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

st: RE: Re: GMM or 2SLS with cluster adjustment or others

From   "Yu-Luen Ma" <>
To   <>
Subject   st: RE: Re: GMM or 2SLS with cluster adjustment or others
Date   Wed, 14 May 2008 16:59:27 -0500

Thank you very much! I have one follow-up question on choosing between xtabond (or xtabond2) and xtivreg2 with cluster-robust SEs plus gmm2s. Under what circumstance would it make sense for me to include lagged dependent variables in the model (and thus considering xtabond)? In my case I am looking at factors affecting insurance premiums across countries (premium is my dependent variable). While I am not interested in how last year's premiums affect current premiums, it is possible that current year's premiums are associated with prior year's premiums, and thus I would like to control for that effect. Would xtivreg2 with cluster-robust SEs be sufficient? Thank you!

From: on behalf of Kit Baum
Sent: Wed 5/14/2008 9:17 AM
Subject: st: Re: GMM or 2SLS with cluster adjustment or others

The Arellano-Bond / Arellano-Bover / Blundell-Bond methodology 
(xtabond, xtabond2) is appropriate if you have lagged dependent 
variables, and unnecessary otherwise.

If you have a panel with endogenous regressors, use xtivreg2 with 
cluster-robust SEs, preferably with the gmm2s option.

See Baum-Schaffer-Stillman, Stata Journal 7:4 2007, for details. 
Preprint available from my homepage below as a BC working paper.

Kit Baum, Boston College Economics and DIW Berlin
An Introduction to Modern Econometrics Using Stata:

On May 14, 2008, at 02:33 , statalist-digest wrote:

> I have panel data and would like to adjust for potential 
> autocorrelation in the dependent variable as well as the 
> endogeneity in one of the independent variables. Would 2SLS with 
> cluster adjustment (i.e. ivreg with cluster option) be sufficient? 
> I was suggested to use the Arellano-Bond (1991)/Arellano-Bover 
> (1995) estimation procedure. I assume it is GMM. Would that be 
> necessary? How do these two methodologies differ? Thanks.

*   For searches and help try:

*   For searches and help try:

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