Panel data characterized by groupwise heteroscedasticity, cross-sectional correlation, and AR(1) serial correlation pose problems for econometric analyses. It is well known that the asymptotically efficient, Feasible Generalized Least Squares (FGLS) estimator (Parks) sometimes performs poorly in...