(完整版)EViews面板数据模型估计教程.docx
EViews6.0beta在面板数据模型估计中的应用来自免费的minixi1、进入工作目录cdd:nklx3,在指定的路径下工作是一个良好的习惯Panel SPeCifiCatjOn Frequency Annual 2、建立面板数据工作文件workfile(1)最好不要选择EViews默认的blanacedpanel类型WorkfileCreateWorkfiiestructuretypebalancedPanelIrregularDatedandPanelWorkfitesmaybemadefromUnstructiredworkfilesbylaterspecifyingdateand/orotheridentifierseries.IOKCancelMoren_panel(2)按照要求建立简单的满足时期周期和长度要求的时期型工作文件3、建立POOl对象(1)新建对象2)选择新建对象类型并命名(3)为新建POOI对象设置截面单元的表示名称,在此提示下(CrossSectionIdentifiers:(Enteridentifiersbelowthisline)输入截面单元名称。,建议采用汉语拼音,例如29个省市区的汉语拼音,建议在拼音名前加一个下划线“,如图关闭建立的pool对象,它就出现在当前工作文件中。4、在PoOl对象中建立面板数据序列双击PooI对象,打开POOI对象窗口,在菜单VieW的下拉项中选择SPreedSheet(展开表)在打开的序列列表窗口中输入你要建立的序列名称,如果是面板数据序列必须在序列名后添加“?”例如,输入GDP?,在GDP后的?的作用是各个截面单元的占位符,生成了29个省市区的GDP的序列名,即GDP后接截面单元名,再在接时期,就表示出面板数据的3维数据结构(1变量2截面单元3时期)了。请看工作文件窗口中的序列名。展开表(类似excel)中等待你输入、贴入数据。5、贴入数据(1)打开编辑(edit)窗2)贴入数据EViewsEiIeEditObjectWeWErOCQuickOtiosWndOWtelpcdd:nklx3Fool:POOL_02LWorkfile:DOCTOR_0l:Untitled-Il×lVieWlPrOi:IobjectIPrintINamelFreezelEdit+/-Order+/-ISrnPi÷-FormatITitleESUmateDeflnePoOIGefr归obsGDP?K?_BEIJING-19865.6519985.000676BEIJING-19875.7830025.281423BEIJING-19885.9815075.435021_BEIJING-19896.1646065.684639BEIJING-19906.2489145.703889_BEIJING-19916.3896005.714875_BEIJING-19926.5395055.967988_BEIJING-19936.7320796.352460_BEIJING-1994勺必g叩IdngSU心gprUjlanXljaoh115gdpjiang×ikanhuikneimenggdpjilink_beijingk_ningxia,UntitIedkNeWPage/hl2dJPath=d:nklx3DB=noneWF=doctor_01/(3)关闭POOl窗口,赶快存盘见好就收6、在POoI窗口对各个序列进行单位根检验选择单位根检验Pool:POOL_01Workfile:DOCTOR_01:Untitled-I11l×lplewlProcObjectPrintNameFreezeEdit+/-1Order+/-1Smpl+/-1FormatTitleEstimateDefinePooIGenrECrossSectionIdentifiersSpreadsheet(stackeddata).DescriptiveStatistics.K?5.000676325.281423LhitRootTest.375.435021CointegrationTest.J65.684639Label145.703889BEIJING-19916.3896JO5.714875BEIJING-19926.5395055.967988_BEIJING-19936.7320796.352460_BEIJING-1994IJ一Ldl设置单位根检验单位根检验结果PooIUnitRootTestonK?Poolunitroottest:SummarySeries:K_BEIJING,K_TIANJIN,K_HEBEI,K_SHANXI_D,K_NEIMENG,K_LiA0NING,KjjiLIN1K-HLJIANG,K_SHANGHAI,K.JiANGSU,K_ZHEJIANG,K_ANHUI,KFUJIAN,K_JIANGXI,K_SHANDONG,K_HENAN,K_HUBEI,K_HUNAN,K_GUANGDONG,K-GANGXI,K-HAINAN,fCsiCHUArTK_GUIZHOU,K_YUNNANJCsHANXAX,KAGANSU,KAQlNGHAI,KJNINGXIA,K_XINJIANG-Date:05/24/07Time:19:57-SampIeJ9862005ExogenousvariablesjndividualeffectsAutomaticselectionofmaximumlagsAutomaticseIectionofIagsbasedonSIC:0to4Newey-WestbandwidthselectionusingBartlettkemelCross-MethodStatisticProb.*sectionsObsNlkUnitroot(assumescommonunitrootprocess)Levin1LinSChu忙-2.660530.003929526NllII:Unitroot(assumesindividualunitrootprocess)Im1PesaranandShinW-Stat4.267011.000029526ADF-FisherChi-Square19.30851.000029526PP-FisherChi-Square14.51641.000029551wwProbabilitiesforFishertestsarecomputedusinganasymptoticChi-Squaredistribution.AllothertestsassumeasymptoticnormaIity.注意检验方法和两种检验的零假设:NUlI:UnitroOt(assumescommonunitrootprocess)各截面有相同的单位根NUlI:Unitroot(assumesindividualunitrootprocess)允许各截面有不同单位根Pool:F4ZHANM(»aWorkfile:SHSH02: 2yongxufazh.其中,LeVin,Lin&Chut*检验拒绝含有单位根的零假设,即拒绝非平稳7、在POOI窗口对面板数据组合进行协整检验选择进行协整检验CrossSectionlctentifiefsRepresentationsEstimationOutputResidualsCOeICOVananCoMaiuxLNK?5.0006765.2814235.4350215.64639LNDH?8.1583398.3201558.3886588.457649LNXH?2.00568H2.02725!a2.04838CoefficientTests>5.7038898.5389732.06907lFixed/RandomEffectsTesting5.7148755.9679888.6090242.08933l2.109208.642618Spreadsheet(Stackeddata).6.3524608.647367DescriptiveStatistics.6.7073698.700082CSoQOUnitRootTest.7.0477068.7005222JW3'2.20300,6.9501428.7032897.0506948.718590Label7.1768528.6765072.220751BEIJlNG¾987.598307.2620738.6882852.238191BEIJlNGT99?7G72087BEIJING200077964137.2490708.7030532.25533,2.27248-!x0.融IPreCiCtieZelEdilCjToflfemTsmoblTFefmailTifelEBtimatelDefinePfldGenfiSBEIJING-2001协整检验设置对话框,注意有3种检验方法(testtype)协整检验结果,同样要注意两种假定(含有AR,即含有单位根,非协整),两种零假设都是非协整,小概率事件发生拒绝非协整。本例题检验的4个序列时协整的,特别提示还要看各个序列的单位根检验是否是同阶单整的,否则单凭协整检验的结果根据不足。PedroniResiduaICointegrationTestSeries:LNY?LNK?LNDH?LNXH?Date:05/24/07Time:20:14SampIeJ9862005lncludedobservations:20Cross-Sectionsineluded:29NullHypothesis:NocointegrationTrendassumption:Nodeterministictrend1.agselectionifixedatlNewey-WestbandwidthseIectionwithBartIettkerneIStatisticProb.Grouprho-Statistic3.8269590.0003GroupPP-Statistic0.1998720.3911GroupADF-Statistic-3.6546000.0005AltemativehypothesisxommonARcoefs.(Within-Climension)StatiStiCProb.StatiStiCProb.WeightedPanelv-Statistic1.4150600.14661.5389000J22Panelrho-Statistic1.8843370.06761.6258930.106PaneIPP-Statistic0.0473930.3985-0.3684740.372PaneIADF-Statistic-3.2473110.0020-3.2811550.001Alternativehypothesis:individualARcoefs.(between-dimension)CroSSSeCtionSPeCifiCreSIIltSPhillips-Peronresults(non-parametric)CrossIDAR(I)VarianceHACBandwidthObs8、建立混合模型在PoOI对象窗口的PrOC(过程)的下拉式菜单中选择估计打开模型设置窗口!Pool:FAZHANMOXIWodcfileSHSH02::yongxufazh.Jl×lViewrircjObjectPrintNameFreezeEdit+/-Order+/TSmPi÷-FormatTitleEStimateDefinePclOlGenrS5.651EstimateMakeResidualsLNY?LNK?LNDH?BMakeGroup.5.6519985.0006768.1±JBMakePeriodStatsseries.5.7830025.28142383-JBMakeModel5.9815075.4350218.3BMakeSystem.6.1646065.6846398.4,BUpdateCoefsfromPool6.2489145.7038898.5:BGeneratePoolseries.DeletePlseries.6.3896005.7148758.6E6.5395055.9679888.6,B6.7320796.3524608.6,!BStQrePlseries(DB).6.9470686.7073698.7BFetchPlseries(DB)7.2089377.0477068.71BImportPooldata(ASen,XLS,WK?)7.3847916.9501428.71BExportPooldata(ASCn,XLS,WK?),.7.4947247.0506948.7BEIJING-1998BEIJING-19987.5983067.17685286BEIJING-1999BEIJING-19997.6726877.2620738.61BEIJING-2000BEIJING-20007.7964137.249070招BEIJING-2001l<lI混合模型的设置PoolEstimationlny?SpecificationOptionsdependentvariableEstimationmethodFixedandRandomPerioconeVeights No weightsCr&ss-sectiINOneJEstimationsettingsMethod:LS-LeastSquares(andAR)Samp 1 e 1986 2005I-Balance二1Saiple取消确定混合模型的结果D即endentVariable:LNY?MethodPooIedLeastSquaresDate:05/24/07Time:20:30Sampled9862005lncludedobservations:20Cross-Sectionsincluded:29Totalpool(balanced)observations:580VariableCoefficientStd.ErrortStatisticProb.C-0.8602430.121396-7.0862770.000nLNK?0.8291890.01305963.49590.000LNDH?0.2613960.01200521.77390.000LNXH?0.2377270.0556284.273470.000LNNE1?-0.1452770.063819-2.2763740.023LNCE1?0.0901580.0561431.605870.108QR-squared0.986459Meandependentvar7.086138AdjustedR-Squared0.986341S.D.dependentvar1.240231FnfrOnrOAQinnn1AAAinfnrrit&rinn-1nidRn9、建立变系数模型这里只建立一次变一个变量且在截面维的变系数模型。当然也可是在时间维的变系数。而且可以一次不止变一个变量的系数。变系数模型的设置变系数模型的估计结果DependentVariabIeiLNY?MethodpooIedLeastSquaresDate:05/24/07Time:20:37Sampled9862005lncludedobservations:20Cross-sectionsincluded:29TOtalPOOI(balanced)observations:580VariableCoefficientStd.Errort-StatisticProb.C-1.0909600.215765-5.0562320.0000LNDH?0.3031280.02609311.617090.0000LNXH?0.1917570.1201801.5955750.111LNNE1?-0.2482310.075168-3.3023300.0010LNCE1?0.6448430.0814977.9124580.0000Beijing-Lnkbeijing0.7414960.02224433.334790.000Tianjin-Lnktianjin0.7727700.02240234.496290.0000hebei-lnkhebei0.7493170.01875839.947250.0000shanxi_d-lnkshanxi_d0.7416180.02142234.619030.000neimeng-lnkneimeng0.7436640.02129534.921500.0000Liaoning-Lnkliaoning0.7869310.01974639.852630.0000jilin-lnkjilin0.7658240.02165935.358240.0000Hljiang-Lnkhljiang0.7833390.02071837.808940.00010、建立截距维的固定效应模型,并检验模型的冗余性(是否比混合模型优?)截面维固定效应模型的设置截面维固定效应模型的估计结果D即endentVariable:LNY?Method:PooledLeastSquaresDate:05/24/07Time:20:42Sampled9862005lncludedobservations:20Cross-Sectionsincluded:29Totalpool(balanced)observations:580VariableCoefficientStd.Errort-StatisticProb.C-0.7852040.267649-2.9337080.003LNK?0.6907670.02006434.427310.000LNDH?0.1038980.0385402.6958280.007LNXH?1.3497620.1780527.5807240.000LNNE1?-0.1652170.072697-2.2726610.023LNCE1?0.3528760.0828214.2606880.000FixedEffects(Cross)BEIJING-O-0.614424TIANJIN-O-0.436907HEBEI-C0.113717Shanxld-C-0.207641NEIMENg-C-0.209332Liaoning-C0.095655截面维固定效应模型的冗余性检验,首先在pool模型的view中选择似然比的检验菜单选项HPookFAZHANMOXIWorkfile:SHSH02:yongxufazhanmoxi-xMiv!ProcObjectPrint!NameiFreezeEstimateDefinePGenrSheetCrossSectionIdentifiersRepresentationsEstimationOutputResiduals匚oef匚OvarianceMatrix匚OefficientTestsRedundantFixedEffects-LikeIihoodRatioCorrelatedRandomEffects-HausmanTestJ40.267649-2.9337080.003370.02006434.427310.000980.0385402.6958280.00732-0.1780527.5807240.0000.1652170.0.072697-2.2726610.0233528760.0828214.2606880.000FiXed/RandomEffectsTesting-Spreadsheet(Stackeddata)-DescriptiveStatistics.UnitRootTest.CointegrationTest.1.abel1.NNE1?1.NCE1?FiveriFfferfjfUrcM/似然比检验的结果,零假设固定效应模型是冗余的,小概率事件发生,拒绝冗余,于是摒弃混合模型:RedundantFixedEffectsTestspookFAZHANMOXITestcross-sectionfixedeffectsEffectsTestStatisticd.f.Prob.Cross-sectionF18.205856(28,546)0.0000Cross-sectionChi-square382.452544280.0000Cross-Sectionfixedeffectstestequation:DendentVariable:LNY?Method:PanelLeastSquaresDate:05/24/07Time:20:47Sampled9862005lncludedobservations:20Cross-sectionsincluded:29Totalpool(balanced)observations:580CoefficientStd.Errort-StatisticProb.CLNK?-0.8602430.121396-7.0862770.00000.8291890.01305963.495910.000011、建立截距维的随机效应模型,并进行HaUSman检验,确定是选择随机效应亦或是固定效应模型,零假设:随机效应模型成立。截面维随机效应模型的设置截面维随机效应模型的估计结果D即endentVariable:LNY?Method:PooledEGLS(Cross-sectionrandomeffects)Date:05/24/07Time:20:56Sampled9862005lncludedobservations:20Cross-sectionsincluded:29Totalpool(balanced)observations:580SWarnyandAroraeStimatOrOfCornPOnentVarianCeSVariablecoefficientStd-Errort-StatisticProb.C-1.2662660.199089-0.0000LNK?0.7567330.01631446.38630.0000LNDH?0.2780840.02008213.84710.0000LNXH?0.4885100.0988494.942000.0000LNNE1?-0.1295320.067090-0.0540LNCE1?0.3901600.0711465.483910.0000RandomEffects(Cross)BEIJING-O-0.243423TIANN“C-0.042283HEBEI-C-0.043812Shanxld-C-0.113288NEIMENG-C-0.073589截面维随机效应模型的HaUSman检验菜单项的选择J三PoolFAZHANMOXIWorkfile:SHSH02:yongxufazhanmoxiPiewliproclObjectIPrintINameFreezeEstimateDefinePoolGenrSheetIcrossSectionIdentifiersactionrandomeffects)RepresentationsEstimationOutputResidualsA匚OefCovarianceMatrix匚OefficientTestsn<:-AAO,IllFiXed/RandomEffectsTesting卜RedundantEixedEffects-LikeIihoodRatioSpreadsheet(Stackeddata).DescriptiveStatistics.UnitRootTest.CointegrationTest.CCrraIatarIRKnriCrn田arts-I-IaUcmanTZt*ntStd.Errort-StatisticProb.360.199089-6.3603000.0000330.01631446.386370.0000Label340.02008213.847110.0000截面维随机效应模型HaUSman检验的结果:HaUSman检验的零假设是应当选择随机效应模型,小概率事件发生拒绝零假设选择固定效应模型CorrelatedRandomEffects-HausmanTestpookFAZHANMOXITestcross-sectionrandomeffectsTestSummaryChi-Sq.StatisticChi-Sq.d.f.Prob.Cross-sectionrandom53,39400450000Cross-Sectionrandomeffectstestcomparisons:VariableFixedRandomVar(Diff.)Prob.LNK?1.NDH?1.NXH?1.NNE1?1.NCE1?0.0000.6907670.7567330.000136n0.1038980.2780840.0010820.0001.3497620.4885100.0219310.000-0.165217-0.1295320.0007840.2020.3528760.3901600.0017980.379212、13在时间维重复10、和11、的工作,确定数据适合采用何种模14、建立截而变截距模型,分析没有观察的截面单元因素的影响截面变截距模型的设置PoolEstimationSpecificationOptions确定I取消截面变截距模型的估计结果DependentVariable:LNY?MethodpooIedLeastSquaresDate:05/24/07Time:21:05Sampled9862005lncludedobservations:20Cross-sectionsincluded:29Totalpool(balanced)observations:580VariableCoefficientStd.Errort-StatisticProb.C-0.7852040.267649-2.9337080.003LNK?0.6907670.02006434.427310.000LNDH?0.1038980.0385402.6958280.007LNXH?1.3497620.1780527.5807240.000LNNE1?-0.1652170.072697-2.2726610.023LNCE1?0.3528760.0828214.2606880.000FixedEffects(Cross)BEIJING-C-0.614424TIANJIN-C-0.436907HEBEI-C0.113717Shanxld-C-0.207641NEIMENG-C-0.209332LIAONING-C0.09565515、建立时期变截距模型,分析没有观察的时期因素的影响时期变截距模型的设置×JSpecification IODtiOnS 确定 I 取消PoolEstimation时期变截距模型的估计结果DgJendentVariable±NY?MethodpooIedLeastSquaresDate:05/24/07Time:21:08Sampled9862005lncludedobservations:20Cross-sectionsincluded:29Totalpool(balanced)observations:580VariableCoefficientStd.Errort-StatisticProb.C-0.6582070.117502-5.6016470.000LNK?0.7467130.01892639.453540.000LNDH?0.3210070.01529420.989580.000LNXH?0.2479490.0528904.6879810.000LNNE1?0.0872650.0856731.0185830.308LNCE1?-0.4359990.080745-5.3997150.000FixedEffects(Period)1986-C-0.1907931987-C-0.1966561988-C-0.1490011989-C-0.1469821990-C-0.1120041991-C-0.10882016、在整个估计、检验构成中养成使用冻结和命名保存的习惯,以便撰写报告时调用。TAbIe:UNTn"LEDWoricfiIe:SHSHo2:Y(mgjciifazhanm甲iewPrci,:IObiaztlPrintlNmmeIEdit+/-CellFmtGrid+/-TitleCeimments+lABCDED即endentVariable:LNY?MethodPooIedLeastSquaresOODate:05/24/07Time:21:08ASampIeJ9862005InCIIJdedobSerVatiOnS:20qaCross-sectionsincluded:29/十11Totalpool(balanced)observations:580VariableCoefficientStd-Errort-StatisticProb.-0.000C-0.6582070.117502crir47-fOLNK?0.746710.01892639.45350.0001.NDH?0.321000.01529420.98950.0001.NXH?0.247940.