一，ARIMA单整阶数识别
平稳性检验（DF检验）
首先检验cpi的平稳性：
ls d(cpi) cpi
|
Variable |
Coefficient |
Std. Error |
t-Statistic |
Prob. |
|
CPI |
-0.007606 |
0.015512 |
-0.490311 |
0.6315 |
ls d(cpi) c cpi
|
Variable |
Coefficient |
Std. Error |
t-Statistic |
Prob. |
|
C |
-29.23275 |
23.49071 |
-1.244439 |
0.2353 |
|
CPI |
0.268076 |
0.222053 |
1.207261 |
0.2488 |
ls d(cpi) c t cpi
|
Variable |
Coefficient |
Std. Error |
t-Statistic |
Prob. |
|
C |
-52.04610 |
27.14520 |
-1.917322 |
0.0793 |
|
T |
0.599964 |
0.401354 |
1.494849 |
0.1608 |
|
CPI |
0.433072 |
0.239195 |
1.810543 |
0.0953 |
3个模型都不能拒绝cpi的系数等于0,所以cpi是不平稳的.
在检验cpi一阶差分的平稳性:
genr c1=d(cpi)
ls d(c1) c1
|
Variable |
Coefficient |
Std. Error |
t-Statistic |
Prob. |
|
C1 |
0.527433 |
0.326352 |
1.616148 |
0.1301 |
ls d(c1) c c1
|
Variable |
Coefficient |
Std. Error |
t-Statistic |
Prob. |
|
C |
1.227163 |
1.669116 |
0.735217 |
0.4763 |
|
C1 |
0.524654 |
0.332298 |
1.578865 |
0.1404 |
ls d(c1) c t c1
|
Variable |
Coefficient |
Std. Error |
t-Statistic |
Prob. |
|
C |
3.820403 |
4.406034 |
0.867084 |
0.4044 |
|
T |
-0.272830 |
0.427133 |
-0.638748 |
0.5361 |
|
C1 |
0.501015 |
0.342815 |
1.461475 |
0.1719 |
同样3个模型都不能拒绝c1的系数等于0,所以c1是不平稳的.
二，在检查cpi的二阶差分的平稳性:
genr c2=d(c1)
ls d(c2) c2
|
Variable |
Coefficient |
Std. Error |
t-Statistic |
Prob. |
|
C2 |
0.772258 |
0.364604 |
2.118070 |
0.0557 |
ls d(c2) c c2
|
Variable |
Coefficient |
Std. Error |
t-Statistic |
Prob. |
|
C |
-1.259522 |
1.958370 |
-0.643148 |
0.5333 |
|
C2 |
0.780742 |
0.374085 |
2.087070 |
0.0609 |
ls d(c2) c t c2
|
Variable |
Coefficient |
Std. Error |
t-Statistic |
Prob. |
|
C |
-5.563968 |
5.686036 |
-0.978532 |
0.3509 |
|
T |
0.430896 |
0.533202 |
0.808129 |
0.4378 |
|
C2 |
0.756281 |
0.381331 |
1.983267 |
0.0755 |
3个模型的c2的t统计量都近似等于2,就是都可以拒绝c2的系数等于0,所以说c2是平稳的.
即cpi二阶差分是平稳的,ARIMA的I=2.
评论