Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, is already up and running.

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

Re: st: Clustering Standard Errors versus Dummies

From   Austin Nichols <>
Subject   Re: st: Clustering Standard Errors versus Dummies
Date   Fri, 2 Jul 2010 09:49:09 -0400

natasha agarwal <> :

No. Clustering of errors can be due to an error components
data-generating process (alone) i.e.
e_{it} = u_i + v_{it}
where v_{it} is i.i.d. but it need not be--the v component can still
be correlated within cluster, whether or not all u_i are zero (if all
u_i are not zero, and they are correlated with explanatory variables,
the point estimates will also differ and FE will be consistent where
OLS is not, as Kit alluded to).  This is why cluster-robust SEs are
usually a good idea--always a good idea if you have a large number of
balanced clusters.  See refs in

If you have a small number of clusters (fewer than 20 as a rule of
thumb) or unbalanced clusters, the cure (cluster-robust SEs) can be
worse than the disease, but to judge that tradeoff, you would need to
do a simulation...

On Fri, Jul 2, 2010 at 5:59 AM, natasha agarwal
<> wrote:
> Does this then imply that if I include year fixed effects, industry
> fixed effects and estimate the same with within estimator then I do
> not need to cluster on industry and year?
*   For searches and help try:

© Copyright 1996–2016 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index