Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.


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

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


From   "Schaffer, Mark E" <[email protected]>
To   <[email protected]>
Subject   RE: st: GMM error (bug in Stata?)
Date   Sun, 30 Oct 2011 11:47:17 -0000

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: [email protected] 
> [mailto:[email protected]] On Behalf Of 
> John Antonakis
> Sent: 30 October 2011 05:54
> To: [email protected]
> 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 
> <[email protected]> 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/
> 


-- 
Heriot-Watt University is a Scottish charity
registered under charity number SC000278.

Heriot-Watt University is the Sunday Times
Scottish University of the Year 2011-2012



*
*   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 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index