A data envelopment analysis approach to clustering operating units for resource allocation purposes

Bodies such as those funding education or health services normally make use of marginal resource levels (e.g. the level of funding per pupil) to allocate resources to operating units. It is typically desired that marginal resource levels should encourage the operating units to use resources as efficiently as possible. This paper develops a method for simultaneously clustering operating units and determining a different set of marginal resource levels for resourcing the units of each cluster, consistent with the foregoing aim. The user controls the number of clusters created, striking a balance between the multiplicity of sets of marginal resource levels estimated and the strength of the incentive offered to operating units to use a resource efficiently. The clusters identified can prove useful in themselves in that operating units sharing similar environmental conditions and/or value systems are identified.