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: FW: appropriateness of cluster option with xtreg, re

From   Federico Belotti <>
Subject   Re: st: FW: appropriateness of cluster option with xtreg, re
Date   Mon, 14 Oct 2013 10:19:05 +0200

Dear Eva,

in my opinion the answer is: you should.

The fact that you are including in your model unit-specific intercepts (which are assumed to be "mean independent" from the exogenous explanatory variables in the case of -xtreg, re-) does not rule out the possibility that regression residuals exhibit heteroskedasticity and/or serial correlation, that is their variance is not constant between panel units and/or is correlated over time. These issues lead to inconsistent covariance matrix estimates and, therefore, to faulty statistical inference. The option -cluster(panelvar)- allows to avoid this risk.

You should take a look at 

Moulton, Brent R., 1986. "Random group effects and the precision of regression estimates," Journal of Econometrics, Elsevier, vol. 32(3), pages 385-397, August.

Hope this help.

On Oct 13, 2013, at 8:07 PM, Eva Wölbert wrote:

> Dear statalist,
>  I would like to know whether I can / should specify cluster(panelvar) when
>  using xtreg, re. 
> This seems to account for the panel structure of the data "twice". Moreover, I am confused because
>  some of my standard errors are going down with the cluster option. 
>  Thanks in advance
>  Eva 		 	   		  
> *
> *   For searches and help try:
> *
> *
> *

Federico Belotti, PhD
Research Fellow
Centre for Economics and International Studies
University of Rome Tor Vergata
tel/fax: +39 06 7259 5627

*   For searches and help try:

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