A proxy approach to dealing with the infeasibility problem in super-efficiency data envelopment analysis

Super-efficiency data envelopment analysis (SE-DEA) models are expressions of the traditional DEA models featuring the exclusion of the unit under evaluation from the reference set. The SE-DEA models have been applied in various cases such as sensitivity and stability analysis, measurement of productivity changes，outliers’ identification，and classification and ranking of decision making units (DMUs). A major deficiency in the SE-DEA models is their infeasibility in determining super-efficiency scores for some efficient DMUs when variable, non-increasing and non-decreasing returns to scale (VRS, NIRS, NDRS) prevail. The scope of this study is the development of an oriented proxy approach for SE-DEA models in order to tackle the infeasibility problem. The proxy introduced to the SE-DEA models replaces the original infeasible DMU in the sample and guarantees a feasible optimal solution. The proxy approach yields the same scores as the traditional SE-DEA models to the feasible DMUs.