st: Re: Cluster Option in Reg command

[Christopher Baum <kit.baum@bc.edu>]

On Sep 4, 2010, at 2:33 AM, Milton wrote:

> 1. Does cluster(v) option after regress comand automatically correct for
> hetroscedasticity as well, or do I have to put in a robust option in
> addition to cluster (v) to ensure it is cluster robust as well as
> hetroscedasticity robust?
> 2. Ecotrix question not STATA specific: Suppose I am not sure that errors
> are correlated within cluster, will putting the option cluster (v) provide
> "wrong" (inconsistent) estimates of standard errors of coefficients? Are
> ther in-built tests in sTaTA to check if rrors are in fact cluster
> correlated?

1. Yes. The cluster-robust VCE subsumes the standard 'robust' VCE. Indeed if every observation is its own cluster they are identical. See my talk from the BOS'10 Stata conference.

2. No. If you are quite sure that within-cluster errors are independently distributed, then it is better to use just robust, but specifying cluster in the absence of strong dependence within-cluster does not give you inconsistent VCE estimates. As always, if you don't have a problem, don't correct for it. 

Mark Schaffer, Austin Nichols and I are working on some code that will provide formal tests of (1-way and 2-way) clustering, but it is not ready yet.

Kit



Kit Baum   |   Boston College Economics & DIW Berlin   |   http://ideas.repec.org/e/pba1.html
                              An Introduction to Stata Programming  |   http://www.stata-press.com/books/isp.html
   An Introduction to Modern Econometrics Using Stata  |   http://www.stata-press.com/books/imeus.html


*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/