EVIEWS在计量经济学教学过程中的演示示例——陈冬冬(川农经管)
Variable C X R-squared
Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat Coefficient -1147.443 0.170052 Std. Error 396.1630 0.011451 t-Statistic -2.896390 14.84990 Prob. 0.0093 0.0000 3080.390 4368.710 17.20981 17.30929 220.5196 0.000000 0.920675 Mean dependent var 0.916500 S.D. dependent var 1262.402 Akaike info criterion 30279518 Schwarz criterion -178.7030 F-statistic 0.688670 Prob(F-statistic)
2、自相关检验 (1)图示法
由上述OLS计算,可直接得到残差RESID,运用GENR命令生成序列E,则在QUICK菜单中选GRAPH,在图形对话框中输入:E E(-1),再点击SCATTER DIOGRAM。得结果如下,从图中可以看出残差et呈线性自回归,表明随机误差ut存在自相关。
400020000E-2000-4000-4000-3000-2000-1000E(-1)010002000
(2)、DW检验
根据OLS计算结果,由:Durbin-Watson stat=0.688670,给定显著性水平a=0.05,查D-W表,n=21,k'(解释变量个数)=1,得下限临界值dL=1.22,上限临界值dU=1.42,因为DW统计量为0.688670
3、自相关的修正
?DW(1)由DW=0.688670,根据??1?,计算得?=0.6556。
2?用GENR分别对X和Y做广义差分。在WORKFILE窗口选择GENR或直接在主窗口输入:
GENR DY=Y-0.6556*Y(-1) GENR DX=X-0.6556*X(-1)
11
EVIEWS在计量经济学教学过程中的演示示例——陈冬冬(川农经管)
然后再用OLS估计参数。结果为:
DY = -585.3252045 + 0.192825853*DX (-1.752142) (8.851754) 2
R=0.813188 DW=1.345597, F=78.35354
这时可以看出使用广义差分法后,DW值有所提高,但仍然存在自相关。
(2)Cochrane-Orcutt迭代法
在QUICK-ESTIMATE EQUATION项,在对话框中输入:Y C X AR(1),OK后得如下结果:
Dependent Variable: Y Method: Least Squares Date: 10/31/05 Time: 18:07 Sample(adjusted): 1979 1998
Included observations: 20 after adjusting endpoints Convergence achieved after 14 iterations
Variable C X AR(1) R-squared
Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat Inverted AR Roots
Coefficient -1876.253 0.198637 0.740777 Std. Error 1975.932 0.049973 0.310956 t-Statistic -0.949554 3.974859 2.382259 Prob. 0.3556 0.0010 0.0292 4428.390 16.84255 16.99191 168.4172 0.000000
0.951955 Mean dependent var 3227.670 0.946303 S.D. dependent var 1026.178 Akaike info criterion 17901699 Schwarz criterion -165.4255 F-statistic 1.451436 Prob(F-statistic) .74
此时DW=1.451436>dU=1.42,认为此时无自相关性。
(3)利用对数线性回归修正自相关。用GENR分别对X和Y生成LogX和LogY,命令为:
GENR LY=LOG(Y) GENR LX=LOG(X)
在OLS估计对话框输入: LY C LX计算结果如下:
Variable C LX R-squared
Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat
Coefficient -7.083917 1.466088 Std. Error 0.337892 0.034876 t-Statistic -20.96504 42.03750 Prob. 0.0000 0.0000 7.043795 1.515819 -0.731905 -0.632427 1767.151 0.000000
0.989363 Mean dependent var 0.988803 S.D. dependent var 0.160400 Akaike info criterion 0.488832 Schwarz criterion 9.685003 F-statistic 1.140571 Prob(F-statistic)
12
EVIEWS在计量经济学教学过程中的演示示例——陈冬冬(川农经管)
同时考虑Cochrane-Orcutt迭代法。在估计对话框里输入: LY C LX AR(1) 计算结果如下:
Dependent Variable: LY Method: Least Squares Date: 10/31/05 Time: 18:19 Sample(adjusted): 1979 1998
Included observations: 20 after adjusting endpoints Convergence achieved after 5 iterations Variable Coefficient Std. Error t-Statistic Prob. C -7.088071 0.604260 -11.73016 0.0000 LX 1.467507 0.061299 23.94022 0.0000 AR(1)
0.425084
0.232705
1.826708
0.0854 R-squared
0.990102 Mean dependent var 7.150795 Adjusted R-squared 0.988937 S.D. dependent var 1.471582 S.E. of regression 0.154782 Akaike info criterion -0.756115 Sum squared resid 0.407278 Schwarz criterion -0.606756 Log likelihood 10.56115 F-statistic 850.2191 Durbin-Watson stat 1.537282 Prob(F-statistic) 0.000000 Inverted AR Roots .43
此时DW=1.537282>dU=1.42,可以认为此时已消除自相关性。
13
百度搜索“77cn”或“免费范文网”即可找到本站免费阅读全部范文。收藏本站方便下次阅读,免费范文网,提供经典小说综合文库EVIEWS在计量经济学教学过程中的演示示例(3)在线全文阅读。
相关推荐: