Alternative measures of environmental technology structure in DEA: An application

The nonparametric data envelopment analysis (DEA) literature on environmental efficiency (EE) considers handling undesirable outputs in two alternative ways: either in their original forms with the assumption that these are weakly disposable or in various translated forms with the assumption that these are strongly disposable. Choosing a particular approach implies adoption of a particular, distinct treatment of undesirable outputs, and hence yields a distinct set of EE estimates. To explore the effects of the interplay between choice of EE measure and specific treatment of undesirable outputs, this paper attempts to generate all possible output-oriented EE measures based on these two alternative approaches. Furthermore, guided by the argument that slacks are important in identifying properly the efficiency behavior of firms, it proposes two new alternative, slacks-based formulations of EE: one based on the range directional model, and the other on the generalized proportional distance function model. Using a confected data set of ten firms and a real-life data set of 22 OECD countries, our empirical analysis reveals that: first, EE scores are influenced not only by the choice of disposability assumption for undesirable outputs but also by the way these are treated in various translated forms; second, the choice of any particular treatment of undesirable outputs plays no role in influencing the rankings of firms; and third, our two new alternative EE formulations are, at the least, viable alternatives to existing EE measures in ranking firms according to their eco-efficiency behavior.