Dear list:

Probably this message shoud be better addressed to a Stata list than to this one. However, I usually use Gretl for my estimations and probably somebody in the list has had the same issue and could be of interest of other Gretl users.

I know that estimations using different statistical packages usually should lead to different results due to different setting defaults, algorithms used, etc.

Recently, I have been introduced in dynamic panel data model. As usual, I first tried Gretl to do my estimations.

The question is that I'm trying to compare the results in gretl and Stata for a dynamic panel data model on the same dataset using the sys-gmm estimator of Blundell and Bond (1998). 

I have tried to do it in Gretl using the gretl GUI with 1 step estimator, no time dummies and including the equation in levels as follows (command log):

dpanel 1 ; Afiliados 0 Coast Aerop70 Walk Pmain Dens DPrs11_2 --system \
  --dpdstyle

where most of the explanatory  variables Coast, Aerop70, Walk, Pmain and Dprs11_2 are time invariant.

For this setting I get the following coefficients:

Model 2: 1-step dynamic panel, using 680 observations
Included 136 cross-sectional units
Including equations in levels
H-matrix as per Ox/DPD
Dependent variable: Afiliados

                  coefficient   std. error       z       p-value
  --------------------------------------------------------------
  Afiliados(-1)     1.04336      0.00244793   426.2      0.0000  ***
  const           100.007       60.8670         1.643    0.1004
  Coast             1.13130      0.554658       2.040    0.0414  **
  Aerop70          16.3194      21.2904         0.7665   0.4434
  Walk             −1.73009      6.48870       −0.2666   0.7898
  Pmain            −2.92214      1.00457       −2.909    0.0036  ***
  Dens             −0.0164583    0.0129767     −1.268    0.2047
  DPrs11_2        −64.0195      31.7635        −2.016    0.0439  **

Sum squared resid    57679996   S.E. of regression   292.9733

Number of instruments = 21
Test for AR(1) errors: z = -3.63181 [0.0003]
Test for AR(2) errors: z = 0.0197901 [0.9842]
Sargan over-identification test: Chi-square(13) = 163.021 [0.0000]
Wald (joint) test: Chi-square(7) = 414893 [0.0000]

When I try to estimate the model in Stata using the following command:

. xtdpdsys Afiliados Coast Aerop70 Walk Pmain Dens residential, lags(1) artests(2)

I get the following result:

note: Coast dropped from div() because of collinearity
note: Aerop70 dropped from div() because of collinearity
note: Walk dropped from div() because of collinearity
note: Pmain dropped from div() because of collinearity
note: residential dropped from div() because of collinearity

System dynamic panel-data estimation            Number of obs     =        680
Group variable: Municipio                       Number of groups  =        136
Time variable: Year
                                                Obs per group:
                                                              min =          5
                                                              avg =          5
                                                              max =          5

Number of instruments =     16                  Wald chi2(6)      =  708786.67
                                                Prob > chi2       =     0.0000
One-step results
------------------------------------------------------------------------------
   Afiliados |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
   Afiliados |
         L1. |   .9912949   .0126633    78.28   0.000     .9664753    1.016114
             |
       Coast |   188.1017   38.27705     4.91   0.000       113.08    263.1233
     Aerop70 |          0  (omitted)
        Walk |  -503.8206   395.3154    -1.27   0.202    -1278.624    270.9833
       Pmain |   293.8372   82.27726     3.57   0.000     132.5768    455.0977
        Dens |  -.8086489   .5405354    -1.50   0.135    -1.868079     .250781
 DPrs11_2l |   4815.363   1926.302     2.50   0.012      1039.88    8590.846
       _cons |  -16164.83   3962.049    -4.08   0.000    -23930.31   -8399.358
------------------------------------------------------------------------------
Instruments for differenced equation
        GMM-type: L(2/.).Afiliados
        Standard: D.Dens
Instruments for level equation
        GMM-type: LD.Afiliados
        Standard: _cons

As it can be seen, the number of observations and entities are the same. However, the number of instruments is different relating to the Gretl output (probably some of the defaults have different behaviour) but, the magnitude of coefficients is very different, and what to me is most worrying is that in Gretl there is a coefficient for the variable "Aerop70", however Stata drops this variable probably due to the collinearity issues with no coefficient estimated.

Somebody has a clue on how to replicate the Gretl setting for this model in Stata? It is in the xtdpdsys command where it should be replicated or in the other command (xtabond2) that I have seen that somebody uses to estimate this kind of models.

Thanks in advance and sorry for inconvenience.
José Perles
University of Alicante
Spain.