re:Re: st: pooled regression vs fixed effects

[Christopher Baum <kit.baum@bc.edu>]

<>
Sami said

I wonder shouldn't -areg- and -xtreg- give the same results:

use grunfeld
areg invest kstock mvalue, absorb(company)
xtreg invest kstock mvalue, fe

or with time dummies

xi: areg invest kstock mvalue i.year, absorb(company)
xi: xtreg invest kstock mvalue i.year, fe

I could be wrong, but what I remember from and old FAQ that the two
commands only differ in R2 calculation.


Yes, those two are the same model. But in the original posting the areg was absorbing industry, not company, and I assume that there are a number
of companies in an industry. In his xtreg, he was computing company fixed effects. So these two models are not the same; the areg with industry-specific constant terms is not the same as a FE model with firm-specific constant terms.

Note also per Rich Williams' posting that areg on industries could be appropriate, because you could gather a larger and larger sample of firms in the same fixed set of industry groupings. 

Kit Baum   |   Boston College Economics and DIW Berlin   |   http://ideas.repec.org/e/pba1.html
An Introduction to Stata Programming   |   http://www.stata-press.com/books/isp.html
An Introduction to Modern Econometrics Using Stata   |   http://www.stata-press.com/books/imeus.html


*
*   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/