《序列相关案例分析.ppt》由会员分享，可在线阅读，更多相关《序列相关案例分析.ppt（23页珍藏版）》请在三一办公上搜索。
1、第二十讲,序列相关案例分析,主要内容：,序列相关的侦查序列相关的处理,考察我国1984年2004年进出口额与国内生产总值之间的关系,数据收集,回归方程,Dependent Variable:YMethod:Least SquaresDate:11/28/10 Time:20:42Sample:1989 2004Included observations:16Variable Coefficient Std.Errort-Statistic Prob.C-11935.44 4575.009-2.608834 0.0206 X 0.632955 0.06010810.53021 0.0000R-s
2、quared 0.887897 Mean dependent var 30494.93Adjusted R-squared 0.879890 S.D.dependent var 25007.71S.E.of regression 8666.900 Akaike info criterion 21.08888Sum squared resid 1.05E+09 Schwarz criterion 21.18545Log likelihood-166.7110 Hannan-Quinn criter.21.09382F-statistic 110.8854 Durbin-Watson stat0.
3、383500Prob(F-statistic)0.000000,序列相关的侦查（1）图解法,（2）DW检验 在给定，查表得，存在着正的序列相关,Null Hypothesis:E has a unit rootExogenous:ConstantLag Length:3(Automatic based on SIC,MAXLAG=3)t-StatisticProb.*Augmented Dickey-Fuller test statistic-1.3587560.5657Test critical values:1%level-4.121990 5%level-3.144920 10%leve
4、l-2.713751,（3）BG检验,Breusch-Godfrey Serial Correlation LM Test:F-statistic 23.01977Prob.F(2,12)0.0001Obs*R-squared12.69191Prob.Chi-Square(2)0.0018Test Equation:Dependent Variable:RESIDMethod:Least SquaresDate:11/28/10 Time:21:23Sample:1989 2004Included observations:16Presample missing value lagged re
5、siduals set to zero.VariableCoefficientStd.Errort-StatisticProb.C-2373.8172800.904-0.8475180.4133X0.0505730.0440251.1487270.2731RESID(-1)1.5764310.2778355.6739860.0001RESID(-2)-0.5729160.393145-1.4572650.1707R-squared 0.793244Mean dependent var5.23E-12Adjusted R-squared0.741556S.D.dependent var8373.
6、021S.E.of regression4256.629Akaike info criterion19.76266Sum squared resid2.17E+08Schwarz criterion19.95581Log likelihood-154.1013Hannan-Quinn criter.19.77255F-statistic 15.34651Durbin-Watson stat1.997979Prob(F-statistic)0.000208,Breusch-Godfrey Serial Correlation LM Test:F-statistic 23.0197 Prob.F(
7、2,12)0.0001Obs*R-squared 12.69191 Prob.Chi-Square(2)0.0018,序列相关的处理（1）可行性GLS法 1、基于d统计量的FGLS，生成，进行估计,Dependent Variable:Y1Method:Least SquaresDate:11/28/10 Time:22:34Sample(adjusted):1990 2004Included observations:15 after adjustmentsVariable Coefficient Std.Error t-StatisticProb.C-8403.595 2756.874-3
8、.0482340.0093 X1 0.977589 0.125228 7.8064760.0000R-squared 0.824184Mean dependent var11108.13Adjusted R-squared 0.810660S.D.dependent var10353.99S.E.of regression 4505.354Akaike info criterion19.78749Sum squared resid 2.64E+08Schwarz criterion19.88189Log likelihood-146.4061Hannan-Quinn criter.19.786
9、48F-statistic 60.94107Durbin-Watson stat0.725518Prob(F-statistic)0.000003,（2）基于残差自回归的FGLS 先估计，得到 再进行GLS,Dependent Variable:EMethod:Least SquaresDate:11/28/10 Time:22:53Sample(adjusted):1990 2004Included observations:15 after adjustmentsVariableCoefficient Std.Error t-Statistic Prob.C1082.892 1442.11
10、8 0.750904 0.4661 E11.038092 0.224787 4.618117 0.0005R-squared0.621289Mean dependent var-359.2454Adjusted R-squared0.592158S.D.dependent var8538.311S.E.of regression5452.780Akaike info criterion20.16920Sum squared resid3.87E+08Schwarz criterion 20.26361Log likelihood-149.2690 Hannan-Quinn criter.20.
11、16820F-statistic 21.32701 Durbin-Watson stat0.626748Prob(F-statistic)0.000482,Dependent Variable:Y1Method:Least SquaresDate:11/28/10 Time:22:57Sample(adjusted):1990 2004Included observations:15 after adjustmentsVariableCoefficient Std.Error t-Statistic Prob.C-1389.632 2248.993-0.617891 0.5473 X11.15
12、3718 0.327927 3.518211 0.0038R-squared0.487741Mean dependent var 5095.562Adjusted R-squared0.448337S.D.dependent var 6718.627S.E.of regression4990.195Akaike info criterion 19.99190Sum squared resid3.24E+08 Schwarz criterion 20.08631Log likelihood-147.9393 Hannan-Quinn criter.19.99090F-statistic 12.3
13、7781Durbin-Watson stat 0.689689Prob(F-statistic)0.003780,3、奥科特-科克伦迭代法 接上面的结果，再次计算,Dependent Variable:EMethod:Least SquaresDate:11/29/10 Time:18:58Sample(adjusted):1990 2004Included observations:15 after adjustmentsVariableCoefficient Std.Error t-Statistic Prob.C1082.892 1442.118 0.750904 0.4661 E(-1
14、)1.038092 0.224787 4.618117 0.0005R-squared0.621289Mean dependent var-359.2454Adjusted R-squared0.592158S.D.dependent var 8538.311S.E.of regression5452.780Akaike info criterion 20.16920Sum squared resid3.87E+08Schwarz criterion 20.26361Log likelihood-149.2690 Hannan-Quinn criter.20.16820F-statistic
15、21.32701 Durbin-Watson stat 0.626748Prob(F-statistic)0.000482,Dependent Variable:Y1Method:Least SquaresDate:11/29/10 Time:19:02Sample(adjusted):1990 2004Included observations:15 after adjustmentsVariable Coefficient Std.Error t-StatisticProb.C-1389.632 2248.993-0.6178910.5473 X1 1.153718 0.327927 3.
16、5182110.0038R-squared 0.487741Mean dependent var 5095.562Adjusted R-squared 0.448337S.D.dependent var 6718.627S.E.of regression 4990.195Akaike info criterion 19.99190Sum squared resid 3.24E+08Schwarz criterion 20.08631Log likelihood-147.9393Hannan-Quinn criter.19.99090F-statistic 12.37781Durbin-Wats
17、on stat 0.689689Prob(F-statistic)0.003780,Dependent Variable:YMethod:Least SquaresDate:11/29/10 Time:19:05Sample(adjusted):1990 2004Included observations:15 after adjustmentsConvergence achieved after 18 iterationsVariableCoefficientStd.Errort-StatisticProb.C-87541.1768965.19-1.2693530.2284X1.171011
18、0.2405014.8690560.0004AR(1)0.9194820.07004413.127290.0000R-squared0.973553Mean dependent var32250.85Adjusted R-squared0.969146S.D.dependent var24843.50S.E.of regression4363.862Akaike info criterion19.77696Sum squared resid2.29E+08Schwarz criterion19.91857Log likelihood-145.3272Hannan-Quinn criter.19
19、.77545F-statistic220.8724Durbin-Watson stat0.879755Prob(F-statistic)0.000000Inverted AR Roots.92ls y c x ar(1),Dependent Variable:YMethod:Least SquaresDate:11/29/10 Time:19:05Sample(adjusted):1990 2004Included observations:15 after adjustmentsConvergence achieved after 18 iterationsVariable Coeffici
20、ent Std.Error t-Statistic Prob.C-87541.17 68965.19-1.269353 0.2284 X 1.171011 0.240501 4.869056 0.0004AR(1)0.919482 0.070044 13.12729 0.0000R-squared 0.973553Mean dependent var32250.85Adjusted R-squared 0.969146S.D.dependent var24843.50S.E.of regression 4363.862Akaike info criterion19.77696Sum squar
21、ed resid 2.29E+08Schwarz criterion19.91857Log likelihood-145.3272Hannan-Quinn criter.19.77545F-statistic 220.8724Durbin-Watson stat0.879755Prob(F-statistic)0.000000Inverted AR Roots.92,ls y c x ar(1)ar(2),Dependent Variable:YMethod:Least SquaresDate:11/29/10 Time:19:08Sample(adjusted):1991 2004Inclu
22、ded observations:14 after adjustmentsConvergence achieved after 19 iterationsVariable Coefficient Std.Error t-Statistic Prob.C-72632.98 55162.18-1.3167170.2173 X 1.217157 0.303008 4.0169120.0025AR(1)1.431497 0.265893 5.3837410.0003AR(2)-0.518707 0.274814-1.8874880.0884R-squared 0.983855Mean dependen
23、t var 34157.34Adjusted R-squared 0.979011S.D.dependent var 24616.31S.E.of regression 3566.295Akaike info criterion 19.43140Sum squared resid 1.27E+08Schwarz criterion 19.61399Log likelihood-132.0198Hannan-Quinn criter.19.41450F-statistic 203.1255Durbin-Watson stat 2.245280Prob(F-statistic)0.000000In
24、verted AR Roots.72-.08i.72+.08i,(3)尼威-韦斯特方法,Dependent Variable:YMethod:Least SquaresDate:11/29/10 Time:19:26Sample:1989 2004Included observations:16Newey-West HAC Standard Errors&Covariance(lag truncation=2)VariableCoefficient Std.Error t-Statistic Prob.C-11935.44 5715.971-2.088086 0.0555X0.632955 0
25、.099324 6.372605 0.0000R-squared0.887897Mean dependent var 30494.93Adjusted R-squared0.879890S.D.dependent var 25007.71S.E.of regression8666.900Akaike info criterion 21.08888Sum squared resid1.05E+09Schwarz criterion 21.18545Log likelihood-166.7110 Hannan-Quinn criter.21.09382F-statistic 110.8854Durbin-Watson stat 0.383500Prob(F-statistic)0.000000,
