddevelopedcountries,preventsallcountriesfromproducingthesamesetofgoods.Thelensconditionisviolatedforagroupofcountriesthatincludesbothdevelopedanddevelopingcountries.Forreference,theendowmentvariationsaremuchmorepronouncedinaninternationalcontext.Thecapital/laborratioinadevelopedcountrysuchasGermanyiseasily15timeshigherthanthecapital/laborratioindevelopingcountriessuchasIndia.Notethatthedifferencebetweenthehighestandthelowestregionalcapital/laborratioisonlyaround2fortheUnitedKingdomandJapan.Inthissection,Iaddressafewissues Fig.3.UKcapital(K)versuslow-skilledlabor(LSL). IfoneaggregatestheregionsofEnglandandifonedrawsthelenstogetherwiththedatafromScotland,WalesandNorthernIreland,onestilldoesnotobtainaviolation.Evenaftercorrectinglaborfiguresfordifferencesinhumancapital,developedcountryratiosarestillfivetimeshigherthanfordevelopingcountries.P.Debaere/JournalofInternationalEconomics64(2004)485–501 thatoftenariseindiscussionsoflenses.Theyshouldgiveanideaoftherobustnessandgeneralityoftheobtainedresult.Acommonconcernisthatthelensesareconstructedwhileitisassumedthatthefactorendowmentsaresimilarenoughnottoinducespecialization.Ifweobtainaviolation,weknowacountryislumpy,butonemaywonderwhatweshouldconcludeifwedonot.IshowinFig.4thatfor(two)regionsthatproducedifferentgoodsbecausetheirendow-mentsaretoodifferent,oneislikelytoobtainaviolationofthelensconditionifonedrawsthelenseswithactualdatathewayIdo.ThetoppanelofFig.4isaLerner–Pearcediagramwithunitvalueisoquantsforthreegoods,andthebottompanelrepresentsthecorrespondingendowmentboxes.Regionone(withendowment)producesgoods1and2,whereasgoods2and3areproducedinthesecondregion(withendowment Fig.4.SpecializedproductionandaviolationoftheFPEproductionlens.P.Debaere/JournalofInternationalEconomics64(2004)485–501 Initially,Itreatthemoreland-intensivegood2,2,anditslesslabor-intensiveversion,2asseparategoodsanddrawforeachregionadifferentgoodslens(indashedlines).Next,Idrawonegoodslensforbothregions.[Inmostcases,thedatamakeoneconsider2and2thesameproductandaddupthefactorinputsof2and2todrawthegoodslens(insolidlines)].Oneclearlyseeshowthisproducesaviolationofthegoodslens.liesoutsidethegoodslens.Inordertoavoidoverburdeningthegraph,IhavenotexplicitlydrawntheregionallensconnectingDavisetal.(1997)arerighttoassumethatfactorpriceequalizationismorelikelywithinacountrythanacrosstheglobe.Still,onewonderswhetherfactor-augmentingproductivitydifferences(asinTrefler,1993)acrossregions,shouldtheyoccur,wouldchangemyresult.Factor-augmentingproductivitydifferencesareconsistentwiththebasicassumptionsofthesetupifonereformulatesendowments,factoruseandfactorpricesinproductivityequivalents.Inparticular,theyrequireonetodrawthelensesinproductivityequivalents.ItcanbeproventhatHicks-neutraldifferencesdonotproduceaviolationofthelensconditioniftherewasnonewithoutadjustments(seeDebaereandDemiroglu,1998).Also,factor-augmentingtechnologydifferencesthatarenotsymmetricacrossfactorsareunlikelytoleadtoaviolationinourcase—justconsiderhowwelltheendowmentlensliesintothegoodslens.Onemaywonderfurtherhowgeneraltheobtainedresultsare.Itisanopenquestionwhethertheobtainedresultsarecharacteristicfordevelopedcountriesonly.IftheregionaldifferencesinpercapitaGDPgiveanyindicationofthevariationinregionalendowments,therecouldbeapresumptionthataviolationofthelensconditionmightbemorelikelyindevelopingcountriesinwhichthereismorevariationinpercapitaGDPacrossstates.ThemaximumregionalpercapitaGDPofaregionis,intheUS,abouttwicetheminimum.ThesameistrueforJapanandfortheUK.ForcountriessuchasBrazilorIndia,weeasilyobtainafactorof4or5.Withduereservationsaboutthequalityofthedata,IreportFig.5thatcangiveanindicationofwhattoexpectindevelopingcountries.ItpresentstheregionalendowmentandgoodslensforIndiafortheyear1988.Thefactorsarecapitalandemployedlabor.Thedatasourceforthisexerciseisthe1998IndustrialDatabookfromtheCentreforIndustrialandEconomicResearch(1998).Itprovidesestimatesofthecapitalstockfor25regionsandalsobreaksdownanestimateofthecountrywidecapitalstockaccordingto20sectors.alsoprovidesemploymentdataattheregionallevelforthesame25regionsandnationwidesectoralemploymentfiguresforthecorrespondingsectors.IreportthedatainTable3.Theregionallenslieswellinsidethegoodslens.Iamcautiouswiththisresultsinceitislikelythattherearetechnologicaldifferencesacrossregions(orother Factor-augmentingproductivitydifferencesrequirethatoneadjuststhesectoralfactorinputsandtheregionalendowmentswitharelativeproductivitymeasureofregionwithrespecttoareferenceregion.Forhigh-skilledlabor,thefactorinputforasectorchangestoHSL.ThefactorendowmentfigureforaregionchangestoHSLTheregionsofIndiaonwhichCIERprovidesdataareJammuandKashmir,Punjab,Haryana,HimachalPradesh,Chandigarth,Delhi,Rajasthan,UtraPradesh,Bihar,Gujarat,Maharashtra,MadhyaPradesh,GoaDamanandDiu,Karnataka,AndhraPradesh,TamilNadu,Kerala,Pondicherry,Orissa,WestBengal,Assam,Manipur,Tripura,Meghalaya,AndamanandNicobar.P.Debaere/JournalofInternationalEconomics64(2004)485–501 violationsofthemodel’sbasicassumptions).IwouldnotinterpretthefindingasevidencesuggestingFPEorFPEinproductivityequivalents,however.Rather,Iseeitasafindingthatsuggeststhatevenwhentechnologywerethesame(andifallotherassumptionsheldup)amongIndianregions,therewouldnotbeenoughregionalvariationinendowmentstotriggerspecializationofproduction.Finally,theremaybeconcernsaboutaggregationandotherfactorsthantheonesthatIstudy.Ihavedrawnthelensesforcapital,skilledandunskilledlabor.Productionfactorssuchasmineraldepositsorarablelandarenotconsideredduetodatalimitations.Itispossible,however,thatsuchotherfactorscouldleadtoaviolation.Thenthereisaggregation.Indeed,thelensesaredrawnforalimitednumberofsectorsonly.Eachsectorisboundtoconsistofmanysubsectors.Onemaywonderhowthisaffectstheresult.Aggregationisboundtomakethegoodslensthinner(forreference,addingupallthesectorsintoonewillyieldthediagonaloftheendowmentbox.)Hence,withmoredisaggregation,thegoodslenswillbecomeevenwider,makingaviolationevenlesslikely.Inaddition,onemaywonder,whethertheendowmentsshouldnotbedisaggregatedandwhethersuchdisaggregationshouldnotworktheotherwayandmakeaviolationofthelensconditionmorelikely.Considerwhatwouldhappenifoneweretousedataatthecountylevel.Inawealthycountywithahighconcentrationofhighlyskilledlaborandvirtuallynolow-skilledlabor,thehigh/lowskilledlaborratioswouldbeveryhigh,whereasinapoorone,theoppositeshouldapply.Insuchacase,onemightthinkthatthemoredispersedisaggregatehigh/lowskilledlaborratios(comparedtotheaverageofbothcounties)shouldmoreeasilyinduceaviolationofthelensconditionforagivengoodslens.Itiscriticaltoremember,however,thatifthepeopleofbothcountiesworkinthesamefirms,theviolationwouldbeentirelyspurious.Inotherwords,thegeographicunitsofanalysiscannotbetoosmallsincetheyshouldincludeallpeoplelivingandworkingin Fig.5.India,labor(L)versuscapital(K).P.Debaere/JournalofInternationalEconomics64(2004)485–501 Table3LaborCapital(A)India:theregionalendowmentsTotalendowmentofIndia=1(fractionoftotal)JammuandKashmir0.0060.004Punjab0.0490.048Haryana0.0310.03HamachalPradesj0.0060.009Chandigarh0.0010.005Delhi0.0170.009Rajasthan0.030.036UttarPradesh0.0960.106Bihar0.050.076Gujarat0.0860.098Maharashtra0.1570.165MadhyaPradesh0.0480.078Goa,DamanandDiu0.0020.003Karnataka0.050.042AndrhaPradesh0.0920.058TamilNadu0.0890.114Kerala0.0320.023Pondicherry0.0020.001Orissa0.0210.043WestBengal0.0940.043Assam0.0130.09Manipur0.0000.000Tripura0.0000.002Meghalaya0.0000.000AndamanandNicobar0.0000.000(B)India:thesectoralfactoruse(fractionoftotal)Basicmetals0.0790.138Metalproducts0.0250.013Chemicalproducts0.0710.118Transportequipment0.0620.060Machineryandtools0.0560.043Rubber0.0270.050Foodproducts0.1280.057Beveragesandtobacco0.0560.010Cottontextiles0.1070.037Woolentextiles0.0400.027Jute,hempandmesta0.0250.004Paperproducts0.0370.027Nonmetallicmineralproducts0.0540.045Textileproducts0.0170.007Leatherandfurproducts0.0100.004Woodfurniture0.0090.003Repairservices0.0220.003Electricity0.1110.323Othermanufacturing0.0100.007CIER(1988).P.Debaere/JournalofInternationalEconomics64(2004)485–501 thesamearea.Withthiscriterioninmind,Davisetal.selectedtheJapaneseregionsandforthesamereason,IusedUKregionaldatafromtheRegionalTrends5.ConclusionInthispaper,Ihaveinvestigatedwhethertheunevendistributionofendowmentsacrosstheregionswithinacountrytriggersspecializationofproductionandwhetheritmakesregionsproducedifferentsetsofgoodsatdifferentfactorprices.Tostudythisquestion,IoperationalizedtheconceptoflumpinessthatCourantandDeardorff(1992)developed.ForJapan,theUnitedKingdomandevenforIndia,Ifoundthatlumpinessshouldnotgivewaytospecializationanddifferentfactorpricesacrossregions.Thissuggeststhatspecializationduetolumpinessisnotaregionalphenomenon.Thisfindingaccordswellwiththeempiricalliteraturethatarguesthatspecializationduetoendowmentdifferencesisaninternationalphenomenon:Allcountriescannotproducethesamesetofgoods.TheresultsalsosuggestthatregionalspecializationismostlikelynotamajorexplanationforwhytradepatternsdonotconformtothebasicpredictionsoftheHeckscher–Ohlinmodel.AcknowledgementsIthankAlanDeardorff,UfukDemiroglu,ScottFreeman,GeneGrossman,WolfgangKeller,GeraldOettingerandJimRauchfortheirhelpfulsuggestions.Twoanonymousrefereesalsoprovidedhelpfulsuggestionsandcomments.DavisandWeinsteingraciouslyprovideddataforJapan.ClaudioBonillafromtheUniversityofTexas,Austin,providedexcellentresearchassistance.ThisprojectwasfundedinpartbyUniversityofTexasSummerResearchfunds.Allremainingerrorsaremine.CentreforIndustrialandEconomicResearchx,1998.IndustrialDatabook.NewDelhi.Courant,P.,Deardorff,A.,1992.Internationaltradewithlumpycountries.JournalofPoliticalEconomy100,198–210.Courant,P.,Deardorff,A.,1993.Amenities,nontradedgoods,andthetradeoflumpycountries.JournalofUrbanEconomics34,299–317.Davis,D.,Weinstein,D.,Bradford,S.,Shimpo,K.,1997.UsinginternationalandJapaneseregionaldatatodeterminewhenthefactorabundancetheoryoftradeworks.AmericanEconomicReview87,421–446.Deardorff,A.,1994.Thepossibilityoffactorpriceequalization,revisited.JournalofInternationalEconomics36,167–175. Fromthisperspective,IconsideritanadvantagethatJapanandtheUKareislandeconomies.Oneshouldnotbeconcernedaboutcross-borderemployment.Notethatwhenregionsproducedifferentgoodsatdifferentfactorprices,theactualpatternoftradeofalumpycountrydiffersfromanypredictionbasedoncountrywidedata.Thelatterobservationhassometimesbeeninvokedtoexplainlong-standingquestionsinthetradeliteraturesuchastheLeontiefparadox(seeCourantandDeardorff,1993P.Debaere/JournalofInternationalEconomics64(2004)485–501 Debaere,P.,Demiroglu,U.,2003.Onthesimilarityofcountryendowments.JournalofInternationalEconomics59,101–136
  Demiroglu,U.,Yun,K.,1999.Thelensconditionforfactorpriceequalization.JournalofInternationalEco-nomics47,449–456.Evenett,S.,Keller,W.,2002.Ontheoriesexplainingthesuccessofthegravityequation.JournalofPoliticalEconomy110,281–316.Flam,H.,June,M.,Flanders,M.,2002.TheYoungOhlinonthetheoryofinterregionalandinternationaltrade.In:Findlay,R.,Jonung,L.,Lundahl,M.(Eds.),BertilOhlin,ACentennialCelebration(1899–1999).MITPress,Cambridge,pp.175–192.Hanson,G.,2000.Scaleeconomiesandthegeographicconcentrationofindustry.JournalofEconomicGeog-raphy1,255–276.Hanson,G.,Slaughter,M.,1999.TheRybczynskitheorem,factor-priceequalization,andimmigration:evidencefromUSStates.NationalBureauofEconomicResearchWorkingPaper7074.Helpman,E.,Krugman,P.,1985.MarketStructureandForeignTrade.MITPress,Cambridge.Kim,S.,1995.Expansionofmarketsandthegeographicdistributionofeconomicactivities.QuarterlyJournalofEconomics110,881–908.Kim,S.,1999.Regions.Resourcesandeconomicgeography:sourcesofUSregionalcomparativeadvantage,1880–1987.RegionalScienceandUrbanEconomics,1–32.Krugman,P.,1991.GeographyandTrade.MITPress,Cambridge.Krugman,P.,Venables,A.,1995.Theseamlessworld:aspatialmodelofinternationalspecialization.NationalBureauofEconomicResearchWorkingPaper5220.Leamer,E.,1987.Pathsofdevelopmentinthethreefactorn-goodgeneralequilibriummodel.JournalofPoliticalEconomy95,961–999.Machin,S.,1996.WageinequalityintheUK.OxfordReviewofEconomicPolicy12,47–64.Melvin,R.,1985.Theregionaleconomicconsequencesoftariffsanddomestictransportationcosts.CanadianJournalofEconomics,238–257.Moroney,J.,Walker,J.,1966.AregionaltestoftheHeckscher–Ohlinhypothesis.JournalofPoliticalEconomy74,573–586.OECD,1996.OECDDataonskills:employmentbyindustryandoccupation,STIWorkingPaper1998/4.Schott,P.,2003.Onesizefitsall?Heckscher–Ohlinspecializationinglobalproduction,AmericanEconomicReview93(2),686–708.Trefler,D.,1993.Internationalfactorpricedifferences:Leontiefwasright!JournalofPoliticalEconomy101,961–987.P.Debaere/JournalofInternationalEconomics64(2004)485–501