Bookmark and Share

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


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

Re: st: fixed effects with multicollinearity


From   daniel klein <klein.daniel.81@googlemail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: fixed effects with multicollinearity
Date   Fri, 29 Jul 2011 12:07:27 +0200

Colin,

you seem surprised that time invariante variables are dropped from the
equation. It is, however straight forward, that effects of time
invariante variables cannot be estimated using the fixed-effects
approach. Substracting the mean of a time invariante variable yields
0, so these variables are not used in the estimation. We only use
within varinace to estimate the effects and if there is none, what
should we estimate?

In fact, the disappearance of time invariant variables is one of the
main reasons we use fixed-effects estimators -- because we are
controlling for observed and _unobserved_ time invariate
heterogeneity.

I do not know what you mean by "OLS with dummies". If you want to
create one dummy for each unit, then the point estimates will be
exactly the same as those returned by -xtreg ,fe-. Although not
desinged for this purpose, you can use Stata's -areg- to verify the
results.

The choice of using OLS, Random-Effects or Fixed-Effects estimators
depends, besides the research interest, on the assumptions you are
willing to make. The correction of standard errors will do little if
your point estimates are inconsistent.


Best
Daniel
*
*   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/


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