Analyzing the concept of super-efficiency in data envelopment analysis: A directional distance function approach (1407.2599v1)

Abstract: Based on the framework of the directional distance function, we conduct a systematic analysis on the measurement of super-efficiency in order to achieve two main objectives. Our primary purpose is developing two generalized directional measures of super-efficiency that completely resolve the crucial infeasibility issue, commonly arisen in the traditional super-efficiency measures. The secondary goal is to demonstrate that our directional super-efficiency models encompass the conventional ones as special cases. The proposed measures are advantageous because they circumvent biases in super-efficiency estimation due to input and output slacks. They are general, and satisfy several desirable properties, such as always feasibility, monotonicity, unit independence, and translation invariance.