Risk Function of Zellner's Extended Melo Estimators and Some Monte Carlo Results

In this paper we theoretically derive the risk of Zellner's extended minimum expected loss function estimator. Using artificial data, we then calculate the risks of known nested estimators that include simple minimum expected loss function, two stage least squares and ordinary least squares. The resulting rankings of these estimators are compared to those generated via a simple Monte Carlo experiment. Both the analytical and the Monte Carlo results suggest that the extended minimum expected loss function estimator performs remarkably well when there is a positive degree of simlutaneity in the system of equations