Re: st: GMM estimation program for simultaneous equations with count data

[Darby Jack <djack@fas.harvard.edu>]

You might want to check out Frank Windmeijer's -ExpEnd- program for  
Gauss.  It's at http://cemmap.ifs.org.uk/publications.php? 
publication_id=2656.

-ExpEnd- will estimate the models described in
[Blundell et al., 2002]    Blundell, R., Griffth, R., and Windmeijer,  
F. (2002). Individual effects and dynamics in count data models.  
Journal of Econometrics, 108:113–131.

[Windmeijer and Santos, 1997]    Windmeijer, F. and Santos, J.  
(1997). Endogeneity in count data models. Journal of Applied  
Econometrics, 12:281–294.

[Mullahy, 1997]    Mullahy, J. (1997). Instrumental variable  
estimation of count data models. Review of Economics and Statistics,  
79(4):586–593.

These models don't coincide exactly with your problem, but it might  
be useful nonetheless.

Here's Windmeijer's description of the code:

"ExpEnd is a Gauss programme for non-linear generalised method of  
moments (GMM) estimation of exponential models with endogenous  
regressors for cross section and panel data. The estimators included  
in this package are simple Poisson pseudo ML; GMM for cross section  
data using moment conditions based on multiplicative or additive  
errors; within groups fixed effects Poisson for panel data; GMM  
estimation using quasi-differenced moment conditions eliminating  
unobserved heterogeneity and allowing for predetermined or endogenous  
regressors; and quasi-differenced GMM for a dynamic linear feedback  
model. This manual describes in detail the various estimators, the  
data and software requirements, and the programme commands."

On Aug 5, 2005, at 5:23 AM, chirag patel wrote:


> Hi
>
> I am trying to solve a system of simultaneous equations where the  
> dependent
> variable in both the equations is count data. I am trying to solve  
> this problem
> by modelling it as GMM type estimation so as to take into  
> consideration the
> exponential mean function.
>
> *The simultaneous equations situation that I have is: *
>
> y = exp (b0 + b1 x1 + b2 x2 + b3 x3 + b4 x4 + b5 x5 + b6 x6)
> z = exp (b7+ b8 x1 + b9 x2 + b10 x3 + b11 x4 + b12 x5 + b13 x6 +  
> b14 x7 + b15 y)
>
> y = count data (0 to 11)
> z = count data (0 to 6)
> x1 = 0,1,2
> x2, x3,x7 = continuous variables
> x4,x5,x6 = nominal (0,1)
>
> I wanted advise on the following  point:
>
> 1)What type of GMM estimation will solve the above system of  
> equations.
>  Has somebody developed a STATA program for such a GMM type  
> estimation?
>
>
> I have been advised that there is a user-written program called - 
> qvf- that will
> estimate IV versions of generalised linear models, including Poisson
> regressions.  However, it estimates only single equations, not  
> systems.
>
> It would be very useful if somebody could help with appropriate  
> STATA code.
>
> Thanks
> Chirag
>
> Chirag PATEL
> Lecturer, Marketing Department
> ESC Rennes (Rennes International School of Business)
> Ph:0033 2 99334847
> *
> *   For searches and help try:
> *   http://www.stata.com/support/faqs/res/findit.html
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/