re: Re: st: GMM with inequality and equality constraints

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

<>
Stas said


The first one looks like the regular GMM equation to me. The second
one is pretty much impossible to implement directly. You can introduce
a Lagrange multiplier in front of it:

E lambda[ gamma y_i + delta z_i ] = 0

You would probably run into identification problems and/or problems
with the standard errors: if the constraint is active, then gamma and
delta are not identified; if it is not, then lambda is identically
zero, and would have to have a missing standard error.


Although I was unaware of this literature until a few minutes ago (there are some rewards to reading dissertation fellowship applications), there appears to be a growing literature dealing with "moment inequalities" of the sort mentioned in this thread. The primary reference is

Pakes A, Porter K, Ho K, Ishii J.,"moment inequalities and their applications", CEMMAP paper CWP16/07, Institute for Fiscal Studies

http://ideas.repec.org/p/ifs/cemmap/16-07.html

Cheers
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/