[Aya Elewa]

Dear Sebastian,

I have another concern. I am wondering whether the order of the variables between brackets affect the results.

xtabond2 tfp_mgtrendrobust lagTFP lropen_constant  lfdi seconary_school_gross year* ,gmmstyle(lropen_constant  lfdi ,lag(2 2) equation(level) collapse ) gmmstyle(lropen_constant  lfdi,lag(2 2) equation(diff) collapse ) gmmstyle(tfp_mgtrendrobust, lag(1 1) eq(diff)) gmm(tfp_mgtrendrobust, lag(1 1) eq(level)) iv( year* seconary_school_gross , eq (level)) iv(seconary_school_gross, eq (diff))robust small two

xtabond2 tfp_mgtrendrobust lagTFP lropen_constant  lfdi seconary_school_gross year*  ,gmmstyle(lropen_constant  lfdi ,lag(2 2) equation(level) collapse ) gmmstyle(lropen_constant  lfdi,lag(2 2) equation(diff) collapse ) gmmstyle(tfp_mgtrendrobust, lag(1 1) eq(diff)) gmm(tfp_mgtrendrobust, lag(1 1) eq(level)) iv(seconary_school_gross year*  , eq (level)) iv(seconary_school_gross, eq (diff))robust small two

I have used these two commands. I have got different results. The only difference is the order of variables between

iv(seconary_school_gross year*  , eq (level))

iv( year* seconary_school_gross , eq (level))

Do you have an explanation?

Many thanks,
Aya

[Sebastian Kripfganz]

1. If lagTFP yields different results compared to L.tfpmgtrendrobust, then I would suspect that you did not create the variable lagTFP properly. There should not be any difference.
2. The statement "two-period lagged levels and first differences of the explanatory variables (FDI and openness) and one-period lagged TFP" does not make clear whether these are instruments for the first-differenced or the level model. One-period lagged TFP is not a valid instrument. In the first-differenced model, you need to use at least the second lag of the dependent variable. Also, I advocate against using the gmmstyle() option without the equation() suboption because it is easy to be confused about which instruments it actually creates for which model equation.
3. Changing the order of the variables in your iv() option should not affect the coefficients of any other variable. If the coefficients of the variables in the iv() option change, then this is due to a multicolinearity problem. Changing the order of the coefficients leads to different year dummies to be dropped, which possibly changes the base year. I would again recommend to use the xtdpdgmm command instead. It has a teffects option that automatically only includes the nonredundant year dummies.

[Annur Wijayakusuma]

Hi Sebastian,

Could you please help me with my coding in Xtabond2 as well?
I am researching the effect of R&D expenditure on firm performance at the firm level. My code is as below

I use "collapse" option

xtabond2 ROATotAsset L.RDI L.ROATotAsset Leverage FirmsSIZE Y2006-Y2017, gmm(RDI, lag(2 4) collapse) gmm(ROATotAsset, lag(1 1)) iv( Leverage FirmsSIZE) twostep robust noconstant

Below is the result

 xtabond2 ROATotAsset L.RDI L.ROATotAsset Leverage FirmsSIZE Y2006-Y2017, gmm(RDI, lag(2 4) collapse) gmm(ROATotAsset, lag(1 1)) 
> iv( Leverage FirmsSIZE) twostep robust noconstant
Favoring space over speed. To switch, type or click on mata: mata set matafavor speed, 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: ASX_ID                          Number of obs      =      2778
Time variable : Year                            Number of groups   =       441
Number of instruments = 27                      Obs per group: min =         1
Wald chi2(16) =     85.86                                      avg =      6.30
Prob > chi2   =     0.000                                      max =        11
------------------------------------------------------------------------------
             |              Corrected
 ROATotAsset |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
         RDI |
         L1. |     .00007   .0000939     0.75   0.456     -.000114     .000254
             |
 ROATotAsset |
         L1. |  -.2104898   .2646949    -0.80   0.426    -.7292823    .3083027
             |
    Leverage |   .1261284   .2067197     0.61   0.542    -.2790348    .5312916
   FirmsSIZE |   3.481166   .9932038     3.50   0.000     1.534522     5.42781
       Y2006 |          0  (omitted)
       Y2007 |  -26.22326   7.115548    -3.69   0.000    -40.16948   -12.27704
       Y2008 |  -26.14865   7.198401    -3.63   0.000    -40.25726   -12.04005
       Y2009 |  -24.79157     6.7919    -3.65   0.000    -38.10345   -11.47969
       Y2010 |   -25.3663   7.153019    -3.55   0.000    -39.38596   -11.34664
       Y2011 |   -25.9895   7.452203    -3.49   0.000    -40.59555   -11.38345
       Y2012 |  -26.10406   7.440849    -3.51   0.000    -40.68786   -11.52026
       Y2013 |  -25.68699   7.298055    -3.52   0.000    -39.99091   -11.38306
       Y2014 |  -25.51451   7.668685    -3.33   0.001    -40.54485   -10.48416
       Y2015 |  -25.96863   7.415399    -3.50   0.000    -40.50255   -11.43471
       Y2016 |  -23.93911   6.994188    -3.42   0.001    -37.64747   -10.23076
       Y2017 |  -24.88442   7.372853    -3.38   0.001    -39.33495   -10.43389
------------------------------------------------------------------------------
Instruments for first differences equation
  Standard
    D.(Leverage FirmsSIZE)
  GMM-type (missing=0, separate instruments for each period unless collapsed)
    L.ROATotAsset
    L(2/4).RDI collapsed
Instruments for levels equation
  Standard
    Leverage FirmsSIZE
  GMM-type (missing=0, separate instruments for each period unless collapsed)
    D.ROATotAsset
    DL.RDI collapsed
------------------------------------------------------------------------------
Arellano-Bond test for AR(1) in first differences: z =  -1.38  Pr > z =  0.167
Arellano-Bond test for AR(2) in first differences: z =  -0.87  Pr > z =  0.382
------------------------------------------------------------------------------
Sargan test of overid. restrictions: chi2(11)   = 304.06  Prob > chi2 =  0.000
  (Not robust, but not weakened by many instruments.)
Hansen test of overid. restrictions: chi2(11)   =   9.13  Prob > chi2 =  0.609
  (Robust, but weakened by many instruments.)

Difference-in-Hansen tests of exogeneity of instrument subsets:
  gmm(RDI, collapse lag(2 4))
    Hansen test excluding group:     chi2(7)    =   8.05  Prob > chi2 =  0.328
    Difference (null H = exogenous): chi2(4)    =   1.09  Prob > chi2 =  0.897
  iv(Leverage FirmsSIZE)
    Hansen test excluding group:     chi2(9)    =   9.02  Prob > chi2 =  0.435
    Difference (null H = exogenous): chi2(2)    =   0.11  Prob > chi2 =  0.944

Without "Collapse" option

xtabond2 ROATotAsset L.RDI L.ROATotAsset Leverage FirmsSIZE Y2006-Y2017, gmm(RDI, lag(2 4)) gmm(ROATotAsset, lag(1 1)) iv( Leverage FirmsSIZE) twostep robust noconstant

the result as follow

 xtabond2 ROATotAsset L.RDI L.ROATotAsset Leverage FirmsSIZE Y2006-Y2017, gmm(RDI, lag(2 4)) gmm(ROATotAsset, lag(1 1)) iv( Lever
> age FirmsSIZE) twostep robust noconstant
Favoring space over speed. To switch, type or click on mata: mata set matafavor speed, 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: ASX_ID                          Number of obs      =      2778
Time variable : Year                            Number of groups   =       441
Number of instruments = 60                      Obs per group: min =         1
Wald chi2(16) =     87.01                                      avg =      6.30
Prob > chi2   =     0.000                                      max =        11
------------------------------------------------------------------------------
             |              Corrected
 ROATotAsset |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
         RDI |
         L1. |   .0000526   .0000513     1.03   0.305    -.0000479    .0001531
             |
 ROATotAsset |
         L1. |  -.2194823   .1209272    -1.81   0.070    -.4564952    .0175307
             |
    Leverage |  -.0055094   .1438171    -0.04   0.969    -.2873856    .2763668
   FirmsSIZE |   2.842752   .7071274     4.02   0.000     1.456808    4.228696
       Y2006 |          0  (omitted)
       Y2007 |  -20.21974   4.994858    -4.05   0.000    -30.00949      -10.43
       Y2008 |  -20.32618   5.074112    -4.01   0.000    -30.27126   -10.38111
       Y2009 |  -20.04415   4.827525    -4.15   0.000    -29.50593   -10.58238
       Y2010 |  -20.63554   5.028991    -4.10   0.000    -30.49218    -10.7789
       Y2011 |   -20.9807   5.189968    -4.04   0.000    -31.15285   -10.80854
       Y2012 |  -20.97483   5.146687    -4.08   0.000    -31.06215   -10.88751
       Y2013 |    -20.564   5.061607    -4.06   0.000    -30.48456   -10.64343
       Y2014 |  -20.45792   5.117973    -4.00   0.000    -30.48896   -10.42688
       Y2015 |  -20.35898   5.072675    -4.01   0.000    -30.30124   -10.41672
       Y2016 |  -20.54744    5.08927    -4.04   0.000    -30.52223   -10.57266
       Y2017 |   -20.8605   5.342503    -3.90   0.000    -31.33161   -10.38938
------------------------------------------------------------------------------
Instruments for first differences equation
  Standard
    D.(Leverage FirmsSIZE)
  GMM-type (missing=0, separate instruments for each period unless collapsed)
    L.ROATotAsset
    L(2/4).RDI
Instruments for levels equation
  Standard
    Leverage FirmsSIZE
  GMM-type (missing=0, separate instruments for each period unless collapsed)
    D.ROATotAsset
    DL.RDI
------------------------------------------------------------------------------
Arellano-Bond test for AR(1) in first differences: z =  -1.46  Pr > z =  0.144
Arellano-Bond test for AR(2) in first differences: z =  -2.24  Pr > z =  0.025
------------------------------------------------------------------------------
Sargan test of overid. restrictions: chi2(44)   =1468.88  Prob > chi2 =  0.000
  (Not robust, but not weakened by many instruments.)
Hansen test of overid. restrictions: chi2(44)   =  36.49  Prob > chi2 =  0.782
  (Robust, but weakened by many instruments.)

Difference-in-Hansen tests of exogeneity of instrument subsets:
  GMM instruments for levels
    Hansen test excluding group:     chi2(23)   =  19.07  Prob > chi2 =  0.697
    Difference (null H = exogenous): chi2(21)   =  17.42  Prob > chi2 =  0.686
  gmm(RDI, lag(2 4))
    Hansen test excluding group:     chi2(7)    =   9.45  Prob > chi2 =  0.222
    Difference (null H = exogenous): chi2(37)   =  27.04  Prob > chi2 =  0.886
  gmm(ROATotAsset, lag(1 1))
    Hansen test excluding group:     chi2(23)   =  12.57  Prob > chi2 =  0.961
    Difference (null H = exogenous): chi2(21)   =  23.92  Prob > chi2 =  0.297
  iv(Leverage FirmsSIZE)
    Hansen test excluding group:     chi2(42)   =  34.22  Prob > chi2 =  0.798
    Difference (null H = exogenous): chi2(2)    =   2.27  Prob > chi2 =  0.322

Please help, whether my coding is correct?

and, how to interpret the AR(2) test? whether >0.05, means that I cannot reject the null hypothesis (there is No autocorrelation )? is it good to have AR(2) >0.05? Here is the result for both code of AR(2) with collapse option, AR(2) is 0.382 while without "collapse" option, it is 0.025.

I look forward to hearing from you. I appreciate your reply

Regards,

Annur

[Sebastian Kripfganz]

It is highly unusual not to include a constant term in the model if you run a system GMM estimation. You should not include the Y2006 dummy. If there are "omitted" coefficients in the regression output, xtabond2 provides incorrect Sargan/Hansen overidentification tests. I recommend to use my xtdpdgmm command instead to avoid this problem.

• Kripfganz, S. (2019). Generalized method of moments estimation of linear dynamic panel data models. Proceedings of the 2019 London Stata Conference.
• XTDPDGMM: new Stata command for GMM estimation of linear (dynamic) panel data models

Ideally, you want to reject the AR(1) test but you do not want to reject the AR(2) test.

The option gmm(ROATotAsset, lag(1 1)) produces invalid instruments! The first lag of the dependent variable is endogenous in the first-differenced model, and the contemporaneous change in the dependent variable is endogenous in the level model. I recommend to explicitly specify the equation() suboption to make sure that you understand which instruments are created by the respective option.

[Annur Wijayakusuma]

Hi Sebastian,

Thank you for your reply. Here I tried to run based on the comment above.
I took off "collapse" option and put equation (diff) suboption on gmm

xtabond2 ROATotAsset L.RDI L.ROATotAsset Leverage FirmsSIZE Y2007-Y2017, gmm(RDI, lag(2 4)) gmm(ROATotAsset, lag(1 1) equation(diff)) iv( Leverage FirmsSIZE) twostep robust

the result:

. xtabond2 ROATotAsset L.RDI L.ROATotAsset Leverage FirmsSIZE Y2007-Y2017, gmm(RDI, lag(2 4)) gmm(ROATotAsset, lag(1 1) equation(d
> iff)) iv( Leverage FirmsSIZE) twostep robust
Favoring space over speed. To switch, type or click on mata: mata set matafavor speed, perm.
Y2015 dropped due to collinearity
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: ASX_ID                          Number of obs      =      2778
Time variable : Year                            Number of groups   =       441
Number of instruments = 50                      Obs per group: min =         1
Wald chi2(14) =    337.53                                      avg =      6.30
Prob > chi2   =     0.000                                      max =        11
------------------------------------------------------------------------------
             |              Corrected
 ROATotAsset |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
         RDI |
         L1. |  -8.36e-06   .0000358    -0.23   0.815    -.0000786    .0000618
             |
 ROATotAsset |
         L1. |  -.7483638   .0708283   -10.57   0.000    -.8871847   -.6095429
             |
    Leverage |  -.3509785   .1351487    -2.60   0.009    -.6158651    -.086092
   FirmsSIZE |   .3596051   .0534211     6.73   0.000     .2549016    .4643086
       Y2007 |  -.1472801   .2917553    -0.50   0.614    -.7191099    .4245497
       Y2008 |  -.0589188   .3205547    -0.18   0.854    -.6871945    .5693569
       Y2009 |   .0314549   .2808925     0.11   0.911    -.5190843     .581994
       Y2010 |   .0480584   .2287607     0.21   0.834    -.4003044    .4964212
       Y2011 |   .0859418    .208583     0.41   0.680    -.3228733    .4947569
       Y2012 |    .060937   .2016827     0.30   0.763    -.3343537    .4562278
       Y2013 |   .0526682   .2065554     0.25   0.799     -.352173    .4575094
       Y2014 |   .0984129   .1049295     0.94   0.348     -.107245    .3040709
       Y2016 |  -.2006608   .1825984    -1.10   0.272    -.5585471    .1572255
       Y2017 |  -.1414778   .2121663    -0.67   0.505    -.5573162    .2743606
       _cons |  -2.751541   .4013387    -6.86   0.000     -3.53815   -1.964931
------------------------------------------------------------------------------
Instruments for first differences equation
  Standard
    D.(Leverage FirmsSIZE)
  GMM-type (missing=0, separate instruments for each period unless collapsed)
    L.ROATotAsset
    L(2/4).RDI
Instruments for levels equation
  Standard
    Leverage FirmsSIZE
    _cons
  GMM-type (missing=0, separate instruments for each period unless collapsed)
    DL.RDI
------------------------------------------------------------------------------
Arellano-Bond test for AR(1) in first differences: z =   0.15  Pr > z =  0.880
Arellano-Bond test for AR(2) in first differences: z =  -2.33  Pr > z =  0.020
------------------------------------------------------------------------------
Sargan test of overid. restrictions: chi2(35)   =  90.24  Prob > chi2 =  0.000
  (Not robust, but not weakened by many instruments.)
Hansen test of overid. restrictions: chi2(35)   =  34.69  Prob > chi2 =  0.483
  (Robust, but weakened by many instruments.)

Difference-in-Hansen tests of exogeneity of instrument subsets:
  GMM instruments for levels
    Hansen test excluding group:     chi2(25)   =  26.50  Prob > chi2 =  0.381
    Difference (null H = exogenous): chi2(10)   =   8.19  Prob > chi2 =  0.610
  gmm(ROATotAsset, eq(diff) lag(1 1))
    Hansen test excluding group:     chi2(25)   =  20.04  Prob > chi2 =  0.745
    Difference (null H = exogenous): chi2(10)   =  14.65  Prob > chi2 =  0.145
  iv(Leverage FirmsSIZE)
    Hansen test excluding group:     chi2(33)   =  32.79  Prob > chi2 =  0.477
    Difference (null H = exogenous): chi2(2)    =   1.90  Prob > chi2 =  0.387

with equation(level) suboption

xtabond2 ROATotAsset L.RDI L.ROATotAsset Leverage FirmsSIZE Y2007-Y2017, gmm(RDI, lag(2 4)) gmm(ROATotAsset, lag(1 1) equation(level)) iv( Leverage FirmsSIZE) twostep robust

the result

xtabond2 ROATotAsset L.RDI L.ROATotAsset Leverage FirmsSIZE Y2007-Y2017, gmm(RDI, lag(2 4)) gmm(ROATotAsset, lag(1 1) equation(l
> evel)) iv( Leverage FirmsSIZE) twostep robust
Favoring space over speed. To switch, type or click on mata: mata set matafavor speed, perm.
Y2015 dropped due to collinearity
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: ASX_ID                          Number of obs      =      2778
Time variable : Year                            Number of groups   =       441
Number of instruments = 50                      Obs per group: min =         1
Wald chi2(14) =     57.10                                      avg =      6.30
Prob > chi2   =     0.000                                      max =        11
------------------------------------------------------------------------------
             |              Corrected
 ROATotAsset |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
         RDI |
         L1. |   7.41e-06   .0000231     0.32   0.749    -.0000379    .0000528
             |
 ROATotAsset |
         L1. |   .1573693   .1235528     1.27   0.203    -.0847897    .3995284
             |
    Leverage |  -.3018306   .1465762    -2.06   0.039    -.5891148   -.0145465
   FirmsSIZE |   .3217667   .0776132     4.15   0.000     .1696477    .4738858
       Y2007 |   .1670106   .3293655     0.51   0.612     -.478534    .8125552
       Y2008 |   .0139232   .2657302     0.05   0.958    -.5068983    .5347448
       Y2009 |   .4400586   .2408315     1.83   0.068    -.0319624    .9120796
       Y2010 |   .2485683   .1904438     1.31   0.192    -.1246947    .6218314
       Y2011 |   .0905946   .1272116     0.71   0.476    -.1587356    .3399248
       Y2012 |   .0791637   .1023267     0.77   0.439    -.1213929    .2797203
       Y2013 |   .0271137   .1020032     0.27   0.790    -.1728089    .2270364
       Y2014 |   .0232287    .067204     0.35   0.730    -.1084888    .1549461
       Y2016 |  -.1628063   .1910053    -0.85   0.394    -.5371698    .2115571
       Y2017 |  -.0169423   .2093019    -0.08   0.935    -.4271665    .3932819
       _cons |  -2.427127   .5544456    -4.38   0.000     -3.51382   -1.340433
------------------------------------------------------------------------------
Instruments for first differences equation
  Standard
    D.(Leverage FirmsSIZE)
  GMM-type (missing=0, separate instruments for each period unless collapsed)
    L(2/4).RDI
Instruments for levels equation
  Standard
    Leverage FirmsSIZE
    _cons
  GMM-type (missing=0, separate instruments for each period unless collapsed)
    DL.ROATotAsset
    DL.RDI
------------------------------------------------------------------------------
Arellano-Bond test for AR(1) in first differences: z =  -2.25  Pr > z =  0.025
Arellano-Bond test for AR(2) in first differences: z =  -0.71  Pr > z =  0.479
------------------------------------------------------------------------------
Sargan test of overid. restrictions: chi2(35)   = 118.57  Prob > chi2 =  0.000
  (Not robust, but not weakened by many instruments.)
Hansen test of overid. restrictions: chi2(35)   =  37.98  Prob > chi2 =  0.335
  (Robust, but weakened by many instruments.)

Difference-in-Hansen tests of exogeneity of instrument subsets:
  GMM instruments for levels
    Hansen test excluding group:     chi2(15)   =   7.89  Prob > chi2 =  0.928
    Difference (null H = exogenous): chi2(20)   =  30.08  Prob > chi2 =  0.069
  gmm(ROATotAsset, eq(level) lag(1 1))
    Hansen test excluding group:     chi2(25)   =  22.51  Prob > chi2 =  0.606
    Difference (null H = exogenous): chi2(10)   =  15.46  Prob > chi2 =  0.116
  iv(Leverage FirmsSIZE)
    Hansen test excluding group:     chi2(33)   =  31.85  Prob > chi2 =  0.524
    Difference (null H = exogenous): chi2(2)    =   6.12  Prob > chi2 =  0.047


.

I will try to run using XTDPDGMM and compare the result to Xtabond2, but at this time, I need your advice whether the above result is still wrong?

Appreciate your reply

Regards,

Annur

[Sebastian Kripfganz]

Still, the first lag of ROATotAsset is not a valid instrument in the first-differenced model because it is correlated with the first-differenced error term by construction. You might specify something like gmm(ROATotAsset, lag(2 2) equation(diff)) gmm(ROATotAsset, lag(1 1) equation(level)). Similarly, I recommend to be explicit about the equation when specifying the iv() option.

[Annur Wijayakusuma]

Hi Sebastian,

Thank you for your reply.
Here is the result for putting equation ( ) suboption in iv

xtabond2 ROATotAsset L.RDI L.ROATotAsset Leverage FirmsSIZE Y2007-Y2017, gmm(RDI, lag(2 4) collapse) gmm(ROATotAsset, lag(2 2) equation(diff)) gmm(ROATotAsset, lag(1 1) equation(level)) iv( Leverage, equation(diff)) iv(FirmsSIZE, equation(diff)) twostep robust

the result:

. xtabond2 ROATotAsset L.RDI L.ROATotAsset Leverage FirmsSIZE Y2007-Y2017, gmm(RDI, lag(2 4) collapse) gmm(ROATotAsset, lag(2 2) e
> quation(diff)) gmm(ROATotAsset, lag(1 1) equation(level)) iv( Leverage, equation(diff)) iv(FirmsSIZE, equation(diff)) twostep ro
> bust
Favoring space over speed. To switch, type or click on mata: mata set matafavor speed, perm.
Y2015 dropped due to collinearity

Dynamic panel-data estimation, two-step system GMM
------------------------------------------------------------------------------
Group variable: ASX_ID                          Number of obs      =      2778
Time variable : Year                            Number of groups   =       441
Number of instruments = 27                      Obs per group: min =         1
Wald chi2(14) =     47.67                                      avg =      6.30
Prob > chi2   =     0.000                                      max =        11
------------------------------------------------------------------------------
             |              Corrected
 ROATotAsset |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
         RDI |
         L1. |   .0000108   .0000241     0.45   0.654    -.0000364     .000058
             |
 ROATotAsset |
         L1. |   .0776424   .0876605     0.89   0.376     -.094169    .2494537
             |
    Leverage |    -.25188   .2217847    -1.14   0.256      -.68657    .1828099
   FirmsSIZE |   .8832735   .2868349     3.08   0.002     .3210874     1.44546
       Y2007 |   .3852702    1.21039     0.32   0.750     -1.98705    2.757591
       Y2008 |   .4201616   .4096403     1.03   0.305    -.3827186    1.223042
       Y2009 |   .2280198   .5453144     0.42   0.676    -.8407768    1.296816
       Y2010 |   .0226068   .2729474     0.08   0.934    -.5123603     .557574
       Y2011 |   .0023614   .2619776     0.01   0.993    -.5111052     .515828
       Y2012 |  -.1043462   .2335706    -0.45   0.655    -.5621361    .3534437
       Y2013 |   .0298407   .2196026     0.14   0.892    -.4005724    .4602539
       Y2014 |   .0978155   .2616853     0.37   0.709    -.4150782    .6107093
       Y2016 |   -.040959   .5536001    -0.07   0.941    -1.125995    1.044077
       Y2017 |  -.0257486   .4690492    -0.05   0.956    -.9450681    .8935709
       _cons |  -6.444881   2.070032    -3.11   0.002    -10.50207   -2.387693
------------------------------------------------------------------------------
Instruments for first differences equation
  Standard
    D.FirmsSIZE
    D.Leverage
  GMM-type (missing=0, separate instruments for each period unless collapsed)
    L2.ROATotAsset
    L(2/4).RDI collapsed
Instruments for levels equation
  Standard
    _cons
  GMM-type (missing=0, separate instruments for each period unless collapsed)
    DL.ROATotAsset
    DL.RDI collapsed
------------------------------------------------------------------------------
Arellano-Bond test for AR(1) in first differences: z =  -2.26  Pr > z =  0.024
Arellano-Bond test for AR(2) in first differences: z =  -1.15  Pr > z =  0.250
------------------------------------------------------------------------------
Sargan test of overid. restrictions: chi2(12)   =  86.77  Prob > chi2 =  0.000
  (Not robust, but not weakened by many instruments.)
Hansen test of overid. restrictions: chi2(12)   =  10.79  Prob > chi2 =  0.547
  (Robust, but weakened by many instruments.)

Difference-in-Hansen tests of exogeneity of instrument subsets:
  GMM instruments for levels
    Hansen test excluding group:     chi2(1)    =   0.09  Prob > chi2 =  0.770
    Difference (null H = exogenous): chi2(11)   =  10.70  Prob > chi2 =  0.469
  gmm(RDI, collapse lag(2 4))
    Hansen test excluding group:     chi2(8)    =   7.46  Prob > chi2 =  0.488
    Difference (null H = exogenous): chi2(4)    =   3.33  Prob > chi2 =  0.505
  gmm(ROATotAsset, eq(diff) lag(2 2))
    Hansen test excluding group:     chi2(2)    =   1.62  Prob > chi2 =  0.445
    Difference (null H = exogenous): chi2(10)   =   9.17  Prob > chi2 =  0.516
  gmm(ROATotAsset, eq(level) lag(1 1))
    Hansen test excluding group:     chi2(2)    =   1.58  Prob > chi2 =  0.453
    Difference (null H = exogenous): chi2(10)   =   9.21  Prob > chi2 =  0.513
  iv(Leverage, eq(diff))
    Hansen test excluding group:     chi2(11)   =   8.24  Prob > chi2 =  0.691
    Difference (null H = exogenous): chi2(1)    =   2.55  Prob > chi2 =  0.111
  iv(FirmsSIZE, eq(diff))
    Hansen test excluding group:     chi2(11)   =   9.36  Prob > chi2 =  0.588
    Difference (null H = exogenous): chi2(1)    =   1.42  Prob > chi2 =  0.233

I am still confused about the equation ( ) suboption in iv ( ). If I do not use the lagged variable as an exogenous variable, do I still need to use equation ( ) suboption?
Appreciate your help

Regards, Annur

[Sebastian Kripfganz]

iv(Leverage) without the equation() suboption is not the same as iv(Leverage, equation(diff)). It is also not the same as specifying both iv(Leverage, equation(diff)) iv(Leverage, equation(level)) jointly. It depends on you what you would like to achieve. If you do not know what iv(Leverage) without the equation() suboption does, do not use it! Technical details can be found in David Roodman's "how to do xtabond2" paper.

[OWUSU ANSAH]

Dear Statalisters, this is my code and and results. Is the specification of the code correct?
Secondly, in the gmm() do you use the lag of the dependent variable or you use the dependent variable?
Code:xtabond2 gdpg gdppc_lag rir hc pop fdi afdi trade inf afdi_fdi i.yearavg5 , gmm(gdpg afdi fdi,laglimits(2 2)eq(level)collapse) gmm(gdpg fdi, laglimits(1 1)eq(diff) collapse) iv(hc i.yearavg5 trade inf rir pop, eq(level)) nodiffsargan robust orthogonal small

Results
xtabond2 gdpg gdppc_lag rir hc pop fdi afdi trade inf afdi_fdi i.yearavg5 , gmm(gdpg afdi
> fdi,laglimits(2 2)eq(level)collapse) gmm(gdpg fdi, laglimits(1 1)eq(diff) collapse) iv(h
> c i.yearavg5 trade inf rir pop, eq(level)) nodiffsargan robust orthogonal small
Favoring space over speed. To switch, type or click on mata: mata set matafavor speed, perm
> .
Warning: Two-step estimated covariance matrix of moments is singular.
Using a generalized inverse to calculate robust weighting matrix for Hansen test.

Dynamic panel-data estimation, one-step system GMM
------------------------------------------------------------------------------
Group variable: countryID Number of obs = 1395
Time variable : year Number of groups = 39
Number of instruments = 19 Obs per group: min = 9
F(18, 38) = 1.98 avg = 35.77
Prob > F = 0.038 max = 44
------------------------------------------------------------------------------
| Robust
gdpg | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
gdppc_lag | 2.573154 1.328282 1.94 0.060 -.1158126 5.262121
rir | .1522238 .0919411 1.66 0.106 -.0339013 .338349
hc | -1.273458 2.101048 -0.61 0.548 -5.526807 2.979891
pop | .0099557 .0487209 0.20 0.839 -.0886746 .108586
fdi | -2.386882 1.432775 -1.67 0.104 -5.287384 .5136195
afdi | -2.594058 2.765893 -0.94 0.354 -8.193317 3.0052
trade | .0665036 .0612053 1.09 0.284 -.0574 .1904072
inf | .0077725 .0465868 0.17 0.868 -.0865375 .1020825
afdi_fdi | .2083395 .1811483 1.15 0.257 -.1583759 .575055
|
yearavg5 |
1 | 0 (empty)
2 | .8940954 2.17788 0.41 0.684 -3.514792 5.302983
3 | 4.283092 3.251295 1.32 0.196 -2.29881 10.865
4 | 5.4885 2.678922 2.05 0.047 .0653067 10.91169
5 | 4.807664 2.980047 1.61 0.115 -1.225126 10.84045
6 | 4.529856 2.804088 1.62 0.114 -1.146723 10.20643
7 | 6.320317 3.686288 1.71 0.095 -1.142183 13.78282
8 | 5.6117 4.030552 1.39 0.172 -2.547725 13.77113
9 | 11.44193 7.670253 1.49 0.144 -4.085684 26.96955
|
_cons | -2.628024 4.004424 -0.66 0.516 -10.73456 5.478508
------------------------------------------------------------------------------
Instruments for orthogonal deviations equation
GMM-type (missing=0, separate instruments for each period unless collapsed)
L.(gdpg fdi) collapsed
Instruments for levels equation
Standard
hc 1b.yearavg5 2.yearavg5 3.yearavg5 4.yearavg5 5.yearavg5 6.yearavg5
7.yearavg5 8.yearavg5 9.yearavg5 trade inf rir pop
_cons
GMM-type (missing=0, separate instruments for each period unless collapsed)
DL2.(gdpg afdi fdi) collapsed
------------------------------------------------------------------------------
Arellano-Bond test for AR(1) in first differences: z = -1.70 Pr > z = 0.090
Arellano-Bond test for AR(2) in first differences: z = 1.02 Pr > z = 0.308
------------------------------------------------------------------------------
Sargan test of overid. restrictions: chi2(0) = 84.34 Prob > chi2 = .
(Not robust, but not weakened by many instruments.)
Hansen test of overid. restrictions: chi2(0) = 12.50 Prob > chi2 = .
(Robust, but weakened by many instruments.)

[Atiqa Rehman]

Dear all,
Hi,

I am using the following code and my Hansen difference test for dependent variable is significant. following is the code I am using. i am using industry year and country dummies. Following is the code and results. I am new in stata blog. I am trying to paste my result table also but I am not able to paste it here in same format. could you please guide me how can I do that. Further, please see my results below and suggest me how can I make my results better and what could be the problem. I will be very thankful to you for your help.

Depnedent variable=RK
Indep variable= FF CA TB LEV SIZE NDF FF_NDF
I have interaction term in my model. will it be also use as instrument?

[xtabond2 win_RK l.win_RK FF CA TB LEV SIZE NDF FF_NDF yr2004a-yr2014a con1a-con19a sic1a-sic48a, gmm (win_RK, lag(2 2)) gmm (TB LEV NDF FF FF _NDF CA SIZE, lag 1 1)) iv(yr2004a-yr2014a con1a-con19a sic1a-sic48a, equation(level)) twostep robust orthogonal small]

Favoring speed over space. To switch, type or click on mata: mata set matafavor space, per
> m.
Warning: Two-step estimated covariance matrix of moments is singular.
Using a generalized inverse to calculate optimal weighting matrix for two-step estimatio
> n.
Difference-in-Sargan/Hansen statistics may be negative.

Dynamic panel-data estimation, two-step system GMM

Group variable: compid Number of obs = 3937
Time variable : year Number of groups = 491
Number of instruments = 259 Obs per group: min = 1
F(86, 490) = 687.82 avg = 8.02
Prob > F = 0.000 max = 12



Instruments for orthogonal deviations equation
GMM-type (missing=0, separate instruments for each period unless collapsed)
L.(TB LEV NDF FF FF_NDF CA SIZE)
L2.win_RK
Instruments for levels equation
Standard
yr2004a yr2005a yr2006a yr2007a yr2008a yr2009a yr2010a yr2011a yr2012a
yr2013a yr2014a con1a con2a con3a con4a con5a con6a con7a con8a con9a
con10a con11a con12a con13a con14a con15a con16a con17a con18a con19a
sic1a sic2a sic3a sic4a sic5a sic6a sic7a sic8a sic9a sic10a sic11a sic12a
sic13a sic14a sic15a sic16a sic17a sic18a sic19a sic20a sic21a sic22a
sic23a sic24a sic25a sic26a sic27a sic28a sic29a sic30a sic31a sic32a
sic33a sic34a sic35a sic36a sic37a sic38a sic39a sic40a sic41a sic42a
sic43a sic44a sic45a sic46a sic47a sic48a
_cons
GMM-type (missing=0, separate instruments for each period unless collapsed)
D.(TB LEV NDF FF FF_NDF CA SIZE)
DL.win_RK

Arellano-Bond test for AR(1) in first differences: z = -3.68 Pr > z = 0.000
Arellano-Bond test for AR(2) in first differences: z = 0.77 Pr > z = 0.442

Sargan test of overid. restrictions: chi2(172) = 563.69 Prob > chi2 = 0.000
(Not robust, but not weakened by many instruments.)
Hansen test of overid. restrictions: chi2(172) = 194.12 Prob > chi2 = 0.119
(Robust, but weakened by many instruments.)

Difference-in-Hansen tests of exogeneity of instrument subsets:
GMM instruments for levels
Hansen test excluding group: chi2(77) = 92.48 Prob > chi2 = 0.110
Difference (null H = exogenous): chi2(95) = 101.64 Prob > chi2 = 0.302
gmm(win_RK, lag(2 2))
Hansen test excluding group: chi2(150) = 151.71 Prob > chi2 = 0.446
Difference (null H = exogenous): chi2(22) = 42.41 Prob > chi2 = 0.006
gmm(TB LEV NDF FF FF_NDF CA SIZE, lag(1 1))
Hansen test excluding group: chi2(11) = 16.64 Prob > chi2 = 0.119
Difference (null H = exogenous): chi2(161) = 177.49 Prob > chi2 = 0.177
iv(yr2004a yr2005a yr2006a yr2007a yr2008a yr2009a yr2010a yr2011a yr2012a yr2013a yr201
> 4a con1a con2a con3a con4a con5a con6a con7a con8a con9a con10a con11a con12a con13a con
> 14a con15a con16a con17a con18a con19a sic1a sic2a sic3a sic4a sic5a sic6a sic7a sic8a s
> ic9a sic10a sic11a sic12a sic13a sic14a sic15a sic16a sic17a sic18a sic19a sic20a sic21a
> sic22a sic23a sic24a sic25a sic26a sic27a sic28a sic29a sic30a sic31a sic32a sic33a sic
> 34a sic35a sic36a sic37a sic38a sic39a sic40a sic41a sic42a sic43a sic44a sic45a sic46a
> sic47a sic48a, eq(level))
Hansen test excluding group: chi2(97) = 118.54 Prob > chi2 = 0.068
Difference (null H = exogenous): chi2(75) = 75.59 Prob > chi2 = 0.459

[Sebastian Kripfganz]

Please see my 2019 London Stata Conference presentation for examples on the correct specification:

• Kripfganz, S. (2019). Generalized method of moments estimation of linear dynamic panel data models. Proceedings of the 2019 London Stata Conference.

Also note the following two major concerns about your use of xtabond2:

• There is a bug that leads to an incorrect computation of the degrees of freedom for the overidentification tests, and therefore also to incorrect p-values, if there are omitted (or "empty") coeffcients in the regression output. This typically occurs when dummy variables are specified with the factor variable notation.
• There is a bug that leads to incorrect coefficient estimates when forward-orthogonal deviations are used.

Both issues are discussed in the above-mentioned presentation. You might want to consider using my xtdpdgmm command instead of xtabond2 to avoid these problems.

[Annur Wijayakusuma]

Hi Sebastian,

I need your help with my coding. I have a problem with the Hansen test which is not >0.05. Can I still use this result or I need to put another coding?
Here is my coding. I specify the year and excluded the Year 2006 and 2007 because it was omitted or emptied.

xtabond2 ROATotAsset L.RDI L.ROATotAsset Leverage FirmsSIZE Y2008-Y2017, gmm(RDI, lag(2 4)) gmm(ROATotAsset, lag(1 1) equation(level)) iv(Leverage FirmsSIZE, equation(level)) orthogonal small twostep robust

below is the result

xtabond2 ROATotAsset L.RDI L.ROATotAsset Leverage FirmsSIZE Y2008-Y2017, gmm(RDI, lag(2 4)) gmm(ROATotAsset, lag(1 1) equation(l
> evel)) iv(Leverage FirmsSIZE, equation(level)) orthogonal small twostep robust
Favoring space over speed. To switch, type or click on mata: mata set matafavor speed, 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: ASX_ID                          Number of obs      =      5349
Time variable : Year                            Number of groups   =       487
Number of instruments = 50                      Obs per group: min =         7
F(14, 486)    =      3.27                                      avg =     10.98
Prob > F      =     0.000                                      max =        11
------------------------------------------------------------------------------
             |              Corrected
 ROATotAsset |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
         RDI |
         L1. |   .0000209   6.82e-06     3.06   0.002     7.49e-06    .0000343
             |
 ROATotAsset |
         L1. |   .3095717    .123973     2.50   0.013     .0659826    .5531609
             |
    Leverage |  -.0085758   .0039637    -2.16   0.031    -.0163639   -.0007877
   FirmsSIZE |   .0452189   .0199084     2.27   0.024     .0061017     .084336
       Y2008 |  -.3519196   .3249019    -1.08   0.279    -.9903053    .2864662
       Y2009 |  -.2019102   .2167678    -0.93   0.352     -.627828    .2240077
       Y2010 |   .0571501   .3583073     0.16   0.873    -.6468727    .7611728
       Y2011 |  -.2643039   .2082071    -1.27   0.205    -.6734011    .1447933
       Y2012 |  -.2047061   .2294542    -0.89   0.373    -.6555508    .2461387
       Y2013 |  -.1699912   .2505895    -0.68   0.498    -.6623638    .3223815
       Y2014 |  -1.189393   .7346716    -1.62   0.106    -2.632918    .2541314
       Y2015 |  -.7970928   .4759323    -1.67   0.095    -1.732232    .1380462
       Y2016 |  -.1173868   .4155981    -0.28   0.778    -.9339778    .6992042
       Y2017 |   -1.08205   1.027388    -1.05   0.293    -3.100721    .9366209
       _cons |  -.0312545   .1951929    -0.16   0.873    -.4147807    .3522716
------------------------------------------------------------------------------
Instruments for orthogonal deviations equation
  GMM-type (missing=0, separate instruments for each period unless collapsed)
    L(2/4).RDI
Instruments for levels equation
  Standard
    Leverage FirmsSIZE
    _cons
  GMM-type (missing=0, separate instruments for each period unless collapsed)
    DL.ROATotAsset
    DL.RDI
------------------------------------------------------------------------------
Arellano-Bond test for AR(1) in first differences: z =  -2.97  Pr > z =  0.003
Arellano-Bond test for AR(2) in first differences: z =  -1.10  Pr > z =  0.273
------------------------------------------------------------------------------
Sargan test of overid. restrictions: chi2(35)   = 256.24  Prob > chi2 =  0.000
  (Not robust, but not weakened by many instruments.)
Hansen test of overid. restrictions: chi2(35)   =  76.51  Prob > chi2 =  0.000 --> this is my concern
  (Robust, but weakened by many instruments.)

Difference-in-Hansen tests of exogeneity of instrument subsets:
  GMM instruments for levels
    Hansen test excluding group:     chi2(15)   =  25.21  Prob > chi2 =  0.047
    Difference (null H = exogenous): chi2(20)   =  51.30  Prob > chi2 =  0.000
  gmm(ROATotAsset, eq(level) lag(1 1))
    Hansen test excluding group:     chi2(25)   =  64.68  Prob > chi2 =  0.000
    Difference (null H = exogenous): chi2(10)   =  11.83  Prob > chi2 =  0.297
  iv(Leverage FirmsSIZE, eq(level))
    Hansen test excluding group:     chi2(33)   =  69.60  Prob > chi2 =  0.000
    Difference (null H = exogenous): chi2(2)    =   6.91  Prob > chi2 =  0.032

Appreciate your help

[Sebastian Kripfganz]

Please notice that xtabond2 has a bug that may lead to incorrect results when using the orthogonal option. I recommend to instead use my xtdpdgmm command to circumvent this problem. For details, see my 2019 London Stata Conference presentation. Therein, you can also find a section on model selection that may be of help to find a specification that does not reject the Hansen test. Further notice that you have specificed time dummies as regressors but not as instruments. Any exogenous regressor must also be included in the set of instruments. Again, see my presentation slides for a treatment of time dummies.

• Kripfganz, S. (2019). Generalized method of moments estimation of linear dynamic panel data models. Proceedings of the 2019 London Stata Conference.
• XTDPDGMM: new Stata command for GMM estimation of linear (dynamic) panel data models

[Annur Wijayakusuma]

Hi Sebastian,

Thank you for your reply.
I changed the code and put years dummy as an instrument variable. First, I run original code for the whole year dummy (2006-2017). Based on the result shown that year 2006 and 2008 is omitted, thus I took those two years from the code. Please give me your feedback


Here is code:

xtabond2 ROATotAsset L.RDI L.ROATotAsset Leverage FirmsSIZE Y2007 Y2009-Y2017, gmm(RDI, lag(2 4) equation(diff)) gmm(ROATotAsset, lag(1 1) equation(level)) iv(Leverage FirmsSIZE Y2007 Y2009-Y2017, equation(diff)) small twostep robust

result

 xtabond2 ROATotAsset L.RDI L.ROATotAsset Leverage FirmsSIZE Y2007 Y2009-Y2017, gmm(RDI, lag(2 4) equation(diff)) gmm(ROATotAsset
> , lag(1 1) equation(level)) iv(Leverage FirmsSIZE Y2007 Y2009-Y2017, equation(diff)) small twostep robust
Favoring space over speed. To switch, type or click on mata: mata set matafavor speed, perm.

Dynamic panel-data estimation, two-step system GMM
------------------------------------------------------------------------------
Group variable: ASX_ID                          Number of obs      =      5349
Time variable : Year                            Number of groups   =       487
Number of instruments = 50                      Obs per group: min =         7
F(14, 486)    =      5.56                                      avg =     10.98
Prob > F      =     0.000                                      max =        11
------------------------------------------------------------------------------
             |              Corrected
 ROATotAsset |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
         RDI |
         L1. |   .0002171   .0002088     1.04   0.299    -.0001933    .0006274
             |
 ROATotAsset |
         L1. |   .3197645   .1536992     2.08   0.038     .0177677    .6217614
             |
    Leverage |  -.0048092   .0020072    -2.40   0.017     -.008753   -.0008653
   FirmsSIZE |   .0521182   .0334832     1.56   0.120    -.0136715    .1179078
       Y2007 |   .0695942   .0460538     1.51   0.131    -.0208949    .1600833
       Y2009 |   -.082469   .0424372    -1.94   0.053    -.1658521    .0009141
       Y2010 |  -.0127756   .0375747    -0.34   0.734    -.0866044    .0610533
       Y2011 |  -.0086805   .0436806    -0.20   0.843    -.0945067    .0771457
       Y2012 |  -.0520049   .0465991    -1.12   0.265    -.1435653    .0395556
       Y2013 |   -.099023   .0532375    -1.86   0.063    -.2036271    .0055812
       Y2014 |  -.1462322   .0518373    -2.82   0.005    -.2480851   -.0443792
       Y2015 |  -.2712174   .1253459    -2.16   0.031    -.5175041   -.0249306
       Y2016 |  -.2246637   .1189399    -1.89   0.060    -.4583636    .0090362
       Y2017 |  -.1497271   .1495898    -1.00   0.317    -.4436497    .1441955
       _cons |  -.3698572   .1965162    -1.88   0.060    -.7559835    .0162691
------------------------------------------------------------------------------
Instruments for first differences equation
  Standard
    D.(Leverage FirmsSIZE Y2007 Y2009 Y2010 Y2011 Y2012 Y2013 Y2014 Y2015
    Y2016 Y2017)
  GMM-type (missing=0, separate instruments for each period unless collapsed)
    L(2/4).RDI
Instruments for levels equation
  Standard
    _cons
  GMM-type (missing=0, separate instruments for each period unless collapsed)
    DL.ROATotAsset
------------------------------------------------------------------------------
Arellano-Bond test for AR(1) in first differences: z =  -2.64  Pr > z =  0.008
Arellano-Bond test for AR(2) in first differences: z =  -0.43  Pr > z =  0.669
------------------------------------------------------------------------------
Sargan test of overid. restrictions: chi2(35)   = 275.50  Prob > chi2 =  0.000
  (Not robust, but not weakened by many instruments.)
Hansen test of overid. restrictions: chi2(35)   =  43.84  Prob > chi2 =  0.145
  (Robust, but weakened by many instruments.)

Difference-in-Hansen tests of exogeneity of instrument subsets:
  GMM instruments for levels
    Hansen test excluding group:     chi2(25)   =  25.64  Prob > chi2 =  0.427
    Difference (null H = exogenous): chi2(10)   =  18.19  Prob > chi2 =  0.052
  gmm(RDI, eq(diff) lag(2 4))
    Hansen test excluding group:     chi2(8)    =   7.32  Prob > chi2 =  0.502
    Difference (null H = exogenous): chi2(27)   =  36.51  Prob > chi2 =  0.104
  gmm(ROATotAsset, eq(level) lag(1 1))
    Hansen test excluding group:     chi2(25)   =  25.64  Prob > chi2 =  0.427
    Difference (null H = exogenous): chi2(10)   =  18.19  Prob > chi2 =  0.052
  iv(Leverage FirmsSIZE Y2007 Y2009 Y2010 Y2011 Y2012 Y2013 Y2014 Y2015 Y2016 Y2017, eq(diff))
    Hansen test excluding group:     chi2(23)   =  21.89  Prob > chi2 =  0.527
    Difference (null H = exogenous): chi2(12)   =  21.95  Prob > chi2 =  0.038


.

Appreciate your help

[Sebastian Kripfganz]

The Hansen test still does not appear entirely satisfying. There could be different reasons, e.g. that Leverage and FirmsSIZE should not be treated as strictly exogenous or that the mean stationarity assumption is not satisfied for the eq(level) instruments.