《國際貿(mào)易(第七版)》Chap

EmpiricalTestsoftheFactorEndowmentsApproach

Chapter9LearningObjectivesAnalyzethefailureofU.S.tradepatternstoconformtoH-Opredictions.ExaminepossibleexplanationsfortheU.S.tradeparadox.Describeissuesarisingfrommulti-countryH-Otests.Assesstheroleoftradeingeneratinggrowingincomeinequalityindevelopedcountries.Leontief’sTest1950s:LeontiefconductsthefirstimportanttestofH-O.UsingU.S.dataLeontiefcalculatedaverageamountofcapitalandlaborembodiedinU.S.exports.averageamountofcapitalandlaborembodiedinU.S.imports.Leontief’sTestPresumably,theU.S.wasrelativelyK-abundantatthattime.Therefore,accordingtotheH-Omodel,theU.S.shouldtendtoexportK-intensiveproducts,andimportL-intensiveproducts.Thatis,thecapital-laborratioforU.S.exportsshouldbegreaterthanthecapital-laborratioforU.S.imports.TheLeontiefParadoxLeontieffoundsomethingsurprising: (K/L)exports=$13,991perperson-year(K/L)imports=$18,184perperson-yearThisistheoppositeofwhattheH-Omodelpredicts.ThisfindingcametobeknownastheLeontiefParadox.TheLeontiefParadoxToseethisfromanotherangle,considertheLeontiefstatistic[(K/L)imp]/[(K/L)exp]IfH-O-Siscorrect,thisstatisticshouldbelessthanonefortheU.S.However,Leontieffoundthestatistictobe($18,184/$13,991)orabout1.3.ExplanationsfortheLeontiefParadoxMuchresearchsinceLeontief’stimehasfocusedontryingtoexplaintheparadox.Doanyoftheseexplanations“rescue”theH-Omodel,oristhemodeljustwrong?Explanation#1:DemandReversals

Recall:whentheK-abundantcountryhasverystrongdomesticdemandfortheK-intensiveproduct,andtheL-abundantcountryhasverystrongdomesticdemandforL-intensiveproducts,therecanbeademandreversal:theK-abundantcountrywillexporttheL-intensiveproductbecauseithastherelativecostadvantageinit.

Explanation#1:DemandReversalsTherefore,theH-Otheorembreaksdown.Ifdemandreversalsarecommonplace,wemightexpecttheU.S.toexportrelativelylabor-intensiveproducts.Explanation#1:DemandReversalsSo:isthereanyevidenceforwidespreaddemandreversals?No.Demandpatternsareactuallyquitesimilar,atleastamongindustrializedcountries.Furthermore,demandreversalsimplythatU.S.wagesshouldbelow.Thiswouldbeahardargumenttosupport.Weneedtolookfurthertoexplaintheparadox.Explanation#2:FactorIntensityReversalsRecall:aFIRoccurswhenagoodisrelativelyK-intensiveatonesetoffactorprices,butrelativelyL-intensiveatanother.IfFIRsoccuroften,theH-Otheoremcannotbevalidforbothcountries,andsowemightexpecttheLeontiefparadox.Explanation#2:FactorIntensityReversalsMinhas(1962)foundevidencethatFIRsarefairlycommonplace.LaterworkbyHufbauer(1966)andBall(1966)suggeststhatMinhasoverstatedthematter;theremaybesomeFIRsintherealworld,butnotasmanyasMinhassuggested.ItwouldseemthatifthereisanexplanationoftheLeontiefparadox,itlieselsewhere.Explanation#3:TheU.S.TariffStructureTheH-Omodelassumesfreetrade,butinfacttherearebarriers(e.g.,tariffs).TheStolper-Samuelsontheoremleadsustoexpectthattheownersofthescarcefactorwillbeprotectionist.IntheU.S.,thiswilllikelymeanthatitisL-intensiveimportsthatarebeingkeptout.Explanation#3:TheU.S.TariffStructureThetariffstructurecouldmaketheLeontiefstatisticartificiallyhigh,andperhapsleadtotheparadox.Consideranexample:Explanation#3:TheU.S.TariffStructure(AnExample)9-15Explanation#3:TheU.S.TariffStructure(AnExample)9-16Explanation#3:TheU.S.TariffStructure(AnExample)Supposethat(K/L)exp=$16,000.ThenthefactthattariffsexistmeansthattheLeontiefstatisticis$18,333/$16,000=1.14;itwouldhavebeen$14,500/$16,000=0.9undertheassumptionoffreetrade.ThismeansthatLeontief’sparadoxmightbetheresultoftariffs,andisn’tevidenceagainsttheH-Omodel.Explanation#3:TheU.S.TariffStructureAstudybyBaldwin(1971)suggeststhat(K/L)impfortheU.S.wouldbeabout5%lowerifweallowforthetariffstructure.ThiswouldlowerLeontief’sstatisticfrom1.3to1.23.Thislessenstheextentoftheparadoxwithoutexplainingitall.Explanation#4:AddingOtherFactorsofProductionKeesing(1966)suggestssubdividinglaborintoeightskillcategories.HefoundthattheU.S.exportsalotofskilledlabor-intensiveproducts;itistheunskilledlabor-intensiveproductsthatwetendtoimport.Explanation#4:AddingOtherFactorsofProductionSincetheU.S.isrelativelyskilledlabor-abundant,thissuggeststhattheH-Omodeldoesexplaintradeaccurately:theLeontiefParadoxdisappears.Laterstudieshavesupportedthisfinding.Explanation#4:AddingOtherFactorsofProductionLeontief(1956)andHartigan(1981)foundthataddingnaturalresourcesasafactorofproductioneliminatestheparadox.However,Baldwin(1971)foundthataddingnaturalresourcesdoesnotcompletelyeliminatetheparadox.TheLeontiefParadox:TheBottomLineAllowingfordemandreversals,FIRs,thetariffstructureandnaturalresourcesasafactorofproductionmaylessentheextentoftheparadox.Allowingfordifferentlevelsofskillinthelaborforcedoesseemtoeliminatetheparadox.TheH-Omodelappearstobeserviceable.TestsoftheH-OModelforOtherCountriesManystudiesprovidesupportforH-OStolperandRoskamp(1961):EastGermanyTatemotoandIchimura(1959):JapanRosefielde(1974):USSROtherstudiesdidnotsupportH-OWahl(1961):CanadaBharadwaj(1962):IndiaMoreRecentTestsofH-OSternandMaskus(1981)lookedatexportsandimportsfor128differentU.S.industries.Theyestimatedthefollowingregressionequation:(X-M)=-18.54-0.08K+0.06H-2.83LMoreRecentTestsofH-O(X-M)=-18.54-0.08K+0.06H-2.83LInterpretation:themoreKanindustryusesthelessisexported.themorelaboranindustryusesthelessisexported.themorehumancapitalanindustryusesthemoreisexported.ThisisbasicallythesamefindingasKeesing’s.MoreRecentTestsofH-OHarknessandKyle(1975)addednaturalresourcestotheregressionequation.foundsimilarresults:theLeontiefparadoxcanberesolvedbyconsideringotherfactorsbesidesjustKandL.MoreRecentTestsofH-OMaskus(1985),Bowenetal.(1987),Gourdon(2009),andMurielandTerra(2009)havealsoaddedmultiplefactorsofproduction.Ingeneral,theirresultsconfor