Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, is already up and running.

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: Overidentification Tests

From   Tirthankar Chakravarty <>
Subject   Re: st: Overidentification Tests
Date   Tue, 30 Aug 2011 15:33:33 -0700

If you note that 3SLS is the GMM estimator under system
homskedasticity, then tranferring calculations to -gmm- produces a
J-statistic directly. This example is taken from the Stata manual
([R]) entry on -gmm-:

webuse klein, clear
// 3SLS via -reg3-
reg3 (consump = wagepriv wagegovt) ///
	(wagepriv = consump govt capital1), ///
	endog(consump wagepriv) exog(wagegovt govt capital1)

// 3SLS via -gmm-
gmm (eq1: consump - {b0} - {xb: wagepriv wagegovt}) ///
	(eq2: wagepriv - {c0} - {xc: consump govt capital1}), ///
	instruments(wagegovt govt capital1) ///
	winitial(unadjusted, independent) ///
	wmatrix(unadjusted) twostep

// J-statistic
estat overid

This way, your estimation can also easily be made robust to
heteroskedasticity. See several old Statalist posts by Christopher
Baum on -reg3-/-sureg- and 3SLS estimation to get an idea of its


On Tue, Aug 30, 2011 at 12:59 PM, Lopez, Andres
<> wrote:
> Hi All,
> If I wanted to do an overidentification test after a REG3 (3SLS) what command/procedure would you use? What are the limitations of this procedure that I should be aware of, given that I am running a 6 equation system with 11 unbalanced panels with about 17 observations per panel/period?
> Thanks,
> Andres
> *
> *   For searches and help try:
> *
> *
> *

Tirthankar Chakravarty

*   For searches and help try:

© Copyright 1996–2015 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index