经济学年会论文投稿领域——教育和福利经济学 105页

经济学年会论文投稿领域——教育和福利经济学经济学年会论文投稿领域教育和福利经济学Overeducationorundereducation—someevidencefromChinesegraduatesJuanYangChangjunYueAbstractWiththeintensiveexpansionofhighereducationinChinainrecentyears,thequestionariseswhetheritsgraduatesarebeingoversuppliedtothelabourmarket.Thispaperemploysthefirstsurveyconductedin45universitiesofChinacoveringthequestionsofindividuals’personalcharacters,academicachievement,employmentinformationand\nself-estimatedqualificationsrequiredinjobsandintendstodiscovertheincidenceofovereducationandundereducation,whatdeterminesovereducationandundereducationandwhetherthetwostylizedfactsonwagesholdinChina.Fromprocessingthesurveyeddatabylogitmodel,wediscoveredthatthereisaround<21%graduatesareovereducatedandthepercentagesofovereducatedmasterandPhDreach9><>36%and4<2%respectively.Sinceforalltheovereducatedgraduates,<>36%ofwhichareapparentlyovereducatedthattheyaresatisfiedwiththeircurrentjobsandwithcomparativelylowhumancapital,andcanbeseparatedawayfrommismatch.Inaddition,duetothelimitationofourdata,wecannotresearchanotherimportantfactoronovereducation—careermobility,whichmaygenerateanupwardbiasintheestimation.Consideringthesetwofactors,overeducationmaybelessthan10%,thatistosay,theexpandedgraduateswerealmostabsorbedbythegraduatelabourmarket.ThestylizedfactsonwageswereholdinChinathatovereducatedgraduatesearnlessthanthesamelevelofgraduateswhosejobneedactuallyacquiredschoolingbutmorethantheemployeeswhodoingthesamelevelofjobsbutnotobtainingmorethanjobneedqualifications,whichinsomerespectexplainsthehighpercentageofapparentlyovereducation.Keywords:overeducation,highereducation,wages内容摘要：\n随着近几年高校扩张加剧，毕业生是否过度供给已越来越引起人们的关注。这篇文章使用了教育部对7个省市45所高校的问卷调查数据，旨在研究中国是否存在过度教育，其影响因素，及与工资的关系。通过运用逻辑斯特模型处理数据，我们发现大约有<21%的本科生存在过度教育，而硕士和博士过度教育的比例高达<>36%和4<2%。但由于在所有毕业生中有<>36%的学生属于”显然过度教育”，因此可以从过度教育人群中划出。此外，由于数据的局限性，我们无法考虑职业发展因素对过度教育的影响，可能使我们的结果高估过度教育的比例。考虑到这两点，总样本过度教育的比例可能不足10%。换句话说，高校毕业生能够基本上被劳动力市场吸收。另外，过度教育毕业生的收入虽然少于同等学历适度教育的毕业生但却高于工作所需学历毕业生的工资，这种现象也可以从一定程度上解释”显然过度教育”毕业生的过高比例。关键字：过度教育，高等教育，工资1.IntroductionSincethepolicyofhighereducationexpansionaimingtoslowdowntheemploymentpressurewasimplemented,therehasbeenarapidexpansionofChina’shighereducationsince1999thattheenrolmentratetotertiaryeducationsoaringmorethanfivetimesasit’sin1998.Figure1clearlydepictsthepositiverelationshipbetweentheuniversitynewentrantsandunemployment.Theattendantstohighereducationexpandwiththeincreasingofunemploymentrate,butatahigherspeed.AsChina’spublicexpenditureandpercentageenrolmenttotertiaryeducationisverylowcomparingtootherdevelopinganddevelopedcountries,besides,educationiscomprehensivelybelievedthatplaysaveryimportantroleinpromotingnationaleconomicgrowth,individualsskillslevel,moreequitabledistributionandnationalconsumption,whetherthedemandinthegraduatelabourmarketcankeeppacewiththesupplyandsotoabsorballtheemployeestoberecruitedinajobrequiring\ntheircorrespondingqualificationsarenotbeconsidered.However,thisyeartheproblembecomeseriousthanever,whenmorethan3milliongraduatesbothfromChinaandabroadcrowdedintothelabourmarket.Concernsareraisedbyeducators,economistsandpoliticiansthatwhethergraduatesintheChineselabourforcehavemoreeducationthanisactuallyrequiredfortheirjobs(i.e.theyareovereducated).Ashighereducationismainlyfundedbypublicexpenditure,overeducationmeansawasterofresources.Itisveryimportanttodigtheproblemout,especiallyforpoorcountries,likeChina.Thereisaconsiderableframeworkandempiricalstudiesonovereducationinthecurrentliterature,whereasnotasinglearticletomyknowledgediscussestheincidenceandextentsofovereducationinChina.Ourpapertriestofillinthisgapandascertainwhethernumerousempiricalevidencesdiscoveredinmanypopularovereducationstudies(e.g.DoltonandVignoles,<200<2)holdforChina.Inthispaper,wewillmainlyexplorethreequestions:theincidenceofovereducationandundereducation.Inparticular,weexaminethegraduatelabourmarketinChinausingarecentsurveyconductedin<2003from45universitiesin7provincesforeachlevelofhighereducationtorepresentthewholeChinesegraduatelabourmarket.Throughself-reportedanswersontheiractualqualificationsandestimatedrequiredqualificationsonjobs,wecanstatisticallysummarizethepercentageofover(under)educatedindividuals.\nWhatdeterminesovereducationandundereducation?Ourdatarecordseachindividual’spersonalcharacters,academicachievementandemploymentinformationwhichallowsustoresearchwhethergender,bornplace,parentseducation,registrationstatus,classofdegree,levelofqualification,workingplace,universityrank,workingsectorsandfirmsizewillplayaroleindeterminingovereducation.Theimpactofover(under)educationonwages.Inthelastissue,twohypotheseswillbeexamined:Theearningsofworkersinoccupationsthatrequirelessschoolingthantheyactuallyhave(i.e.theyareovereducated)arelessthantheearningsofworkerswiththesamelevelofeducationasthemselves.However,thoseovereducatedworkersearnmorethantheworkerswithsimilarjobsbutnotacquiringmorethanjobneededqualifications.Theearningsofworkersinoccupationsthatrequiremoreschoolingthantheyactuallyhave(undereducated)aremorethantheearningsofworkerswiththesamequalificationbutinajobjustrequiringtheirlevelofschooling.But,undereducatedworkersreceivelessearningsthantheworkerswhoobtainedjobneededqualifications.TheabovestylizedfactswerefoundbyDuncanandHoffman(1981);Hartog(1986);Rumberger(1987);HartogandOosterbeek(1988);Sicherman(1991);Sloaneetal(1999)andDoltonandVignoles(<2000)inUK,USA\nandotherdevelopedcountries.Therestpartofthepaperisorganizedasfollows:partIIintroducestheovereducationtheoryandmeasurementissues.PartIIIbrieflyreviewsthedatasetandvariablesusedandthenliststhestatisticsontheincidenceofundereducationandovereducationfromourdata.PartIVanalyzesaffectingfactsofovereducationandundereducation.Thefinalpartofthepaperdiscussestheimpactofovereducationandundereducationonwages.Source:China’seducationbulletinandChina’sStatisticsYearbookTable1InternationalcomparisonCountryPublicexpenditureonHighereducationasapercentageofGDPPercentageofPopulationon<25-34haveattainedhighereducationUK1.0<29USA<2.738\nJapan1.147India0.71<2Mexico1.018China0.59410Source:OECDdatabaseandChina’sStatisticsYearbook<2004.Allthedatabasedon<2000.<2.OvereducationtheoryandmeasurementissuesOvereducationcanbedefinedinanumberofways(Rumberger,1981).Mostresearchershavedefinedindividualsasbeingovereducatedif(s)hehasmoreeducationthanisrequiredtodohis(her)jobandthereisa\ndeclineinthepecuniaryreturnscomparetothejobsthatrequirecorrespondingqualifications.Overeducationtheorycontradictsthegeneralhumancapitaltheory(Becker,1993;Mincer,1974),whichassumesalltheschoolingcanbetransformedtoproductivityandwagesarethereturntomarginalproductivity.ButinsomepartreflectsSpence(1973)signaltheoryandStiglitz(1975)screeningtheorythatwagesarereturntoeducationsignals.Themajorreasonsandexplanationsonovereducationcanbegroupedintosixbranches:Firstly,trade-offbetweenschoolingandothercomponentsofhumancapital.Sicherman(1991)foundovereducatedpeopletendtohavelessworkingexperience(oron-the-jobtraining),lowerqualityofschoolingorgeneralability.Withtheincreasingofmarketexperience,someoftheovereducatedindividualswillbeemployedingraduatelabourmarket.Battuetal(1997)andDoltonandVignoles(1997)alsofoundsimilarresultsSecondly,careermobilitywhichbroughtforwardbySichermanandGalor(1990)andSicherman(1991).Theysuggestthatemployeeswhoworkinjobsthatrequirelesseducationthantheyactuallyhavearemorelikelytomovetoahigher-leveloccupationandahigherwagelevel.Thetheoryimpliesinordertopursueagoodprospectincareers,someindividualswouldliketobeemployedinjobswheretheschoolingrequirementarelowerthantheirlevelofschooling.Thisbranchcanalsoincludeindividualswhoarelookingforjobswithmoreflexibletime,stability,goodenvironmentorwelfare.\nThirdly,individuals’heterogeneityonability.Theexpansionofhighereducationhasincreasedtheheterogeneityoftheskillsofnewgraduatesenteringthelabourmarket(Chevalier,<2003).Inotherwords,thecandidates’overallskillslevelhasdeclinedandgraduatescannotreachtherequiredgraduationskilllevelleadingthememployedinnon-graduatejobs.Alargenumberofgraduateswhobelongtothetaillevelofskillsareemployedinjobswhichdonotneedtheyactuallyacquiredqualificationsareapparentlyovereducated.Fourthly,education-employmentmismatch.Somekindofovereducationisduetoqualifiedgraduatesemployedinanon-specialty-relatedcareerresultssomeemployeeswouldbeover-educatedforthetypeofworkforwhichtheyhired.Inthegraduatelabourmarkettherearemainlytwosortsofspecialty-employmentmismatch:oneisgraduatesinsciencesubjectswereemployedineconomicsorbusinessfields;theotherisgraduateswithhistory,laworothersubjectmaybeemployedinclericaloradministrativejobs.Employersfrombanksorotherfinancialcompaniesrecruitgraduatesinphysicsorchemistry,notbecauseoftheirgoodknowledgeinphysicsorchemistry,buttheirgoodmathematicsbackgroundandanalysisability.Graduatesbelongingtothisbranchdonotthinktheyareovereducatedtotheircurrentjobs.However,theotherbranchisnotsofortunatethattheyfindjobsaccordingtotheirlinguisticsskillsandtheybelievetheyareovereducated.\nFifthly,the‘ratrace’theorymayexplainpartofthereasonsforovereducation.Whennumerousgraduatesrushedintothelabourmarket,theentrystandardofsomecompetitivepositionsinthelargeandprofitablefirmsmayincreaseandforcegraduatestoreceivemoreeducation.Sixthly,familycommitments,labourmarketrigidityorimperfectinformationmayresultovereducation.Limitedgeographicmobilityofmarriedwomen,disabledpersonorotherconstraintscanalsoleadtoovereducation.Intheliteraturesofar,therearethreewaysinwhichover-educationhasbeenmeasured:externalmethods,self-assessmenttechniqueandstatisticalmethod.Externalmethodsdependonsystematicevaluationbyexpertjobanalystswhodefinetheeducationrequiredbyaparticulartypeofjob.GenerallytheyuseU.S.DictionaryofOccupationalTitles(DOT),whichprovidesinformationontheeducationalrequirementsofawiderangeofoccupations,asareference(e.g.Rumberger,1987).OnesimilarbookinEuropeancountryisARBIcodedevelopedbytheDutchDepartmentofSocialAffairs(SeeHartogandOosterbeek,1988).Self-assessmenttechniqueisasubjectevaluationmethod.Respondentswereaskedwhatistheminimumeducationlevelrequiredincurrentjobsandwhatlevelofformalqualificationstheyalreadyhad.Thenresearcherscandrawtheconclusionsbycomparingtheanswersofthesetwo.StatisticalmethoddevelopedbyVerdugoandVerdugo(1989)whocalculatedthemeaneducationlevelacrossarangeofoccupationsanddefinedanindividualas\novereducatedifhe/shehadmorethanonestandarddeviationabovethemeaneducationlevelforhisorheroccupation.Itisarguedbymanyresearchersthateachmethodhasitslimitationsandmayresultdifferentevaluateoutcomes.Externalmethodsmaybeobjective,butitisalmostimpossiblethatthedictionarycandefineeveryparticularjobineachestablishmentforanyenvironment.Technologieschangingveryfast,therequiredskillsandknowledgeembodiedinthejobsmayincreaseaccordingly,butthecollectedinformationforeachjobmightnotbereleasedimmediatelyleadtoanupbiasedestimationofovereducation.Whatismore,thedefinedoccupationaltitleusuallyaggregatesimilarjobattrition,whichcanbealengthdistributionofrequirededucation.Anotherdrawbackofthismethodisassumingindividualsarehomogeneous.Infact,individualsaredifferentbytheirability,familybackground,tenure,sex,raceandothercharactersandmayreceivevariedqualityofschooling.Consequently,peoplewiththesameeducationmayprocessdifferentskillsandthesamejobcanbecompletedbythehighabilitypersonwithlessschooling.DoltonandSilles(<2001)showsovereducatedpeoplearepronetodismissalandgenerallywithalowergradeineducation.Inotherwords,overeducatedindividualstendtohavelowerabilityandsurpluseducationisanecessaryrequirementinordertoperformthework,thereforeutilizethestandardruletodefineavarietyofpeopleisunreasonable.Self-assessmentcanovercomethedrawbacksofheterogeneityof\nindividualsandspecialskillsneededforeachjob,butmaybebiasedbecauseofsubjectiveandunsymmetricalinformation.Sincetherespondentsdonotknowtheexacteducationorskillsrequirementfromemployer,theymayreporttheirsubjectideas,whichmayreflecttheirownprejudiceandleadtoclassificationerror.Statisticalmethodcanworkwellifthemajorityofworkerwithinacertainoccupationhavetheappropriateeducationalbackgrounds.However,whenitisusedonexaminingtheovereducatedmarket,itmaygenerateadownwardorupwardbias,whichdependsontheperiodofbasemarket.Iftheapproachisbasedoncurrentlabourmarket,partoftheovereducatedpeoplemaybelongtotheuppertailoftheeducationdistributionandcannotderivecorrectovereducationrate.Ontheverse,ifitisbasedonpreviouslabourmarket,itcannotadjustthetechnologyandenvironmentdevelopments.Chevalier(<2003)revealsstatisticalmethodistheleastdesirablemethod,becauseitisverysensitivetothecohorteffectsandlabourmarketdevelopment.Nomatterwhattherealsituation(overeducationorundereducation)isinthelabourmarket,theywillalwayscertainamountofpeopleovereducated.Wewillusesubjectiveself-assessmentmethodtoestimatetheincidenceofovereducationintheChinesegraduatelabourmarket.Thoughthismethodasanalyzedabovehasitsownlimitations,itisthemostaccurateandpopularmethodinestimatingovereducationinthecurrentliteratureonovereducation.\n3.TheincidenceofovereducationandundereducationThispaperemploysthesurveydataconductedbyaprojectfundbytheMinistryofEducationofChinain<2003.Thesurveywasdesignedtousesampleselectionmethodandchoosethreeprovincesineachrepresentativeeconomicdevelopmentarea(theEast,theMiddleandtheWest).Ineachprovinces,6highereducationinstitutionswerechosen,wherethereare<2eliteuniversities,<2commonuniversitiesand<2polytechniccolleges.Howeverduetotheinequalityofuniversitiesineachprovinceandotherpracticalproblems,therealinvestigatedprovincesandhighereducationinstitutionsareasfollows:Beijing(4),Shangdong(6),Guangdong(6),Hunan(6),Shannxi(4),Yunnan(17),Guangxi(1).Altogether,thereare7provincesand45universitiestookpartinthesurveyandthetotalsamplenumberis187<2<2.Amongthesesamples,39.3%percentindividualsacquiredcollegeorequivalentqualificationsandgraduateswhoobtainedbachelor,master,PhDoccupies57%,3.0%and0.6%percentrespectively.Maleproportionforthesefourlevelsofdegreesare5<2.7%,65.4%,59.<2%and73.9%respectively.Genderratioisalmostbalancedineachlevelofqualification,exceptdoctorallevel.Astheinvestigationtakesplacejustafterindividuals’graduation,only40.7%percentrespondentshavefoundajob,4%individualsplantobeself-employed,15.1%percentcandidatesplantocontinuesstudyandtheresthaven’tfoundajoborhaveotherplans.Thereforetheefficient\nsamplestoourresearchreduceto6786.Inthequestionnaire,studentsareasked“whatisyourcurrentqualifications?(Fourpossibleselectionsaregivenfromcollegegraduatetodoctor)andwhatistheminimumformalqualificationrequiredinyourcontractedjobs?”(Six-pointscalefromJuniorschooltoPhD).Throughmatchingthetwoanswers,wesummarizethefollowingstatistics.Table<2Incidenceofmismatchacrosseducationandsubjects(%)EducationlevelcombinedMalesFemalesunderacquiredOverunderacquiredOverunderacquiredOver\ndiploma41.146.01<2.9<21.8<23.87.119.3<2<2.<25.8Bachelor1<2.466.5<21.18.74<2.414.33.7<24.16.8Master7.3\n57.035.83.833.5<21.83.5<23.514.0PhD---58.04<2.0---43.530.4---14.511.6Total17.46<2.1<20.510.8\n38.513.56.6<23.67Econ18.958.9<2<2.<210.334.511.68.6<24.410.6Law<2<2.058.319.61<2.9<27.911.09.1\n47.39.5Art18.664.117.16.9<25.09.011.739.<28.1Medicine17.<265.517.<27.631.75.59.633.811.7\nScience1<2.<269.618.38.537.81<2.83.731.85.5Engineering16.864.718.513.150.014.93.714.73.7agriculture8,363.5\n<28.17.343.8<21.91.019.76.<2Fromtable<2wecanclearlyseethatoveralltherearearound<20percentgraduatesarebelievedtheyareovereducatedintheChina’slabourmarket,amongwhichcollegegraduate,bachelor,masteranddoctortakes1<2.9%,<21%,<>36%and4<2%respectively.Itseemsobviousthatthemorequalificationyouacquiredtheeasieryouareovereducated,whichisalsoconsistentwithGroot(1996).However,thestudiesofFrenette(<2004)inCanadianlabourmarketstellsthedifferentstorythatgraduatesfrommasterdegreesarethemostlikelytobeovereducatedandfollowedbycollegegraduates.Thereisareversegendertrendonovereducationfortheundergraduatescomparingtootherdevelopedcountries(suchasUKandUSA).Theratioofmalestofemalesinthefirstdegreeare1.89:1,buttheratioofgenderfortheovereducatedgraduatesinthesamequalificationbecomesto<2.1:1,whichmeanstoundergraduatesmalesaremorelikelytobeovereducatedinChina.However,inUSAandUK,evidenceshowsthatfemalehavemorefamilycommitmentorgeographic\nimmobilityandiseasiertobeovereducated.Thegendertrendtothehigherdegreeisthesameasthefindingsinotherdevelopedcountries,thatwomenaremorelikelytobeovereducated.Thefindingsonfieldsarealsocontradicttothefindingsinthedevelopedcountries,whichhasahighvariationintheratesacrossfields,thedistributionofovereducated(undereducated)graduatesinChinaamongmajorofsubjectsisequable,exceptagriculture,whichistheeasiesttobebothovereducatedandundereducated.Exceptagriculture,graduatesfromEconomicssubjectaretheeasiesttobeovereducatedandgraduatesfromlawsubjecthavethehighprobabilitytobeundereducated.Theovereducatedindividualscanbefurtherdividedintoseveralsubgroupsaccordingtotheirspecialtyandwhethersatisfiedwiththeircurrentjobs,whichcomesfromtheideaofChevalier(<2003).Thecolumnintable3lists:thetotalnumberofrespondentsforeachlevelofqualification(total),percentageofovereducatedrespondentsineachlevelofqualification(over),thepercentageofovereducatedrespondentswhoaresatisfiedwiththeircurrentjobs(oversatis),thepercentageofovereducatedrespondentswhosespecialtiesmatchorrelatedwiththeircurrentjobs(overmatch),thepercentageofovereducatedandspecialtymatchedrespondentswhoarenotsatisfiedwiththeircurrentjobs(ousm),thepercentageofovereducatedindividualswhoaresatisfiedwithcurrentjobsbutthejobdoesn’tmatchtheirspecialty(mismatsatis).\nTable3Furtherdivisionofovereducatedindividualstotaloveroversatisovermatchousmmismatsatisdiploma1<2771<2.9%3<2.7%77.6%64.1%4.8%bachelor51<24<21.1%34.8%90.9%60.0%<2.1%\nmaster31635.8%46.9%9<2.0%51.0%1.8%PhD694<2.0%65.5%93.1%37.0%6.9%Total6786<20.5%<>36.<2%89.4%6<2.3%<2.5%\nChevalier(<2003)groupedtheindividualswhoareovereducatedbutsatisfiedwiththeircurrentjobsasapparentlyover-educatedandshetreatedsuchtypeofovereducationasmatchorappropriate.Weassumethistypeofindividualsnormallyholdlessamountofhumancapitalthanrequiredbutcomefromgoodfamilybackground,whichcanprovidethemsufficientfinancialsupporttoachievetheiroptimumeducationlevel.InadditionthestylizedfactmustholdinChinathatthoseovereducatedworkersearnmorethantheworkerswithsimilarjobsbutnotacquiringmorethanjobneededqualifications.Wewillexaminethesetwoassumptionsinthenexttwosections.Sincethistypeofovereducationtakesaround<>36%oftotalovereducatedindividuals,excludingthisamountofovereducatedgraduatestheovereducationproblemisnolongeraveryseriousissueinChina.Thefourthcolumnintable3clearlyshowsspecialtyandjobs’mismatchisnotanobviousproblemforovereducation.Inotherwords,notaparticularsubjectissignificantlyoversuppliedthandemandinthelabourmarketleadingtomismatch.Columnfivefurtherdemonstratestheopinionthatmorethanhalfofthematchedindividualsareunhappyabouttheirjob.Thismeansthereareotherreasonsnotsubjectmismatchthatinducethemnotsatisfiedwiththeircurrentjobs.Ofcoursetheindividualswhoaresatisfiedwiththeircurrentjob,butnotmarchingthespecialtyareveryrareandshowedincolumnsix.\nTable4SamplemeansandstandarddeviationofselectedvariablesforundergraduatesVariablesCombinedMalesFemalesunderacquiredoverunderacquiredoverunderacquiredoverUnivrank<2.<253(0.77)<2.33<2\n(0.74)<2.39<2(0.77)<2.<274(0.78)<2.330(0.76)<2.395(0.77)<2.<207(0.73)<2.3<>36(0.69)<2.401(0.75)Univgrad1.915\n(0.91)<2.006(0.90)<2.069(0.91)<2.019(0.91)<2.11<2(0.9<2)<2.185(0.9<2)1.656(0.85)1.814(0.8<2)1.809\n(0.83)English1.591(0.77)1.671(0.8<2)1.818(0.87)1.511(0.75)1.676(0.85)1.854(0.90)1.600(0.78)1.661\n(0.77)1.750(0.81)Regis<2.490(1.<23)<2.590(1.<2<2)<2.4<28(1.<20)<2.606(1.<25)<2.763(1.<21)<2.550(1.<21)\n<2.<203(1.15)<2.<27<2(1.18)<2.161(1.13)Pquali4.15<2(1.38)4.<279(1.31)4.049(1.5<2)4.<207(1.40)4.<>365\n(1.31)4.168(1.5<2)4.011(1.30)4.117(1.<29)3.848(1.45)Msalary1449(966)1<>369(847)1144(864)1493\n(101<2)1379(8<29)1143(8<29)1358(845)1355(879)1177(931)Table4listssamplemeanandstandarddeviationofselectedvariablesforundergraduatessoastogiveusanintuitiononfamilybackgroundandinstitutioninformationforundereducatedgraduates,overeducatedgraduatesandgraduateswhoacquiredtheexactqualificationrequiredbythecurrentjob.Thestatisticsclearlyshowsthatovereducatedpeoplearemorelikelytobefromcitieswithacomparativelygoodfamily\nbackground,butwithalowerrankofuniversity,lowerclassofdegree,worseEnglishskillsandlesssalary.Whereastheaverageuniversityrank,classofdegree,Englishlevel,evenfamilybackgroundforundereducatedgraduatesarebetterthangraduateswhoacquirednomoreandnolessthanjobrequiredschooling.Self-believedundereducatedgraduatesarenormallyveryhumble.Sometimesitisnottheyarereallyundereducatedtotheirjobs,buttheymaynotstandinanadvantagepositionthatcanbenefitcareerpromotionandthinkthemselvesareundereducated.Nomatterfromregistrationstatusorparentseducation,femalegraduatesarebetterthanmalegraduates.Thereasonbehindthisisacultureproblem:intheonehandparentswouldliketospendallthelimitedfinancialresourcesonboysinthevillageandontheotherhandcityboyslackself-disciplineandareeasytodropoutinthestudy.Asaresult,theproportionsofmalegraduatesfromvillageorcountryarehigherthanfemalegraduates.Despitetheimbalanceddistributionofovereducationandundereducationbetweenmaleandfemale,thesexwagesdifferentialisnotverylarge.4.Whatdeterminesovereducation(undereducation)?Asourdatasetisacross-sectiondata,wecannotobservedurationofovereducationandwhethertheovereducationwilldeclinewiththeincreasingofworkingexperiences.Howeverourdatasetrecordseachindividual’scomprehensiveinformationonpersonalcharacters,academicachievementandemploymentinformationwhichallowsustoexaminewhatkindoffactorsmayaffectovereducationandundereducation.Inorderto\nachievethisgoal,overeducationandundereducationwereestimatedbyalogitmodelseparatelyconditioningonindividuals’qualificationlevel,registrationstatus,parentseducation,ethnic,partymember,cadre,universityrank,gradeofdegree,Englishlevel,specialty,companyownership,firmlocatedplace.Threegroupsofcoefficientswerelistedinthetable5forcombined,maleandfemaleindividuals.Theresultsshowthatforthegroupofvariablesonacademicachievement,themostsignificantvariableisqualificationlevel，whichexplainsquitealotinbothundereducationandovereducation.Universityrankandclassofdegreeallhaveanegativeeffectonindividuals’undereducationandpositiveeffectonovereducation,whichoppositethefindingsbyBattu(1997)inUKthatselectiveuniversityplaynopartinexplainingovereducation.InChina,theaveragemarginaleffectsofuniversityrankshowthatahigherlevelofuniversitywillincrease3.1%probabilitytobeundereducated(<2.7%formaleand3.5%forfemale)andsimilarly,iftheclassofdegreedeclinesalevel,graduateswilldecrease0.9%probabilitytobeundereducated(0.1%formaleand1.3%forfemale).EnglishskillwhichmeasuredbyCollegeEnglishTest(CET)isevenmoreimportantindeterminingindividuals’overeducation(undereducation),especiallyformen.Theeffectsofsubjectsareveryinteresting.Whenweaddsubjectsintothelogitmodeltoexaminetheirinfluencetoovereducationorundereducationwithoutcontrollinguniversityrank,noneofthemissignificant.WhereastheevidencefromUSA,UKandCanadashowsthatgraduatesfrommedicine,law\nandtechnicaldegreeareleasttobeovereducated.Thereasonbehindthisisthedifferenceofeducationsystemthatthepercentageofenrolmentformedicine,lawandtechnicaldegreeisveryhighinChina.Inaddition,mathematicsisacompulsorycoursesforalltheseniorschoolstudents,butnotinUSA,UKandsomeotherwesterncountries.VariablePmember(partymember)standsoutfromthepersonalcharacters’variables.Comparingtonon-party-member,partymemberwillincrease51%probabilitytobeundereducatedanddecrease30%probabilitytobeovereducated.Theincidenceofovereducationandundereducationmayalsodependonparents’education,ethnic,andregistrationstatus.Onethingneedstobenotedisthecoefficientsignsofpqualiforbothundereducationandovereducationarenegative,butnotsignificant.Thereasoncanbeexplainedfromtwoangels.Firstly,itiseasytounderstandthatgraduatesfromlesseducatedfamilyarelesspossibletobeovereducatedbecausetheywouldn’tinvestmoreineducationunlesstheyaresuretheycanobtainmorefromeducation.Andinthesametime,thewelleducatedfamilywilllettheirchildrenaccesssufficienteducationandlesslikelytobeundereducated.Secondly,graduatesfrompoorfamilybackgroundaremorelikelytoappreciatethecurrentstudyopportunityandspendmoretimeoneducationinordertoacquiregoodclassdegreeandfindajobinwhichdegreeisrequired.Thegroupofvariablesforemploymentcharacteristicsaffectovereducationandundereducationmeaningfully.Inordertoestimatethe\ndeterminedfactorsaccurately,weincludefirmlocationandcompanyownershipintotheregressionmodel.Workplacelocationisverysignificanttobothundereducationandovereducationforthreegroupsofsamples.ComparingtoworkingintheWest,workingintheEastorMiddleismorelikelytobeundereducatedratherthanovereducated.Relativetoworkersinthe“otherownership”,workersinthestate-ownedcompanyareovereducatedmorefrequently,whereasworkersinthegovernmentorjoint-venturecompaniesarelessfrequentlyovereducated.Ingeneral,forallthethreegroupsofaffectedvariables,thegroupofacademicachievementvariablesoccupiestheleastproportion,whichmaypartlyreflectdrawbackofSelf-assessmentmethod:subjective.Withtheincreasingofuniversityrankandgrade,graduates’standardonovereducationandundereducationmayincreasegraduallyresultstheinsignificantdeterminantsofacademicvariables.Liketheevidenceinsomewesterncountriesthatfemalearemorelikelytobeovereducatedfromtheregressionresults..Inthefinalpartofthissection,wewillexaminewhetherfamilybackgroundhasasignificanteffecton“apparently”overeducatedindividuals.Table6liststhelogitregressionresultsforthreegroupsofrespondents.Thecoefficientsforparents’education(Pquali)aresignificantat1%or5%andforregisteredstatus(Regis)aresignificantat10%,buttheregressioncoefficientsforPqualiandRegisarenotverysignificantintable5.Theresultsshowthatfamilybackgroundplaysan\nimportantrolefor“apparently”overeducatedgraduatestofindagoodjobimplyingtheycomefromcomparativelygoodfamilybackground.Table5DeterminantsofovereducationandundereducationVariableCombinedMaleFemaleUnderOverUnderOverUnderOverGender-0.<20<2**(0.09)-0.176**（0.08）--\n------Ethnic0.049(0.15)0.<255*(0.14)-0.01<2(0.<21)0.<274(0.17)0.018(0.<24)0.3<21(0.<24)cadre\n-0.1<2<2(0.09)0.<27<2***(0.08)-0.168(0.11)0.307***(0.10)0.0<27(0.15)0.176(0.14)Pquali-0.058*(0.03)-0.044\n(0.03)-0.044(0.04)-0.050(0.04)-0.093(0.06)-0.017(0.06)Regis-0.038(0.04)-0.056(0.03)-0.0<28(0.05)-0.058\n(0.04)-0.08<2(0.07)-0.04<2(0.06)pmember0.511***(0.10)-0.<298***(0.09)0.441***(0.1<2)-0.<2<25**(0.11)0.581***(0.16)\n-0.434***(0.16)univrank-0.031(0.0<2)0.03<2(0.0<2)-0.0<27(0.03)0.043*(0.0<2)-0.035(0.05)0.003(0.04)Univgrad\n-0.009**(0.05)0.066**(0.04)-0.001(0.06)0.1<20**(0.05)-0.013(0.09)0.064(0.08)English-0.084*(0.05)0.163***\n(0.04)-0.161***(0.06)0.187***(0.05)0.07<2(0.09)0.096(0.09)qualileve-1.393***(0.10)0.797***(0.09)-1.34<2***(0.1<2)0.788***\n(0.11)-1.547***(0.18)0.850***(0.16)east0.314*(0.17)-1.089***(0.1<2)0.410**(0.<20)-1.17<2***(0.14)0.075(0.3<2)\n-0.890***(0.<23)middle0.511***(0.17)-0.96<2***(0.13)0.437**(0.<21)-0.91<2***(0.15)0.585*(0.33)-1.066***(0.<26)Govern\n0.<267**(0.1<2)-0.<239**(0.11)0.310**(0.15)-0.005(0.14)0.<245(0.19)-0.654***(0.18)Statown-0.<211*(0.1<2)0.043\n(0.11)-0.065(0.15)0.147(0.13)-0.574**(0.<23)-0.096(0.19)Joinvent0.199(0.17)-0.353**(0.17)0.3<26(0.<21)-0.337\n(0.<2<2)-0.087(0.<29)-0.<>364(0.<27)constant0.857**(0.<>36)-1.906***(0.3<2)0.704**(0.43)-<2.093***(0.37)1.<215*(0.66)\n-1.930(0.61)Loglikelihood-1980.6766-<2<298.4139-1301.9751-1548.50<27-666.60375-737.<230<2PseudoR<20.07850.053<20.06350.05680.1<2170.0545Note:theresultswereestimatedbylogitmodelinStata8.*meansthecoefficientissignificantat10percentinterval,**meansthecoefficientissignificantat5percentintervaland***meansthecoefficientissignificantat1percentinterval.\nTable6DeterminantsofapparentlyovereducatedindividualsvariableCombinedMaleFemaleCoeff.Std.Err.Coeff.Std.Err.Coeff.Std.Err.gender-0.19<2**0.09--------qualileve\n0.853***0.090.797***0.110.981***0.17Pquali-0.067***0.03-0.068**0.04-0.053*0.06Regis-0.063*0.04-0.067*0.04-0.046*0.07cadre0.338***\n0.080.353***0.100.<279**0.15pmember-0.<263***0.10-0.198*0.1<2-0.397**0.17ethnic0.<2<240.150.<2400.180.<2600.<25Univrank-0.0340.0<2\n-0.0380.03-0.0130.04Univgrad0.091**0.050.13<2***0.05-0.0140.09English0.15<2***0.050.167***0.060.10<20.09east-1.014***0.1<2-1.080***\n0.14-0.833***0.<24middle-0.894***0.13-0.86<2***0.15-0.9<>36***0.<27govern-0.1880.1<20.0960.15-0.69<2***0.19Statown0.1580.110.<296**0.14\n-0.0450.<20Joinvent-0.<2890.18-0.<2060.<23-0.4190.<20constant-<2.<264***0.33-<2.397***0.39-<2.375***0.65Loglikelihood-<21<24.<260<2-1447.<2<26<2-666.90<267PseudoR<2\n0.05730.05650.065<25.TheimpactofovereducationandundereducationonwagesInthissection,wewillexaminetherelationshipbetweenwagesandovereducation,particularlythetwohypothesesthatwelistedinthefirstsection.Inordertoexaminethetwostylizedfactsthattheeffectsofmismatchonindividuals’wageswecreateatwo-stepmethod.Firstly,examinewhetherovereducation(undereducation)mayhaveanegative(positive)effectonwagesforindividuals’acquiredthesamelevelofqualificationsbyregressinglogwagesonindividuals’actuallyobtainededucation,plustwodummyvariablesofovereducationandundereducation.Secondly,wetestwhetherovereducation(undereducation)mayhaveapositive(negative)effectonwagesforindividuals’doingthejobsthatrequiredthesamelevelofeducation.FromthesetworegressionresultswecanclearlyknowwhetherthetwohypothesesholdinChina.Thefirstpurposecanberealizedbyregressingequation(1),whereSmeansactuallyacquiredqualifications,Xrepresentsalltheother\nvariablesthatmayaffectindividuals’wagesincludingpersonalcharacters,academicachievementandemploymentinformation.(istheInverseMillsratio,whichintendstoovercomeself-selectionbiasandinefficiencyofOLSestimationbecauseoftruncateddatasetand(isthedisturbanceterm.Overandunderaretwodummyvariables,whichcanexplainwhetherovereducated(undereducated)graduatesmayearnless(more)thanthegraduateswiththesamequalification,butexactjobrequiredEMBEDEquation.3(1)Table7OLSLogwages’regressingonactuallyacquiredqualificationsvariablesCombinedMaleFemaleCoef.Std.Err.Coef.Std.Err.Coef.\nStd.Err.gender0.00040.013--------college-0.183**0.090.0090.11-0.443***0.13degree0.1<24*0.080.0750.110.384***0.13\nMaster0.545***0.030.588***0.040.53<2***0.05PhD-0.0160.070.0860.08-0.5<24***0.15Over-0.05<2***0.0<2-0.069***0.0<2-0.0150.03\nunder0.0<23*0.0<20.04<2**0.0<2-0.0150.03cadre0.078***0.010.069***0.0<20.105***0.0<2ethnic0.0040.0<20.0<270.03-0.0150.04regis\n-0.044***0.01-0.038***0.01-0.049***0.01pmember-0.0010.01-0.00010.0<20.0040.0<2univrank-0.0<24***0.003-0.031***0.004-0.0050.01univgrade0.005\n0.010.0050.010.01<20.01English-0.0<2<2***0.01-0.034***0.010.0070.01Pquali-0.0<21***0.01-0.0<21***0.01-0.0<27***0.01Govern-0.048***0.0<2\n-0.097***0.0<20.0330.03statown-0.00<20.0<2-0.0190.0<20.0<2<20.03joinvent0.147***0.0<20.130***0.030.167***0.04East0.0390.03-0.003\n0.030.143***0.05Middle-0.165***0.03-0.158***0.03-0.159***0.06law-0.108***0.03-0.099**0.05-0.0610.05medicine-0.<205***0.04-0.166***0.06\n-0.169***0.05science-0.095***0.0<2-0.0<>360.04-0.139***0.04engineering-0.0<>36*0.0<2-0.0160.03-0.0<280.03agriculture-0.<237***0.05-0.183***0.06-0.347***\n0.10econ0.0<2<20.030.0380.040.0330.05history-0.16<2***0.05-0.178*0.07-0.143**0.07education0.0190.090.<2060.17-0.0840.11\nmanagement-0.04<2*0.0<2-0.0140.03-0.0460.03philosophy-0.0840.08-0.1100.10-0.0580.13constant7.771***0.107.643***0.137.770***0.15\nAdjustedR-squared0.<26<230.<25660.3084Table7tellsusthattherearepositiveandsignificantreturnstobachelorsandmasters,especiallyformastersthatmaycontribute55%percentincreasetowagescomparingtobachelordegree(formale59%andforfemale53%).Thenegativereturntocollegeismainlybecausethesamplesizeisspeciallyfocusedongraduates,butforPhDstudents,itisprobablytrue.AccordingtotheSurveybyagraduaterecruitingwebsite,themeanandmedianwagesformasterareallhigherthanthewagesforPhD.TheresultsarealsoconsistentwithFlenette(<2004)’sstudyinCanadianlabourmarket.Asexpected,overeducation(undereducation)maygenerateanegative(positive)effecttothewagesofbothmaleandfemale.Buttheeffectstomalewagesaremuchstrongerthanfemale.ContrasttothefindingsinUKandUSA,universityrank,Englishgrade,familybackgroundandregistrationstatusinChinaplayanimportantroleinexplainingwagesforbothmaleandfemale.Themarginaleffectsofuniversityrank,Englishgrade,registrationstatusandfamilybackgroundtowagesare<2.4%,<2.<2%,4.4%and<2.1%,respectively.Inaddition,beingacadreintheclassoruniversitymaycontribute8%percenttowages.\nHowever,classgradewhichisakeyfactorinsomewesterncountriesisnolongeressentialinChina.Companyownershipinfluencesindividuals’earningsdramatically,especiallyformen.Relativetoworkinginothersectors,workinginthegovernmentandstate-ownedfirmshavetheprobabilitytoearnalowerlevelofwages,butthewagesinjointventurearesignificantlyhigher.However,thewelfareinthegovernmentandstate-ownedfirmsaremuchbetterthanotherfirms,besidestheworkersinsuchkindsofcompaniesenjoythestableworkingpositionandlesspressingworkload,whichmaytrade-offthelowwages.Workingplacewillalsoaffectindividuals’expectedwages.ComparingtoworkingintheWest,workingintheMiddlewillreceivelessfavourablepayments,butnottrueforworkingintheEast.ThereturnstosubjectsarealsoquitedifferenttotheresultsintheUSAandUK.Comparingtothereturntoliterature,onlyeducationandeconomicssubjectsarepositive.Thehighreturnsubjectsinotherwesterncountries,suchaslawandmedicineallhaveanegativeeffectonwages.Oneparticularreasonforlowreturntolawandmedicinesubjectsmaybecauseofthehighreturntoexperienceinsuchsubjectsinducingthelowbeginningwages.Equation(1)examinesthereturnstoeachlevelofqualificationinaMincerianequationexceptweaddingtwodummiesofovereducationandundereducation.IfthehumancapitaltheorybyBecker(1993)andMincer(\n1974)holdsinChina,thatwagesarereturnstoeducation,thenthecoefficientforthetwodummiesshouldequaltozero.However,theresultsshowtheyarenotequaltozeroandsignificant,especiallyformen,whichmayreflectSpence(1973)’ssignaltheorythatwagesarereturnstoacertainamountofskillsthatarerequiredbyaparticularjobandhumancapital(education)isonlyasignal.Stimulatedbythisidea,weregressthelogearningequationonthejobrequiredqualifications,overeducated(undereducated)partineachlevelofqualificationandothervectorsinequation(<2),inordertoevaluatethepremiumforovereducationandpenaltyforundereducation.EMBEDEquation.3(<2)whereqrrepresentsthejobrequiredqualification,qodenotesovereducatedpartsandqumeansundereducatedparts.Table8OLSLogwages’regressiononjobrequiredqualificationsvariableCombinedMaleFemaleCoef.\nStd.Err.Coef.Std.Err.Coef.Std.Err.Gender-0.0010.01--------College-0.0350.04-0.0570.040.0130.07Degree0.07<2***0.03\n0.100***0.030.0510.04Master0.455***0.030.45<2***0.040.485***0.05PhD-0.0490.05-0.0040.06-0.<218**0.11Ovcoll0.0140.07-0.016\n0.080.0080.11Ovdeg0.0050.03-0.0010.040.0<250.05Ovmas0.506***0.050.5<29***0.070.503***0.09Ovphd0.<231**0.100.<299***0.11\n-0.0600.<2<2Uncoll-0.064**0.03-0.0540.04-0.085**0.04Unmas-0.1130.110.0880.14-0.<2440.18Undeg-0.4<>36***0.04-0.435***0.04-0.463***\n0.06univrank-0.0<25***0.003-0.033***0.004-0.0060.01univgrade-0.0010.010.0010.01-0.00010.01English-0.0<20***.0.01-0.033***0.010.0140.01\nPquali-0.0<2<2***0.01-0.0<21***0.01-0.0<28***0.01Regis-0.048***0.01-0.04<2***0.01-0.057***0.01Ethnic0.0010.0<20.0<250.03-0.0180.04\nCadre0.071***0.010.065***0.0<20.09<2***0.0<2Pmember0.0040.01-0.00030.0<20.0140.0<2Govern-0.056***0.0<2-0.101***0.0<20.0<240.03statown\n0.0050.0<2-0.0110.0<20.0300.03joinvent0.154***0.0<20.1<29***0.030.180***0.04east0.0370.03-0.0010.030.1<20**0.06middle-0.176***\n0.03-0.164***0.03-0.194***0.06constant7.605***0.067.688***0.077.340***0.1<2AdjustedR-squared0.<24480.<23930.<2871Onesignificantdifferencebetweentable7andtable8istherewardtoalllevelsofovereducationispositiveandtoundereducationisnegativeforallthreegroups(intable7,thereturntoovereducationisnegativeandtoundereducationispositive).Throughregressingthe\nwagestothejobrequiredqualifications,thereturntobachelorandmasterreducesto7%and46%forthecombinedgroup.Table8alsotellsusthewagepremiumforthosewhohaveacquiredahigherlevelofdegreethenthecurrentjobrequiredispositiveandsignificant,especiallyformastersanddoctors.Similarly,thewagepenaltyfortheundereducatedpartofacquireddegreeorcollegequalificationishugeandessential.Otheraffectingvariablesarerobustinbothtables.Table7andtable8canclearlyexplainthetwohypothesesthatwebroughtforwardinthefirstsection:(1)theearningsofworkersinoccupationsthatrequirelessschoolingthantheyactuallyhave(overeducated)arelessthantheearningsofworkerswiththesamelevelofeducationasthemselves.(<2)theearningsofworkersinoccupationsthatrequiremoreschoolingthantheyactuallyhave(undereducated)aremorethantheearningsofworkerswiththesamequalificationbutinajobjustrequiringtheirlevelofschooling.Becauseifwetreatactualqualificationstheyobtainedasrepressors,thenthereturntoovereducationisnegativeandtoundereducationispositive,whichmeansifyouareovereducated(undereducated)toyourjob,youmayearnbelow(above)theaverageofthisqualification.However,intable8,iftherequiredqualificationbythejobsareindependents,thecoefficient(effects)ofovereducation(undereducation)willbepositive(negative),implyingifyougethigher(lower)educationthanpracticallyrequiredinthejob,youcangetapositive(negative)premium\nfortheextra(less)education.However,thepremium(penalty)isimbalancedamongqualificationsandgenders.6ConclusionsFromprocessingthesurveyeddata,wediscoveredthattherearearound<20%graduatesareovereducatedandthepercentagesofovereducatedmasterandPhDreach35.8%and4<2.0%respectively.Amongalltheovereducatedgraduates,<>36%ofwhichareapparentlyovereducated.Sincetheyaresatisfiedwiththeircurrentjobsandwithcomparativelylowhumancapital,theycanbeseparatedfrommismatch.Besides,ourdataisnotapaneldatarestrictingustoresearchthecareermobility,whichmaygenerateanupwardbiasintheestimation.Consideringthesetwofacts,overeducationmaybelessthan10%,inotherwords,theexpandedgraduateswerealmostabsorbedbythegraduatelabourmarket.Forthereasonsthatmaydetermineovereducationamongallthepossiblevariables,partymember,universitygradeandEnglishgradeareparticularimportant.Thatistosay,thoughtherapidexpansionofhighereducationintherecentyears,graduatesfromtoprankeduniversitywithagooduniversityandEnglishgradeislesslikelytobeovereducated.Contrastingtothefindingsinotherdevelopedcountries,subjectsdonotexplainmuchonovereducationinChina.Asexpected,overeducatedgraduatesearnlessthanthesamelevelof\ngraduateswhosejobneedactuallyacquiredschoolingbutmorethantheemployeeswhodoingthesamelevelofjobsbutnotobtainingmorethanjobneedqualifications.Thesamearguesuitsforundereducatedgraduates.Inthecontextofrewards(loss)toextra(less)educationofjobrequired,theovereducated(undereducated)partcomparingtoamaster(college)orPhD(first)degreeobtainthehighestpremium(penalty).Thoughfemalearemorelikelytobeovereducated,thereisnoevidenceongenderdiscriminationonwages.Amongallthequalifications,thereturntothemasteristhehighest,attains59%formenand53%forwomen.Inaddition,universityrank,Englishgrade,familybackground,registrationstatus,cadre,companyownershipandworkingplaceallinfluenceindividuals’wagessignificantly.7.ReferenceBattu,H.,Belfield,C.R.&SloaneP.J.(1999).Overeducationamonggraduates:acohort6><#00aa00'>view.EducationEconomics,7(1),<21-38.Becker,G.S.(1993).HumanCapital:aTheoreticalandEmpiricalAnalysiswithSpecialReferencetoEducation,UniversityofChicagoPress,USA.Chevalier,A(<2003).Measuringover-education.Economica,70,509-531.\nChina’sstatisticsyearbook<200<2-<2004.Dearden,L.,Mclntosh,S.,Myck,M.&Vignoles,A.(<200<2).TheReturnstoAcademicandVocationalQualificationsinBritain.TheBulletinofEconomicResearch,54(3),<249-<274.Duncan,G.J.&Hoffman,S.D.(1981).Theincidenceandwageeffectsofovereducation.EconomicsofeducationReview,1(1),75-86.Dolton,P&Silles,M.(<2001).Over-educationinthegraduatelabourmarket:someevidencefromalumnidata,CentrefortheEconomicsofEducation,LondonSchoolofEconomicsandPoliticalScience.Dolton,P&Vignoles,A.(<2000).IncidenceandeffectsofovereducationintheUKgraduatelabourmarket.EconomicsofEducationReview,19(<2),179-198.Frenette,M.(<2004).TheoverqualifiedCanadiangraduate:theroleoftheacademicprogramintheincidence,persistenceandeconomicreturnstooverqualification.EconomicsofEducationReview,<23,<29-45.Greene,W.(<200<2).EconometricAnalysis,fifthedition,NewYorkUniversity.\nGroot,W.(1996).Theincidenceof,andreturnstoovereducationintheUK.AppliedEconomics,<28,1345-1350.Hartog,J.(1985).Earningsfunctions:testingforthedemandside.EconomicLetter,19,<281-<285.Hartog,J.(1986).Allocationandtheearningsfunction.EmpiricalEconomics,11,97-100.Hartog,J&Oosterbeek,H.(1988).Education,allocationandearningsintheNetherlands:overschooling?EconomicsofEducationReview,7(<2),185-194.Heckman,J.(1979).SampleSelectionBiasasaSpecificationError.Econometrica,47(1),153-161.Kroch,E.A.,&Sjoblom,K.(1994).Schoolingashumancapitalorasignal:someevidence.JournalofHumanResources,<29(1),156-180.Mayston,D.(<200<2).EvaluatingtheBenefitsofCareerGuidance,CentreforPerformanceEvaluationandResourceManagement,York,andCentreforGuidanceStudies.McGoldrick,K&Robst,J.(1996).Genderdifferencesinovereducation:atestofthetheoryofdifferentialoverqualification.\nTheAmericanEconomicReview,86(<2),<280-<284.Mincer,J.(1974).Schooling,experienceandearnings,NewYork:ColumbiaUniversityPress.O’Leary,C.N&Sloane,P.(<2005),Thechangingwagesreturntoanundergraduateeducation,TheGraduateLabourMarket:RatesOfReturnAndOvereducationConference,CentrefortheEconomicsofEducation.Rumberger,R.W.(1981).TherisingincidenceofovereducationintheU.S.labormarket.EconomicsofEducationReview,1(3),<293-314.Rumberger,R.W.(1987).Theimpactofsurplusschoolingonproductivityandearnings,JournalofHumanResource,<2<2(1),<24-50.Sicherman,N.(1991).“Overeducation”inthelabourmarket.JournalofLabourEonomics,9(<2),101-1<2<2.Sicherman,N&Galor,O(1990).Atheoryofcareermobility,ThejournalofPoliticalEconomy,98(1),169-19<2.Sloane,P.J.,Battu,H.&Seaman,P.T.(1999).Overeducation,undereducationandtheBritishlabourforce.AppliedEconomics,31(11),1437-1453.\nSpence,M.(1973).Jobmarketsignalling.QuarterlyJournalofEconomics,87(3),355-374.Stiglitz,J.E.(1975).Thetheoryof“screening”educationandthedistributionofincome.TheAmericanEconomicReview,LXV,<283-300.Verdugo,R.R.&Verdugo,N.T.(1989).Theimpactofsurplusschoolingonearnings:someadditionalfindings.TheJournalofHumanResources,<24(4),6<29-643Webbink,D&Hartog,J(<2004).Canstudentspredictstartingsalaries?Yes!EconomicsofEducationReview,<23,103-113Wooldridge,J(<200<2).Econometricanalysisofcrosssectionandpaneldata,TheMITPress,USAAppendixACadre-Whetheryouareamemberofcadreintheclassorintheuniversity?1.Yes0NoEast—YourworkplaceislocatedintheeastofChina.(weinclude\nBeijing,Tianjin,Shanghai,Liaoning,Hebei,Shandong,Guangdong,Jiangsu,Zhejiang,Fujian,Hainan)1.Yes0NoEnglish---EnglishqualificationintheUniversity(Haveyouevergotthefollowingqualification?)1.CET-6<2CET-43.NoneEthnic---Doyoubelongtoethnicminority?1.Yes0NoGovern—Whetheryouworkingovernmentoreducationinstitutions?1.Yes0NoJoinvent---Whetheryouworkinajoint-venturecompany?1.Yes0NoMiddle---YourworkplaceislocatedintheeastofChina.(weincludeHeilongjiang,Jilin,Anhui,Jiangxi,Henan,Huibei,Hunan,Chongqing)1.Yes0No\nOvcoll----Youthinkyouareovereducatedtothecollegequalificationsaccordingtoyourcurrentjob1.Yes0NoOvdeg----Youthinkyouareovereducatedtothefirstdegreeaccordingtoyourcurrentjob1.Yes0NoOvmas----Youthinkyouareovereducatedtothemasterqualificationsaccordingtoyourcurrentjob1.Yes0NoOvPhD----YouthinkyouareovereducatedtothePhDqualificationsaccordingtoyourcurrentjob1.Yes0NoPmember---AreyouamemberofCommunistParty?1.Yes0No\nPquali---Parentseducation(father’shighestqualification):1.higherdegree<2.degree3.diploma4.seniorschool5.juniorschool6.primaryschool7.noqualificationQualileve--.Qualificationlevel1.diplomaorequivalent<2.Bachelor3.Master4.PhDRegis---Registrationstatus(whereareyoufrom?):1.middleorlargecities<2.smallcities3.town4.villageStatown---Whetheryouworkinastate-ownedcompany?1.Yes0NoUncoll----Youthinkyouareundereducatedtothecollegequalificationsaccordingtoyourcurrentjob.1.Yes0NoUndeg----Youthinkyouareundereducatedtothefirstdegreeaccordingtoyourcurrentjob.\n1.Yes0NoUnivgrad---classofdegree1.firstclass<2.Uppersecond3.second4.PassUnivrank---Universityrank(therankofgraduationuniversity):1.top<2<2.<2-153.16-504.51-1005.101-1506.151-<2007.<201-3008.301-4009.401---(includingpolytechniccollege)Unmas----Youthinkyouareunderducatedtothemasterqualificationsaccordingtoyourcurrentjob1.Yes0NoWest---YourworkplaceislocatedinthewestofChina.(weincludeShanxi,Neimenggu,Guangxi,Sichuan,Guizhou,Yunnan,Shannxi,Gansu,Qinghai,Ningxia,Xinjiang)1.Yes0NoSeetable1fordetail.Unlikeothercountries,ChineseparentswouldliketocontributeasignificantparttotheirChildren’s\neducation.Thenumbersinthebracketarethenumbersofinvestigateduniversitiesinthisprovince.AlsoincludesgraduatesfrompolytechniccollegeOursamplesizefordoctorsareverysmall,whichmaybiasourestimation.SeeMc-GoldrickandRobst(1996),Sloaneetal(1999)fordetail.Theevidencefromwesterncountriesshowsthatgraduatesfromscience,engineering,medicineandlawarelesslikelytobeovereducatedthanothergraduates.SeeFrenette(<2004),DoltonandSilles(<2001)fordetail.Inthequestionnaire,thestudentswereasked‘Accordingtoyourunderstanding,howmuchdoyouthinkyourcurrentjobrelatedwithyourspecialty?’andtheanswerwasseparatedintofourranks:veryrelated,almostrelated,somerelatedandnonerelated.Weincludethefirstthreerankasmatchinthestatistics.ThemeaningofvariablevalueislistedintheappendixA.\nThesamplesizeforotherqualificationsiscomparativelysmall,soweusethemostrepresentationalbachelorsampletoanalyze.Werankthe45universitiesbystudents’qualitymeasuredbytheentranceexaminationtohighereducation.Weclassifythe45universitiesinto9ranksaccordingtothestudents’qualitywhichmeasuredbyentranceexaminationscoretohighereducation.DoltonandSilles(<2001),DoltonandBignoles(<2000)andFrenette(<2004).SeeAppendixAforvariablenamesweusefourdummyvariablestorepresentdiploma,bachelor,masterandPhDseparately,whosevalueequalsto1ifheorshereachesthatlevel.Forexampleforamasterdegree,thedummyvariablediploma,bachelorandmasterequalto1,butPhDdummyequalsto0.Experiencehereisequalto0,astherespondentsarejustgraduatingfromuniversities.Inmanyotherdevelopingcountries,itdoesnothold.SeeKrochandSjoblom(1994)fordetail.\nSloaneetal(1999)employsthesimilarmethodtoexaminethereturntoovereducation.However,itmayalsooffsetthebiasinfootnote14.PAGEPAGE<25Chart10.6659<2.50.6386<2.30.6496<2.30.7874<2.60.9661<2.8\n0.9509<2.90.97731.0<2543.11.06373.11.15613.11.6893.1<2.33463.6<2.84843.40764.34.09064.<2numbersofuniversitynewentrants(units:million)\nunemploymentrateYearnumbersofuniversitynewentrantsandunemploymentrateFigure1.ThegrowthtrendofhighereducationattendantswithacontrastofunemploymentrateSheet1GDPnumbersGDPuniversitynewentrants794038<2080451780810.119<2846001.34850.1380795145-0.0789908395-0.0433715<2<2184717100850.<25\nYearnumbersofuniversitynewentrants(units:million)unemploymentrate85896440900.097665900-0.041900.0969117404-0.0409971467-0.481900.6659<2.5901854790910.1660.017<2<2518016386000.017910.16551199870.017<2<251801-1.4<2910.6386<2.391<21617809<20.<23<20.<21<213054196496000.<21<29<20.<23<2<2<299<2160.<21<2130541911.3159<20.6496<2.39<2<2663810910.30.<2<2694945397874000.<2<27930.30018<28<2080.<2<2694945390.07930.7874<2.6933463440940.35-0.015733<>3609966100-0.016940.350085464<2-0.015733<>3609-1.069940.9661<2.8944675940950.<2510.0<2744768119509000.0<27950.<250616988<20.0<274476811-<2.745\n950.9509<2.9955847810960.1610.04953940639770000.05960.16085508930.04953940630.805960.9773966788460970.0970.037351<277610<254000.037970.09689973870.037351<2776-0.<246971.0<2543.1977446<260980.05<20.086866597710637000.087980.05<214161<20.08686659771.3<26981.06373.19878345<20990.0480.460946<284911561000.461990.0475115<2590.460946<28494.306991.15613.1998<206750000.090.38<2<238010716890000.38<2000.09017698850.38<2<2380107-0.171001.6893.1008946810010.0880.<21990919<2<2<23346000.<2<2010.08770388550.<21990919<2<2-0.4<25\n01<2.33463.60197314800<20.0810.196488764<28480000.1960<20.08074311410.196488764-0.1070<2<2.84840<210517<230030.1150.<2004343<23334076000.<2030.11485533740.<2004343<2330.0<2033.40764.303117<251904090600044.09064.<204GDPuniversitynewentrants910.16551199870.017<2<2518019<20.<23<2<2<299<2160.<21<21305419910.30018<28<2080.<2<269494539\n940.350085464<2-0.015733<>3609950.<250616988<20.0<274476811960.16085508930.0495394063970.09689973870.037351<2776980.05<214161<20.0868665977990.0475115<2590.460946<2849000.09017698850.38<2<2380107010.08770388550.<21990919<2<2\n0<20.08074311410.196488764030.11485533740.<2004343<233Sheet100000000000000\n000000000000GDPuniversitynewentrantsYearsPercentagegrwothofGDPanduniversitynewentrantsFigure1.ThegrowthtrendofhighereducationattendantwithacontrastofGDPgrowth\nSheet<20000000000000000000000\n00000000numbersofuniversitynewentrants(units:million)unemploymentrateYearnumbersofuniversitynewentrantsandunemploymentrateFigure1.ThegrowthtrendofhighereducationattendantswithacontrastofunemploymentrateSheet3\n