如果直接建立VAR模型，AIC、SC数值为什么会与通过lag length criteria显示的同阶数（1阶）下的 AIC、SC数值不一样？
如下，表一为通过VAR(1,1)出来的结果， 表二是经VAR代出 lag length criteria在1-7阶的显示结果 ，相应 AIC、SC已用下划线标出。
表一
| | AA | GG | CC |
| AA(-1) | | | |
| GG(-1) | | | |
| CC(-1) | | | |
| C | | | |
| R-squared | 0.223623 | 0.423853 | 0.237276 |
| Adj. R-squared | 0.158924 | 0.375841 | 0.173715 |
| Sum sq. resids | 0.054576 | 0.319372 | 0.02049 |
| S.E. equation | 0.038936 | 0.094188 | 0.023857 |
| F-statistic | 3.4564 | 8.828022 | 3.733074 |
| Log likelihood | 75.18312 | 39.84802 | 94.7767 |
| Akaike AIC | -3.55916 | -1.7924 | -4.53884 |
| Schwarz SC | -3.39027 | -1.62351 | -4.36995 |
| Mean dependent | 0.034579 | 0.027974 | 0.004397 |
| S.D. dependent | 0.042455 | 0.11922 | 0.026245 |
| | | |
Determinant resid covariance (dof adj.) | 5.00E-09 |
Determinant resid covariance | 3.65E-09 |
Log likelihood | 218.3059 |
Akaike information criterion | -10.3153 |
Schwarz criterion | -9.80863 |
表二
| Lag | LogL | LR | FPE | AIC | SC | HQ |
0 | 199.0045 | NA | 1.97E-09 | -11.52968 | -11.395 | -11.48375 |
1 | 220.5385 | 38.00123 | 9.47E-10 | -12.26697 | -11.72826* | -12.08326 |
2 | 236.1821 | 24.84572 | 6.50E-10 | -12.65777 | -11.71502 | -12.33627 |
3 | 244.9171 | 12.33167 | 6.85E-10 | -12.64218 | -11.29539 | -12.18289 |
4 | 252.9057 | 9.868315 | 7.79E-10 | -12.58269 | -10.83186 | -11.98561 |
5 | 270.596 | 18.73094* | 5.26E-10 | -13.09388 | -10.93902 | -12.35901 |
6 | 279.7408 | 8.068949 | 6.32E-10 | -13.1024 | -10.5435 | -12.22975 |
7 | 297.3063 | 12.39917 | 5.18e-10* | -13.60626* | -10.64332 | -12.59581* |
扫码加好友，拉您进群
全部版块
我的主页
收藏
