Re: st: xttest2 for N<T

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

On Aug 16, 2011, at 2:33 AM, Elii wrote:

> I have a panel of N=20 and T=28 and I want to test for cross sectional dependence. I read that Pesaran's test is valid only for N>T and LM test is applicable in my case. However, I have a problem after estimating with FE. I take the following:
> 
> 
> xttest2
> 
> Correlation matrix of residuals is singular.
> not possible with test

Elii and I communicated privately about this. It turns out that the covariance matrix of residuals from her FE model is nearly singular, and xttest2 circa 2004 had a criterion for the determinant that was flagged. I relaxed the criterion, so that it now works on her dataset. The revised version is available from SSC. Ultimately the routine should be rewritten in Mata.

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/