Not sure why this has not shown up.
--Alex Cavallo
Managing Consultant
Navigant Consulting, Inc.
----- Forwarded by Alexander Cavallo/NCI on 12/14/2004 12:56 PM -----
Alexander Cavallo/NCI wrote on 12/14/2004 10:13:16 AM:
> Bin,
>
> Blundell, Griffith and Windmeijer (2000) show that with panel data,
> the possion fixed effect estimator is consistent. In a private
> email, Windmeijer suggested the use of Stata's xtpois with the fe
option.
>
> I think you could use xtgee for the models you propse.
>
> Model 1: xtgee y x1 x2, family(poisson) link(log) corr(ar1)
>
> Create dummies for your id variable
> tab(id), gen(dum)
>
> Model 2: xtgee y x1 x2 dum*, family(poisson) link(log) corr(ar1)
>
> Here's a reference on the consistency of poisson fixed effects
> Blundell, Griffith, Windmeijer "Individual effects and dynamics in
> count data models" (Institute for Fiscal Studies Working Paper W99/3).
>
> --Alex Cavallo
>
> Managing Consultant
> Navigant Consulting, Inc.
>
> Subject: st: regress nonlinear panel data model with AR(1) error term
>
> Hi,
> I have some questions on regression of two models. I have PANEL COUNT
data.
> I want two models by two different error term specification.
> 1. the error term is AR(1) and no individual effect.
> 2. the error term is AR(1) plus the individual effect.
>
> Does anyone know how I can regress these two models
>
> Best,
> Bin Sun
*
* 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/