Christopher Gigoriou wrote:
> I have run the drukker test implemented on stata to detect serial
> correlation in fixed effect panel data models...it told me I have some
> serial correlation...then, to correct it, am I allowed to use the command
> xtgls then p(h) and c(a)??(since my model is specified with a FE)
You don't necessarily _have_ to use FGLS. You could, for instance, try
David Roodman's user-written -newey2-, downloadable from SSC. With this
routine, you can also fit a fixed-effects model _and_ obtain OLS parameter
estimates corrected for both autocorrelation and heteroscedasticity. For
instance:
. xi: newey2 y x1 x2 i.fe xk, lag(#) force
I was reminded about all this off-list by another Statalist member last
week. So long as you _don't_ have a lagged dependent variable, this works
just as well as -areg- or -xtreg, fe-.
CLIVE NICHOLAS |t: 0(044)191 222 5969
Politics |e: [email protected]
Newcastle University |http://www.ncl.ac.uk/geps
*
* 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/
