RE: st: -xtreg,fe-, whether to drop out individuals participated once

["Schaffer, Mark E" <M.E.Schaffer@hw.ac.uk>]

Ekaterina,


> -----Original Message-----
> From: owner-statalist@hsphsun2.harvard.edu 
> [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of 
> Ulrich Kohler
> Sent: 22 May 2007 07:46
> To: statalist@hsphsun2.harvard.edu
> Subject: Re: st: -xtreg,fe-, whether to drop out individuals 
> participated once
> 
> Ekaterina Selezneva wrote:
> > There is a 5 round panel datset, which is unbalanced. Some cases 
> > contain missing values for several variables (in the 
> variable of interest as well).
> >
> > Let's assume, I want to estimate something like -xtreg y x* 
> time*,fe 
> > i(id)-
> >
> > if I exploit the unbalanced dataset "as it is",  Stata 
> estimates the 
> > model showing a number of observations per group being from 1 to 5. 
> > (in the dataset used in Stata 9 manual it is from 1 to 15).
> >
> > Reading a panel econometrics textbook, one would find (for 
> example, in 
> > Wooldridge, p.267) a rather logical phrase like "when at least two 
> > time periods available,... " before the formulas of the 
> fixed effects 
> > transtormation.
> >
> > If following this logic, I would need to drop all the 
> individuls, who 
> > participated just once and than estimate the model.  If I 
> do so, the 
> > estimate results are slightly different.
> >
> > Am I right, that it happends due to the fact that in 
> -xtreg,fe- Stata 
> > not only substracts the individual means, but also the 
> 'pooled sample mean'?
> >
> > What would be the best strategy to choose: leave it as it 
> is, or drop 
> > those participated once?
> >
> > And if the answer is the second option.. should I drop out 
> those two 
> > have just one anwser to my question of interest, OR, at 
> Stata performs 
> > in the complete cases analysis style, I need leave in my 
> sample just 
> > those people who have answered the question at least twice, and 
> > moreover, only if they have complete information for all 
> the explanatory variables.
> 
> There might be a difference in the constant and in sigma_u, 
> but there shouldn't be a difference in the coefficients of 
> the independent variables. 
> In a fixed effects model the 1-observation-only respondents 
> do not contribute to the calculation of the coefficients. As 
> long as you only interpret the coefficients of the 
> independent variables it really doesn't matter whether your 
> drop these observations, or not,
> 
> The sigma_u is the standard deviance of the dummies for the 
> individuals. You can interpret this statistic as a measure 
> for how much the individuals differ in their average value of 
> the dependent variable. If you want to interpret it that way 
> it is your choice wheter you want to drop the 
> 1-observation-only respondents, or not. If you only want to 
> talk about people, who are observed more than once, drop them. 
> 
> Uli

Just a couple of things to add to what Uli said above.

- If I'm not mistaken, dropping the singletons (one-observation groups)
will affect the estimate of the constant.  This is because Stata's
xtreg,fe uses between-group variation to get the constant (I find this
approach a little odd, but maybe it's just me).

- If you want fixed effects estimates that drop singletons, and you
don't need an estimate of the constant, you can use -xtivreg2-.  As well
as panel IV, -xtivreg2- will do fixed effects estimates for the case
where all regressors are exogenous.

Cheers,
Mark


Prof. Mark Schaffer
Director, CERT
Department of Economics
School of Management & Languages
Heriot-Watt University, Edinburgh EH14 4AS
tel +44-131-451-3494 / fax +44-131-451-3296
email: m.e.schaffer@hw.ac.uk
web: http://www.sml.hw.ac.uk/ecomes
 

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/