If you are wedded to fixed effects, you have no choice--you can't
include year dummies and any variable that varies only by year (not by
company). I'm inclined to think that's the way to go, but I have no
idea what the distribution of your dependent variable looks like
(remember fixed effects means that every company has the same slope
for every variable, but differ only in their constant terms, so you
are sort of running OLS for each company and constraining the
coefficients to be identical across firms--if your data or theory
don't match this description, you should rethink your strategy). As
for including the dGDP variable, you could interact it with something
that varies by company, if you have a theoretical reason that the
impact differs by some other trait of companies...
On 2/11/06, Julius Fr�d�ric Andr� <[email protected]> wrote:
> I am using a fixed-effects panel data regression model on a balanced panel
> of 110 firms observed for 12 years, all are in the same country. The
> dependent variable is an insolvency predictor for the respective firm in the
> respective year, the independent variables are a time dummy, to evaluate the
> effect of enactment of a certain law, and other controlling variables
> unrelated to the time period.
>
> As one of the controlling variables, I believe that the general economic
> situation in that country, as expressed by the GDP growth rate, has an
> influence on the insolvency predictor, i.e. the dependent variable, and
> therefore include a variable containing the growth rate for the 12 years
> under observation. Since all firms are situated in the same country, these
> 12 values for GDP growth rate are of course the same for all 110 firms. I
> include the GDP variable to control for influence outside the law effect
> that I am researching on using the time dummy and am not particularly
> interested in its coefficient.
>
> My question: Is it sensible to include the GDP growth rate variable (which
> turns out to be significant in the regression model), or is this merely a
> year dummy, since it is the same for all firms? Should it therefore be left
> out?
*
* 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/
