Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
| From | Austin Nichols <austinnichols@gmail.com> |
| To | statalist@hsphsun2.harvard.edu |
| Subject | Re: st: Clustered standard errors with panel-specific trends |
| Date | Tue, 28 Aug 2012 09:55:50 -0400 |
Jing Xing <Jing.Xing@sbs.ox.ac.uk>: You cannot estimate SEs on cluster-specific coefficients such as fixed effects or cluster-specific time trends with a cluster-robust SE. See e.g. http://www.stata.com/meeting/13uk/nichols_crse.pdf and refs therein. On Tue, Aug 28, 2012 at 9:05 AM, Jing Xing <Jing.Xing@sbs.ox.ac.uk> wrote: > Hi all, > > I am trying to estimate a simple linear regression model using a panel data of around 80 countries. I included country-specific time trends in the estimations, and I also clustered the standard errors for each country. I noticed that once I cluster the standard errors, the standard errors of those country-specific linear time trends become extremely small and essentially zero. As a result, the t-statistics become very large. I find these results suspicious and puzzling. If I only calculate the robust (but non-clustered) standard errors on these trends, I obtain reasonable standard errors. Can anyone suggest me why this is the case? Should I cluster the standard errors if I include country-specific time trends in the model? > > Thank you very much. > > Best wishes, > Cindy Xing * * 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/