11. Measuring Environmental Efficiency: An Application to U.S. Electric Utilities

This chapter highlights limitations of some DEA (data envelopment analysis) environmental efficiency models, including directional distance function and radial efficiency models, under weak disposability assumption and various return-to-scale technology. It is found that (1) these models are not monotonic in undesirable outputs (i.e., a firm’s efficiency score may increase when polluting more, and vice versa), (2) strongly dominated firms may appear efficient, and (3) some firms’ projection points derived from the optimal environmental efficiency scores are strongly dominated, thus they cannot be the right direction for the improvement. To address these problems, we propose a weighted additive model, i.e., the Median Adjusted Measure (MAM) model. An application to measuring the environmental efficiency of 94 U.S. electric utilities is presented to illustrate the problems and to compare the existing models with our MAM model. The empirical results show that the directional distance function and radial efficiency models may generate spurious efficiency estimates, and thus it must be with caution.