We employ a wide range of parametric and non–parametric cost frontiers’ efficiency estimation methods to estimate economic efficiency and economies of scale, using the same panel data of 22 Taiwanese commercial banks over the period 1982–97. According to our empirical implementation, the two methodologies yield similar average efficiency estimates, yet they come to very dissimilar results pertaining to the efficiency rankings, the stability of measured efficiency over time, the consistency between frontier efficiency and conventional performance measures, and the estimates of scale economies. Thus, the choice of an estimation approach can result in very different conclusions and policy implications regarding cost efficiencies and cost economies. These findings suggest that making policy decisions and evaluations relies on multiple techniques and specifications.