Controlling for time vs. clustered/robust SE

[M Mullahey]

Conducting a regression on panel data using xtreg... can someone help me out with the theory behind clustered SEs as relates to controlling for time? It's country/year data and I had originally included year as a variable to account for the effects of time. Would it ALSO be advisable to cluster SEs or is that essentially duplicating the same concept and therefore it would be best to do one or the other? thanks.

[Clyde Schechter]

The use of cluster-robust standard errors does not adjust for time effects, nor does adjustment for time effects deal with within-country non-independence of options or heteroscedasticity. In fact, it's hard for me to think of what either has to do with the other. In any case, each should be used when it is (independently of the other) appropriate to the data. Depending on circumstances one might use either, neither, or both.