Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at

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

Re: st: two-way cluster-robust covariance matrix estimates

From   Christopher Baum <>
To   "" <>
Subject   Re: st: two-way cluster-robust covariance matrix estimates
Date   Sat, 28 Dec 2013 15:23:57 +0000

On Dec 28, 2013, at 2:33 AM, Anat wrote:

> My question is, does employing (xtivreg2, fe cluster (id year),
> equivalent to running one way clustering  (xtreg, fe vce(cluster id))
> including years dummies in the regression?
> Any hint of what should guide me in choosing one option over the other?
> I will really appreciate any advice

The short answer: no. Clustering by year allows cross-sectional observations' errors to be arbitrarily correlated -- thus affectijng the VCE -- but does nothing to alter the conditional mean that regression estimates. If your analysis shows that time effects are jointly significant, they should be included in the regression to avoid specification error. I would suggest using xtivreg2 with two-way clustering and time effects. You can evaluate how much difference it makes to the VCE to cluster by the second (time) dimension.


Kit Baum
Professor of Economics and Social Work, Boston College, Chestnut Hill MA, USA
DIW Research Professor, Department of Macroeconomics, DIW Berlin, Berlin, Germany  |

*   For searches and help try:

© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index