re: st: Why do i get different results with year robust clusters vs year dummies

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

<>
I am using areg and
when I specify model as areg export political tension ( list of
control variables), robust cluster(year) absorb(country) I find my
coefficient to be significant. However, when I specifiy it with year
dummies as areg export political tension (list of control variables)
y1990 y1991 y1992......., absorb(country), my coefficient turns
insignificant. Can someone help me why I see this inconsistency and
let me know if one model is better than the other? I mean, what I
observe seems counterintuitive because the results should be the same
given that I am controlling for the year effect in both
specifications.

No, you're not. It is a common misconception that estimating a cluster-robust VCE 'controls for year effects'. It does not do anything of the sort. It gives you a different estimate of the VCE, but it does not alter the coefficients' point estimates. Therefore a model with year dummies is not the same as a model excluding them, and if the data say they are jointly significant, you should include them.

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/