[Limdep Nlogit List] FEM Hetero is BLUE but not consistent; what is the t-statistic?

[William Greene]

Dear Mr. Lester.  The t statistic for the alpha^s in a FE model has degrees
of freedom NT-K-N.  Assuming normality, the t distribution has plenty of
degrees of freedom.  This is an exact, finite sample result.  The t statistic
is not unreliable at all.  The issue of small T relates to the asymptotic
properties.  The conditional variance of the constant term is O(1/T), so it
does not converge to the true alpha unless T is "growing."  This is what it
means to be inconsistent - it is still unbiased..  This is the issue of "unreliability," 
I suppose, though that would be a misnomer.  I don't think reliability has a
relationship to the points here. Interesting question.  
Regards,
Bill Greene

************************************************
Professor William Greene
Department of Economics
Stern School of Business
New York University
44 West 4th St., Rm. 7-78
New York, NY   10012
Ph. 212.998.0876
Fax. 212.995.4218
URL. http://www.stern.nyu.edu/~wgreene
Email. wgreene at stern.nyu.edu
************************************************

----- Original Message -----
From: Laurence Lester <laurence.lester at flinders.edu.au>
Date: Wednesday, July 19, 2006 7:20 pm
Subject: [Limdep Nlogit List] FEM Hetero is BLUE but not consistent;	what is the t-statistic?

> 
> 
> It the one or two-way FEM panel model the estimated fixed effect 
> (heteroalpha) is BLUE (Hsaio 1986) but inconsistent because
> Time (waves) does not go to infinity but is fixed. LIMDEP provides a
> t-statistic for the alphas; is LIMDEP assuming T is large enough for
> asymptotic consistency, so that if T is small (<5) the t-statistic is
> unreliable?
> 
> Thanks 
> 
> Laurence Lester 
> National Institute of Labour Studies 
> Flinders University, SA 
> Ph: (08) 8201 2002 
> Email:  <Laurence.lester at flinders.edu.au>
> Laurence.lester at flinders.edu.au 
> Mail: GPO Box 2100, Adelaide, SA 50001 
> 
> 
> _______________________________________________
> Limdep site list
> Limdep at limdep.itls.usyd.edu.au
> http://limdep.itls.usyd.edu.au
>