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, statalist.org is already up and running.
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: st: RE: AR test
Austin Nichols <email@example.com>
Re: st: RE: AR test
Fri, 6 May 2011 11:42:54 -0400
Dan Bergov <firstname.lastname@example.org>:
The AR test is easy, but confidence regions, not so much; see
and also read the recent paper:
On Wed, May 4, 2011 at 5:01 AM, Dan Bergov <email@example.com> wrote:
> Eric, thanks a lot. It just I do not get a thing. I read in Mr. Baum's book that the first stage regression is to test the relevance of the instruments. The AR test I am talking about (Anderson and Rubin 1949) tests the whether we have a confidence region for the estimated parameter even if the presence of weak instruments. Am I correct?
> So I am wondering: I estimate with generalised method of moments, I assume that I have one endogenous regressor and after that test if the parameter I estimated for the endogenous regressor has a confidence region in the case that some of my instruments are not relevant? If someone can tell me if I am wrong or right I would deeply appreciate it. Thnks. Dan.
> ----- Original Message -----
> From: DE SOUZA Eric <firstname.lastname@example.org>
> To: "email@example.com" <firstname.lastname@example.org>
> Sent: Tuesday, May 3, 2011 7:18 PM
> Subject: Re: st: RE: AR test
> add the option -first- to your -ivreg2- command.
> Have a look at the help file for ivreg2: -help ivreg2- for more details.
> Eric de Souza
> College of Europe
> Brugge (Bruges), Belgium
> -----Original Message-----
> From: email@example.com [mailto:firstname.lastname@example.org] On Behalf Of Dan Bergov
> Sent: 03 May 2011 17:57
> To: email@example.com
> Subject: st: AR test
> Hi! I am somehow new to this forum, and stata:). I was wondering if someone could tell what package is needed to perform Anderson Rubin test for weak instruments, I mean the one that is robust to weak instruments not the one reported bi ivreg2 for overidentyfing conditions. I downloaded condivreg package but I have more than one endogenous regressors so I don't need CLR test, but AR test. Thanks! Hope to get a reply!
* For searches and help try: