Combining Directional Distances and ELECTRE Multicriteria Decision Analysis for Preferable Assessments of Efficiency

Traditional nonparametric frontier models used to asses technical, allocative, cost, and scale efficiencies, based on Data Envelopment Analysis (DEA), reflect not only the most favorable way of weighing outputs over inputs but also tradeoffs of compensations among the many production possibilities. When managers or policy makers have an explicit preference over some production resources or products, such tradeoffs and the resulting estimated efficiency measure may not represent the most appropriate scenario of evaluation. The good performance of some decision units on some production variables may offset the bad performance on others, and this may be sufficient to qualify such decision units as efficient (or less inefficient) in most DEA rankings, but not under the subjective perspective of the decision maker. This is particularly important in the case of non-discretionary inputs, bad outputs, or less desirable production configurations. In this chapter, we discuss this issue offering a perspective on how we can advance in this avenue by developing multicriteria non-compensatory directions for the expansion of outputs or contraction of inputs. A numerical example is reported at the end of this discussion.