st: xtabond

[Kit Baum <baum@bc.edu>]

P. Garcia said

I want to know if it is better to use a dynamic or a static panel model.
I have estimated a dynamic panel model with two stage GMM estimator,  
and I
have found that there is second order correlation.
In this way, the estimations are not consistent.
Could I say that the dynamic approach is not valid and it is better  
use a
static panel?


If the lagged dependent variable is significant (even in the presence  
of a possibly misspecified model) a static model, by restricting that  
coefficient to zero, would be misspecified and inconsistent.

The fact that there is meaningful AR(2) in the residuals is more  
likely to signal invalidity of the instruments. See David Roodman's  
"how to do xtabond2" working paper:
http://ideas.repec.org/p/boc/asug06/8.html
and G. Bruno's paper
http://ideas.repec.org/p/boc/usug06/06.html


Kit Baum, Boston College Economics
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/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/