[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
RE: st: some unexplained issues with coefficient comparisons
> -----Original Message-----
> From: firstname.lastname@example.org
> [mailto:email@example.com] On Behalf Of
> Jeff Pitblado, StataCorp LP
> Sent: 27 June 2006 16:40
> To: firstname.lastname@example.org
> Subject: Re: st: some unexplained issues with coefficient comparisons
> Narasimhan Sowmyanarayanan
> <email@example.com> is comparing results
> of -suest- after -areg- with an equivalent -regress- model
> with indicator variables:
> > I am facing one situation that I have not been able to intuit. I ran
> > two different fixed effect panel regressions for different groups
> > compared the coefficients across the two groups. The test yeilded
> > there was no significant difference when I used 'suest' to compare
> > coefficients.
> > Then I tried to run the same models in OLS using explicit dummy
> > variables. My estimates are very close (some difference possibly
> > because of some additional dummy variables that I tried to put in
> > I could not use in my fixed effects model). The coefficients that I
> > compared were very close. But when I ran the same comparison using
> > I found that the coefficients were significantly different.
> > To summarize the same coefficients when I used suest after areg gave
> > me insignificant difference as compared to the usage of suest after
> > simple regression with dummy variables which indicated the
> > was significant.
> We've recently discovered that -suest- yields incorrect
> results when used after -areg-. This is not something that
> can easily be fixed given that the meat of the sandwich
> estimator of variance (Robust/Huber/White VCE) cannot be
> properly computed due to the fact that the coefficient
> estimates for the absorbed categories in -areg- are not
> present in -e(b)- (and the corresponding indicators are not
> present in the dataset).
> In the next ado-file update, -suest- will report an error
> message when used with -areg- (something it should have done
> from the start).
> Note that -suest- was not listed as a supported
> postestimation command after
> -areg- in [R] areg postestimation.
Jeff's comment offers a clue on how you might be able to proceed. A
fixed effects model can be estimated using OLS on transformed data. The
transformation is "demeaning", which is conveniently done by Ben Jann's
-center- package. You might be able to do the following:
(1) Create demeaned variables for each of your fixed effect models. The
variables may differ from model to model because the samples on which
they are estimated may differ, hence the possible need to create
different sets of demeaned variables.
(2) Estimate the fixed effects models using OLS on the demeaned
(3) Do the degrees-of-freedom adjustment for the N fixed effects on the
var-cov matrix saved in e(V) by hand and repost using Ben's -eret2-
routine (I *think* this will work). The dof adjustment is standard and
is discussed in various textbooks as well as in the Stata manual's
entries for xtreg etc.
(4) Use -suest- to compare the coefficients.
Rather laborious, but I think it should work.
Hope this helps.
Prof. Mark Schaffer
Department of Economics
School of Management & Languages
Heriot-Watt University, Edinburgh EH14 4AS
tel +44-131-451-3494 / fax +44-131-451-3296
* For searches and help try: