Statalist The Stata Listserver

[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: Generalized Method of Moments (GMM) and applications

From   "Robert Duval" <>
Subject   Re: st: Generalized Method of Moments (GMM) and applications
Date   Sun, 12 Nov 2006 15:36:34 -0500

Stata does not have a GMM general routine. This has been discussed
before on Statalist.

If you have a non-linear problem, I think you can use the ML
optimization routines to solve for the parameters.

However, you'll have to calculate your varcov matrix by hand, and if
you really care about the accuracy of your s.e.'s then you'll better
use computationally sound numerical matrix operations (e.g.
decompositions, etc.) together with Mata.
My own experience with the simple matrix operators provided in Stata
is that they are not very accurate (e.g. try solving an OLS problem
"by hand" with mat accum, inv, etc. and then compare it with the
actual results from reg).


On 11/12/06, Thomas Erdmann <> wrote:

I was looking at Stata's GMM capabilities due to a specific task (see
below). What I found was basically the command family around -ivreg- and

-ivreg2- (-overid-, -ivendog-, -ivhettest-, -ivreset-, -condivreg-,
-ivgmm0-) as well as extension for panel use, like -xtivreg2- .

Any general comments on GMM and Stata would be welcome.

To be more specific about what I was looking for: a more general GMM setup
to assess the equality of portfolios using the moment conditions
e1= R1 - MR; e2= R2 - MR; ...; e10 = R10 - MR
where R1 to R10 are the return of ten decile portfolios and MR is the Mean
Return parameter to be estimated (over-identified equation with 10 moment
conditions and one parameter to be estimated)*.

[Note: I am aware of the following alternative commands to check for
equality for portfolios available in Stata: -anova-, -oneway-, -kwallis-,
-kwallis2-, -nptrend-]


* This method is for example described in "Filter rules based on price and
volume in individual security overreaction", M. Cooper, Review of Financial
Studies, (p.907) Vol.12 1999.

*   For searches and help try:

*   For searches and help try:

© Copyright 1996–2017 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index