I tried to perform a xttest2, and this was the output:

Error: too few common observations across panel.
no observations

I have about 20 observations for each year (for 4 years).

So I performed a reg controlling for the time variable and using robust estimator. I need to test if the errors are correlated,

Any help?