[Limdep Nlogit List] bivariate probit selection

William Greene wgreene at stern.nyu.edu
Wed Nov 28 11:28:34 EST 2007

There is an appendix in the original Heckman (1979) Econometrica
paper on selection that deals with simultaneous equations. It's
not hardwired in LIMDEP now. But, it should not be too hard to
program the computations he describes there. Do note that in the
original paper, Jim did not use several now well known matrix 
results that greatly simplify the problem of computing the standard
errors, so the results in his paper look more complicated than
they really are.
I would note, the application you mentioned has a binary endogenous
variable (marital status). In such a case, the Heckman results on
2sls with the selection correction won't apply. You would want an
MLE instead.  I have never worked this out, but I suspect it is
moderately, though not extremely, high on the difficulty scale.
Maybe a two (two + two?) step estimator that begins with a probit
model for the endogenous dummy variable and the computation of
predicted probabilities... (just thinking "out loud")
Bill Greene

----- Original Message -----
From: "Brendan Cushing-Daniels" <bdaniels at gettysburg.edu>
To: "Limdep and Nlogit Mailing List" <limdep at limdep.itls.usyd.edu.au>
Sent: Tuesday, November 27, 2007 4:38:54 PM (GMT-0500) America/Bogota
Subject: RE: [Limdep Nlogit List] bivariate probit selection

Actually, while I'm at it, may I ask if anyone can point me in the
direction of a sample selection model that corrects for an endogenous
RHS variable at the same time?  For example, suppose I am estimating a
wage equation controlling for selection bias into full-time work, but I
also want to allow for the possibility that marital status is

Thanks, again.


-----Original Message-----
From: limdep-bounces at limdep.itls.usyd.edu.au
[mailto:limdep-bounces at limdep.itls.usyd.edu.au] On Behalf Of Brendan
Sent: Tuesday, November 27, 2007 4:31 PM
To: Limdep and Nlogit Mailing List
Subject: [Limdep Nlogit List] bivariate probit selection

I am trying to estimate a model in which there are two selection
problems.  Intuitively, it seems to me that a bivariate probit selection
rule is the way to handle this.  I see in the LIMDEP help file under
SELECTION that another model based on the original framework is the
Bivariate Probit Selection Rule, but I can't find further information on
this either in the electronic help menu or in my LIMDEP manuals.  Can
anyone point me in the right direction for the specification of this
model and for the technical details?



Limdep site list
Limdep at limdep.itls.usyd.edu.au
Limdep site list
Limdep at limdep.itls.usyd.edu.au

Professor William Greene
Department of Economics
Stern School of Business
New York University
44 West 4th St., Rm. 7-78
New York, NY   10012

More information about the Limdep mailing list