Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

re:st: pooled regression vs fixed effects

From   Christopher Baum <>
To   "" <>
Subject   re:st: pooled regression vs fixed effects
Date   Tue, 7 Feb 2012 08:10:47 -0500

I am trying to address the multiple directorships association with firm performance however having difficulties to find which method to use. for example, when I do the following test on my panel data which has companies and years as index.
xi : areg tobin_q market_value logboard_size average_directorships_per_director i.year, robust absorb(industry)
I got all the independent variables significant however when I use the fixed effect model which seems more convient than the random effects based on the hausman test, 

xi : xtreg tobin_q market_value logboard_size average_directorships_per_director  i.icb_suprsectr_code i.year, robust fe

I got most of them insignificant. I am having difficulties to understand the reason and any help will be appreciated. I expect that the results should be similar in both cases as I have the same dummies across the year. however, do not have sufficient information about what areg really do ?   kind regardsOzgur Ozdemir 		 	   		  

You don't have the same model. areg is just a convenience command that allows you to partial out the effect of dummies included in absorb(). So your areg model
is pooled OLS with industry and time dummies. All firms share the same constant term. In the xtreg, you are fitting fixed effects with the within estimator, which gives
each firm its own constant term. That estimator will have explanatory power i-f-f the movements of firm-level Tobin's Q around the firm mean are correlated with firm-level deviations of regressors around their firm means.

If the F-test at the foot of the xtreg,fe (which may be suppressed if you say 'robust' --try removing that option) is significant, then the areg model is misspecified in terms of unobserved heterogeneity at the firm level.


Kit Baum   |   Boston College Economics & DIW Berlin   |
                             An Introduction to Stata Programming  |
  An Introduction to Modern Econometrics Using Stata  |

*   For searches and help try:

© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index