re: Re: st: simultaneous equations with panel data, endogenous regressors and

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

<>
many thanks for your suggestion. I still have a question, though.
Since I have two simultaneous equations, I should control for the
dependency in the two error terms.
Does your suggestion control for the latter as well?

It is a common misconception that this is necessary. It is not, as is often pointed out on the list. That is, 3SLS is never necessary unless
you are trying to impose cross-equation constraints. 2SLS on each equation will give you consistent estimates if their specification
is valid. 3SLS can potentially improve efficiency, but will also cause all equations' estimates to be inconsistent if one of them
is misspecified. Thus limited-info (single-equation) estimation has much to recommend it.

Here endeth the lesson.  ( I gave this same lecture to my class on Wednesday. )

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/