The Objective Setting in DEA Models with a Dual-Role Factor: Measuring the Radial Efficiency

Abstract

Data Envelopment Analysis (DEA) modeling is challenging when a factor’s role can be considered as both an input and an output simultaneously, termed the dual-role factor. This paper investigates the objective setting in DEA models with a dual-role factor, which can be subjective and heavily relies on domain expertise and personal belief. We focus on the movement of the dual-role factor to decide whether to move together with the inputs or the outputs on measuring radial efficiency, particularly providing a feedback (validating) mechanism for DEA modeling under these circumstances. We develop an analytical framework based on axiomatic production theory, and bridges intuitive practical sense, i.e., the benchmark suggested by the model should be consistent with the model’s objective setting. We also propose a model that can adjust the possible misspecification of the movement.