Re: RE: st: RE: GMM estimation: restricting parameter estimates

[Nick Cox <njcoxstata@gmail.com>]

Kit already told you how to do this.

 gmm (y-{b0} - {b1}*a-{b2}*k-{b3}*h+ (1 - {b2} - {b3})*v)

Nick
njcoxstata@gmail.com


Cynthia a.k.a. Bley N'Dede

> Thank you very much Kit;
> I have tried to that before ut were not able to find the right way to set it up or the right command; Here is my original command for the gmm estimation (solow growth model):
> method 1: gmm (y-{b0} - {b1}*a-{b2}*k-{b3}*h-{b4}*v), instruments(z1 z2 z3 z4 z5)
> method 2: xtabond y a k h v lags(1) artests(2)
>
> now my question is how to put the constraints on the command above? also, is it possible to run a gmm estimation and fixed effects (FE) model? because tafter running the Hausman test, I found FE more appropriate.
>
> I have found the constrained regression but I am not familiar
> constraint 1 _b[k]  + _b[h] + _b[v] = 1
> cnsreg y a k h v, constraints (1)
>
> when using the above comments, all the coefficients sum up to one. But it's not a GMM estimation.
>
> How do I put a constraint on the parameters if I want to use a GMM estimation?
>
> Thanks for your help.
>
> Notes: I am using STATA 12
>            Hausman test report FE more efficient

behalf of Christopher Baum [kit.baum@bc.edu]

 On Jun 25, 2013, at 8:33 AM, Bley wrote:
>
>> I am trying to estimate a growth model using also gmm estimation. I would like to set the sum of three parameters equal to one. could you please help me in this matter.
>
> That one's easy. If b1 + b2 + b3 = 1, then you can only estimate two of them, and the third you substitute as (1 - b1 - b2).
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/