发帖
雷达卡
其一：检验疑问：
| Redundant Fixed Effects Tests | |
| Pool: LOAN |
|
| Test cross-section and period fixed effects | |
|
|
|
|
|
|
|
|
|
|
|
|
| Effects Test | Statistic | d.f. | Prob. | |
|
|
|
|
|
|
|
|
|
|
|
|
| Cross-section F | 37.456022 | (8,166) | 0.0000 | |
| Cross-section Chi-square | 204.225630 | 8 | 0.0000 | |
| Period F | 1.065008 | (21,166) | 0.3905 | |
| Period Chi-square | 25.026154 | 21 | 0.2460 | |
| Cross-Section/Period F | 12.041072 | (29,166) | 0.0000 | |
| Cross-Section/Period Chi-square | 224.244915 | 29 | 0.0000 | |
|
|
|
|
|
|
|
|
|
|
|
|
这是不是表明只有CROSS- Section是固定的，通过了检验，但是PERIOD 不是固定的？
然后这个检验只用看这里通过与否，做了检验带来的R值啊那些变化不是模型的变化吧?
其三：加入ar(4)可以使得所有解释变量通过T检验，但是拟合优度稍微下降，AR（2）拟合优度高，但是AR（2）自身通不过T检验(P值
为0.51)，AR（5）拟合优度就下降的多了，就不考虑之后
| Included observations: 18 after adjustments |
| |||
| Cross-sections included: 9 |
|
| ||
| Total pool (balanced) observations: 162 |
| |||
| Iterate coefficients after one-step weighting matrix | ||||
| Convergence achieved after 15 total coef iterations | ||||
|
|
|
|
|
|
| | ||||
| Variable | Coefficient | Std. Error | t-Statistic | Prob. |
|
|
|
|
|
|
|
|
|
|
|
|
| C | 137.6160 | 4.839253 | 28.43744 | 0.0000 |
| DR? | 212.4604 | 97.05961 | 2.188968 | 0.0301 |
| DCJL? | -14.98573 | 3.574350 | -4.192576 | 0.0000 |
| AR(4) | -0.260654 | 0.102454 | -2.544099 | 0.0120 |
| Fixed Effects (Cross) |
|
|
|
|
| | ||||
|
| Effects Specification |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| Cross-section fixed (dummy variables) |
| |||
|
|
|
|
|
|
|
|
|
|
|
|
|
| Weighted Statistics |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| R-squared | 0.671463 | Mean dependent var | 0.828570 | |
| Adjusted R-squared | 0.647370 | S.D. dependent var | 1.324255 | |
| S.E. of regression | 0.877917 | Sum squared resid | 115.6107 | |
| F-statistic | 27.86994 | Durbin-Watson stat | 2.026346 | |
| Prob(F-statistic) | 0.000000 |
|
|
|
|
|
|
|
|
|
| | ||||
不太明白有时候是拟合优度至上，哪怕有解释变量通不过T检验，
还是先让所有变量通过T检验较为重要？
然后这是之前不加任何AR时候的 结果：
|
| ||||||||||||||||||||||||||||||||||||||||||||||
小白求帮助，到底哪一个作为最终结果比较恰当
京公网安备 11010802022788号
