[Limdep Nlogit List] hyper parameters a b in random effects negbin

Thornburg, Steven sthornburg at wsu.edu
Thu Jul 1 09:15:23 EST 2010


Thank you for your quick response!

Hauseman et al state...parameters a and b from the beta distribution [which] describe the distribution of the δ [variance?] across firms.
And ...Since these are unobservable random variables, the scale parameter merely serves as a normalization.

A mathematical convenience?

Is there any insight to be gained by looking at the ratio of a & b?
If not, why are they reported?

I guess any suggestion on how I should describe these parameters in my paper would help.

-------------------------------------
Steven Thornburg, Ph.D. CPA
Assistant Professor
Todd 237F
Department of Accounting
Washington State University
Pullman, WA 99164
Voice:  (509) 335-2222
Fax:      (509) 335-4275


-----Original Message-----
From: limdep-bounces at limdep.itls.usyd.edu.au [mailto:limdep-bounces at limdep.itls.usyd.edu.au] On Behalf Of William Greene
Sent: Wednesday, June 30, 2010 2:17 PM
To: Limdep and Nlogit Mailing List
Subject: Re: [Limdep Nlogit List] hyper parameters a b in random effects negbin

Steven. I don't think that is true at all. They are the two parameters of some
underlying beta distribution, and so far as I know, do not have a direct
interpretation. You should check the 1984 Hausman, Hall and Griliches (Econometrica)
paper where that model was created for more details. For my money, the random
effects NB model should be created by having a random parameters NB model in which
the constant term is a random parameter. That puts it on the same platform as other
familiar random effects models.
Regards,
Bill Greene

----- Original Message -----
From: "Steven Thornburg" <sthornburg at wsu.edu>
To: "Limdep and Nlogit Mailing List" <limdep at limdep.itls.usyd.edu.au>
Sent: Wednesday, June 30, 2010 5:12:37 PM GMT -05:00 US/Canada Eastern
Subject: [Limdep Nlogit List] hyper parameters a b in random effects negbin

Dr Greene
Is it safe to say that the hyper parameters in the random effects negative binomial model [a, b ] can be interpreted as [mean dispersion across groups, variance of dispersion across groups] ?


-------------------------------------
Steven Thornburg, Ph.D. CPA
Assistant Professor
Todd 237F
Department of Accounting
Washington State University
Pullman, WA 99164
Voice:  (509) 335-2222
Fax:      (509) 335-4275

_______________________________________________
Limdep site list
Limdep at limdep.itls.usyd.edu.au
http://limdep.itls.usyd.edu.au
_______________________________________________
Limdep site list
Limdep at limdep.itls.usyd.edu.au
http://limdep.itls.usyd.edu.au


More information about the Limdep mailing list