st: instrument in the Heckman sample selection model

[Agnieszka Kanas <a.kanas@maw.ru.nl>]

Dear all,
I have a question about a model specification in the Heckman sample 
selection model. My problem is a bit more complicated but the essence 
can be explained as follows. I test the hypothesis that immigrants into 
regions within a country receive larger (earning) returns to their 
schooling in one region vs. the other region. One important issue that I 
need to address is the self-selection of immigrants into the two 
regions. I estimate the Heckman sample selection model. It is 
recommended that at least one variable that is not included in the 
outcome (earnings) equation to be included in the selection equation. I 
want to use the first region upon arrival to the host-country as an 
instrument in the selection equation. I assume that original location 
choice influences the current location, but net of the current location, 
it should not affect immigrants’ earnings. Thus, by including the first 
region variable into the selection model I control for the unobserved 
characteristics that influence immigrants’ current region. Is this 
approach - using the lagged selection variable as an instrument - 
statistically sound? I appreciate your comments. Agnes

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/