带有固定效应且n和T都很大的动态面板模型的诊断检验

2009,24(3):266-274nT(,310018):.nT,,.,,..:;;;:O212:A:1000-4424(2009)03-0266-09x1,,.,,,,.,.,NLS(theNationalLongitudinalSurveysofLaborMarketExperience)PSID(theUniversityofMichigan'sPanelStudyofIncomeDynamics).,.,[1-2],.,,.,JournalofEconometrics,,.[3-4](),.:2007-08-15:(07JJD790154);(Q09-12):nT267,.,Yit=®+¯0Xit+it;i=1;2;¢¢¢;n;t=1;2;¢¢¢;T;(1)it=¹i+uit(2)it=¹i+´t+uit;(3)Yit,Xit,¹i,´t,uit,nT.(1)-(2)(one-way),(1)-(3).,,.,,(),,nT.,,,,.[3-4].,.,??,?,,.,,.,,[5].,,.,.,,,.,,..,[6](),,()..[6]BreucshandPangan,,,.,,,,[7-10].uit,.,,,[11-13].,[14],,,,.,.,.[14],.,268243.,,,,BaltagiBH,HongYM..,.,.Xit=°0Xit(q)+¯0Wit+¹i+uit;i=1;¢¢¢;n;t=1;¢¢¢;T;(4)Xit(q)=(Xi;t¡1;¢¢¢;Xi;t¡q)0,°¯q-p-,Witp£1(),¹i,,uit.,,(GMM),[15],[3](Chap8).(),.nT.nT,,.nT,Penn-Worldtables,,.Baltagi[3](p237),,nT,nTnT.,[3](Chap12)[4](p295).nT.,Xit=°Xi;t¡1+¹i+uit;i=1;¢¢¢;n;t=1;¢¢¢;T;(5)¹i,uit.(5),,,(5).,,.,.x2,.,(CV)(OLS)(MLE)(IV)(GMM),.,,.,[4].(Econometrica)[16].,(5),[16],,,.:nT269,,.(5),[16],.,.(5),°^°,¹i+uit(ºit),.(5),^ºit=^¹i+^uit¢^ºit:=^ºit¡^ºi;t¡1=^uit¡^ui;t¡1.,¢^ºi;2j+1=¸1¢^ºi;2j¡1+¸2¢^ºi;2j¡3+¢¢¢+¸d¢^ºi;2j+1¡2d+²i;2j+1;j=d;¢¢¢;k;(6)k=[T¡12]([¢]¢),d.¢ºi;2j+1(d)=(¢ºi;2j¡1;¢¢¢;¢ºi;2j+1¡2d)0,¢^ºi;2j+1(d)=(¢^ºi;2j¡1;¢¢¢;¢^ºi;2j+1¡2d)0.¢^ºi;2j+1(d)f¢^ºi;2j+1¡2lg0·l·d.(6)¢^ºi;2j+1=¸0¢^ºi;2j+1(d)+²i;2j+1;j=d;¢¢¢;k;¸=(¸1;¢¢¢;¸d)0.^¸(OLS).¢ºi;2j+1(¢ºi;2j¡1;¢¢¢;¢ºi;2j+1¡2d),OLS~¸.(),(idea).[17],,.,(A1):uiti:i:d:»N(0;¾2).(A2):0lim(n=T)1.(A3):¡1°1.(A4):1nPni=1X2i0=O(1);1nPni=1¹2i=O(1).(A5):~¸^°.1(A1){(A4)^°,[16].(A5),,[17].(A1){(A5),OLS^¸,n12T12^¸D¡!N(0;P¡1QP¡1);P=E(¢ºi;2j+1(d)¢ºi;2j+1(d)0);S=¡Eh1n(k¡d)Pni=1Pkj=d+1¢ºi;2j+1(d)¢Xi;2ji,R=E(¢º2i;2j+1);Q=2RP¡(1¡°2)SS0..,T2d=nT^¸0^P^Q¡1^P^¸;(7)^P=1n(k¡d)nXi=1kXj=d+1¢^ºi;2j+1(d)¢^ºi;2j+1(d)0;^R=1nknXi=1kXj=1¢^º2i;2j+1;^Q=2^R^P¡(1¡^°2)^S^S0;^S=1n(k¡d)nXi=1kXj=d+1¢^ºi;2j+1(d)¢Xi;2j;270243^¸=^P¡11n(k¡d)nXi=1kXj=d+1¢^ºi;2j+1(d)¢^ºi;2j+1:2.1(5)(A1){(A5),T2dD¡!Â2d;Â2dd.,[16]°^°=T+1TPni=1PTt=1(Xit¡¹Xi)(Xi;t¡1¡¹Xi¡)Pni=1PTt=1(Xi;t¡1¡¹Xi¡)2+1T;¹Xi=PTt=1Xit=T¹Xi¡=PTt=1Xi;t¡1=T.(A1)¡(A4)[16],^°n12T12(^°¡°)D¡!N(0;1¡°2):^¸.(6),(OLS)^¸=³1n(k¡d)nXi=1kXj=d+1¢^ºi;2j+1(d)¢^ºi;2j+1(d)0´¡11n(k¡d)nXi=1kXj=d+1¢^ºi;2j+1(d)¢^ºi;2j+1;1n(k¡d)Pni=1Pkj=d+1¢^ºi;2j+1(d)¢^ºi;2j+1(d)0°^°P=E(¢ºi;2j+1(d)£¢ºi;2j+1(d)0).^°°n.1n(k¡d)nXi=1kXj=d+1¢^ºi;2j+1(d)¢^ºi;2j+1(d)0=P+op(1):¸¤=1n(k¡d)Pni=1Pkj=d+1¢ºi;2j+1(d)¢ºi;2j+1^¸¤=1n(k¡d)nXi=1kXj=d+1¢^ºi;2j+1(d)¢^ºi;2j+1:n12(k¡d)12¸¤D¡!N(0;RP);R=E(¢º2i;2j+1).T!1,Tk¡d,n12T12¸¤D¡!N(0;2RP):,~¸¢ºi;2j+1¢ºi;2j+1(d)OLS,~¸=(1n(k¡d)nXi=1kXj=d+1¢ºi;2j+1(d)¢ºi;2j+1(d)0)¡1£1n(k¡d)nXi=1kXj=d+1¢ºi;2j+1(d)¢ºi;2j+1=P¡1¸¤:(A5),¸d^°,n12T12ø¤^°¡°!D¡!N³0;Ã2RPV12V121¡°2!´:[18],n12T12^¸¤D¡!N(0;2RP¡(1¡°2)SS0);S=limn;T¡!1E(@¸¤@°)=¡Eh1n(k¡d)nXi=1kXj=d+1¢ºi;2j+1(d)¢Xi;2ji::nT271n12T12^¸=(1n(k¡d)nXi=1kXj=d+1^¢ºi;2j+1(d)^¢ºi;2j+1(d)0)¡1n12T12^¸¤=(P+op(1))¡1n12T12^¸¤D¡!N(0;P¡1QP¡1);Q=2RP¡(1¡°2)SS0.nT^¸0PQ¡1P^¸d.P;Q,T2d=nT^¸0^P^Q¡1^P^¸.Slutsky,T2dd.22.1,d,d.,d.3nT,,(4),,.x300.10.20.30.40.500.20.40.60.81c(d=1)power00.10.20.30.40.500.20.40.60.81c(d=2)power00.10.20.30.40.500.20.40.60.81c(d=3)power00.10.20.30.40.500.20.40.60.81c(d=4)power11T2d(d=1;2;3;4)c.1Xit=°Xi;t¡1+¹i+c´t+uit;i=1;¢¢¢;n;t=1;¢¢¢;T;(8)°=0:4,uiti:i:d:»N(0;1),´ti:i:d:»N(0;42),¹i().c=0;0:1;¢¢¢;0:5,c=0,c6=0.(n;T)=(100;101).d=1;2;3;4,(6),T2d.1000,.27224300.10.20.30.40.500.20.40.60.81c(d=1)power00.10.20.30.40.500.20.40.60.81c(d=2)power00.10.20.30.40.500.20.40.60.81c(d=3)power00.10.20.30.40.500.20.40.60.81c(d=4)power22T2d(d=1;2;3;4)cT2d.,d(d·4),.(1).2Xit=°Xi;t¡1+¹i+uit;(9)uit=cui;t¡1+²iti=1;¢¢¢;n;t=1;¢¢¢;T;(10)°=0:4,²iti:i:d:»N(0;1),¹i().c=0;0:1;¢¢¢;0:5,c=0,c6=0.(n;T)=(100;101).d=1;2;3;4,(6),T2d.1000.T2d.,,d(d·4),.(2).3Xit=°Xi;t¡1+cXi;t¡2+¹i+uit;i=1;¢¢¢;n;t=1;¢¢¢;T;(11)°=0:4,uiti:i:d:»N(0;1),¹i().c=0;0:05;0:1;0:15;0:2;0:25,c=0,c6=0.(n;T)=(50;51).d=1;2;3;4,(6),T2d.1000.T2d.,,d(d·4),.:nT273(3).(100;101),,,(),,.00.050.10.150.20.2500.20.40.60.81c(d=1)power00.050.10.150.20.2500.20.40.60.81c(d=2)power00.050.10.150.20.2500.20.40.60.81c(d=3)power00.050.10.150.20.2500.20.40.60.81c(d=4)power33T2d(d=1;2;3;4)c,,..(),..,.,,,,()...:[1]MundlakY.Empiricalproductionfunctionfreeofmanagementbias[J].JFarmEconomics,1961,43:44-56.[2]BalestraP,NerloveM.Poolingcross-sectionandtime-seriesdataintheestimationofadynamicmodel:Thedemandfornaturalgas[J].Econometrica,1966,34:585-612.[3]BaltagiBH.EconometricAnalysisofPanelData[M].NewYork:JohnWileyandSons,2005.[4]HsiaoC.AnalysisofPanelData[M].Cambridge:CambridgeUniversityPress,2002.274243[5]CaiZongwu,LiQi.Nonparametricestimationofvaryingcoe±cientdynamicpaneldatamodels[J].EconomTheory,2008,24:1321-1342.[6]BreuschTS,PaganAR.TheLagrangemultipliertestanditsapplicationstomodelspeci-¯cationineconometrics[J].RevEconomStud,1980,47:239-253[7]HondaY.Testingtheerrorcomponentsmodelwithnon-normaldisturbances[J].RevEconomStud,1985,52:681-690.[8]MoultonBR,RandolphWC.Alternativetestsoftheerrorcomponentsmodel[J].Econo-metrica,1989,57:685-693.[9]BaltagiBH,ChangYJ,LiQi.MonteCarloresultsonseveralnewandexistingtestsfortheerrorcomponentmodel[J].JEconometrics,1992,54:95-120.[10]BeraAK,Sosa-EscuderoW,