st: Probit analysis with panel data and endogeneous explicative variables

["Elena Giarda" <elena.giarda@prometeia.it>]

Dear Stata-listers,

I have a very large data-set of individual earnings, with a panel data 
structure, covering the period 1985-2004. I would like to estimate the 
probabilities of transition from the status of temporary to long-term 
employee. So I would regress the variable y (y=1 if the individual changes 
status, y=0 if the individual does not change status) over his/her level of 
earnings, age and other demographic characteristics. In this set-up, though, 
earnings are an endogenous variable.

I know of a routine (PROBITIV) which estimates probit models with endogenous 
explicative variables, but not in a panel-data framework (am I right?). With 
the command XTPROBIT I can estimate probit models within panel data. Is 
there a way of combining probit estimation with panel data and endogenous 
variables?

I wonder whether I could run a 2-stage regression, such as:

1) XTREG (FE) y instruments, to get y(hat)
2) XTPROBIT y(hat) exogenous variables

Would this approach be correct? If not, do you have any suggestions? I use 
Stata 8.

Many thanks!

Cheers,
Elena Giarda 
*
*   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/