Measuring hospital performance. A non-parametric approach.

In this paper a technique for assessing the relative performance of firms is introduced and applied to a sample of hospitals in California. Hospitals are compared on the basis of their relative technical efficiency. The reference technology is constructed from observed outputs and inputs (including physicians) using programming techniques, and efficiency is assessed relative to the frontier of the technology using measures similar to those suggested by Farrell. The technique used here imposes no prespecified functional form, allows for multiple outputs and inputs (allowing for differences in case mix), and yields information on the productive performance of individual hospitals. This technique can also be used to determine whether different types of hospitals use different technologies. Our results, although illustrative rather than definitive, suggest that ownership affects 'efficiency' as measured here: public and NFP hospitals have 'different' best practice frontiers, and public hospitals appear to use relatively fewer resources. These results could reflect differences in quality of care by ownership.