Objective: To discuss the usefulness of efficiency measures as instruments of monitoring and resource allocation by analyzing their invariance to changes in the operationalization of hospital production.
Study setting: Norwegian hospitals over the three-year period 1989-1991.
Study design: Efficiency is measured using Data Envelopment Analysis (DEA). The distribution of efficiency and the ranking of hospitals is compared across models using various distribution-free tests.
Data collection: Input and output data are collected by the Norwegian Central Bureau of Statistics.
Principal findings: The distribution of efficiency is found to be unaffected by changes in the specification of hospital output. Both the ranking of hospitals and the scale properties of the technology, however, are found to depend on the choice of output specification.
Conclusion: Extreme care should be taken before resource allocation is based on DEA-type efficiency measures alone. Both the identification of efficient and inefficient hospitals and the cardinal measure of inefficiency will depend on the specification of output. Since the scale properties of the technology also vary with the specification of output, the search for an optimal hospital size may be futile.