st: re: system estimation with dynamic panel

[Kit Baum <baum@bc.edu>]

< >
Hewan said

I used the example from Nicola so as to not introduce a whole different
example but to re-ask Nicola's question. In my particular case, my
equations didn't have a dynamic nature altogether, but they did have
endognous vars on the right hand sides (e.g. one equations dependent var
was the other's regressor and vice versa). I had, at the time, wanted to
use a "panel version" of reg3, i.e. use 3SLS to estimate a simultaneous
equation system, only that each of three equations wasn't a cross- 
section
(i.e. along the lines of y_i = a + b*x_i + e_i; consider the x_it a  
vector
of regressors) but rather panel (ie. y_it = a + b*x_it + e_it).

Is it correct then that there is no panel version of reg3, and one would
have to use a more "manual" approach in imposing the desired  
restrictions
on the parameter and error matrices?


(1) There is to my knowledge no systems estimator that has panel- 
specific features.

(2) What are the 'desired restrictions on the parameter and error  
matrices'? Do you have cross-equation restrictions or restrictions on  
the VCE in mind?

If not, I see no reason why equation-by-equation -xtivreg2- would not  
yield the desired estimates, as it handles limited-information  
estimation for a panel with endogenous regressors.


Kit Baum, Boston College Economics and DIW Berlin
http://ideas.repec.org/e/pba1.html
An Introduction to Modern Econometrics Using Stata:
http://www.stata-press.com/books/imeus.html


*
*   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/