re:Re: RE: st: How to run 2SLS twice

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

<>
Thank you for your replies. I am just a bit confused and maybe it's
because of the way I have put my equations. Equation 3) below is
actually the first stage regression, equation 2) is a second stage or
intermediate stage and then equation 1) is the last stage or
structural eq.
Now when you suggest to run equation 1) with ivreg2, do you have in
mind to use z2, z2hat (fitted values of ivreg2 from equations 2) and
3)) or x3 as the instrument for the endogenous variable z1?


There is no need--ever--to construct fitted values and plug them in. As Eric suggests, given the recursive nature of your model, it would actually be possible (and preferable, by the usual arguments of OLS vs IV) to estimate all three equations with OLS. There is no simultaneity in this system.

But even if there was, you would not want to be saving predicted values from one equation and plugging them in to another. If you need to use IV (and here you don't), let the program take care of that for you. 

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/