st: RE: -nbreg, robust- not displaying any standard errors

["Lachenbruch, Peter" <Peter.Lachenbruch@oregonstate.edu>]

It could also be that some of your variables are not always recorded.
When I've had problems like this, I go on a hunting expedition dropping
a variable to see when it becomes non-singular.  Not much help I
suppose.

Tony

Peter A. Lachenbruch
Department of Public Health
Oregon State University
Corvallis, OR 97330
Phone: 541-737-3832
FAX: 541-737-4001


-----Original Message-----
From: owner-statalist@hsphsun2.harvard.edu
[mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Brian
Karfunkel
Sent: Monday, October 20, 2008 1:54 PM
To: statalist@hsphsun2.harvard.edu
Subject: st: -nbreg, robust- not displaying any standard errors

Hello,

I am using a dataset of workplace injuries per month for a panel of
different facilities over time, trying to estimate the effect of
certain characteristics of the facility on the monthly count of
injuries. The characteristic I am focusing on is a certain policy
change at the company, and I use a dummy variable, new_policy, which
is 1 for facilities where the new policy has been implemented and 0
otherwise. I also am using period (i.e. month) dummies.

I am using a negative binomial model of the form:

nbreg injuries new_policy ln_hours_worked _I_period*, cluster(facility)
And I get the following message:

Warning:  variance matrix is nonsymmetric or highly singular

And though coefficient estimates are returned, there are no standard
errors (they're all "."). I try doing unclustered, robust standard
errors and get the same message. Only when using non-robust standard
errors are the standard errors displayed.