# Re: st: duration analysis with endogenous regressors

 From Antoine Terracol To statalist@hsphsun2.harvard.edu Subject Re: st: duration analysis with endogenous regressors Date Thu, 14 Jun 2007 10:45:07 +0200

Hello,

You might want to have a look at
Abbring, J.H. & van den Berg, G.J. "The nonparametric identification of treatment effects in duration models", Econometrica, 2003, 71, 1491-1517

and at some applications of their method, for example :

van den Berg, G.J.; van der Klaauw, B. & van Ours, J. "Punitive Sanctions and the Transition Rate from Welfare to Work", Journal of Labor Economics, 2004, 22, 211-241

Best,
Antoine

Aysegul Aydin wrote:

Dear Colleagues,

I apologize if this message goes a little too long but rather than just
asking what the Stata command should look like, I want to ask you how to
answer my research question with Stata as Maarten Buis has pointed out a few
days ago. I started with the most basic idea in the civil war literature
that third-party interventions affect civil war duration. I estimated a
weibull model and used a dataset with conflict month as the unit of
analysis. Let's say the model simply looks like this;

War Duration = Constant + Third-Party Intervention + Natural Resources + GDP
of Civil War Country + Error Term

Since weibull is used, the dependent variable is really the likelihood that
the conflict will end in the next month given that an intervention has
happened in the present month. It is of course possible to estimate this
model with something other than duration analysis such as OLS if we can
collapse the data and have one line of observation for each conflict in the
dataset.

Third-party intervention is a time-variant dummy variable that takes on a
value of one for months in which an intervention has taken place and 0
otherwise. Interestingly, the intervention variable is - at least
theoretically - an endogenous regressor in the sense that natural resources,
GDP of the civil war country and most importantly war duration also affects
the likelihood that a potential third-party state will intervene. So,

Third-Party Intervention = Constant + War Duration + Natural Resources + GDP
of Civil War Country + Error Term

The main argument is actually that war duration and intervention
simultaneously affect each other, so the causal arrow runs both ways. That
said, I think that this is a simultaneous equations model but the duration
part complicates the analysis and the nature of the dependent variables is
different (but as I said the first equation can be estimated with something
other than event history analysis if this makes things easier). I remember
reading a 1989 article by Randall Olsen and George Farkas about endogenous
covariates in duration models but frankly I do not even know what the model
that I need right now should look like or how to apply their solution in
Stata. Rather than going further with what I think the model is and sound
I will greatly appreciate any suggestions that you might have. Many thanks
All the best,

Aysegul
=======================

Aysegul Aydin Assistant Professor
Department of Political Science
University of Colorado at Boulder
Ketchum, 123 333 UCB
Boulder, CO 80309-0333
Voice: (303) 492-7802
Fax: (303) 492-0978

*
* 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/

```
--
Ce message a ete verifie par MailScanner
pour des virus ou des polluriels et rien de
suspect n'a ete trouve.

*
*   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/
```