Closest target for the orientation-free context-dependent DEA under variable returns to scale

An important branch of data envelopment analysis (DEA) is context-dependent DEA, which evaluates efficiency by combining the attractiveness and progress for a particular decision-making unit (DMU). Traditionally, context-dependent DEA models are based on the assumption of constant returns to scale. Two limitations are found when directly extending original radial context-dependent DEA (ORCD-DEA) models into variable returns to scale versions. One is that it may not be possible to determine the attractiveness of a DMU that logically must be attractive in that context. The other problem is that the progress measure cannot ensure an inefficient DMU projects to a Pareto-efficient frontier. A small numerical example is used to illustrate these two issues. In order to overcome these deficiencies, the concept of closest target is introduced to determine the attractiveness and progress for each DMU. The closest target method can further improve DMUs' performance with less wastes in inputs or underproduction in outputs. Finally, a practical application involving computer printers is presented.