Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: Differences in results for ivreg2 and ivgmm0

From   "Mark Schaffer" <[email protected]>
To   [email protected]
Subject   Re: st: Differences in results for ivreg2 and ivgmm0
Date   Tue, 16 Mar 2004 15:22:39 -0000


Date sent:      	Tue, 16 Mar 2004 10:06:37 -0500
Subject:        	Re: st: Differences in results for ivreg2 and ivgmm0
To:             	[email protected]
From:           	"Min Y. Tan" <[email protected]>
Send reply to:  	[email protected]

> Hi Mark, 
> I have reestimated using ivreg, it gives me the same result as ivreg2. 
> In this case of exact identification, why would ivgmm0 produce different
> results?

I think this might be a bug in the version of ivgmm0 that you are 
using.  You are estimating a rather special case from a programming 
point of view, namely one explanatory variable that is endogenous and 
no exogenous regressors - not even a constant.  If you add a 
constant, you might find all three agree.  Also, do you have the most 
recent version of ivgmm0?  Try updating it and see what happens.

> Just a clarifying programming question of ivreg2: If I want to estimate the
> following model: dum * fe-square - alpha fe-square = error; and my instruments
> is lag fe; the objective is to estimate alpha by minimizing the error term. how
> would I program this in ivreg2 using gmm?

To be honest, I'm not sure what you are trying to estimate here.  I 
guess you want to minimize the *squared* error, no?  Is your 
instrument lagged fe or lagged fe-square?  Etc.

But the answer to your question may be short and less than ideal from 
your perspective.  -ivreg2- estimates linear models.  If you can 
express your model in linear terms, then you can use -ivreg2-.  If 
you can't, you'll have to use some other program or do the 
programming yourself.  Sorry about that!

Best wishes,

> I have programmed it in SAS but SAS
> does not allow for cluster analysis which ivreg2 has.
> minyen
> Min, Have a look at my previous posting: > Try estimating with Stata's built-in
> ivreg, and the coefficient > estimates it reports should match one or the
> other. (I think they'll > match -ivreg2-, because I recently spent some time on
> the programming > involved in a similar special case.) You report the output of
> -ivreg2- and -ivgmm0-. What is reported by Stata's official -ivreg- with the
> -robust- option? I.e., ivreg y (x=lfe), robust nocon --Mark 
> *
> *   For searches and help try:
> *
> *
> *

Prof. Mark E. Schaffer
Centre for Economic Reform and Transformation
Department of Economics
School of Management & Languages
Heriot-Watt University, Edinburgh EH14 4AS  UK
44-131-451-3494 direct
44-131-451-3008 fax
44-131-451-3485 CERT administrator
*   For searches and help try:

© Copyright 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index