時間序列相關性檢驗-自相關

1、序列相關性檢驗(一)一元線性回歸結果：Depe ndent Variable: YMethod: Least SquaresDate: 06/01/12 Time: 14:16Sample: 1981 2007In cluded observati ons: 27VariableCoefficie ntStd. Errort-StatisticProb.C4276.3621079.7863.9603800.0005X0.8716680.02944829.600120.0000R-squared0.972258Mean depe ndent var24869.44Adjusted R-squar

2、ed0.971149S.D.dependent var25261.92S.E. of regressi on4290.920Akaike info criteri on19.63758Sum squared resid4.60E+08Schwarz criteri on19.73356Log likelihood-263.1073F-statistic876.1668Durb in -Watson stat0.174669Prob(F-statistic)0.000000= =（二）拉格朗日乘數檢驗：含二階殘差項的回歸結果：Breusch-Godfrey Serial Correlation

3、LM Test:F-statistic120.8648Probability0.000000Obs*R-squared24.65421Probability0.000004Test Equati on:Depe ndent Variable: RESIDMethod: Least SquaresDate: 06/01/12Time: 13:50VariableCoefficie ntStd. Errort-StatisticProb.C361.5102372.64610.9701170.3421X-0.0256970.013222-1.9433980.0643RESID(-1)1.477525

4、0.1936207.6310490.0000RESID(-2)-0.4852980.229297-2.1164590.0453R-squared0.913119Mean depe ndent var-2.29E-12Adjusted R-squared0.901787S.D.dependent var4207.593S.E. of regressi on1318.618Akaike info criteri on17.34251Sum squared resid39991346Schwarz criteri on17.53449Log likelihood-230.1239F-statisti

5、c80.57655Durb in -Watson stat1.772240Prob(F-statistic)0.000000含三階殘差項的回歸結果：Breusch-Godfrey Serial Correlation LM Test:F-statistic77.16026Probability0.000000Obs*R-squared24.65663Probability0.000018Test Equati on:Dependent Variable: RESIDMethod: Least SquaresDate: 06/01/12Time: 13:53VariableCoefficie n

6、tStd. Errort-StatisticProb.C340.4064405.78320.8388870.4106X-0.0246880.015080-1.6371600.1158RESID(-1)1.4649820.2146826.8239740.0000RESID(-2)-0.4417890.371964-1.1877210.2476RESID(-3)-0.0391990.260256-0.1506180.8816R-squared0.913208Mean depe ndent var-2.29E-12Adjusted R-squared0.897428S.D.dependent var

7、4207.593S.E. of regressi on1347.559Akaike info criteri on17.41555Sum squared resid39950151Schwarz criteri on17.65552Log likelihood-230.1100F-statistic57.87019Durb in -Watson stat1.751706Prob(F-statistic)0.000000= =序列相關性消除(一)二階迭代法回歸結果：Depe ndent Variable: YMethod: Least SquaresDate: 06/01/12 Time: 15

8、:22Sample(adjusted): 1983 2007In eluded observati ons: 25 after adjusti ng en dpo intsCon verge nee not achieved after 100 iterati onsVariableCoefficie ntStd. Errort-StatisticProb.C921803.7585832910.0157350.9876X0.6036150.0879456.8635190.0000AR(1)1.5195610.1896688.0116780.0000AR(2)-0.5200790.203612-

9、2.5542640.0185R-squared0.998698Mean depe ndent var26697.20Adjusted R-squared0.998512S.D.dependent var25384.37S.E. of regressi on979.2577Akaike info criteri on16.75711Sum squared resid20137857Schwarz criteri on16.95213Log likelihood-205.4639F-statistic5368.622Durb in -Watson stat1.759566Prob(F-statis

10、tic)0.000000In verted AR Roots1.00.52再用拉格朗日乘數檢驗自相關是否已消除 含二階殘差項回歸結果：Breusch-Godfrey Serial Correlation LM Test:F-statistic0.831638Probability0.450575Obs*R-squared2.011998Probability0.365679Test Equati on:Depe ndent Variable: RESIDMethod: Least SquaresDate: 06/01/12 Time: 15:27VariableCoefficie ntStd.

11、 Errort-StatisticProb.C33620447893174890.3764150.7108X-0.0383090.101577-0.3771430.7102AR(1)-0.0096630.725955-0.0133100.9895AR(2)0.0286470.7577680.0378040.9702RESID(-1)0.1573620.7324180.2148520.8322RESID(-2)-0.2659740.409538-0.6494480.5238R-squared0.080480Mean depe ndent var3.544879Adjusted R-squared-0.161499S.D.dependent var916.0045S.E. of regressi on987.2043Ak