I have problems with heteroscedasticity in my panel.
Then,I can't use GLS to correct for it,as my panel has T<N.
I can use robust version of areg or Price_Winston regression for panel data. I cannot choose between them, as Stata does not provide any information about assumptions of these two methods. Even worse, if I choose,say, P-W regression, I also have to choose type of correlation within groups and heteroscedasticity, Tests show I have all kinds of them. So, if not understanding how this method works, I cannot secure myself against low degrees of freedom. The main question is: how the variance-covariance matrix is estimated in the franework of P-W regression? Is it possible, say, estimate panel of 214x6=1284 observations with 10 coefficients, AR(1) within correlation, contemprorary correlation and heteroscedasticity? I am not sure in this, as covariance matrix has 23000 elements...
*
* 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/
