st: Cluster xtpoisson at level other than that of panel?

[Garret Christensen <garretc@gmail.com>]

I have longitudinal count data with monthly county observations, I'm
running poisson regression, and I include monthly and county fixed
effects, so I'm using the xtpoisson command. My question is this: is
it possible to cluster standard errors at a level other than that of
the panel ID variable? I have set county as the panel variable, so
including the vce(robust) option clusters standard errors at the
county level. However, due to concerns about spatial correlation in my
data, some have suggested to me that I try clustering at a larger
level such as the state. Is there an easy way to implement this? The
cluster option is not allowed.


[Side note: since there are two sets of fixed effects, I just include
the larger set (county) via the built-in xtpoisson, fe command, and I
include the smaller set of fixed manually: xtpoisson y x
monthfe1-monthfe58, fe vce(robust)]

Thanks,
Garret Christensen
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/