Probabilistically constrained models for efficiency and dominance in DEA

This paper proposes a stochastic model for data envelopment analysis (DEA), based on the theory of joint probabilistic constraints, which can be used with general multivariate distribution functions. The key assumption is that the random variables representative of the uncertain data follow a discrete distribution or that a discrete approximation of continuous distribution is available. Under this assumption, mixed integer linear models are formulated to tackle, rather originally, dependencies among DMUs inputs, outputs and inputs-outputs through the theory of joint probabilistic constraints. The features of the model are illustrated through an application for the performance evaluation of screening units.