Data Envelopment Analysis with Robust and Closest Targets:Integrating Full-Dimensional Efficient Facets for Risk-Resilient Benchmarking

Abstract

As the external environment become increasingly volatile and unpredictable, the selection of benchmarking targets in data envelopment analysis should account for their ability to consider risks; however, this aspect has not received sufficient attention. We propose a robust benchmarking target defined by the intersection of the maximum number of full-dimensional efficient facets, each representing a unique marginal substitution relationship. These targets can serve as robust projections for decision making units that are lacking prior risk information because they incorporate the maximum number of marginal substitution relationships. This enables decision makers to adjust their production through these relationships, thereby maximizing the likelihood of achieving globally optimal outcomes. Furthermore, we propose a novel, well-defined efficiency measure based on robust and closest targets. Finally, we demonstrate the application of the proposed measure using a dataset comprising 38 universities from China's 985 Project.