st: Overidentifying restriction test for 2sls with clustered error and for any ML

[Suyin Chang <suyinlchang@gmail.com>]

Dear list,

A few weeks ago I came to you with a more general question on
post-estimation for system of equations. This time I try to refine the
questions a little bit, going directly to issue I am having trouble
with: I need to test for the over-identification restriction of a
non-recursive system of two equations estimated by 2sls with clustered
se and by ML.

The first problem is that Stata 12 doesn't seem to have estat overid
implemented for 2sls with clustered error, only for GMM with clustered
error. Does anyone know a solution or a workaround for that? I've
found a couple of past threads in the list's archives, but quite
inconclusive on the matter.

The second problem is that Stata 12 does not have any post-estimation
test for instrumentalization of systems estimated with the comand sem,
i.e. via ML. Once I am dealing with a nonrecursive model, it is
crucial for me to verify the overidentification restriction and maybe
find a way to assess the strength of instruments also under ML. Any
ideas here would be also very welcome.

Please, give me any insights or suggestions you may find useful.

Thanks for your time,

Suyin
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/