Dear Statalist community,

1) After running every required test (F-test, Breusch Pagan, Hausman) I determine that FE Model is the most suitable one for my panel regression. Data description: N (120) > T (3)

2) Nonetheless, diagnostic tests for panel data detect that (a) heteroscedasticity, (b) autocorrelation, and (c) cross-sectional dependency are present.

While I have studied the most recent literature (e.g. Hoechle (2007)) and a number of threads on the Stata Forum, I am not quite sure yet which robust standard error to choose that solves all the issues from the diagnostic tests (Heteroscedasticity, autocorrelation, dependency).

Do I use or -cluster()- / -vce(robust)-?

Or any other such as -xtgls-, -xtpcse- or -xtscc-?

I am looking forward to your response.

Much appreciated and thank you in advance