检验过度识别约束，xtabond2里面sargan and hansen检验老是矛盾，问题到底出在哪里啊？下面用例子说明：

第一个模型不进行标准差调整，第二个模型进行调整。两个模型中sargan and hansen检验结果都矛盾。我做了其他的手册上的模型，发现这是一个普遍问题。一直没有想通。相关性检验又没有什么问题，而且这个例子是David Roodman最后给出的例子，复制DPD for OX中对应的结果。

请高手指点一下，谢谢！

1. 没有放robust

use http://www.stata-press.com/data/r7/abdata.dta
. xtabond2 n l.n l(0/1).(w k) yr1978-yr1984, gmm(L.(w k n)) iv(yr1978-yr1984, eq(lev
> el)) h(2) two small
Dynamic panel-data estimation, two-step system GMM
------------------------------------------------------------------------------
           n |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
           n |
         L1. |    .872881   .0067325   129.65   0.000     .8595697    .8861923
           w |
         --. |  -.7797449   .0087665   -88.95   0.000    -.7970779   -.7624119
         L1. |   .5268032   .0145408    36.23   0.000     .4980536    .5555529
           k |
         --. |   .4700773   .0144778    32.47   0.000     .4414521    .4987024
         L1. |  -.3576081    .014154   -25.27   0.000    -.3855931   -.3296232
      yr1978 |   .0058018   .0055501     1.05   0.298    -.0051717    .0167753
      yr1979 |   .0188977   .0054704     3.45   0.001     .0080817    .0297137
      yr1980 |   .0028196   .0054678     0.52   0.607    -.0079911    .0136304
      yr1981 |  -.0200226   .0065488    -3.06   0.003    -.0329708   -.0070744
      yr1982 |   .0152802   .0047388     3.22   0.002     .0059107    .0246498
      yr1983 |    .031731   .0053602     5.92   0.000     .0211329    .0423291
      yr1984 |   .0224206   .0049654     4.52   0.000      .012603    .0322381
       _cons |   .9484881   .0626819    15.13   0.000     .8245548    1.072421
------------------------------------------------------------------------------
Warning: Uncorrected two-step standard errors are unreliable.
------------------------------------------------------------------------------
Arellano-Bond test for AR(1) in first differences: z =  -6.30  Pr > z =  0.000
Arellano-Bond test for AR(2) in first differences: z =  -0.15  Pr > z =  0.881
------------------------------------------------------------------------------
Sargan test of overid. restrictions: chi2(100)  = 186.75  Prob > chi2 =  0.000

Hansen test of overid. restrictions: chi2(100)  = 111.59  Prob > chi2 =  0.201

2.放robust

. xtabond2 n l.n l(0/1).(w k) yr1978-yr1984, gmm(L.(w k n)) iv(yr1978-yr1984, eq(lev
> el)) h(2) two small robust
Dynamic panel-data estimation, two-step system GMM
------------------------------------------------------------------------------
             |              Corrected
           n |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
           n |
         L1. |    .872881   .0457559    19.08   0.000     .7824136    .9633485
           w |
         --. |  -.7797449   .1177744    -6.62   0.000    -1.012606   -.5468839
         L1. |   .5268032   .1637713     3.22   0.002     .2029983    .8506082
           k |
         --. |   .4700773   .0806911     5.83   0.000     .3105366    .6296179
         L1. |  -.3576081   .0808642    -4.42   0.000    -.5174911   -.1977251
      yr1978 |   .0058018   .0199152     0.29   0.771    -.0335741    .0451777
      yr1979 |   .0188977   .0230045     0.82   0.413    -.0265862    .0643816
      yr1980 |   .0028196   .0243216     0.12   0.908    -.0452685    .0509078
      yr1981 |  -.0200226   .0277278    -0.72   0.471    -.0748454    .0348002
      yr1982 |   .0152802   .0235491     0.65   0.517    -.0312804    .0618409
      yr1983 |    .031731   .0237422     1.34   0.184    -.0152116    .0786736
      yr1984 |   .0224206    .031398     0.71   0.476    -.0396589    .0845001
       _cons |   .9484881   .3814834     2.49   0.014     .1942277    1.702749
------------------------------------------------------------------------------

------------------------------------------------------------------------------
Arellano-Bond test for AR(1) in first differences: z =  -5.81  Pr > z =  0.000
Arellano-Bond test for AR(2) in first differences: z =  -0.15  Pr > z =  0.883
------------------------------------------------------------------------------
Sargan test of overid. restrictions: chi2(100)  = 186.75  Prob > chi2 =  0.000

Hansen test of overid. restrictions: chi2(100)  = 111.59  Prob > chi2 =  0.201

Roodman的working paper见下面第2楼：

https://bbs.pinggu.org/thread-349848-1-1.html

[此贴子已经被作者于2008-8-21 22:52:32编辑过]