a question for panel data

[deleted_user]

I am using Proc panel to regress an unbalanced panel data. I wonder before doing random effect vs fixed effect choice, do I need check heteroskedasticity and autocorrelation? If I need test them, which options in proc panel can help to correct them? Some people say proc panel can automatically give the best estimate. Is it true? Thank you.