st: Problem with multicollinearity in panel data fixed effects regressions

[Rolf Lyneborg Lund <starpen@mac.com>]

Hello everybody.
 
I am new to this list so I hope I don't break any unwritten rules here.
 
My problem is probably very complex but can be summed up quickly. I have
balanced panel data containing 42 countries over 12 years and I measure
different types of segregation in education (measured by an index resembling
the dissimilarity index) and I run regressions with other macro level
variables; one is GDP (transformed by ln). My problem is, that all of the
other macro level independent variables are highly correlated and this is
confirmed by not only a VIF-test but also just by observing the change in
coefficients in my pooled models. Are there any way of correcting this so
that I am able to run multiple regressions with the variables? I've found a
few different fixed effects regressions that is supposed to handle this but
I can't get any to work. Most literature tells me to just drop variables
that cause multicollinearity, but I would really like to be able to run some
pooled models.
 
Thank you for your help!
 
Best
-Rolf


*
*   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/