1、外文翻译原文INTERINDUSTRYWAGEDIFFERENCESANDINDIVIDUALHETEROGENEITYMATERIALSOURCEOXFORDBULLETINOFECONOMICSKRUEGERANDSUMMERS,1987,1988KATZANDSUMMERS,1989THESEDIFFERENTIALSEXHIBITAHIGHDEGREEOFSTABILITYOVERTIMEANDAPPEARTOHOLDACROSSAVARIETYOFCOUNTRIESWITHDISTINCTINSTITUTIONALANDSTRUCTURALARRANGEMENTSGITTLEMANA
2、NDWOLFF,1993KAHN,1998THEYALSOPERSISTACROSSDIFFERENTTYPESOFWORKERSANDESTABLISHMENTSIZESUCHDISPARITIESAREDIFFICULTTOEXPLAINFULLYBYTHEDISTRIBUTIONOFHUMANCAPITALACCUMULATIONACROSSINDUSTRIESORBYCOMPENSATINGDIFFERENTIALSFORNONPECUNIARYJOBATTRIBUTESAFFECTINGTHEUTILITYOFWORKERSTHEIRPERSISTENCEISALSOINCONSIS
3、TENTWITHTHENOTIONOFTRANSITORYDISEQUILIBRIUMPHENOMENABROUGHTABOUTBYADJUSTMENTSTOLABOURSUPPLYORDEMANDINTHEPRESENCEOFIMPERFECTSHORTRUNLABOURMOBILITYACCORDINGLY,ANUMBEROFNONCOMPETITIVEEXPLANATIONSHAVEBEENPROPOSEDINSTANDARDHUMANCAPITALTHEORY,ITISARGUEDTHATJOBATTRIBUTESTHATDONOTDIRECTLYAFFECTTHEUTILITYOFW
4、ORKERSSHOULDHAVENOEFFECTONTHEDETERMINATIONOFINDIVIDUALWAGESINCONTRAST,NONCOMPETITIVETHEORIESOFWAGEDETERMINATIONPOSITTHATSUCHATTRIBUTESCANHAVEASYSTEMATICEFFECTONWAGESBECAUSETHEYINFLUENCETHEOPTIMALWAGEFORFIRMSTOSETTHEPURPOSEOFTHESETHEORIESISTODETERMINEWHYFIRMSMAYFINDITPROFITABLEANDAREABLETOPAYWAGESHIG
5、HERTHANTHEMARKETCLEARINGRATEINADDITION,SUCHEXPLANATIONSMUSTALSOEXPLAINWHYTHEIMPORTANCEOFSUCHFACTORSDIFFERSBYINDUSTRYANUMBEROFPOSSIBILITIESHAVEBEENSUGGESTEDEFFICIENCYWAGEMODELSEMBODYTHENEEDFORHIGHWAGESTOELICITWORKEREFFORTWEISS,1980AKERLOF,1984SHAPIROANDSTIGLITZ,1984STIGLITZ,1985UNIONTHREATMODELSDICKE
6、NS,1986UNDERLINETHETHREATOFCOLLECTIVEACTION,WHILSTINSIDEROUTSIDERMODELSSOLOW,1985LINDBECKANDSNOWER,1986STRESSTHEINCUMBENTPOWEROFEMPLOYEESWHENBARGAININGFORASHAREOFINDUSTRYRENTSEACHOFTHESEMOTIVATIONSPREDICTSCORRELATIONSBETWEENINDUSTRYWAGEPREMIAANDINDUSTRYCHARACTERISTICSCONSISTENTWITHTHEAVAILABLEEVIDEN
7、CEHOWEVER,NONEOFTHESEMODELSISABLETOATTESTTOALLOFTHEEVIDENCEANALTERNATIVEEXPLANATIONISTHATOBSERVEDWAGEDIFFERENTIALSREFLECTUNOBSERVEDDIFFERENCESINWORKERQUALITYSEVERALSTUDIESMURPHYANDTOPEL,1987KRUEGERANDSUMMERS,1988GIBBONSANDKATZ,1992KEANE,1993SHIPPEN,1999ABOWD,KRAMARZANDMARGALIS,1999HAVEUTILIZEDLONGIT
8、UDINALDATAANDFIXEDEFFECTSMODELSTOTESTTHISHYPOTHESIS,WITHMIXEDRESULTSKRUEGERANDSUMMERS1988,FOREXAMPLE,PRESENTLITTLEEVIDENCETOSUPPORTTHEIMPORTANCEOFUNOBSERVEDHETEROGENEITYINTHEDETERMINATIONOFINDUSTRYPAYINCONTRAST,MURPHYANDTOPEL1987,KEANE1993,ANDABOWDETAL1999FINDTHATUNOBSERVEDHETEROGENEITYEXPLAINS66,84
9、,ANDAPPROXIMATELY90,RESPECTIVELY,OFTHEAPPARENTDIFFERENTIALINLOGWAGESACROSSINDUSTRIESGIBBONSANDKATZ1992PROVIDEANEXPERIMENTTHATDOESNOTRULEOUTANUNOBSERVEDABILITYEXPLANATION,BUTANOTHEREXPERIMENTISSYMPATHETICTOTHEKRUEGERANDSUMMERS1988THESISTHATTRUEINDUSTRYDIFFERENTIALSEXISTACROSSINDUSTRIES,EVENFORIDENTIC
10、ALWORKERSIDENTIFYINGTHETRUENATUREANDSOURCESOFOBSERVEDINTERINDUSTRYWAGEDIFFERENTIALSISIMPORTANTONBOTHRESEARCHANDPOLICYGROUNDSANDALSOWITHREGARDTOINDIVIDUALWELFARERENEWEDINTERESTINTHESTRUCTUREOFWAGESHASOCCURREDATATIMEWHENWAGEINEQUALITYINBOTHTHEUSANDTHEUKISHIGHERTHANATANYOTHERTIMEINTHEPAST50YEARSKATZ,LO
11、VEMANANDBLANCHFLOWER,1995MACHIN,1996THISINCREASEDDISPERSIONHASOCCURREDBOTHBETWEENANDWITHINGROUPSWITHTHESAMEOBSERVABLECHARACTERISTICSCOMPETITIVEMODELSOFTHELABOURMARKETIMPLYTHATCHANGESINDISPERSIONSHOULDBELARGELYTRANSITORYINNATUREINCREASEDINEQUALITY,INTHISREGARD,MAYBECONSIDEREDASBEINGSHAREDAMONGSTINDIV
12、IDUALSINCONTRAST,NONCOMPETITIVEMODELSOFTHELABOURMARKETIMPLYTHATSUCHCHANGESMAYBELARGELYPERMANENTTHEACCEPTANCEOFSUCHEXPLANATIONSTHUSHASSIGNIFICANTPOSITIVEANDNORMATIVEIMPLICATIONSTHISPAPERSETSOUTTOEXAMINETHEEXISTENCE,MAGNITUDEANDSTABILITYOFINTERINDUSTRYWAGEDIFFERENTIALSFORBRITAINUSINGLONGITUDINALDATADR
13、AWNFROMTHEFIRSTEIGHTWAVESOFTHEBRITISHHOUSEHOLDPANELSURVEYBHPSTHESURVEYENABLESUSTOEXAMINECROSSSECTION,POOLEDANDLONGITUDINALEVIDENCEFORBRITISHINTERINDUSTRYWAGEDIFFERENTIALSBETWEEN1991AND1998THEREAREEXTREMELYFEWPANELSTUDIESOFTHEINTERINDUSTRYWAGESTRUCTUREINTHEEXISTINGLITERATUREAND,TOOURKNOWLEDGE,ONLYTWO
14、FORBRITAINHILDRETH1995PRESENTSCROSSSECTIONALANDLONGITUDINALEVIDENCEBASEDONTHEFIRSTTWOWAVESOFTHEBHPSBENITO2000EXTENDSTHISANALYSISACROSSTHEFIRSTFOURWAVESOFTHEBHPSWHILSTPROVIDINGSOMEADDITIONALINNOVATIVEANALYSISCONCERNINGINDUSTRYPROFITABILITYBOTHOFTHESESTUDIESPRESENTEVIDENCETHATMUCHOFTHECROSSSECTIONALVA
15、RIATIONINWAGESATTRIBUTEDTOINDUSTRYAFFILIATIONCANBEACCOUNTEDFORBYUNOBSERVEDINDIVIDUALHETEROGENEITYCAPTUREDINFIXEDEFFECTSESTIMATIONHOWEVER,BOTHSTUDIESFOCUSSOLELYONONEDIGITINDUSTRYAFFILIATIONAND,COMBINEDWITHTHEFACTTHATTHEYHAVERELATIVELYFEWWAVESOFDATA,RELYONARELATIVELYSMALLNUMBEROFINDUSTRYSWITCHERSTHEYA
16、LSODONOTACCOUNTFORMEASUREMENTERRORWHICHISKNOWNTOINTERFEREINPANELAPPROACHESTOESTIMATIONMUCHMORESEVERELYTHANCROSSSECTIONALESTIMATION,ESPECIALLYWHENTHENUMBEROFSWITCHERSISSMALLSEE,EGCHOWDHURYANDNICKELL,1985SWAFFIELD,2001THISSTUDYADVANCESTHEANALYSISOFTHISTOPICBYEXPLICITLYADDRESSINGTHEROLEOFUNOBSERVEDHETE
17、ROGENEITYASANEXPLANATIONOFOBSERVEDINTERINDUSTRYDIFFERENTIALS,WHILSTTAKINGACCOUNTOFBOTHINDUSTRYAGGREGATIONANDMEASUREMENTERRORUTILIZINGEIGHTWAVESOFBHPSDATAFORBOTHONEANDTWODIGITINDUSTRYAFFILIATIONPERMITSAGREATERNUMBEROFINTERINDUSTRYTRANSITIONSTOBEOBSERVEDANDHENCEALLOWSUSTOESTIMATETHECRUCIALINDIVIDUALEF
18、FECTSMOREACCURATELYBYUSINGAMETHODOFESTIMATIONATTRIBUTEDTOHAISKENDENEWANDSCHMIDT1997WEAREALSOABLETOPROVIDEAUNIQUEESTIMATEOFINTERINDUSTRYWAGEDISPERSIONINCONTRAST,INTHEKRUEGERANDSUMMERS1988METHODOLOGYUSEDBYBENITO2000ANDHILDRETH1995INTERALIA,THEMEASUREMENTOFINTERINDUSTRYWAGEDISPERSIONISAFUNCTIONOFTHEOMI
19、TTEDINDUSTRY,WHICHMEANSTHATSTUDIESWHICHUSETHISMETHODAREOFTENNOTCOMPARABLEBECAUSEDIFFERENTBASEINDUSTRIESHAVEBEENCHOSENBYTHERESEARCHERSINADDITION,WEEXPLICITLYINVESTIGATETHEIMPACTTHATMEASUREMENTERRORSININDUSTRYAFFILIATIONMAYHAVEONOURRESULTSTHEADDITIONALWAVESPROVIDEAMOREROBUSTTESTFORTHEPERSISTENCEANDSTA
20、BILITYOFTHEINTERINDUSTRYDIFFERENTIALSOVERTIME,ANDHENCEALLOWUSTOINVESTIGATEANUMBEROFALTERNATIVEEXPLANATIONSDEEMEDIMPORTANTINTHEEMPIRICALDEBATEOURRESULTSSHOWTHATDIFFERENCESINWAGESCANBELARGELYEXPLAINEDBYOBSERVEDANDUNOBSERVEDINDIVIDUALEFFECTS,ALONGWITHAROLEFORINTERINDUSTRYDIFFERENTIALS,ANDJOBSPECIFICCHA
21、RACTERISTICSTHEKEYTOTHISSUBSTANTIALIMPROVEMENTINEXPLANATIONOVERTHECROSSSECTIONLITERATUREONINDIVIDUALWAGESISTHEFIXEDEFFECTSMETHODOLOGYTHATCONTROLSFORUNOBSERVEDINDIVIDUALHETEROGENEITYINADDITION,THECONTROLOFUNOBSERVEDINDIVIDUALTRAITSCONSIDERABLYREDUCESTHEEXTENTOFINTERINDUSTRYWAGEDIFFERENTIALSTHEIMPORTA
22、NCEOFUNOBSERVEDHETEROGENEITYINWAGEDETERMINATIONISDIFFICULTTOASCERTAINKRUEGERANDSUMMERS1988ADOPTTWOMETHODSTOTESTFORTHEPRESENCEOFUNOBSERVEDDIFFERENCESINLABOURQUALITYFIRST,THEYCOMPARECROSSSECTIONWAGEEQUATIONSFORMAY1979CURRENTPOPULATIONSURVEYCPSDATABOTHWITHANDWITHOUTCONTROLSFORHUMANCAPITALAGE,GENDER,RAC
23、E,EDUCATION,TENUREANDOCCUPATIONTHEINCLUSIONOFLABOURCONTROLSISREPORTEDTOHAVENOIMPACTONESTIMATEDINDUSTRYDIFFERENTIALSSECONDLY,THEYUSEMATCHEDCPSDATASETSALONGWITHTHE1984DISPLACEDWORKERSSURVEYDWSTOESTIMATEFIRSTDIFFERENCEEQUATIONSONINDUSTRYMOVERSTHEDWSPROVIDESASAMPLEWHERETHESEVERANCEFROMTHEPREVIOUSJOBISEX
24、OGENOUSTOTHEEMPLOYEECAUSEDBYPLANTCLOSURE,FOREXAMPLETHESEESTIMATESAPPEARBROADLYSIMILARTOTHEIRCROSSSECTIONRESULTSBOTHMETHODSTHUSLEADTHEAUTHORSTOREJECTARGUMENTSFORUNOBSERVEDDIFFERENCESINLABOURQUALITYGIBBONSANDKATZ1992ADDSOMESUPPORTTOTHISCONCLUSIONTHEYESTIMATEFIRSTDIFFERENCEMODELSUSINGDATAFROMTHEDWSFORT
25、HEPERIOD198486ANDREPORTTHATINDUSTRYSWITCHERSAPPEARTOEARN,ONAVERAGE,97OFTHERELEVANTCROSSSECTIONDIFFERENTIALTHEAUTHORSACCORDINGLYREJECTUNOBSERVEDABILITYEXPLANATIONSANDCONCLUDETHATINDUSTRYEFFECTSAREINDEEDIMPORTANTINEXPLAININGINTERINDUSTRYWAGESOURAIMISTOINVESTIGATETHEROLEOFUNOBSERVEDHETEROGENEITYASANEXP
26、LANATIONOFTHEOBSERVEDINTERINDUSTRYDIFFERENTIALSWEUTILIZEGENUINEPANELDATARATHERTHANMATCHEDDATAANDTHEIMPROVEDMETHODOLOGYOFHAISKENDENEWANDSCHMIDT1997TOSHOWEXPLICITLYTHATMUCHOFTHEOBSERVEDCROSSSECTIONINTERINDUSTRYWAGEDIFFERENTIALSCANBEACCOUNTEDFORBYOBSERVEDANDUNOBSERVEDINDIVIDUALSPECIFICEFFECTSUSINGTHEBH
27、PSIMPLIESTHATWECANALSOCONTROLFOROBSERVEDJOBANDWORKPLACECHARACTERISTICSOFCOURSE,ITREMAINSTOBEEXPLAINEDWHYTHESEUNOBSERVEDINDIVIDUALEFFECTSSHOULDBECORRELATEDWITHINDUSTRYAFFILIATION,ALTHOUGHWESUGGESTSOMEPOSSIBLEEXPLANATIONSINOURDISCUSSIONANDCONCLUSION译文行业间工资差异和个体间差异资料来源OXFORDBULLETINOFECONOMICSKRUEGERAN
28、DSUMMERS,1987,1988KATZANDSUMMERS,1989。这些差别久而久之具有高度稳定性,并似乎在许多不同的国家都存在，这些国家有不同的制度和结构形式GITTLEMANANDWOLFF,1993KAHN,1998。并且它们将在不同类型的员工和公司规模中持续存在，这些（工资）不一致很难完全地被行业间人力资本积累或者不能用金钱来衡量的工作的分配不同而引起的工人效用的差异所解释。同时差异性的持续和以下的见解相矛盾（产业和劳动力市场完全竞争的假设下）当短期劳动力流动不合理时，短暂的非均衡现象将自发引起劳动力供给和需求的调整。因此，一些非竞争性的解释相继提出。根据标准人力资本理论,不直
29、接影响工人效用的工作特性也不能决定个人的工资。相反,非竞争性理论则假设这些工作特性会系统的影响工资，因为它们影响公司设立的最佳工资。运用这些理论目的在于确定为什么当公司盈利时它们有能力支付高于市场出清率的工资。此外，这些理由也必须能够解释不同行业产生的种种因素间差异的重要性。效率工资模型体现了高工资的需要来引起工人的努力，WEISS,1980AKERLOF,1984SHAPIROANDSTIGLITZ,1984STIGLITZ,1985，工会风险模型DICKENS,1986强调集体行动的风险,而内外模型SOLOW,1985LINDBECKANDSNOWER,1986强调员工的现任能力当为行业租
30、金的份额讨价还价时的作用。每一种动机预示着行业工资奖金和行业特征的的关系，这种关系同现有的证据相一致。然而，这些模型中没有一个能够证明所有的证据。另一种解释是已被发觉的工资差异反映未被发觉的工人素质差异。一些研究MURPHYANDTOPEL,1987KRUEGERANDSUMMERS,1988GIBBONSANDKATZ,1992KEANE,1993SHIPPEN,1999ABOWD,KRAMARZANDMARGALIS,1999已经利用纵向数据和固定影响模型来证实这个假设，得到混合结果。KRUEGERANDSUMMERS1988,例如提出一点证据支持未被发现的差异在决定行业工资的重要性。相比
31、，MURPHYANDTOPEL1987,KEANE1993,ANDABOWD1999等发现未被发觉的差异分别解释66，84，和大约90对于行业间工资的明显差异。GIBBONSANDKATZ1992提供一个实验这并没有排除未被注意的能力解释，但是另一种试验证实了KRUEGERANDSUMMERS1988的论文，即使对相同的工人，依然存在行业间工资差异。识别确定性的行业间工资差异的真正特性和来源对研究和政策导向以及个人福利都有重要意义。更新的工资结构数据显示在美国和英国的工资不平等比过去50年中的任何时候都要高。KATZ,LOVEMANANDBLANCHFLOWER,1995MACHIN,1996
32、这种增加的发生在群体间的离散具有相同可见的特点。劳动力市场下的竞争模型暗示实际上在短期内的离差变化应该更大。不平等加剧,在这方面,将被视为个体间的分配，相反，劳动力市场的非竞争性模型暗示了这样的变化可能在很大程度上是持久的，接受这些解释因而具有重要的正面积极和规范的影响。本文检验英国行业间工资差异的存在,大小和稳定，纵向数据来自前英国家庭专门研究小组调查BHPS。调查使我们能够检查英国从1991年至1998年间横向、纵向全方位的行业间工资差异的证据。在现存的文献中有极少专门研究小组研究行业工资结构,据我们所知,在英国只有两个。HILDRETH1995提出了基于第一个两阶段的有代表性和纵向的证据
33、，BENITO（2000）提供一些额外的关于工业利润的创新分析，把研究分析扩展到第一个四阶段。这两种研究证明,其中工资中大部分有代表性的差异归因于行业相互关系。然而，这两项研究都仅仅只从数字上关注产业的相互联系，运用相关的几个阶段的数据，并依靠一小部分行业转换者的分析。它们也不能解释的测量误差尤其是在行业转换者数量很小时例如CHOWDHURYANDNICKELL,1985SWAFFIELD,2001。这个研究通过指明未被发觉的差异作为已被发觉的行业间工资差异的解释推进了这个主题的分析，同时考虑到行业聚集和测量误差。利用BHPS数据,确保了更多的产业的转型能够被观察到，因此我们能够更加精确的估计
34、到具有决定性的个体作用。既允许所属行业两位数有更多的产业被观察的转型,因此让我们估计关键的独特作用更为精确。运用HAISKENDENEWANDSCHMIDT1997的估计法，我们也能够提供一个独特的行业工资离散的模型。相反，BENITO2000ANDHILDRETH1995利用KRUEGERANDSUMMERS1988的方法来测量行业工资离散被视为一个新的产业研究，这意味着运用这种分析方法的研究往往不具有可比性，因为研究者为根据自己的需要选取不同的基础行业。此外，我们明确研究中行业聚集的测量误差会影响所得的结果。额外的阶段提供了对于久而久之持久而稳定的行业间差异更固定的检验，因此允许我们调查在
35、实证上重要的可供选择的诸多解释。我们的结果表明，工资的差异能在很大程度上被已现和未现的个体差异所解释，以及行业定位的差异、具体的工作特点。在文献中所提及个体工资的解释的实质性改进关键在于对决定着未被发现的个人多样化的固定因素进行分析。此外,对于没被发现的个人特点的控制在相当程度上降低了行业间的工资差距。在工资决定中未察觉的差异重要性是难以断言的。KRUEGERANDSUMMERS1988采取两种方法来检验劳动力素质中未察觉的差异在存在。首先,他们为1979年5月的当期人口调查CPS数据分别在考虑人力资本年龄、性别、种族、教育、终身职位及职业和排除人力资本的影响时比较跨部门工资。结果报告称人工控
36、制对预估的行业差异没有影响。其次，他们用相应的CPS数据集和1984年的失业工人调查（DWS）上的数据对行业运动建立一阶差分方程。DWS提供一个把雇员的失去第一项工作作为外生变量的样本，比如是因为工厂的倒闭而丢失工作。这些估计同结果相当相似。这两种方法因此使作者拒绝了在劳动力素质考察上未被察觉的参数的影响。GIBBONSANDKATZ1992对这个结论给出了一些支持论点，他们用DWS198486年的数据建立一阶差分的模型，得到行业转换平均可以解释97的相关部门的差异。作者因此排除了为观察到的能力的解释性,并得出这样的结论产业效应本身才是是解释行业工资很重要的因素。我们的目的是调查未被发觉的不均作为解释已经被观察到的行业间差异。我们对真实的固定样本数（而不是与其配套的数据）用改良的HAISKENDENEWANDSCHMIDT方法进行检验，明确的得到已观察到的和未观察到的特殊因素可以准确地解释显现的横向行业间的工资差异。使用BHPS意味着，我们也可以把观测到的工作和工作地点的特点作为控制变量。当然，未被考察到的个体差异对工业集聚具有相关性还有待解释，除此之外我们在讨论和研究结果中还是得到了一些确定的可能解释。
