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

Re: st: testing endogeneity in a two-equation model with censored and binary dependent variables.


From   "Mark Schaffer" <M.E.Schaffer@hw.ac.uk>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: testing endogeneity in a two-equation model with censored and binary dependent variables.
Date   Wed, 9 Jul 2003 14:29:05 +0100

Ricardo,

> Dear Statalist readers,
> 
> I am estimating the following two-equation model using a two-stage procedure
> suggested by Maddala (1983):
> 
> Y1 = a1X1 + B1Y2 + e1
> 
> Y2*= a2X2 + e2
> 
> where Y2=1 if Y2*>0
> 	Y2=0 otherwise
> 
> Y1 is censored at zero and Y2 is binary (the realised value of the latent
> variable
> Y2*). Since Y2 is assumed to be endogenous, I would like to test the
> endogeneity of Y2. I checked the Durbin-Wu-Hausman test but it is not
> appropriate when one of the dependent variable is binary.

This last statement confuses me.  The standard DWH test of 
endogeneity is used when estimating a single equation using 
instrumental variables.  The endogeneity relates to one or more 
regressors in this single equation.  These regressors can be binary - 
it doesn't matter for either the IV estimation or the DWH test.

If I understand correctly, you can't use the DWH test for a different 
reason - Y1 is censored and you can't us straightforward IV to 
estimate the Y1 equation.

Would the following work?

1. Estimate the Y1 equation using Joe Harkness' -ivtobit-, i.e., 
treat Y2 as endogenous.  This is your efficient but possibly 
inconsistent estimator.

2. Estimate the Y1 equation using -tobit-, i.e., treat Y2 as 
exogenous.  This is your inefficient but consistent estimator.

3. Use -hausman- to perform a Hausman test.  It should be distributed 
as chi-squared with 1 degree of freedom (because you are testing 1 
variable for endogeneity).  The output of -hausman- may suggest that 
there are 2 degrees of freedom (because -hausman- tries to be clever 
in working out the df and occasionally doesn't get it right), but 
df=1.

Comments, anybody?

--Mark

> Thus, does anyone
> know if there is an alternative way to test for endogeniety in such a
> models.
> 
> Any help will be much appreciated.
> 
> 
> Ricardo Henríquez
> Chile
> 
> *
> *   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/


Prof. Mark E. Schaffer
Director
Centre for Economic Reform and Transformation
Department of Economics
School of Management & Languages
Heriot-Watt University, Edinburgh EH14 4AS  UK
44-131-451-3494 direct
44-131-451-3008 fax
44-131-451-3485 CERT administrator
http://www.som.hw.ac.uk/cert

*
*   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