[Limdep Nlogit List] klein spady - binary choice

[William Greene]

Wiji. This is an identification issue. The constant must equal zero
because of estimation of the density using a kernel estimator. But, 
because there is also no internal normalization on the coefficient
vector - e.g., it is robust to heteroscedasticity - a coefficient is
normalized at one.  The estimator will pump out numbers if you don't
fix a coefficient at 1. I suspect, however, that the model is not
identified in that case.

Any other wisdom out there?

/Bill Greene



----- Original Message -----
From: "Wiji Arulampalam" <Wiji.Arulampalam at warwick.ac.uk>
To: "Limdep and Nlogit Mailing List" <limdep at limdep.itls.usyd.edu.au>
Sent: Monday, November 9, 2009 10:56:57 AM GMT -05:00 US/Canada Eastern
Subject: [Limdep Nlogit List] klein spady - binary choice

I am trying to estimate a binary choice model using Klein-Spady semi
parametric estimator.  Limdep sets the intercept=0.  There is a user
written program in Stata which also sets the intercept=0 (stata program
does not ask for any other restriction).  However, when this technique
is applied, I find the authors setting the intercept=0 as well as
another slope coefficient=1.  I.e. an additional normalisation is
applied.
 
Could someone please tell me what Limdep is doing here?  Is there
another normalisation that is applied here?
 
Many thanks.
 
Yours,
Wiji
 
 
Wiji Arulampalam
Department of Economics
University of Warwick
Coventry CV4 7AL
UK
email: wiji.arulampalam at warwick.ac.uk
<mailto:wiji.arulampalam at warwick.ac.uk> 
Tel: +44 (0)24 76523471
Fax: +44 (0)24 76523032
Web: http://www.warwick.ac.uk/go/arulampalam
<http://www.warwick.ac.uk/go/arulampalam> 
_______________________________________________
Limdep site list
Limdep at limdep.itls.usyd.edu.au
http://limdep.itls.usyd.edu.au