re: st: IVreg2 on interaction term

[Kit Baum <baum@bc.edu>]

<>


I would very much appreciate if you could help me with the following 
concern on IVreg2.

I have an interaction term of two dummy variables (d1, d2), the first is 
endogenous, d2 is very probably not. I have 1 instrument for d1. I read 
that an IV-regression in this case must include at least two instruments: 
my IV for d1 and IV*d2 for the interaction term. My problem is now a 
technical one: I do not know how to write an ivreg2-command including two 
IVs. The following does not work and replies "Invalid syntax"

Ivreg2   d2   (d1  =  iv)   (d1_d2  =    iv_d2)

I would very much appreciate if you could point out how such a regression 
is written.

If you allow, I would like to ask a second question. Is it possible to use 
newey2 in an IV-regression (I need autocorrelation robust standard errors; 
data is an unbalanced panel).


Re the syntax examples in the help file, what you want is

ivreg2 depvar d2 (d1 d1_d2 = iv iv_d2)

I presume that you inadvertently omitted the name of the dependent variable.

ivreg2 with -robust bw()- options will provide HAC standard errors. But if you have a panel,
what you want is Mark Schaffer's xtivreg2, which is a 'wrapper' for ivreg2. Same options apply.

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/