Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down at the end of May, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: GMM error (bug in Stata?)


From   Maarten Buis <maartenlbuis@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: GMM error (bug in Stata?)
Date   Sun, 30 Oct 2011 17:18:39 +0100

1.7e-33 is 0.00000000000000000000000000000000017, so practically 0, so
I would answer yes rather than no.

On Sun, Oct 30, 2011 at 4:51 PM, John Antonakis <John.Antonakis@unil.ch> wrote:
> Hi:
>
> No; the output I gave was from when I don't stack. Here it is again:
>
> Step 1
> Iteration 0:   GMM criterion Q(b) =  12.045213
> Iteration 1:   GMM criterion Q(b) =  2.209e-26
> Iteration 2:   GMM criterion Q(b) =  1.786e-32
>
> Step 2
> Iteration 0:   GMM criterion Q(b) =  1.883e-33
> Iteration 1:   GMM criterion Q(b) =  1.656e-33
>
> GMM estimation
>
> Number of parameters =  14
> Number of moments    =  14
> Initial weight matrix: Unadjusted                     Number of obs  =
>  3344
> GMM weight matrix:     Cluster (lead_n)
>
>                               (Std. Err. adjusted for 418 clusters in
> lead_n)
> ------------------------------------------------------------------------------
>             |               Robust
>             |      Coef.   Std. Err.      z    P>|z|     [95% Conf.
> Interval]
> -------------+----------------------------------------------------------------
>         /c1 |   .9598146   .0517448    18.55   0.000     .8583967
>  1.061232
>         /c2 |   .9256337   .0535588    17.28   0.000     .8206605
>  1.030607
>         /c3 |   .8305105   .0582733    14.25   0.000     .7162969
>  .9447241
>         /c4 |    .956631   .0482825    19.81   0.000     .8619991
>  1.051263
>         /c5 |   .9736638    .053159    18.32   0.000     .8694742
>  1.077853
>         /c6 |   .9493385   .0541098    17.54   0.000     .8432853
>  1.055392
>         /c7 |   .8518398   .0555893    15.32   0.000     .7428867
>  .9607929
>         /c8 |   .8813955    .051279    17.19   0.000     .7808906
>  .9819004
>         /c9 |   .9793823   .0518981    18.87   0.000      .877664
>  1.081101
>        /c10 |   .9923967   .0533734    18.59   0.000     .8877868
>  1.097007
>        /c11 |   .8911549   .0555809    16.03   0.000     .7822183
>  1.000092
>        /c12 |   -.865805   .0502334   -17.24   0.000    -.9642607
> -.7673493
>        /c13 |  -.8909156   .0537489   -16.58   0.000    -.9962615
> -.7855697
>         /c0 |  -.1046114   .0564682    -1.85   0.064    -.2152871
>  .0060643
> ------------------------------------------------------------------------------
> Instruments for equation 1: x_clus1 x_clus2 x_clus3 x_clus4 x_clus5 x_clus6
> x_clus7 x_clus8
>    x_clus9 x_clus10 x_clus11 x_clus12 x_clus13 _cons
>
> And the second model:
>
> Step 1
> Iteration 0:   GMM criterion Q(b) =  13.184222
> Iteration 1:   GMM criterion Q(b) =  1.497e-26
> Iteration 2:   GMM criterion Q(b) =  4.387e-32
>
> Step 2
> Iteration 0:   GMM criterion Q(b) =  4.314e-33
> Iteration 1:   GMM criterion Q(b) =  4.314e-33  (backed up)
>
> GMM estimation
>
> Number of parameters =  14
> Number of moments    =  14
> Initial weight matrix: Unadjusted                     Number of obs  =
>  3344
> GMM weight matrix:     Cluster (lead_n)
>
>                               (Std. Err. adjusted for 418 clusters in
> lead_n)
> ------------------------------------------------------------------------------
>             |               Robust
>             |      Coef.   Std. Err.      z    P>|z|     [95% Conf.
> Interval]
> -------------+----------------------------------------------------------------
>         /b1 |   1.049204   .0549893    19.08   0.000     .9414266
>  1.156981
>         /b2 |   1.078344   .0586466    18.39   0.000     .9633983
>  1.193289
>         /b3 |   .9043237   .0616768    14.66   0.000     .7834394
>  1.025208
>         /b4 |    1.04687   .0528909    19.79   0.000     .9432057
>  1.150534
>         /b5 |   1.043876   .0569363    18.33   0.000     .9322833
>  1.155469
>         /b6 |    1.01851   .0592967    17.18   0.000     .9022906
>  1.134729
>         /b7 |   .9258437   .0602654    15.36   0.000     .8077256
>  1.043962
>         /b8 |   .9485584   .0553715    17.13   0.000     .8400322
>  1.057085
>         /b9 |   1.066044   .0601146    17.73   0.000     .9482216
>  1.183867
>        /b10 |   1.075929   .0577217    18.64   0.000     .9627967
>  1.189062
>        /b11 |   1.017601   .0614807    16.55   0.000     .8971007
>  1.138101
>        /b12 |  -.9610472   .0526738   -18.25   0.000    -1.064286
> -.8578085
>        /b13 |  -.9627249   .0589321   -16.34   0.000     -1.07823
> -.8472202
>         /b0 |  -.1096011   .0587362    -1.87   0.062    -.2247219
>  .0055198
> ------------------------------------------------------------------------------
> Instruments for equation 1: x_fe1 x_fe2 x_fe3 x_fe4 x_fe5 x_fe6 x_fe7 x_fe8
> x_fe9 x_fe10
>    x_fe11 x_fe12 x_fe13 _cons
>
> Best,
> J.
>
> __________________________________________
>
> Prof. John Antonakis
> Faculty of Business and Economics
> Department of Organizational Behavior
> University of Lausanne
> Internef #618
> CH-1015 Lausanne-Dorigny
> Switzerland
> Tel ++41 (0)21 692-3438
> Fax ++41 (0)21 692-3305
> http://www.hec.unil.ch/people/jantonakis
>
> Associate Editor
> The Leadership Quarterly
> __________________________________________
>
>
> On 30.10.2011 16:34, Schaffer, Mark E wrote:
>>
>> John,
>>
>> When you don't stack, do you get nonzero values for the maximized GMM
>> objective functions?
>>
>> --Mark
>>
>>> -----Original Message-----
>>> From: owner-statalist@hsphsun2.harvard.edu
>>> [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of
>>> John Antonakis
>>> Sent: 30 October 2011 15:01
>>> To: statalist@hsphsun2.harvard.edu
>>> Subject: Re: st: GMM error (bug in Stata?)
>>>
>>> Hi Mark:
>>>
>>> I am unsure, particularly because gmm works when I don't
>>> stack the models. I have send the dataset and code to the
>>> Stata people to look at (also, forgot to mention, I am using
>>> Stata 11).
>>>
>>> Best,
>>> J.
>>>
>>> __________________________________________
>>>
>>> Prof. John Antonakis
>>> Faculty of Business and Economics
>>> Department of Organizational Behavior
>>> University of Lausanne
>>> Internef #618
>>> CH-1015 Lausanne-Dorigny
>>> Switzerland
>>> Tel ++41 (0)21 692-3438
>>> Fax ++41 (0)21 692-3305
>>> http://www.hec.unil.ch/people/jantonakis
>>>
>>> Associate Editor
>>> The Leadership Quarterly
>>> __________________________________________
>>>
>>>
>>> On 30.10.2011 12:47, Schaffer, Mark E wrote:
>>>  >  John,
>>>  >
>>>  >  I see from the output that after an iteration, the value
>>> of the GMM>  objective function becomes very small, e.g.,
>>> 1.656e-33 ... in other>  words, zero.
>>>  >
>>>  >  This could happen if the model is exactly identified, or
>>> (if I remember>  the discussion in Hall's GMM book
>>> correctly) if the rank of the VCV of>  moment conditions is
>>> PSD instead of PD.  Could either of these be the>  explanation?
>>>  >
>>>  >  Cheers,
>>>  >  Mark
>>>  >
>>>  >>  -----Original Message-----
>>>  >>  From: owner-statalist@hsphsun2.harvard.edu
>>>  >>  [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf
>>> Of>>  John Antonakis>>  Sent: 30 October 2011 05:54>>  To:
>>> statalist@hsphsun2.harvard.edu>>  Subject: Re: st: GMM error
>>> (bug in Stata?)>>   >>  Hi Stas (and Cam):
>>>  >>
>>>  >>  Thanks for the follow-up but its not the number of
>>> clusters>>  that is causing the problem; I have 418 of them
>>> (refers to>>  the output of the first note:
>>>  >>
>>>  >>  Here is the output from the first gmm estimation:
>>>  >>  Step 1
>>>  >>  Iteration 0:   GMM criterion Q(b) =  13.184222
>>>  >>  Iteration 1:   GMM criterion Q(b) =  1.497e-26
>>>  >>  Iteration 2:   GMM criterion Q(b) =  4.387e-32
>>>  >>
>>>  >>  Step 2
>>>  >>  Iteration 0:   GMM criterion Q(b) =  4.314e-33
>>>  >>  Iteration 1:   GMM criterion Q(b) =  4.314e-33  (backed up)
>>>  >>
>>>  >>  GMM estimation
>>>  >>
>>>  >>  Number of parameters =  14
>>>  >>  Number of moments    =  14
>>>  >>  Initial weight matrix: Unadjusted
>>> Number of obs
>>>  >>  =    3344
>>>  >>  GMM weight matrix:     Cluster (lead_n)
>>>  >>
>>>  >>                                  (Std. Err. adjusted for 418
>>>  >>  clusters in
>>>  >>  lead_n)
>>>  >>  --------------------------------------------------------------
>>>  >>  ----------------
>>>  >>                |               Robust
>>>  >>                |      Coef.   Std. Err.      z    P>|z|
>>>  [95% Conf.
>>>  >>  Interval]
>>>  >>  -------------+------------------------------------------------
>>>  >>  ----------
>>>  >>  -------------+------
>>>  >>            /b1 |   1.049204   .0549893    19.08   0.000
>>>  >>  .9414266
>>>  >>  1.156981
>>>  >>            /b2 |   1.078344   .0586466    18.39   0.000
>>>  >>  .9633983
>>>  >>  1.193289
>>>  >>            /b3 |   .9043237   .0616768    14.66   0.000
>>>  >>  .7834394
>>>  >>  1.025208
>>>  >>            /b4 |    1.04687   .0528909    19.79   0.000
>>>  >>  .9432057
>>>  >>  1.150534
>>>  >>            /b5 |   1.043876   .0569363    18.33   0.000
>>>  >>  .9322833
>>>  >>  1.155469
>>>  >>            /b6 |    1.01851   .0592967    17.18   0.000
>>>  >>  .9022906
>>>  >>  1.134729
>>>  >>            /b7 |   .9258437   .0602654    15.36   0.000
>>>  >>  .8077256
>>>  >>  1.043962
>>>  >>            /b8 |   .9485584   .0553715    17.13   0.000
>>>  >>  .8400322
>>>  >>  1.057085
>>>  >>            /b9 |   1.066044   .0601146    17.73   0.000
>>>  >>  .9482216
>>>  >>  1.183867
>>>  >>           /b10 |   1.075929   .0577217    18.64   0.000
>>>  >>  .9627967
>>>  >>  1.189062
>>>  >>           /b11 |   1.017601   .0614807    16.55   0.000
>>>  >>  .8971007
>>>  >>  1.138101
>>>  >>           /b12 |  -.9610472   .0526738   -18.25   0.000
>>> -1.064286
>>>  >>  -.8578085
>>>  >>           /b13 |  -.9627249   .0589321   -16.34   0.000
>>>  -1.07823
>>>  >>  -.8472202
>>>  >>            /b0 |  -.1096011   .0587362    -1.87   0.062
>>>  >>  -.2247219
>>>  >>  .0055198
>>>  >>  --------------------------------------------------------------
>>>  >>  ----------------
>>>  >>  Instruments for equation 1: x_fe1 x_fe2 x_fe3 x_fe4 x_fe5
>>> x_fe6 x_fe7>>  x_fe8 x_fe9 x_fe10
>>>  >>       x_fe11 x_fe12 x_fe13 _cons
>>>  >>
>>>  >>  Here's the output from the second gmm estimation:
>>>  >>
>>>  >>  Step 1
>>>  >>  Iteration 0:   GMM criterion Q(b) =  12.045213
>>>  >>  Iteration 1:   GMM criterion Q(b) =  2.209e-26
>>>  >>  Iteration 2:   GMM criterion Q(b) =  1.786e-32
>>>  >>
>>>  >>  Step 2
>>>  >>  Iteration 0:   GMM criterion Q(b) =  1.883e-33
>>>  >>  Iteration 1:   GMM criterion Q(b) =  1.656e-33
>>>  >>
>>>  >>  GMM estimation
>>>  >>
>>>  >>  Number of parameters =  14
>>>  >>  Number of moments    =  14
>>>  >>  Initial weight matrix: Unadjusted
>>> Number of obs
>>>  >>  =    3344
>>>  >>  GMM weight matrix:     Cluster (lead_n)
>>>  >>
>>>  >>                                  (Std. Err. adjusted for 418
>>>  >>  clusters in
>>>  >>  lead_n)
>>>  >>  --------------------------------------------------------------
>>>  >>  ----------------
>>>  >>                |               Robust
>>>  >>                |      Coef.   Std. Err.      z    P>|z|
>>>  [95% Conf.
>>>  >>  Interval]
>>>  >>  -------------+------------------------------------------------
>>>  >>  ----------
>>>  >>  -------------+------
>>>  >>            /c1 |   .9598146   .0517448    18.55   0.000
>>>  >>  .8583967
>>>  >>  1.061232
>>>  >>            /c2 |   .9256337   .0535588    17.28   0.000
>>>  >>  .8206605
>>>  >>  1.030607
>>>  >>            /c3 |   .8305105   .0582733    14.25   0.000
>>>  >>  .7162969
>>>  >>  .9447241
>>>  >>            /c4 |    .956631   .0482825    19.81   0.000
>>>  >>  .8619991
>>>  >>  1.051263
>>>  >>            /c5 |   .9736638    .053159    18.32   0.000
>>>  >>  .8694742
>>>  >>  1.077853
>>>  >>            /c6 |   .9493385   .0541098    17.54   0.000
>>>  >>  .8432853
>>>  >>  1.055392
>>>  >>            /c7 |   .8518398   .0555893    15.32   0.000
>>>  >>  .7428867
>>>  >>  .9607929
>>>  >>            /c8 |   .8813955    .051279    17.19   0.000
>>>  >>  .7808906
>>>  >>  .9819004
>>>  >>            /c9 |   .9793823   .0518981    18.87   0.000
>>>  >>  .877664
>>>  >>  1.081101
>>>  >>           /c10 |   .9923967   .0533734    18.59   0.000
>>>  >>  .8877868
>>>  >>  1.097007
>>>  >>           /c11 |   .8911549   .0555809    16.03   0.000
>>>  >>  .7822183
>>>  >>  1.000092
>>>  >>           /c12 |   -.865805   .0502334   -17.24   0.000
>>> -.9642607
>>>  >>  -.7673493
>>>  >>           /c13 |  -.8909156   .0537489   -16.58   0.000
>>> -.9962615
>>>  >>  -.7855697
>>>  >>            /c0 |  -.1046114   .0564682    -1.85   0.064
>>>  >>  -.2152871
>>>  >>  .0060643
>>>  >>  --------------------------------------------------------------
>>>  >>  ----------------
>>>  >>  Instruments for equation 1: x_clus1 x_clus2 x_clus3
>>> x_clus4 x_clus5>>  x_clus6 x_clus7 x_clus8
>>>  >>       x_clus9 x_clus10 x_clus11 x_clus12 x_clus13 _cons
>>>  >>
>>>  >>  Best,
>>>  >>  J.
>>>  >>
>>>  >>  __________________________________________
>>>  >>
>>>  >>  Prof. John Antonakis
>>>  >>  Faculty of Business and Economics
>>>  >>  Department of Organizational Behavior>>  University of
>>> Lausanne>>  Internef #618>>  CH-1015 Lausanne-Dorigny>>
>>> Switzerland>>  Tel ++41 (0)21 692-3438>>  Fax ++41 (0)21
>>> 692-3305>>  http://www.hec.unil.ch/people/jantonakis
>>>  >>
>>>  >>  Associate Editor
>>>  >>  The Leadership Quarterly
>>>  >>  __________________________________________
>>>  >>
>>>  >>
>>>  >>  On 30.10.2011 00:40, Stas Kolenikov wrote:
>>>  >>  >  John,
>>>  >>  >
>>>  >>  >  how many clusters do you have? May be you are running
>>> out>>  of clusters>  in estimation of the weight matrix if
>>> you have>>  fewer clusters than>  parameters.
>>>  >>  >
>>>  >>  >  On Sat, Oct 29, 2011 at 11:56 AM, John Antonakis>>
>>> <John.Antonakis@unil.ch>  wrote:
>>>  >>  >>  The goal of my estimation procedure is to make cross
>>>>>
>>>>> model comparison, where>>  the model have different>>
>>>
>>> instruments ( and given the clustering I have, I>>  want to
>>>>>
>>>>> have a generalized Hausman test hence the use of gmm). I
>>>
>>> want>>  to>>  show that the second stage estimates don't
>>> change when>>  I change the>>  instruments....it's a
>>> simulation study I am>>  working on, hence the>>  "strangeness".
>>>  >>
>>>  >>  *
>>>  >>  *   For searches and help try:
>>>  >>  *   http://www.stata.com/help.cgi?search
>>>  >>  *   http://www.stata.com/support/statalist/faq
>>>  >>  *   http://www.ats.ucla.edu/stat/stata/
>>>
>>> *
>>> *   For searches and help try:
>>> *   http://www.stata.com/help.cgi?search
>>> *   http://www.stata.com/support/statalist/faq
>>> *   http://www.ats.ucla.edu/stat/stata/
>>>
>>
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/
>



-- 
--------------------------
Maarten L. Buis
Institut fuer Soziologie
Universitaet Tuebingen
Wilhelmstrasse 36
72074 Tuebingen
Germany


http://www.maartenbuis.nl
--------------------------

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/


© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index