Test Equation:
Dependent Variable: RESID^2
Method: Least Squares
Date: 12/26/04 Time: 15:29
Sample(adjusted): 1991 2003
Included observations: 13 after adjusting endpoints
Variable Coefficient Std. Error t-Statistic Prob.
C 0.000253 0.000196 1.291025 0.2232
RESID^2(-1) 0.071351 0.288059 0.247695 0.8089
R-squared 0.005547 Mean dependent var 0.000278
Adjusted R-squared -0.084858 S.D. dependent var 0.000585
S.E. of regression 0.000609 Akaike info criterion -11.82788
Sum squared resid 4.08E-06 Schwarz criterion -11.74096
Log likelihood 78.88120 F-statistic 0.061353
Durbin-Watson stat 2.036923 Prob(F-statistic) 0.808933
Obs*R-squared为0.0721,根据其检验公式(n-p)*R2=13*0.0721=0.9373<(0.05)=7.87944,说明其异方差不严重。
(3)D-W值:
根据回归结果,D-W值为1.1415,说明自相关不严重。
(4)扩展的迪克-富勒检验:
ADF Test Statistic -3.223814 1% Critical Value* -4.0681
5% Critical Value -3.1222
10% Critical Value -2.7042
*MacKinnon critical values for rejection of hypothesis of a unit root.
Augmented Dickey-Fuller Test Equation
Dependent Variable: D(R1)
Method: Least Squares
Date: 12/26/04 Time: 14:01
Sample(adjusted): 1991 2003
Included observations: 13 after adjusting endpoints
Variable Coefficient Std. Error t-Statistic Prob.
R1(-1) -0.928649 0.288059 -3.223814 0.0081
C 0.000253 0.000196 1.291025 0.2232
R-squared 0.485813 Mean dependent var -6.78E-05
Adjusted R-squared 0.439068 S.D. dependent var 0.000814
S.E. of regression 0.000609 Akaike info criterion -11.82788
Sum squared resid 4.08E-06 Schwarz criterion -11.74096
Log likelihood 78.88120 F-statistic 10.39298
Durbin-Watson stat 2.036923 Prob(F-statistic) 0.008104
ADF值说明在显著性为1%的情况下模型是非平稳的;但在5%-10%的显著性水平情况下,是平稳的。
<二>、滞后一期的回归与检验:
以下将在方程中引入滞后变量(-1),希望通过这样的方式,以期望在上述几个不足之处,得到提高:
Dependent Variable: Y2
Method: Least Squares
Date: 12/25/04 Time: 10:00
Sample(adjusted): 1990 2003
Included observations: 14 after adjusting endpoints
Variable Coefficient Std. Error t-Statistic Prob.
X1 0.634182 0.153133 4.141369 0.0025
X2 1.098138 1.094926 1.002933 0.3421
X3 0.707316 0.138436 5.109333 0.0006
X5 -13.59997 5.005682 -2.716906 0.0237
X1(-1) -0.513131 0.171368 -2.994330 0.0151
R-squared 0.764610 Mean dependent var 0.092857
Adjusted R-squared 0.659993 S.D. dependent var 0.027363
S.E. of regression 0.015956 Akaike info criterion -5.165568
Sum squared resid 0.002291 Schwarz criterion -4.937333
Log likelihood 41.15897 Durbin-Watson stat 2.207147
可决系数达到0.7646;的t 检验值为1.0029,尽管还存在问题,但的确有所改进,与无滞后回归结果相比较来看。
有滞后的简单相关系数矩阵:
X1 X1(-1) X2 X3 X5
X1 1 0.861788095785 -0.305074860295 0.288901418405 0.709362441472
X1(-1) 0.861788095785 1 -0.194910058875 0.367472471048 0.734308549284
X2 -0.305074860295 -0.194910058875 1 -0.0560525422067 -0.128645904536
X3 0.288901418405 0.367472471048 -0.0560525422067 1 0.752410491281
X5 0.709362441472 0.734308549284 -0.128645904536 0.752410491281 1
简单相关系数矩阵的结果,从一个侧面说明了多重共线性不严重。
ARCH检验(1)
ARCH Test:
F-statistic 0.717218 Probability 0.415103
Obs*R-squared 0.795738 Probability 0.372371
Test Equation:
Dependent Variable: RESID^2
Method: Least Squares
Date: 12/26/04 Time: 15:34
Sample(adjusted): 1991 2003
Included observations: 13 after adjusting endpoints
Variable Coefficient Std. Error t-Statistic Prob.
C 9.65E-05 7.59E-05 1.270708 0.2300
RESID^2(-1) 0.223359 0.263741 0.846887 0.4151
R-squared 0.061211 Mean dependent var 0.000135
Adjusted R-squared -0.024134 S.D. dependent var 0.000216
S.E. of regression 0.000218 Akaike info criterion -13.88190
Sum squared resid 5.24E-07 Schwarz criterion -13.79498
Log likelihood 92.23235 F-statistic 0.717218
Durbin-Watson stat 2.074059 Prob(F-statistic) 0.415103
计算:(n-p) =120.795738=9.548856<,接受原假设,表明模型中随机误差项不存在异方差。
(2).图示法:
由图直观观测可知,异方差不存在。
4、自相关:
同样,此图显示,自相关不存在。
ADF Test Statistic -11.38700 1% Critical Value* -4.2207
5% Critical Value -3.1801
10% Critical Value -2.7349
*MacKinnon critical values for rejection of hypothesis of a unit root.
Augmented Dickey-Fuller Test Equation
Dependent Variable: D(TABLE1)
Method: Least Squares
Date: 12/27/04 Time: 12:20
Sample(adjusted): 1993 2003
Included observations: 11 after adjusting endpoints
Variable Coefficient Std. Error t-Statistic Prob.
TABLE1(-1) -0.991444 0.087068 -11.38700 0.0000
D(TABLE1(-1)) 0.044339 0.078928 0.561764 0.5918
D(TABLE1(-2)) 0.019404 0.056557 0.343096 0.7416
C 9.92E-05 4.51E-05 2.202218 0.0635
R-squared 0.976688 Mean dependent var -0.000192
Adjusted R-squared 0.966698 S.D. dependent var 0.000615
S.E. of regression 0.000112 Akaike info criterion -15.07585
Sum squared resid 8.82E-08 Schwarz criterion -14.93117
Log likelihood 86.91720 F-statistic 97.75970
Durbin-Watson stat 1.592054 Prob(F-statistic) 0.000004
由上表知由该模型回归产生的残差序列在扩展的迪克-富勒检验下是平稳的,从而可以保证该回归结果的真实性.
<三>、对于计量经济检验的评价:
1.无滞后的回归结果是: