xtgee with fixed effects

[Gustav Egede Hansen]

Hi everybody

Is there a way to run xtgee with fixed effects? From this post (https://www.statalist.org/forums/for...e-and-xtreg-re), I see that using - independent - yields POLS estimates, while - exchangeable - is random effects. I have been looking into - xtpoisson yvar xvar, vce(robust) fe -but my code is running extremely slow, so I am seeking an alternative in xtgee.

Best,
Gustav

[Fei Wang]

If you feel time-consuming to estimate poisson FE using -xtpoisson-, you may try Mundlak device which essentially controls for within-panel mean of all time-varying regressors instead of the full set of panel dummies. Jeff Wooldridge gives an alternative to -xtoprobit- in https://www.statalist.org/forums/for...for-panel-data and simply replacing "oprobit" in #4 with "poisson" will solve your problem.