Planning improvements through data envelopment analysis (DEA) benchmarking based on a selection of peers

Abstract

Incorporating preferences on suitable peers into benchmarking analyses may ensure the setting of appropriate targets, which enable designing plans for improving performance that are aligned with management. This paper deals with target setting in situations where decision makers (DMs) have previously made a selection of peer candidates for the benchmarking of a given organization. A first approach is developed within the framework of conventional Data Envelopment Analysis (DEA), which is the technology mostly used in non-parametric frontier analysis. It provides targets from reference sets consisting of peer candidates that span a face of the strong efficient frontier of the production possibility set (PPS). These targets result from solving a DEA-like model, thus preventing from the need to identify all of the maximal efficient faces (MEFs) of the DEA frontier. We also propose a second approach where the convexity in DEA is somehow relaxed to allow additionally for reference sets consisting of candidates that are Pareto-efficient, provided that their convex hull is not dominated by other units. In that sense, the targets found can be seen as representing best practices. This approach broadens the range of alternatives when planning improvements, and may eventually provide closer targets.