Models and measures for efficiency dominance in DEA - Part I: Additive models and MED measures

Abstract

The usual models in DEA (Data Envelopment Analysis) employ a postulate of continuity to obtain comparison points for the entities known as DMUs (Decision Making Units) whose input-output behavior is to be evaluated. In some applications, it may be desired to restrict attention to actual DMUs and hence to drop (or modify) the continuity assumptions in DEA. Using the concept of efficiency dominance, this is accomplished in the present paper in the form of mixed integer programming models which restrict the efficiency evaluations to comparisons with actually observed performances. Simple and easily interpreted scalar measures of efficiency are provided while retaining the ability to identify the sources and amounts of inefficiency in each DMU that is evaluated.