This sounds like a case for two-part models if your zeros are
identifiable - e.g., cost of hospitalization for someone who isn't
hospitalized is zero. If your response looks like a zero-inflated
Poisson or negative binomial you can use those routines.
Tony
Peter A. Lachenbruch
Department of Public Health
Oregon State University
Corvallis, OR 97330
Phone: 541-737-3832
FAX: 541-737-4001
-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Jon O'Brien
Sent: Thursday, January 03, 2008 4:20 PM
To: [email protected]
Subject: st: Dynamic Tobit with endogenous regressors?
Are there any Stata programs (i.e., ado files) to perform a dynamic
Tobit
model (with panel data, obviously) and endogenous regressors?
For those of you who want more info on my data, here it is:
1. Dependent variable is essentially linear, but with a big cluster at
zero.
The zeros are essentially unobserved on the dependent variable, but this
is
due to external factors rather than self-selection (thus, Tobit should
be
more appropriate than a Heckman, I believe)
2. It is panel data, where the lagged value of the dependent variable
should
have a direct causal influence on the current value
3. There are potentially endogenous regressors. However, I think I have
a
good instrument for the key [potentially endogenous] regressor.
Thank you for any help you can offer,
Jon O'Brien
University College Dublin
Belfield, Dublin 4, Ireland
*
* 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/
