st: RE: RE: within estimator as "phyrric-victory" in corporate finance?

["Millimet, Daniel" <millimet@mail.smu.edu>]

Ditto to Mark's response.  Easiest way to tell that no cross-section variation is used is to note that the coefficients on time invariant regressors are not identified in fixed effects models (despite the fact that they may vary wildly cross-sectionally).

More helpful, though, is the proper interpretation of the results you are finding.  If there is little temporal variation, you will most likely find statistical insignificance due to large std errors.  This should not be interpreted as these covariates do not matter, rather that they are too highly co-linear with the fixed effects.  Thus, you cannot distinguish between the effects of the covariates that are slow moving over time and the (unobservable) time invariant attributes of firms. 

****************************************************
Daniel L. Millimet, Professor
Department of Economics
Box 0496
SMU
Dallas, TX 75275-0496
phone: 214.768.3269
fax: 214.768.1821
web: http://faculty.smu.edu/millimet
****************************************************

-----Original Message-----
From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Schaffer, Mark E
Sent: Thursday, January 20, 2011 8:34 AM
To: statalist@hsphsun2.harvard.edu
Subject: st: RE: within estimator as "phyrric-victory" in corporate finance?

Erasmo,

> -----Original Message-----
> From: owner-statalist@hsphsun2.harvard.edu 
> [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of 
> Erasmo Giambona
> Sent: 20 January 2011 14:09
> To: statalist
> Subject: st: within estimator as "phyrric-victory" in 
> corporate finance?
> 
> Dear Statalist,
> 
> There is a large tendency in empirical corporate finance 
> research to rely on the within estimator. Now, this seems 
> appropraite to many on statistical ground. Empirical 
> corporate finance research relies on panel data for firms and 
> using the within estimator allows to control for for 
> time-invariate firm heterogeniety. However, once we "subtract"
> firm-level averages with the within estimators, it seems that 
> the coefficient estimates (within estimates) are ONLY 
> measuring the effect of a change in the RHS variables within 
> firm on changes in the dependent variable. This is 
> problematic because firms change very little and very slowly 
> and with fixed-effects many independent variables could 
> "appeear" statistically and/or economically insignificant, 
> while they might still be very powerful in explaining 
> cross-sectional variation.
> 
> Perhaps I am missing something on the interpretation of the 
> "within estimator"? That is, are these estimators preserving 
> cross-sectional variation in any way?

Short answer - no.  You've got it right.  All panel data econometrics
mainstream textbook stuff.

The classic article on this in the econometrics literature is Griliches,
Zvi & Hausman, Jerry A., 1986. "Errors in variables in panel data,"
Journal of Econometrics, Elsevier, vol. 31(1), pages 93-118, February.

Cheers,
Mark

> 
> I would appreaciate any reaction on this issue.
> 
> Best regards,
> 
> Erasmo
> *
> *   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/
> 


-- 
Heriot-Watt University is a Scottish charity
registered under charity number SC000278.


*
*   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/

*
*   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/