Estimating the cost efficiency of public service providers in the presence of demand uncertainty

Abstract

Public service managers generally make input choices in the face of uncertainty about future demands for service. This is generally not taken into account when estimating cost efficiency. In the operations research literature, for example, the standard approach to estimating cost efficiency is based on the assumption that managers choose inputs to minimise the cost of producing realised (i.e., observed) outputs. However, when outputs are unknown at the time input decisions are made, most managers will instead choose inputs to minimise the cost of producing output targets (e.g., minimum service levels, predicted maximum demands). In this paper, we explain how data envelopment analysis (DEA) estimators can be used to estimate cost, technical and allocative efficiency in these situations. The methodology is applied to Australian data on hospital and health service providers. We obtain estimates of efficiency that are quite different from estimates obtained using a standard approach that ignores uncertainty. Our work has important implications for performance evaluation and improvement programs in many public service settings.