搜索
Verypreliminaryandincomplete.Pleasedonotquote.Commentsarewelcome.TheEffectofHRMPracticesandR&DInvestmentonWorkerProductivityFredrikAndersson,CornellUniversityandU.S.CensusBureauClairBrown,UniversityofCalifornia,BerkeleyBenjaminCampbell,WhartonSchool,UniversityofPennsylvaniaHyowookChiang,UniversityofMarylandandU.S.CensusBureauYookiPark,UniversityofCalifornia,BerkeleyMay23,2005ThisdocumentincludestheresultsofresearchandanalysisundertakenbytheU.S.CensusBureaustaffandco-authors.IthasundergoneaCensusBureaureviewmorelimitedinscopethanthatgiventoofficialCensusBureaupublications,andisreleasedtoinforminterestedpartiesofongoingresearchandtoencouragediscussionofworkinprogress.ThedatausedareapartoftheU.S.CensusBureau’sLongitudinalEmployer-HouseholdDynamicsProgram(LEHD),whichispartiallysupportedbytheNationalScienceFoundationGrantSES-9978093toCornellUniversity(CornellInstituteforSocialandEconomicResearch),theNationalInstituteonAging(R01-AG18854-01),andtheAlfredP.SloanFoundation.SupportwasalsoprovidedbytheInstituteofIndustrialRelations,UCBerkeley.Theviewsexpressedhereinareattributableonlytotheauthor(s)anddonotrepresenttheviewsoftheU.S.CensusBureau,itsprogramsponsorsordataproviders.SomeorallofthedatausedinthispaperareconfidentialdatafromtheLEHDProgram.TheU.S.CensusBureauispreparingtosupportexternalresearchers’useofthesedata;pleasecontactU.S.CensusBureau,LEHDProgram,FB2138-3,4700SilverHillRd.,Suitland,MD20233,USA.WehavebenefitedfromdiscussionsandfeedbackfromCharlieBrown,PeterCappelli,EricaGroshen,AndrewHildreth,DanielParent,LindaSattler,andEdwardWolff,seminarparticipantsatBerkeleyandWharton,andparticipantsattheNBERSummerInstitute.ABSTRACTUsingdataonalargesampleofelectronicsfirmsinsevenlargestatesfromanewlydevelopedemployer-employeematcheddatabase(LongitudinalEmployerHouseholdDynamics,LEHD),weexaminetheimpactofhumanresourcemanagement(HRM)practicesandtechnologyonworkerproductivity.WeidentifyHRMclustersforfirmsbasedonfirm-levelobservationsofninemeasuresofHRMoutcomes.Next,weuseprincipalcomponentsanalysistoexaminetherelationshipsbetweentheHRMmeasures.ThenweusetheseprincipalcomponentsandtheirinteractionswithR&Dinvestmentasexplanatoryvariablesinaworkerproductivityregression.Wefindthattherearelargedifferencesontheimpactofhumanresourcepracticesonlaborproductivityacrosslevelsoftechnologicalinvestment.OurpreliminaryresultsindicatethatfirmswithhighlevelsofR&DinvestmentandHRMsystemswithmultipleportsofentry,performanceincentives,andlowerturnoverhavehigherworkerproductivitythancomparablehigh-R&DfirmswithouttheseHRMpractices.Similarly,firmswithlowR&DthatimplementHRMsystemswithperformanceincentiveshavehigherproductivitythanlowR&Dfirmswithoutperformanceincentives.TheseresultssuggestthathighR&DfirmsaremorelikelytobuynewskillscomparedtolowR&Dfirms,andyetthesehighR&Dfirmssufferiftheylosetoomanyexperiencedworkers.Thesefindingsareconsistentwiththeimplicationsofour“makeversusbuy”modelofworkforceskilladjustmentasaresponsetotechnologicalchange.1.IntroductionAsthepaceoftechnologicalchangehasquickened,andasglobalcompetitionhasshortenedproductlifecycles,firmshavehadtorethinktheirtechnologyinvestmentstrategiesandtheirhumanresourcemanagementpracticesinordertoremaincompetitive.Thispaperexaminestherelationshipbetweenfirm-levelresearchanddevelopmentinvestment(R&D)andfirms’humanresourcemanagement(HRM)practicesinahigh-techindustry.ThereareseveralchannelsthroughwhichfirmR&DandHRMdecisionsmayberelated.Iftechnologyandlaborforceskillsarecomplementsinfirms’productionfunctions,andifHRMsystemsimpactthecostofacquiring,developing,andretainingtheportfolioofskillsinafirm,thenfirms’choiceofHRMsystemaffectstheirabilitytoadjustworkerskilllevelstomaximizethevalueoftheirtechnologicalinvestments.Forexample,iffirmsneedtoaugmenttheskilloftheirworkforcetocomplementaninvestmentintechnology,theyfaceatraditional“makeversusbuy”problem.FirmscanstructuretheirHRMsystemtodevelopthenecessaryskillsin-houseortheycanstructuretheirHRMtoattractworkerswiththenecessaryskillsontheexternalmarket.Also,workersgainskillsdirectlyfromlearning-by-doingintheirR&Dactivities,whichtriggersthefirmtoenactHRMpoliciesthatretaintheincreasinglyproductiveknowledgeworkers.Additionally,firms’productstrategiesdirectlyaffectboththeirR&DchoicesandtheirHRMchoice(Lazear(1998)andBaronandKreps(1999)).Althoughtherelationshipoftechnologicalchange,compensationandtenureattheindividuallevelhasbeenwell-studied,surprisinglylittleisknownabouttherelationshipbetweentechnologicalchangeandfirms’HRMdecisions.Previousresearchonthistopichasbeeneithercasestudyorientedorhasutilizeddatafrombroadestablishment-levelsurveys.Thisprojectconnectsthesemicroandmacroapproachesbyusingdatathatallowsustocapitalizeonthestrengthsofeachtypeofresearch.UsingdatafromtheLongitudinalEmployer-HouseholdDynamics(LEHD)Program,weareabletoexaminetheHRMpracticesandfirm-levelcharacteristicsformanyfirmsinsevenstates,whichallowsustobuildonthebreadthoftheestablishment-levelsurveyresearch.Additionally,wecantracktheoutcomesoftheuniverseofworkerswithineachestablishment.Weu