Random output and hospital performance.

Abstract

Many countries are under pressure to reform health care financing and delivery. Hospital care is one part of the health system that is under scrutiny. Private management initiatives are a possible way to increase efficiency in health care delivery. This motivates the interest in developing methodologies to assess hospital performance, recognizing hospitals as a different sort of firm. We present a simple way to describe hospital production: hospital output as a change in the distribution of survival probabilities. This output definition allows us to separate hospital production from patients' characteristics. The notion of "better performance" has a precise meaning: (first-order) stochastic dominance of a distribution of survival probabilities over another distribution. As an illustration, we compare, for an important DRG, private and public management and find that private management performs better, mainly in the range of high-survival probabilities. The measured performance difference cannot be attributed to input prices or to economies of scale and/or scope. It reflects pure technological and organisational differences.