Ilaria Tucci <[email protected]>:
How are you measuring the productivity regressor?
What happens to the coef on productivity when you leave out area
and/or sector dummies and cluster errors on the most inclusive of
these sets?
In general, if you identify the "effect" of region off migrants only,
and migrants have different productivity and different effects of
productivity on wages, anything can happe. Try running the regressions
separately for migrants and non-migrants, and of course in the latter
the dummies will drop out--are the coefs on productivity the same? Is
it because of omitted variable bias or differences in the sample?
Perhaps comparing the coef on productivity when you leave out area
and/or sector dummies and run regressions on only migrants, clustering
errors as before, will help answer that question.
On 10/30/07, Ilaria Tucci <[email protected]> wrote:
> Hi,
>
> I have a question regarding the fixed effects estimations. Indeed I am performin a regression of
>
> Log wages= A*productivity+B*sex+C*age*....+Area DUMMY+sector DUMMY+...
>
> I am estimating this equation with fixed effects. My question is the foolowing.
> some of the workers do not change area and sector during the period of the analysis, while others do this. In this case the estimates for dummies are identified for those workers who change region and sector. ON the contrary, the productivity estimate regard all workers of course.
> Is that possible that such a way of getting the estimates of the dummies impacts the estimates concerning the other variables (sucha s productivity)?
>
> I red on Wooldridge that when a variable is time variant for some workers and while it is not for others it should be excluded from the model. Is that right?
>
> I think the area and sector dummies should be in the model.
>
*
* 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/
