Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: RE: SE's in Fixed Effects Model


From   "Mark Schaffer" <M.E.Schaffer@hw.ac.uk>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: RE: SE's in Fixed Effects Model
Date   Wed, 23 Jun 2004 17:01:49 +0100

Just to add to Richard's explanation,

Subject:        	st: RE: SE's in Fixed Effects Model
Date sent:      	Wed, 23 Jun 2004 10:54:42 -0500
From:           	"Boylan, Richard" <rboylan@cba.ua.edu>
To:             	<statalist@hsphsun2.harvard.edu>
Send reply to:  	statalist@hsphsun2.harvard.edu

> In method (i) you need to adjust the degree of freedom.  If you compare
> the standard errors in the two estimates you will notice that they are
> identical up to a scalar (approx = 1.46).
> Richard

The adjustment is necessary because in the second method (de-meaning 
by hand) -regress- doesn't realize you have used up about 1k degrees 
of freedom with your 1k fixed effects.  Notice that the (correct) F-
stat of -areg- is reported as F(  2, 10062) whereas the (incorrect) F-
stat from -regress- is F(  2, 21598).  Sqrt(21598/10062) = appx 1.46.

--Mark

> 
> -----Original Message-----
> From: owner-statalist@hsphsun2.harvard.edu
> [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Tim R. Sass
> Sent: Wednesday, June 23, 2004 10:45 AM
> To: statalist@hsphsun2.harvard.edu
> Subject: st: SE's in Fixed Effects Model
> 
> 
> I am estimating a fixed-effects panel model using two methods: (i)
> manually 
> demeaning the data and running regress, (ii) running areg on the
> original 
> data.  I am able to get identical estimated slope coefficients from both
> 
> models.  However, the standard errors are very different.  The standard 
> errors should be asymptotically equivalent, so given my large sample
> size 
> (21,000+ observations) the reported standard errors ought to be quite 
> close.  I tried applying the standard error correction noted by Wiggins
> and 
> Gould in the Stata 6 FAQ "How can I estimate a fixed-effects regresion
> with 
> instrumental variables?", but as one would expect (given my large sample
> 
> size) it made very little difference.  Any ideas what may be going on 
> here?  My output is given below.
> 
> Tim
> 
> 
> 
> . areg  nrtrgain nschools chgschl, absorb(student) ;
> 
>                                                         Number of obs =
> 21601
>                                                         F(  2, 10062) =
> 219.77
>                                                         Prob > F      =
> 0.0000
>                                                         R-squared     =
> 0.3732
>                                                         Adj R-squared =
> -0.3455
>                                                         Root MSE      =
> 31.382
> 
> ------------------------------------------------------------------------
> ------
>      nrtrgain |      Coef.         Std. Err.          t        P>|t|
> -------------+----------------------------------------------------------
> -------------+------
>     nschools |  -2.487567   1.967418      -1.26   0.206
>       chgschl |  -11.67239   .5572808    -20.95   0.000
>          _cons |   20.07552   2.067988       9.71   0.000
> 
> 
> . reg   de_nrtrgain de_nschools de_chgschl;
> 
>        Number of obs =   21601
>        F(  2, 21598) =  471.74
>        Prob > F      =  0.0000
>        R-squared     =  0.0419
>        Adj R-squared =  0.0418
> 
> ------------------------------------------------------------------------
> ------
>     de_nrtrgain |      Coef.        Std. Err.         t        P>|t|
> -------------+----------------------------------------------------------
> -------------+------
>   de_nschools |  -2.487567   1.342864     -1.85   0.064
>     de_chgschl |  -11.67239   .3803728   -30.69   0.000
>              _cons |   20.07552   1.411508     14.22   0.000
> ------------------------------------------------------------------------
> ------
> 
> . display _se[de_chgschl]*sqrt(e(df_r)/(e(df_r)-3+1));
> .38039039
> 
> 
> 
> 
> Tim R. Sass
> Professor                               Voice:   (850)644-7087
> Department of Economics         Fax:      (850)644-4535
> Florida State University                E-mail:   tsass@coss.fsu.edu
> Tallahassee, FL  32306-2180     Internet:
> http://garnet.acns.fsu.edu/~tsass
> 
> 
> *
> *   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/
> 
> *
> *   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/

Prof. Mark E. Schaffer
Director
Centre for Economic Reform and Transformation
Department of Economics
School of Management & Languages
Heriot-Watt University, Edinburgh EH14 4AS  UK
44-131-451-3494 direct
44-131-451-3008 fax
44-131-451-3485 CERT administrator
http://www.som.hw.ac.uk/cert

*
*   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/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index