求助：关于GMM结果的解读(版主，没人理我，麻烦您看下可以不)

xtabond2 booklever lagbooklever_fd2 lagbooklever_fd lagbooklever laggrowth lagppe lagprofit lagvar lagsize lagunique la
> gshield lagmean_ml, gmmstyle(lagbooklever_fd2 lagbooklever_fd lagbooklever) ivstyle(laggrowth lagppe lagprofit lagvar l
> agsize lagunique lagshield lagmean_ml) twostep robust small
Favoring speed over space. To switch, type or click on mata: mata set matafavor space, perm.
Warning: Two-step estimated covariance matrix of moments is singular.
  Using a generalized inverse to calculate optimal weighting matrix for two-step estimation.
  Difference-in-Sargan/Hansen statistics may be negative.
Dynamic panel-data estimation, two-step system GMM
------------------------------------------------------------------------------
Group variable: stkcd                           Number of obs      =      2478
Time variable : trdynt                          Number of groups   =       546
Number of instruments = 141                     Obs per group: min =         1
F(11, 545)    =    102.05                                      avg =      4.54
Prob > F      =     0.000                                      max =         9
------------------------------------------------------------------------------
             |              Corrected
   booklever |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
lagbooklev~2 |  -.4242184   .1917901    -2.21   0.027    -.8009567   -.0474802
lagbooklev~d |   .1862471   .0784806     2.37   0.018     .0320856    .3404086
lagbooklever |   .8010709   .0448344    17.87   0.000     .7130016    .8891402
   laggrowth |  -.0070225   .0049706    -1.41   0.158    -.0167863    .0027414
      lagppe |   .0772451   .0212869     3.63   0.000     .0354307    .1190594
   lagprofit |   .1792112   .0724134     2.47   0.014     .0369676    .3214548
      lagvar |   .0389165   .1106138     0.35   0.725     -.178365    .2561981
     lagsize |   -.009736   .0039505    -2.46   0.014     -.017496    -.001976
   lagunique |  -.0555462   .0367504    -1.51   0.131     -.127736    .0166437
   lagshield |  -.6006677   .2181873    -2.75   0.006    -1.029259   -.1720765
  lagmean_ml |   -.017447   .0461081    -0.38   0.705    -.1080184    .0731244
       _cons |   .2596013   .0797927     3.25   0.001     .1028624    .4163402
------------------------------------------------------------------------------
Instruments for first differences equation
  Standard
    D.(laggrowth lagppe lagprofit lagvar lagsize lagunique lagshield
    lagmean_ml)
  GMM-type (missing=0, separate instruments for each period unless collapsed)
    L(1/.).(lagbooklever_fd2 lagbooklever_fd lagbooklever)
Instruments for levels equation
  Standard
    _cons
    laggrowth lagppe lagprofit lagvar lagsize lagunique lagshield lagmean_ml
  GMM-type (missing=0, separate instruments for each period unless collapsed)
    D.(lagbooklever_fd2 lagbooklever_fd lagbooklever)
------------------------------------------------------------------------------
Arellano-Bond test for AR(1) in first differences: z =  -9.06  Pr > z =  0.000
Arellano-Bond test for AR(2) in first differences: z =   0.05  Pr > z =  0.957
------------------------------------------------------------------------------
Sargan test of overid. restrictions: chi2(129)  = 293.69  Prob > chi2 =  0.000
  (Not robust, but not weakened by many instruments.)
Hansen test of overid. restrictions: chi2(129)  = 174.89  Prob > chi2 =  0.004  
(Robust, but can be weakened by many instruments.)

Difference-in-Hansen tests of exogeneity of instrument subsets:
  GMM instruments for levels
    Hansen test excluding group:     chi2(105)  = 139.10  Prob > chi2 =  0.015
    Difference (null H = exogenous): chi2(24)   =  35.78  Prob > chi2 =  0.058

  iv(laggrowth lagppe lagprofit lagvar lagsize lagunique lagshield lagmean_ml)
    Hansen test excluding group:     chi2(121)  = 162.86  Prob > chi2 =  0.007
    Difference (null H = exogenous): chi2(8)    =  12.02  Prob > chi2 =  0.150




问题：
Sargan test of overid. restrictions: chi2(129)  = 293.69  Prob > chi2 =  0.000
  (Not robust, but not weakened by many instruments.)
Hansen test of overid. restrictions: chi2(129)  = 174.89  Prob > chi2 =  0.004  
(Robust, but can be weakened by many instruments.)


Difference-in-Hansen tests of exogeneity of instrument subsets:
  GMM instruments for levels
    Hansen test excluding group:     chi2(105)  = 139.10  Prob > chi2 =  0.015
    Difference (null H = exogenous): chi2(24)   = 35.78  Prob > chi2 =  0.058


  iv(laggrowth lagppe lagprofit lagvar lagsize lagunique lagshield lagmean_ml)
    Hansen test excluding group:     chi2(121)  = 162.86  Prob > chi2 =  0.007
    Difference (null H = exogenous): chi2(8)    =  12.02  Prob > chi2 =  0.150

（1）这三块检验量（红绿蓝）分别用来说明什么，特别是下面两部分（绿色那块和蓝色那块）
（2）（绿色那块和蓝色那块中的）Hansen test excluding group和Difference (null H = exogenous)有什么区别，
（3）139.10+35.78＝162.86+12.02＝174.89；105+24＝121+8＝129

详细点啊，谢谢了，如果不想打字，通过信箱告诉我其它的联系方式，我直接问你，上面的金钱也会直接给你

问题一：用什么来判断其整个模型的拟合度，即类似于OLS回归中的“调整R平方”?
回答：无法判断

问题二：差分hansen检验中“Hansen test excluding group:     chi2(105)  = 139.10  Prob > chi2 =  0.015”和“Difference (null H = exogenous): chi2(24)   =  35.78  Prob > chi2 =  0.058”分别用来说明什么？
同样，“ Hansen test excluding group:     chi2(121)  = 162.86  Prob > chi2 =  0.007”和“Difference (null H = exogenous): chi2(8)    =  12.02  Prob > chi2 =  0.150”又分别用来说明什么？

答：我只知道这是用来判断附加工具变量过度识别的，但为什么是4个，谁能告诉我。


问题三：如果Hansen检验通不过，要怎么处理(换工具变量吗？)；还是说这种方法不适用于这个数据集？

答：可以通过改变工具变量得到好的结果

问题四：对于模型Yt=a0+a1 Yt-1+a2 Mt *Yt-1+a3 Xt, Mt *Yt-1 是Mt和Yt-1的交乘项，那么正常情况下，估计时的命令是
（1）“xtabond2 Yt Yt-1 Mt*Yt-1 Xt, gmmstyle(Yt-1) ivstyle(Xt)”
还是（2）“lagbooklever_fd2 lagbooklever_fd 即 Mt *Yt-1要不要与Yt-1一样视为前定变量？
还是把其当成外生变量（3）“xtabond2 Yt Yt-1 Mt*Yt-1 Xt, gmmstyle(Yt-1) ivstyle(Mt*Yt-1 Xt)”
这三个要选哪一个

答：具体问题具体分析，看交互项的性质，如果是0或1哑变量，则把其与lagbooklever相当；如果是其它的外生变量，则当其为外生变量。

这些是我后来自己理解的，不过怕有问题，大家为什么不讨论一下呢？

楼主，你的问题自己解决了吗？我也遇到同样的问题。比如，判断附加工具变量过度识别的4个结果中，应该汇报哪两个呢？另外，如何通过改变工具变量得到好的结果？着急中，请不吝赐教。

Arellano-Bond test for AR(1) in first differences: z =  -8.40  Pr > z =  0.000
Arellano-Bond test for AR(2) in first differences: z =      .  Pr > z =      .
------------------------------------------------------------------------------
Sargan test of overid. restrictions: chi2(26)   = 191.94  Prob > chi2 =  0.000
  (Not robust, but not weakened by many instruments.)
Hansen test of overid. restrictions: chi2(26)   =  46.11  Prob > chi2 =  0.009
  (Robust, but can be weakened by many instruments.)

Difference-in-Hansen tests of exogeneity of instrument subsets:
  GMM instruments for levels
    Hansen test excluding group:     chi2(11)   =  14.58  Prob > chi2 =  0.203
    Difference (null H = exogenous): chi2(15)   =  31.54  Prob > chi2 =  0.007
  iv( var1 var2 var3 var4 var5 _Iyear_2001 _Iyear_2002 _Iyear_2003 _Iyear
> _2004, eq(level))
    Hansen test excluding group:     chi2(19)   =  25.82  Prob > chi2 =  0.135
    Difference (null H = exogenous): chi2(7)    =  20.30  Prob > chi2 =  0.005

麻烦哪位高手帮我看一下上面结果，为什么AR(2)的P值没有汇报呢？另外，哪个结果是说明整个工具变量有效性的？哪个结果又是说明GMM类和IV类工具变量子集有效性的？麻烦具体一些，先谢谢了。

希望这篇文章对你们有用

劳动力抚养负担对居民储蓄率的影响研究  钟水映李魁

好的，谢谢你

冒昧地说一句。做出来都不知道结果什么意思，谁敢相信你的结果？呵呵。

为什么没有人愿意回答啊，我可以增加金钱 的，只要有人能够解答