EBSCOhost_ Trade Liberalization, Exit, and Product

TradeLiberalization,Exit,andProductivityImprovements:EvidencefromChileanPlantsNINAPAVCNIKDartmouthCollegeandNBERFirstversionreceivedDecember1998;finalversionacceptedJune2001(Eds.)ThispaperempiricallyinvestigatestheeffectsofliberalizedtradeonplantproductivityinthecaseofChile.Chilepresentsaninterestingsettingtostudythisrelationshipsinceitunderwentamassivetradeliberalizationthatsignificantlyexposeditsplantstocompetitionfromabroadduringthelate1970sandearly1980s.Methodologically,Iapproachthisquestionintwosteps.Inthefirststep,Iestimateaproductionfunctiontoobtainameasureofplantproductivity.Iestimatetheproductionfunctionsemiparametricallytocorrectforthepresenceofselectionandsimultaneitybiasesintheestimatesoftheinputcoefficientsrequiredtocon-structaproductivitymeasure.Iexplicitlyincorporateplantexitintheestimationtocorrectfortheselectionprobleminducedbyliquidatedplants.Thesemethodologicalaspectsareimportantinobtainingareliableplant-levelproductivitymeasurebasedonconsistentestimatesoftheinputcoefficients.Inthesecondstep,Iidentifytheimpactoftradeonplants’productivityinaregressionframeworkallowingvariationinproductivityovertimeandacrosstraded-andnontraded-goodssectors.Usingplant-levelpaneldataonChileanmanufacturers,Ifindevidenceofwithinplantproductivityimprovementsthatcanbeattributedtoaliberalizedtradefortheplantsintheimport-competingsector.Inmanycases,aggregateproductivityimprovementsstemfromthereshufflingofresourcesandoutputfromlesstomoreefficientproducers.1.INTRODUCTIONDuringthe1980s,manydevelopingcountriesabandonedtheirinward-lookingdevelop-mentstrategiesfordrastictradeliberalizationprogrammes.Thesupportersofthesereformsclaimedthattheexposureofhomeproducerstoadditionalimportcompetitionandeasieraccessofplantstoforeigntechnologywouldenhanceproductivityindomesticindustries.Despitethehighprofileofthistopic,surprisinglylittleisknownabouttheactualeffectsoftradepolicychangesonplant’sproductivity.Thetheoreticaltradeliteratureoffersconflictingpredictionsabouttheevolutionofplant-levelproductivityfollowingatradeliberalizationepisode,especiallyincaseswhereimperfectcompetitionispresent.Rodrik(1988,1991)providesanexcellentsurveyofthemainissues.Ononehand,tradeliberalizationexposesdomesticproducerstoforeigncompetition,reducestheirmarketpower,andmightforcethemtoexpandoutputandmovedowntheaveragecostcurve;thismightresultintheexploitationoftheeconomiesofscale.Gainsfromscaleeconomiesarenotverylikelyindevelopingcountries,wheretheincreasingreturnstoscaleareusuallyassociatedwiththeimport-competingindustries,whoseoutputislikelytocontractasaresultofintensifiedforeigncompetition.InmodelssuchasRodrik(1988),whereplantsinvestinsuperiortechnologytoreducetheircost,theirincentivetocutcostsmightincreasewiththeirmarketshare.Iftradeliberalizationreducesthedomesticmarketsharesofunshieldeddomesticproducerswithoutexpandingtheirinternationalsales,theirincentivestoinvestinimprovedtechnologywilldecreaseasReviewofEconomicStudies(2002)69,245–2760034-6527/02/00100245$02.00
2002TheReviewofEconomicStudiesLimited245protectionceases.Thiseffectreducesthebenefitsoftariffreductionsthatlowertherelativepricesofimportedcapitalgoodsandeaseaccesstoforeigntechnologyfordomesticplants.Althoughtradeliberalizationfacilitatesprocurementofforeigntechnology,itisquestionablewhetherdomesticplantsactuallyadoptbettertechnology.ArecentseriesofpapersbyEatonandKortum(1996,1997)modelshowthebenefitsofinnovationarespreadfromonecountrytoanothereitherthroughdiffusionoftechnologyorthroughtheexchangeofgoods.Theyfindthattheimpactofdiffusionofknowledgeonproductivitydependscruciallyontheproximityofacountrytothetechnologysourceandtheflexibilityofthedomesticlabourforce.Inlightofmanymodelspredictingthattradediffusesinnovationandknowledge,itisalsopuzzlingthatstudiesofconvergenceofproductivityacrosscountriessuchasBernardandJones(1996)findconvergenceinserviceratherthanmanufacturingsector,whichisextensivelyaffectedbyinternationaltrade.Whiletradetheoryhasconsideredintraindustrygainsfromtradeliberalizationthroughexpansionofeconomiesofscale,ithassofarnotexploredtheimplicationsofplantheterogeneitywithinanindustryasmostofthetraditionaltrademodelsrelystronglyonarepresentativeplantassumption.Recentworkexplainingplant-leveldatabyOlleyandPakes(1996),RobertsandTybout(1996),andAw,ChenandRoberts(1997),introducesevidenceofasignificantdegreeofplant-levelheterogeneitywithinanindustry.Thepresenceofplant-levelheterogeneitysuggeststhattradeliberalizationmayyieldproductivityimprovementsbyreshufflingtheresourcesamongplantswithinthesameindustryandthatplantdynamicssuchasexitmaycontributesignificantlytothisprocess.Inparticular,highlevelsofprotectionmayaccommodatethecoexistenceofproducerswithdifferentlevelsofproductivity.Byreducingprotection,tradeliberalizationlowersdomesticprices,potentiallyforcinghighcostproducerstoexitthemarket.Thiswouldleadtoareallocationofoutputfromlessefficienttomoreefficientproducers.Thesepro-ductivitygainsemergeonlyiftheirreversibilityofinvestmentincapitalequipmentdoesnotimpedetheexitofthelessproductiveplants.Eveniftradeliberalizationenha