Simulation of nonparametric efficiency shows that even when the number of firms is large, defining ten or more inputs results in most firms being efficient. Comparison of empirical with simulated results suggests that the dimension of the problem rather than actual efficiencies has the greater effect for some empirical results.
