A New Class of BoD Models Accommodating Forward and Reverse Component Indicators

In this paper we present a new class of Benefit-of-Doubt (BoD) models that can accommodate both forward (desirable) and reverse (undesirable or bad) component indicators proposes by relying on different disposability assumptions, namely, weak, constrained weak and extend strong disposability. We first show that, regardless of the orientation chosen, BoD models with constrained weak and extended strong disposability are equivalent to each other. Then we analyse eight models: the first one is a radial input-oriented model with weak disposability; the second and the third are radial forward-output-oriented models with respectively weak an extended strong disposability; the fourth and the fifth are radial reverse-output-oriented models with respectively weak and extended strong disposability; the sixth is a directional output-oriented model with extended strong disposability; and the last two are hyperbolic output-oriented models with respectively weak and extended strong disposability. These are then compared and contrasted, using the same data set, with two relevant models existing in the literature, namely, a radial input-oriented model with extended strong disposability and a directional output-oriented model with weak disposability.