一、背Yt=1+2

+其中????表示進(jìn)口總額，????表示GDP，????表示隨機(jī)誤差項(xiàng)數(shù)指國內(nèi)生產(chǎn)總進(jìn)口額(匯地頻年年單億億模型估DependentVariable:YMethod:LeastSquaresDate:06/26/15 Time:18:24Sample:19912014Includedobservations:Std.t-CXMeandependentAdjustedR-S.D.dependentS.E.ofAkaikeinfoSumsquaredSchwarzLog-Hannan-QuinnF-用GDP來解釋商品進(jìn)口的變化。因此，回歸模型應(yīng)該為：其中：Y為國內(nèi)生產(chǎn)總值，X為GDP的線性函數(shù)；EX為兌民幣的匯率 殘差圖為模型中DW=0.515763ExchangeGDP已經(jīng)按從小到大順序排列，因此，重新計(jì)算d統(tǒng)計(jì)量。對N=24，K’=1,α=5%，德賓-d統(tǒng)計(jì)量的臨界值為dL=1.273和dU=1.466,表明存在顯著的遺漏變量現(xiàn)象。DependentVariable:YMethod:LeastSquaresDate:06/26/15 Time:18:40Sample(adjusted):1992Includedobservations:23afterStd.t-C--X(---MeandependentAdjustedR-S.D.dependentS.E.ofAkaikeinfoSumsquaredSchwarzLog-Hannan-QuinnF-LM檢驗(yàn)步驟，首先生成殘差序列（E表示E對全部解釋變量（包括遺漏變量）進(jìn)行回DependentVariable:EMethod:LeastSquaresDate:06/26/15 Time:18:41Sample(adjusted):1992Includedobservations:23afterStd.t-C--X(-----MeandependentAdjustedR-S.D.dependentS.E.ofAkaikeinfoSumsquaredSchwarzLog-Hannan-QuinnF- (2)=7.37776;因?yàn)?.51544<7.37776;不接受無約束回歸模型的假設(shè)，不存在遺漏變量。兩個備選模型：研究收入、影響消費(fèi)和收入、消費(fèi)習(xí)慣影響消費(fèi)一、數(shù)據(jù)城二、回歸如下DependentVariable:PPCEMethod:LeastSquaresDate:06/26/15 Time:21:52Sample(adjusted):1992Includedobservations:23afterStd.t-CPDPL(-MeandependentAdjustedR-S.D.dependentS.E.ofAkaikeinfoSumsquared.SchwarzLog-Hannan-QuinnF-DependentVariable:PPCEMethod:LeastSquaresDate:06/26/15 Time:21:53Sample(adjusted):1992Includedobservations:23afterStd.t-CPPCE(-MeandependentAdjustedR-S.D.dependentS.E.ofAkaikeinfoSumsquared.SchwarzLog-Hannan-QuinnF-從預(yù)測的目的來看，R平方相差不大，選擇B。但也未必合適。模型A：ppce=1080.892+0.669129pdpl+0.190376pdpl^-1模型B：ppce=704.5361+0.511754pdpl0.427629ppce^-三、J檢假定模型AB是備擇假設(shè)，使用模型B的PPCEA中的一個新增回DependentVariable:PPCEMethod:LeastSquaresDate:06/26/15 Time:22:13Sample(adjusted):1992Includedobservations:23afterStd.t-C----PDPL(---MeandependentAdjustedR-S.D.dependentS.E.ofAkaikeinfoSumsquared.SchwarzLog-Hannan-QuinnF-因?yàn)樽兞縋PCEFB的系數(shù)在統(tǒng)計(jì)上是顯著的，按照J(rèn)檢驗(yàn)程序，模型A接受模型B假定模型B是虛擬假設(shè)而模型AA的PPCEBDependentVariable:PPCEMethod:LeastSquaresDate:06/26/15 Time:22:14Sample(adjusted):1992Includedobservations:23afterStd.t-CPPCE(---MeandependentAdjustedR-S.D.dependentS.E.ofAkaikeinfoSumsquaredSchwarzLog-Hannan-QuinnF-因?yàn)樽兞縋PCEF的系數(shù)在統(tǒng)計(jì)上是不顯著的，按照J(rèn)檢驗(yàn)程序，不模型B四、結(jié)果與