Re: st: Panel data estimation with constant variables

[Maarten buis <maartenbuis@yahoo.co.uk>]

--- On Tue, 22/3/11, Sirak wrote:
> I have some variables which are constant over all the
> years, eg. altitude, latitude and distance, and thus
> droped from the estimation. Does the fixed effects
> pannel data estimation allow contant variables.
> If not, what could be the solution?.

The solution depends on whether you want to just control 
for those time constant variables or whether you are 
substantively interested in the effects of these 
variables. You are not able to include those variables 
because you already control for everything that is constant
within each country and you cannot control for the same
thing twice. So if all you want to do is control for these
time constant variables, than you just leave those 
variables out of your model and you'll be fine (and if you
don't leave them out, Stata will do it for you, and there
is still no problem). If you are substantively interested
in the effects of these time constant variables and you 
want to use fixed effects regression then you are in 
trouble as the two are fundamentally incompatible.
Basically, you will then need to revert to a random effects
model. There are tricks you can sometimes apply to get 
fixed effects estimates within that model on the time 
changing variables, but the parameter of the time constant 
variables will always remain random effects estimates.

Hope this helps,
Maarten

--------------------------
Maarten L. Buis
Institut fuer Soziologie
Universitaet Tuebingen
Wilhelmstrasse 36
72074 Tuebingen
Germany

http://www.maartenbuis.nl
--------------------------


      

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/