This article seeks to explain the productivity differences observed in France between public hospitals, private non-profit hospitals (French acronym: PSPH), and private for-profit clinics. We examine whether the composition of the patient population and the types of stays influence hospital productivity. If so, the introduction of competition between hospitals via procedure-based invoicing (French acronym: T2A) not only promotes efficiency but also creates strong pressures within hospitals to redirect healthcare supply. Our database consists of a nearly exhaustive panel of 1,604 hospitals offering acute care. Our observation period, 1998-2003, covers the six years prior to the introduction of T2A in France. This enables us to examine the situation before the implementation of new incentives. We show that the assessment of public hospitals' productive efficiency depends on the definition of the production frontier. With a classic function, public hospitals' efficiency scores are lower than those of PSPHs, which in turn are lower than those of private clinics. But the performance ranking is reversed when we factor in the characteristics of the patient population and the composition of hospital stays. On this criterion, with the exception of small establishments, public hospitals and PSPHs outperform private clinics. These results should be interpreted in the light of the different terms of reference applying to the activities of public and private hospitals. A final breakdown shows that the lower productivity of public hospitals is mainly due to their size, the composition of their patient population, and that of their stays, characterized by a small proportion of surgery stays. It is not due to a lesser efficiency of public hospitals.