re:st: Is there a way to center factor variables

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

<>
If using the grunfeld standard data in Stata and running:

use grunfeld
reg invest kstock mvalue i.company i.time

is there a way to make Stata use all members in company and constrain
there coefficient to sum up to ZERO, and so the constant becomes the
grand mean and not the base category. Things even look harder with
time effect as well.



Doesn't work for a second set of time dummies, but what you describe above is what -xtreg, fe- does automatically for the panel unit dummies. The constant reported is the grand mean of the u_i terms, and the (unreported) coefficients for the panel dummies are deviations from that grand mean.

As described in section 7.1.1 of IMEUS (p. 164-165), you can create such a set of time dummies by subtracting the excluded one from each of the included ones, and using those (T-1) transformed dummies (which n in the xtreg, fe model. If you want the excluded dummy coefficient in point and interval form, -lincom- will do it for you (example on p. 165).

Kit


Kit Baum   |   Boston College Economics & 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/