黃河泉 发表于 2019-2-6 16:39
我搞不太懂为何你要比较 Hansen 检验与 0.25?你有用 collapse 之选项吗?
我的语句是先用xtabond2 manu1 l.manu1 l2.manu1 l1lnpgdp a a2 lnlc phw year4-year16 , gmm(manu1,lag(2 .) collapse) iv(l1lnpgdp year3-year16 , eq(level)) iv( a a2 lnlc phw ) twostep robust,输出结果中
Dynamic panel-data estimation, two-step system GMM
------------------------------------------------------------------------------
Group variable: _ID Number of obs = 377
Time variable : year Number of groups = 29
Number of instruments = 32 Obs per group: min = 13
Wald chi2(19) = 5857.98 avg = 13.00
Prob > chi2 = 0.000 max = 13
------------------------------------------------------------------------------
| Corrected
manu1 | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
manu1 |
L1. | .9052043 .1870041 4.84 0.000 .538683 1.271726
L2. | -.1956644 .066926 -2.92 0.003 -.326837 -.0644917
|
l1lnpgdp | .5206942 1.23098 0.42 0.672 -1.891982 2.93337
a | .2218914 .800566 0.28 0.782 -1.347189 1.790972
a2 | -.1669151 .3524306 -0.47 0.636 -.8576663 .5238362
lnlc | -27.19269 13.51177 -2.01 0.044 -53.67527 -.7101158
phw | -.3206022 .1765244 -1.82 0.069 -.6665836 .0253792
year4 | -3.486217 1.52771 -2.28 0.022 -6.480473 -.4919615
year6 | 1.827706 1.318853 1.39 0.166 -.7571992 4.412611
year7 | 3.933458 2.759256 1.43 0.154 -1.474583 9.3415
year8 | 8.486659 4.084415 2.08 0.038 .481354 16.49196
year9 | 10.28791 5.402165 1.90 0.057 -.3001372 20.87596
year10 | 11.50536 7.216229 1.59 0.111 -2.638189 25.64891
year11 | 14.99454 8.014553 1.87 0.061 -.7136937 30.70277
year12 | 19.71353 8.906814 2.21 0.027 2.256494 37.17056
year13 | 19.10681 10.74584 1.78 0.075 -1.954642 40.16827
year14 | 21.25112 11.833 1.80 0.073 -1.941139 44.44338
year15 | 22.74845 12.83755 1.77 0.076 -2.412687 47.9096
year16 | 25.85722 14.15606 1.83 0.068 -1.888145 53.60259
_cons | 277.0422 142.5951 1.94 0.052 -2.439025 556.5235
------------------------------------------------------------------------------
Instruments for first differences equation
Standard
D.(a a2 lnlc phw )
GMM-type (missing=0, separate instruments for each period unless collapsed)
L(2/14).manu1 collapsed
Instruments for levels equation
Standard
a a2 lnlc phw
l1lnpgdp year3 year4 year5 year6 year7 year8 year9 year10 year11 year12
year13 year14 year15 year16
_cons
GMM-type (missing=0, separate instruments for each period unless collapsed)
DL.manu1 collapsed
------------------------------------------------------------------------------
Arellano-Bond test for AR(1) in first differences: z = -3.67 Pr > z = 0.014
Arellano-Bond test for AR(2) in first differences: z = -0.06 Pr > z = 0.707
------------------------------------------------------------------------------
Sargan test of overid. restrictions: chi2(12) = 89.65 Prob > chi2 = 0.000
(Not robust, but not weakened by many instruments.)
Hansen test of overid. restrictions: chi2(12) = 10.43 Prob > chi2 = 0.578
(Robust, but weakened by many instruments.)
Difference-in-Hansen tests of exogeneity of instrument subsets:
GMM instruments for levels
Hansen test excluding group: chi2(11) = 8.90 Prob > chi2 = 0.631
Difference (null H = exogenous): chi2(1) = 1.54 Prob > chi2 = 0.215
iv(a a2 lnlc phw )
Hansen test excluding group: chi2(8) = 9.25 Prob > chi2 = 0.322
Difference (null H = exogenous): chi2(4) = 1.19 Prob > chi2 = 0.880
我记得我看文献说,hansen Test的 p值不能太大,一般超过0.25就要注意可能是工具变量太多了。我分析主要是加入了year4-16这些时间变量。
另外请问是否 xtabond2 …… twostep robust两步法 结果主要用来看这些检验的p值,而具体估算出的回归系数等还是看一步法xtabond2 …… robust的结果比较稳健?
麻烦您了。