Introduction

One of the main objectives of DEA (Data Envelopment Analysis) is to measure the efficiency of a DMU (Decision Making Unit, e.g. school, public agencies and banks). One of the ways for determining efficiency score of DMUs is to apply the Charnes, Cooper, Rhodes (CCR) model [1] that deals with a ratio of multiple outputs and inputs. One of the interesting research subjects is to discriminate between efficient DMUs. Several authors have proposed methods for ranking the best performers ([2–۷] among others).

For a review of ranking methods, readers are refereed to Adler et al. [8]. In some cases, the models purposed by [3,6] can be infeasible. In addition to this difficulty, the Andersen and Petersen [3] model may be unstable because of extreme sensitivity to small variations in the data when some DMUs have relatively small values for some of their inputs.

The objective of this work is to propose a new ranking system for extreme efficient DMUs based on the work of Hibiki and Sueyoshi [2]. Our approach does not have the difficulties arising from Andersen and Petersen [3] and Mehrabian et al. [6] models. The main methodological difference of our model in relation to the one of Hibiki and Sueyoshi [2] is that while their approach suggests a measure of efficiency for extreme efficient DMU named DSS (DEA selfsensitivity) which is not dependent of the inefficient DMUs, our methodology proposes a measure that is totally dependent of the inefficient DMUs.