[面板数据求助] PVAR模型估计GMM估计好多系数不显著会影响模型建立吗

3801

收藏 2020-07-24

SATA小白一枚，最近在摸索pvar，但是做出来的回归系数不显著，想问这个会影响后续的脉冲和方差分解部分吗？理论上是不是后者就没有存在的必要？另外，还想问一下PVAR估计出来的GMM和通过xtabond2估计出来的有区别吗？求高人指点，谢谢！
EQ1: dep.var    : h_BSP

b_GMM se_GMM t_GMM
L.h_BSP 0.2856654 0.59618303 0.47915721
L.h_GLR -0.00159899 0.00301133 -0.53099258
L2.h_BSP -0.10537139 0.22320947 -0.472074
L2.h_GLR -0.00066068 0.00198667 -0.33255599

EQ2: dep.var    : h_GLR

b_GMM se_GMM t_GMM
L.h_BSP -17.833455 20.560817 -0.86735146
L.h_GLR -0.11053433 0.10673184 -1.0356266
L2.h_BSP 6.3993342 6.5922502 0.97073594
L2.h_GLR -0.11947298 0.05881784 -2.0312373

just identified - Hansen statistic is not calculated

symmetric uu[2,2]
         BSP       GLR
BSP    0.00972212
GLR    0.06993311 26.189009

Residuals correlation matrix

         |    u1    u2
-------------+------------------
      u1 | 1.0000
         |
         |
      u2 | 0.1344 1.0000
         | 0.1917
         |

D[6,3]
s BSP GLR
BSP 10 0.99155163 0.00844837
GLR 10 0.1349631 0.8650369
BSP 20 0.99155163 0.00844837
GLR 20 0.1349631 0.8650369
BSP 30 0.99155163 0.00844837
GLR 30 0.1349631 0.8650369