Performance evaluation with the entropy-based weighted Russell measure in data envelopment analysis

Conventional Data envelopment analysis is based on the Debreu-Farrell optimal solution in evaluating the decision-making unit's efficiency. Even if Farrell efficiency is achieved, there may exist slacks in individual input or output. To solve this problem, the Russell measure can be used to address the inherent shortcomings of the Farrell measure and devise an optimal solution for the Pareto-Koopmans concept of efficiency. However, the non-proportional radial measure may lead to distorted efficiency measurement of inefficient decision-making units, due to assumptions about its implicit importance of inputs and outputs. Therefore, this paper uses a simple method in calculating the weighting measurements, in order to override this assumption using the concept of entropy. By introducing entropy, the Russell measure easily uses input, output, and system weighting to evaluate performance. In addition, we also use the entropy-concepts applied to slack-based measure. Moreover, we illustrate this entropy-based Russell measure using data gathered from 24 of Taiwan's commercial banks in order to rank and compare it with the conventional Russell measure.