This paper presents an extension to the Zero Sum Gains DEA model (ZSG-DEA) useful for the cases in which the output reduction (sum of the output is constant) obliges reductions in the inputs values. Instead of displacing the frontier, the DMUs move themselves along the efficient frontier (or iso-efficiency layers). The bidimensional and multidimensional cases are illustrated. For the latter case, due to the complexity of the algorithms that compute all the faces of the efficient frontier, we propose the use of the smoothed DEA frontier model.
DEA; zero sum gains; movement along the efficient frontier; smoothed frontier
