Portion Reduction procedure in the Two-Stage Network DEA

Abstract

Data Envelopment Analysis (DEA) is an improved technique for optimization and performance evaluation of peer decision-making units (DMU) s especially in real-world network systems for instance. In a two-stage network structure, undesirable factors are jointly produced with desirable ones, either as the final outputs or as the intermediate products. An appropriate selection of the DEA-based model has a significant role in solving the optimization problems parallel to conducting the undesirable factors. On the other hand, any increment in the desirable output applies the equivalent change with the undesirable outputs. Likewise, it is not possible to reduce the production of undesirable outputs without any cost. In the presence of the additional inputs in the second stage, in a two-stage structure as an example, this approach searches the proportional abatement in desirable and undesirable factors. By keeping input factors invariant and selecting the leader-follower structure the scenario of this paper is fair treatment of the undesirable outputs, besides the less sacrifice of the desirable. In addition, the theoretical contributions of the proposed model have been illustrated in two examples which are13 poultry industry and 25 power plants in Iran.