Statalist The Stata Listserver

[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: Fixed effect vs. random effect or something else?

From   Guido Heineck <>
Subject   Re: st: Fixed effect vs. random effect or something else?
Date   Fri, 02 Feb 2007 08:57:35 +0100


have a look at Stata's xthtaylor, its background is given by Wooldridge (2002), 325-328; Verbeek, 2004, 353-355 or Baltagi, 2005, 124-133 (incl. Stata-example).


Date: Thu, 1 Feb 2007 20:35:29 -0800 (PST)
From: tabreez shams <>
Subject: st: Fixed effect vs. random effect or something else?

Dear Statalisters,
I am sure somebody out there will be kind enough to
provide some thoughts on the following issue.

I want to control for unobserved fixed effect along
with endogeneity in regressors for my exhaustive
unbalanced panel sample of 212 firms spanning over 8
years. Many of the explanatory variables of interests
are not that time-variant and one variable is

In this regard, my understanding is that Fixed Effect
(FE) estimation (with xtreg, fe) would control for
both unobserved and endogeneity in the regressors and
consistent and efficient for large T and if there is
sufficient within panel variation. Alternatively,
random effect (RE) estimation (xtreg, re) would only
control for random unobserved heterogeneity but not
the endogeneity issue.

Given the low within panel variation of the regressors
as well as presence of time-variant regressor, I am
not sure which way to pursue: FE or RE? I am aware
that Hausman test would provide some indication in
this regard still concern of the time-invariant nature
of my data.

Sorry if I am missing something here and any helping
hands will be highly appreciated.

PhD candidate,
Accounting and Finance Dept.
Monash University

*   For searches and help try:

© Copyright 1996–2017 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index