Test Equation:
Dependent Variable: RESID^2
Method: Least Squares
Date: 09/12/00 Time: 22:52
Sample(adjusted): 1984 2001
Included observations: 18 after adjusting endpoints
Variable Coefficient Std. Error t-Statistic Prob.
C 0.002490 0.001439 1.730717 0.1055
RESID^2(-1) 0.410769 0.264166 1.554965 0.1423
RESID^2(-2) -0.268047 0.273208 -0.981111 0.3432
RESID^2(-3) 0.069684 0.266371 0.261605 0.7974
R-squared 0.160318 Mean dependent var 0.003144
Adjusted R-squared -0.019613 S.D. dependent var 0.003114
S.E. of regression 0.003144 Akaike info criterion -8.493484
Sum squared resid 0.000138 Schwarz criterion -8.295623
Log likelihood 80.44135 F-statistic 0.890996
从以上结果看Probability的值,拒绝H0反错误概率较大,同时残差序列的系数的t值并不显著,应该接受残差序列系数为零的原假设,即为模型不存在异方差. 另一方面,从White检验看,也不存在异方差.
White Heteroskedasticity Test:
F-statistic 0.845106 Probability 0.516929
Obs*R-squared 3.662919 Probability 0.453542
Test Equation:
Dependent Variable: RESID^2
Method: Least Squares
Date: 09/12/00 Time: 22:54
Sample: 1981 2001
Included observations: 21
Variable Coefficient Std. Error t-Statistic Prob.
C -0.143065 0.152989 -0.935134 0.3636
LOG(Y) 0.033727 0.035255 0.956644 0.3530
(LOG(Y))^2 -0.001709 0.001740 -0.982254 0.3406
LOG(100*R) -0.021206 0.035407 -0.598914 0.5576
(LOG(100*R))^2 0.005768 0.009131 0.631705 0.5365
R-squared 0.174425 Mean dependent var 0.003023
Adjusted R-squared -0.031969 S.D. dependent var 0.003022
S.E. of regression 0.003070 Akaike info criterion -8.530027
3 自相关检验
模拟结果显示DW值为1.178723,而通过查表得到dL的值为1.125,du的值为1.538.DW的值正好落在无决定区域,因此需要对自相关进行修正.利用Cochrane-Orcutt 法对自相关性进行修正,得到以下结果.
Dependent Variable: LOG(M)
Method: Least Squares
Date: 09/12/00 Time: 22:56
Sample(adjusted): 1982 2001
Included observations: 20 after adjusting endpoints
Convergence achieved after 6 iterations
Variable Coefficient Std. Error t-Statistic Prob.
C -3.580921 0.234682 -15.25862 0.0000
LOG(Y) 1.371571 0.022439 61.12427 0.0000
LOG(100*R) -0.162553 0.046406 -3.502824 0.0029
AR(1) 0.399709 0.220168 1.815473 0.0882
R-squared 0.998604 Mean dependent var 10.04683
Adjusted R-squared 0.998342 S.D. dependent var 1.339765
S.E. of regression 0.054558 Akaike info criterion -2.802242
Sum squared resid 0.047625 Schwarz criterion -2.603096
Log likelihood 32.02242 F-statistic 3813.850
Durbin-Watson stat 1.964503 Prob(F-statistic) 0.000000
Inverted AR Roots .40