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

st: Re: RE: Heckman self selection bias - selection variable


From   "R.E. De Hoyos" <redeho@hotmail.com>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: Re: RE: Heckman self selection bias - selection variable
Date   Thu, 6 Oct 2005 13:41:04 +0100

Lilach,

If I understood you right, you want to control for selectivity using a transformed (originally continuous) binary variable. I think the answer to your question will depend on the nature of your first and second stage dependent variables. The whole motivation of Heckman's procedure is the truncated nature of the objective function's dependent variable. You must make the case for a truncated dependent variable whose "observability" depends upon the "critical value" of a second continuous variable. One can argue that determining such a "critical value" in a continuous variable is highly subjective ---unless there is a natural way of determining it.

There are some examples. Consider wage equations for part-time and full-time employees. Although hours worked is a continuous variable, it has a "natural" critical value. If you plot the distribution of hours worked per week, very likely you will see two agglomeration points around 20 and 40 hours. Hence it seems reasonable to transformed this continuous variable into a binary one and then use it to control for selectivity a la Heckman.

Hope this helps,

Rafa
________________________
R.E. De Hoyos
Faculty of Economics
University of Cambridge
CB3 9DE, UK
www.econ.cam.ac.uk/phd/red29/

----- Original Message ----- From: "Maarten Buis" <M.Buis@fsw.vu.nl>
To: <statalist@hsphsun2.harvard.edu>
Sent: Thursday, October 06, 2005 1:15 PM
Subject: st: RE: Heckman self selection bias - selection variable



Dear Lilach

Consider the classic example: you want to explain the wage of women. This is a case where you might expect self-selection, since a considerable proportion of women "choose" not to work. The wage is a continuous variable, but it is only observed when the woman works. The selection equation has one dependent variable, which is dichotomous: is the wage observed (1) or not (0), i.e. does the woman work or not. So the selection equation does not have a continuous dependent variable, but a dichotomous dependent variable indicating whether the case is selected or not.

Hope this helps,
Maarten

-----Original Message-----
From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu]On Behalf Of Lilach_Nachum@baruch.cuny.edu
Sent: donderdag 6 oktober 2005 4:43
To: statalist@hsphsun2.harvard.edu
Subject: st: Heckman self selection bias - selection variable



I need to conduct Heckman style self-selection bias test.
For the first stage of this test, the selection probit model, I have a
number of alternative DVs, all of whom are continuous variables. Can I
dichotomize a continuous variable, so that it can be used as a DV in the
probit model?

Will appreciate very much any advice on this.

*
* 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/

*
*   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