On the Use of DEA Models with Weight Restrictions for Benchmarking and Target Setting

This chapter discusses the use of DEA models with weight restrictions for purposes of benchmarking and target setting. Weight restrictions have been used in the literature to incorporate into the analysis both value judgments (managerial preferences or prior views about the relative worth of inputs and outputs) and technological judgments (assumptions on production trade-offs). An important consideration in the use of restricted models for the benchmarking is that they may provide targets that are outside the production possibility set (PPS). Such difficulties are overcome if weight restrictions represent production trade-offs, because in those cases restricted models lead to a meaningful expansion of the production technology. However, if weight restrictions are only used as a way of incorporating preferences or value judgments, then there could be no reason to consider the targets derived from those models as attainable. Despite the classical restricted DEA formulations may yield targets within the PPS, it is claimed here that an approach based on a more appropriate selection of benchmarks would be desirable. We develop some restricted models which provide the closest targets within the PPS that are Pareto-efficient. Thus, if weight restrictions represent value judgments, the proposed approach allows us to identify best practices which show the easiest way for improvement and are desirable (in the light of prior knowledge and expert opinion) in addition to technically achievable.