Re: st: Issues with xtabond2

[Christopher Baum <kit.baum@bc.edu>]

<>
On Nov 28, 2011, at 2:33 AM, Alan wrote:

> I am trying to perform a regression using the GMM estimator with the
> xtabond2 function.  I have panel data with about 25,000 observations, 430
> groups (unbalanced), and 132 time periods.  My model is of the form:
> s(i,t) = s(i,t-1) + p(i,t) + x(i,t) + error (without coefficients)
> Where s is my dependent variable, p are my variables of interest, and x are
> my control variables.  My Stata command to implement xtabond is essentially
> as follows (after implementing xtset for i and t):
> xtabond2 s l.s p x, gmm(l.s) iv(p x) robust

Take some time out and read "How to do xtabond2" by Roodman, cited in the help file for xtabond2, as well as his "Too many instruments" paper. The DPD approach is appropriate for small T, large N panels. You have relatively large T, so the xtabond2 approach is trying to create hundreds of instruments. You need to use the lag( ) suboption to limit the number of instruments generated as gmm-style instruments.

Kit

Kit Baum   |   Boston College Economics & DIW Berlin   |   http://ideas.repec.org/e/pba1.html
                             An Introduction to Stata Programming  |   http://www.stata-press.com/books/isp.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/