Robust estimation under error cross section dependence

•Propose a variance estimator for fixed effects and mean group estimators in panels.•Robust to various forms of serial and cross sectional dependence in errors.•Useful in applied work for large N short T panels when little is known about the process generating the errors.•Shown to be consistent for N going to infinity, with T fixed.

We propose a robust, partial sample estimator for the covariance matrix of the fixed effects and mean group estimators of the slope coefficients in a short T  panel data model with group-specific effects and errors that are weakly cross sectionally dependent and serially correlated.