Measuring group performance using stochastic frontier analysis

Abstract

In this paper, a global out-line frontier model is proposed to compare each group’s frontier as well as each group’s efficiency average. The suggested model is also compared to the metafrontier model. Instead of estimating the global envelope, which encompasses all groups’ frontiers, the outline of all frontiers is used as a reference frontier set. Unlike the metafrontier, the outline frontier contains no infeasible area. A new concept index, called frontier share, can be calculated using the proposed method. Frontier share is the percentage of best practice or highest parts in a frontier compared to the other groups’ frontiers. Frontier share can show the relative frontier positions of various groups. With a sufficient number of entities in each group, the frontier share can be a good comparison index for a frontier’s relative position. To extend the model, some environmental variables are included in the global out-line frontier. However, owing to the limit of the estimation step, i.e., that each group has its own frontier parameters, the environmental variables cannot be compared to other groups’ environmental variable levels. Thus, the variables should be used in mean deviation form to express the level changes within the group. The global out-line frontier suggested here can compare each group’s frontier as well as each group’s efficiency average. Empirical results for five European countries using the basic global outline frontier model show that the German banking industry’s frontier is the most efficient. Germany’s frontier share ranks first, with a value of 73 percent, and that of the UK ranks next, with 17 percent. The global out-line frontier seems to be shaped mostly by German and UK banks. However, some German banks in frontier share area are considerably inefficient. High return on assets (ROA) and return on equity (ROE) values for the UK and the high standard deviation of German ROA and ROE values may explain these results. With many banks w...