Re: st: ivreg2: interacting the endogenous regressor

[Christopher Baum <kit.baum@bc.edu>]

<>
On May 24, 2012, at 8:33 AM, Jana wrote:

> Hi -- thanks for the idea, Kit (Go Blue!). Unfortunately, even decent  
> instruments are very hard to find. Essentially, I am trying to test  
> whether the effect of the instrumented variable depends on the values  
> that the variable x1 takes on. Is there some way to do this that  
> doesn't involve trying to find a second instrument (which is near- 
> impossible)? I've split x1 into "low" and "high" and run separate  
> regressions, but that doesn't allow cross-estimation comparisons, right?

I didn't say you needed to find two instruments. You need an instrument z1 and you need to interact it with x1. That adds up to two instruments, but ony one excluded variable will be used to define them both.  There should not be any difficulty testing for an interaction effect using this strategy.

Kit


Kit Baum   |   Boston College Economics & DIW Berlin   |   http://ideas.repec.org/e/pba1.html
                             An Introduction to Stata Programming  |   http://www.stata-press.com/books/isp.html
  An Introduction to Modern Econometrics Using Stata  |   http://www.stata-press.com/books/imeus.html


*
*   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/