Compensating for non-homogeneity in decision-making units in data envelopment analysis

Abstract Data envelopment analysis (DEA) assumes homogeneity among the decision-making units (DMU) in terms of the nature of the operations they perform, the measures of their efficiency, and the conditions under which they operate. When the DMU are not homogeneous, the efficiency scores may reflect the underlying differences in environments rather than any inefficiencies. One strategy to overcome this is to separate DMU into homogeneous groups. However, one needs large numbers of DMU to do this. Another strategy is to adjust for non-homogeneity. This paper presents three adjustment techniques along with the basic Charnes, Cooper, and Rhodes (CCR) model. We do two simulation experiments where we know the underlying efficiencies of the DMU and the parameters of the non-homogeneities. We also compare the four models using actual data from a DEA evaluation of municipal reverse logistics channels. The results show none of the adjustment mechanisms are clearly superior to the unadjusted CCR model. Consequently, what makes a good adjustment mechanism is open at this time.