st: Do I need to cluster errors?

["Jim O'Grady" <jimogr@gmail.com>]

Some help would be greatly appreciated.

I have a specification in which I'm interested in the effect of a
country level time invariant characteristic (X) on a country level
outcome which is time variant across years (Y).

So each country has the same X in every year.

I'm using year fixed effects and robust standard errors.

Do I need to cluster the errors? And if so do I need to do it by year
or by country.

I know what if I had observations of something within the country all
getting the same country level treatment errors would be correlated
among them and  I would cluster by country. But here the country is
getting the same treatment in different years so not sure what
applies?

Thanks,

J.
*
*   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/