<>
John said
One difference is that x1 is entirely dependent on endogenous variables;
so my naive question here is: which predicted values of x1 and x2 are
included in Eq. 2 and 3 respectively (also knowing that x1 and x2
predict each other and that x1 has no unique instruments that predict it
directly)?
I believe that, following up on Eric's comments, the appropriate estimation can be carried out as
ivreg2 y (x1 x2 = n1 n2 q1 q2 p1 p2)
ivreg2 x1 (m1 m2 x2 = n1 n2 q1 q2 p1 p2), endog(m1 m2)
ivreg2 x2 (x1 = q1 q2 p1 p2) n1 n2
reg m1 q1 q2
reg m2 p1 p2
I include the endog option on the second eqn to determine whether m1 and m2 must be considered endogenous. Other options, such as gmm2s robust, are probably advisable.
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/
