計(jì)量經(jīng)濟(jì)學(xué)EVIEWS自相關(guān)檢驗(yàn)與修正

1、一、數(shù)據(jù)來源數(shù)據(jù)：國家統(tǒng)計(jì)局（19812010年國內(nèi)生產(chǎn)總值與固定資產(chǎn)投資）軟件版本：EVIEWS7.2二、回歸結(jié)果1、一元線性回歸:=Equation:UNTITLEDTorkfile;J9812010:_BXViewProcObjectPrintNameFreezeEstimateForecastStatsResidsDependentVariable:XMethod:LeastSquaresDate:05705/14Time:19:04Sample:19312010Includedobservations:30VariableCoefficientStd.Errort-Statistic

2、Prob.Y0.5933250.02262226.249600.0000R-squared0.939936Meandependentvar50354.74AdjustedR-squared0.939936S.D.dependentvar74104.55S.E.ofregression10161.52Akaikeinfocriterion22.48476Sumsquaredresid9.57E+09Schwarzcriterion22.53147Loglikelihood-336.2714Hannan-Quinncrite匚22.43970Durbin-Watsonstat0.290755三、模

3、型診斷與修正階正自d二1.35DW檢驗(yàn)：相關(guān)系數(shù)5=0.8546，查表得，十人門經(jīng)檢驗(yàn)，DWV1.35,自變量呈d二1.49U相關(guān)四、廣義差分法修正后的結(jié)果對(duì)E進(jìn)行滯后一期的自回歸，可得回歸方程：E=0.9337E(-l)二Equation:WJTITLEDTorkfile:19812010:；:._XViewjProcObjectPrintNameFreezeEstimateForecastStatsResidsDependentVariable:EMethod:LeastSquaresDate:05/05/14Time:21:40Sample(adjusted):19322010Inclu

4、dedobservations:29afteradjustmentsVariableCoefficientStd.Errort-StatisticProb.Et-1)0.9336710.1120210.3347630.0000R-squared0.64-2304Jeandependentvar-3056.903Adjusted尺-和uared0.642304S.D.dependentvar16&60.61S.E.ofregression9904.523Akaikeinfocriterion21.27324Sumsc|uaredresid2.75E+09Schwarzcriterion21.32

5、039Loglikelihood-307.4621Hannan-Quinncrite匚21.23001Durbin-Watsonstat1.S59544對(duì)原模型進(jìn)行廣義差分，輸出結(jié)果為：=Equation:UKTITLEDTorkfile;；：_芒ViewProcObjectPrintNameFreezeEstimateForecastStatsResidsDependentVariable:Y-0.9337*Y(-1Method:LeastSquaresDate:05/05/14Time:21:48Sample(adjusted):19322010Includedobservations:2

6、9afteradjustmentsVariableCoefficientStd.Errort-StatisticProb.C69S1.7232445.4362.S549430.0032X-0.9337X(-1)1.0027490.1105209.0730260.0000R-squared0.753013Meandependentvar19425.70AdjustedR-squared0.743371S.D.dependentvar21543.32S.E.ofregression10903.16Akaikeinfocriterion21.49796Sumsquaredresid3.21E-H09

7、Schwarzcriterion21.59226Loglikelihood-309.7205Hannan-Quinncrite匚21.52750F-statistic02.31900Durbin-Watsonstat2.314136Prob(F-statistic0.000000y*=6981.723+1.002749X*tt由于使用廣義差分?jǐn)?shù)據(jù)，樣本容量減少了1個(gè)，為29個(gè)。查5%的顯著性水平的DWd二134統(tǒng)計(jì)表可知，l-，模型中的4-DUDWDU,所以廣義差分模型已無序列相關(guān)。d=1.48U根據(jù)0(1-0)=6981.723，可得*105305.023。因此，原回歸模型應(yīng)為y=10530

8、5.023+1.002749xtt采用普萊斯-文斯滕變換后第一個(gè)觀測值變?yōu)閥J1-62為1750.7019和xJ1-6211為344.1377，變換后普通最小二乘結(jié)果為$*=7555503+1-0611x*，根據(jù)*(1-0)=7555-503，得=113959.321,由此，最終模型是y=113959.321+1.0611xttEquation:UNTITLEDTorkfileUHTITLED:U._lviewProcObjectPrintNameFreezeEstimateForecastStatsResidsDependentVariable:Y-0.90527Y(-1)Method:Le

9、astSquaresDate:0&/06/14Time:13:34Sample(adjusted):19322010Includedobservations:29afteradjustmentsVariableCoefficientStd.Errort-StatisticProb.C7555.5032503.4333.0130500.0055X-0.9052?*X(-1)1.0610550.1037S310.223790.0000R-squarEd0.794717Meandependentvar22041.22AdjustedR-squared0.737114S.D.dependentvar24080.00S.E.ofregression11114.46Akaikeinfocriterion21.53635Sumsquaredresid3.3