stata中差分GMM后的检验

vapory 发表于 2010-5-17 18:31
estat sargan
我记错了，不好意思

非常感谢！！！

xiongjerry 发表于 2015-2-28 16:52
第一个warning，可以加入collapse或lag（）来限制工具变量个数。
第二个warning，搞不懂，什么是广义逆矩 ...

你好，能帮我看下下面这个结果有什么问题吗？不胜感激
xtabond2 ik lik lik2 lyk lcfk cfkhhid yrdum*,gmm(lik lik2,lag(1 1) collapse) gmm(lyk lcfk cfkhhid,lag(1        2)        collapse)        iv(yrdum*)        ro
> bust
Favoring speed over space. To switch, type or click on mata: mata set matafavor space, perm.
yrdum1 dropped due to collinearity
yrdum8 dropped due to collinearity
Warning: Two-step estimated covariance matrix of moments is singular.
Using a generalized inverse to calculate robust weighting matrix for Hansen test.
Difference-in-Sargan statistics may be negative.

Dynamic panel-data estimation, one-step system GMM

Group variable: firm                            Number of obs      =      2780
Time variable : year                            Number of groups   =       632
Number of instruments = 20                      Obs per group: min =         1
Wald chi2(11) =    303.87                                      avg =      4.40
Prob > chi2   =     0.000                                      max =         7

Robust
ik       Coef.   Std. Err.      z    P>z     [95% Conf. Interval]

lik    .5760357   .0964315     5.97   0.000     .3870334    .7650379
lik2   -.6061712   .1851619    -3.27   0.001    -.9690818   -.2432605
lyk   -.0143977   .0080182    -1.80   0.073    -.0301132    .0013177
lcfk     .061705    .022245     2.77   0.006     .0181057    .1053043
cfkhhid    .8439348   .2751188     3.07   0.002     .3047119    1.383158
yrdum2    .0732113   .0111842     6.55   0.000     .0512907    .0951319
yrdum3    .0536512   .0093608     5.73   0.000     .0353044    .0719979
yrdum4    .0289772   .0079563     3.64   0.000      .013383    .0445713
yrdum5     .059661    .007043     8.47   0.000     .0458569     .073465
yrdum6    .0464262   .0060819     7.63   0.000     .0345059    .0583465
yrdum7    .0112769   .0051102     2.21   0.027     .0012612    .0212927
_cons    .0176373   .0102865     1.71   0.086    -.0025238    .0377985

Instruments for first differences equation
Standard
D.(yrdum1 yrdum2 yrdum3 yrdum4 yrdum5 yrdum6 yrdum7 yrdum8)
GMM-type (missing=0, separate instruments for each period unless collapsed)
L.(lik lik2) collapsed
L(1/2).(lyk lcfk cfkhhid) collapsed
Instruments for levels equation
Standard
_cons
yrdum1 yrdum2 yrdum3 yrdum4 yrdum5 yrdum6 yrdum7 yrdum8
GMM-type (missing=0, separate instruments for each period unless collapsed)
D.(lik lik2) collapsed
D.(lyk lcfk cfkhhid) collapsed

Arellano-Bond test for AR(1) in first differences: z =  -9.05  Pr > z =  0.000
Arellano-Bond test for AR(2) in first differences: z =   1.29  Pr > z =  0.199

Sargan test of overid. restrictions: chi2(8)    =  44.34  Prob > chi2 =  0.000
(Not robust, but not weakened by many instruments.)
Hansen test of overid. restrictions: chi2(8)    =  10.92  Prob > chi2 =  0.206
(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(3)    =   4.84  Prob > chi2 =  0.184
Difference (null H = exogenous): chi2(5)    =   6.08  Prob > chi2 =  0.299
gmm(lik lik2, collapse lag(1 1))
Hansen test excluding group:     chi2(4)    =   5.83  Prob > chi2 =  0.212
Difference (null H = exogenous): chi2(4)    =   5.09  Prob > chi2 =  0.279
iv(yrdum1 yrdum2 yrdum3 yrdum4 yrdum5 yrdum6 yrdum7 yrdum8)
Hansen test excluding group:     chi2(2)    =   4.32  Prob > chi2 =  0.115
Difference (null H = exogenous): chi2(6)    =   6.60  Prob > chi2 =  0.360

你好，能帮我看下下面这个结果有什么问题吗？不胜感激
xtabond2 ik lik lik2 lyk lcfk cfkhhid yrdum*,gmm(lik lik2,lag(1 1) collapse) gmm(lyk lcfk cfkhhid,lag(1        2)        collapse)        iv(yrdum*)        ro
> bust
Favoring speed over space. To switch, type or click on mata: mata set matafavor space, perm.
yrdum1 dropped due to collinearity
yrdum8 dropped due to collinearity
Warning: Two-step estimated covariance matrix of moments is singular.
Using a generalized inverse to calculate robust weighting matrix for Hansen test.
Difference-in-Sargan statistics may be negative.

Dynamic panel-data estimation, one-step system GMM

Group variable: firm                            Number of obs      =      2780
Time variable : year                            Number of groups   =       632
Number of instruments = 20                      Obs per group: min =         1
Wald chi2(11) =    303.87                                      avg =      4.40
Prob > chi2   =     0.000                                      max =         7

Robust
ik       Coef.   Std. Err.      z    P>z     [95% Conf. Interval]

lik    .5760357   .0964315     5.97   0.000     .3870334    .7650379
lik2   -.6061712   .1851619    -3.27   0.001    -.9690818   -.2432605
lyk   -.0143977   .0080182    -1.80   0.073    -.0301132    .0013177
lcfk     .061705    .022245     2.77   0.006     .0181057    .1053043
cfkhhid    .8439348   .2751188     3.07   0.002     .3047119    1.383158
yrdum2    .0732113   .0111842     6.55   0.000     .0512907    .0951319
yrdum3    .0536512   .0093608     5.73   0.000     .0353044    .0719979
yrdum4    .0289772   .0079563     3.64   0.000      .013383    .0445713
yrdum5     .059661    .007043     8.47   0.000     .0458569     .073465
yrdum6    .0464262   .0060819     7.63   0.000     .0345059    .0583465
yrdum7    .0112769   .0051102     2.21   0.027     .0012612    .0212927
_cons    .0176373   .0102865     1.71   0.086    -.0025238    .0377985

Instruments for first differences equation
Standard
D.(yrdum1 yrdum2 yrdum3 yrdum4 yrdum5 yrdum6 yrdum7 yrdum8)
GMM-type (missing=0, separate instruments for each period unless collapsed)
L.(lik lik2) collapsed
L(1/2).(lyk lcfk cfkhhid) collapsed
Instruments for levels equation
Standard
_cons
yrdum1 yrdum2 yrdum3 yrdum4 yrdum5 yrdum6 yrdum7 yrdum8
GMM-type (missing=0, separate instruments for each period unless collapsed)
D.(lik lik2) collapsed
D.(lyk lcfk cfkhhid) collapsed

Arellano-Bond test for AR(1) in first differences: z =  -9.05  Pr > z =  0.000
Arellano-Bond test for AR(2) in first differences: z =   1.29  Pr > z =  0.199

Sargan test of overid. restrictions: chi2(8)    =  44.34  Prob > chi2 =  0.000
(Not robust, but not weakened by many instruments.)
Hansen test of overid. restrictions: chi2(8)    =  10.92  Prob > chi2 =  0.206
(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(3)    =   4.84  Prob > chi2 =  0.184
Difference (null H = exogenous): chi2(5)    =   6.08  Prob > chi2 =  0.299
gmm(lik lik2, collapse lag(1 1))
Hansen test excluding group:     chi2(4)    =   5.83  Prob > chi2 =  0.212
Difference (null H = exogenous): chi2(4)    =   5.09  Prob > chi2 =  0.279
iv(yrdum1 yrdum2 yrdum3 yrdum4 yrdum5 yrdum6 yrdum7 yrdum8)
Hansen test excluding group:     chi2(2)    =   4.32  Prob > chi2 =  0.115
Difference (null H = exogenous): chi2(6)    =   6.60  Prob > chi2 =  0.360

L火中取栗 发表于 2015-4-17 11:38
你好，能帮我看下下面这个结果有什么问题吗？不胜感激
xtabond2 ik lik lik2 lyk lcfk cfkhhid yrdum*,g ...

sargan没通过，hansen可以。你看是否存在异方差。

xiongjerry 发表于 2015-4-17 13:30
sargan没通过，hansen可以。你看是否存在异方差。

恩，可是我不管怎么调整工具变量个数，总得不到好的结果，看有的外文文章里只采用了Hansen检验的结果，我在想能不能只看Hansen的，会不会不妥？

L火中取栗 发表于 2015-4-17 14:41
恩，可是我不管怎么调整工具变量个数，总得不到好的结果，看有的外文文章里只采用了Hansen检验的结果，我 ...

这也是苦恼我之处，大部分文章是采用sargan检验的，但是我觉得如果存在异方差，用hansen是合适的。

我做的结果也和楼主的差不多，求问该怎么解决？