st: RE: fixed effects estimation with unbalanced panel

["Jacobs, David" <jacobs.184@sociology.osu.edu>]

One problem you may be having is that your matrix isn't full rank.  If you cluster by caseid ID to remove the effects of serial correlation on your standard errors, you remove the over time portion of your N.  

If you then have more coefficients to estimate than your effective N (after clustering), the program can't calculate your standard errors.  Is this your problem?

Dave Jacobs

-----Original Message-----
From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Eric Walcott
Sent: Tuesday, March 26, 2013 12:46 PM
To: statalist@hsphsun2.harvard.edu
Subject: st: fixed effects estimation with unbalanced panel

I'm having an issue with fixed effects estimation of an unbalanced panel.
Due to various issues, mostly related to the smallness of my dataset, when I run fixed effects including the year dummies and cluster robust standard errors, STATA won't calculate the F statistic for the model. If I remove the year dummies or I run the regression without cluster-robust SE it works, but then I'm losing important parts of the model. The year dummies are jointly significant, so I don't want to remove them, but I also need cluster robust standard errors to control for heteroskedasticity and serial correlation that I have in my data. Any suggestions on how I should handle this? Is there a way around this problem, or do I have to choose between year dummies and cluster-robustness?

Thanks,
Eric
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/