DEA-Based Incentive Regimes in Health-Care Provision

A major challenge to legislators, insurance providers and municipalities will be how to manage the reimbursement of health-care on partially open markets under increasing fiscal pressure and an aging population. Although efficiency theoretically can be obtained by private solutions using fixed-payment schemes, the informational rents and production distortions may limit their implementation. The healthcare agency problem is characterized by (i) a complex multi-input multi-output technology, (ii) information uncertainty and asymmetry, and (iii) fuzzy social preferences. First, the technology, inherently nonlinear and with externalities between factors, yield parametric estimation difficult. However, the flexible production structure in Data Envelopment Analysis (DEA) offers a solution that allows for the gradual and successive refinement of potentially nonconvex technologies. Second, the information structure of healthcare suggests a context of considerable asymmetric information and considerable uncertainty about the underlying technology, but limited uncertainty or noise in the registration of the outcome. Again, we shall argue that the DEA dynamic yardsticks (Bogetoft, 1994, 1997, Agrell and Bogetoft, 2001) are suitable for such contexts. A third important characteristic of the health sector is the somewhat fuzzy social priorities and the numerous potential conflicts between the stakeholders in the health system. Social preferences are likely dynamic and contingent on the disclosed information. Similarly, there are several potential hidden action (moral hazard) and hidden information (adverse selection) conflicts between the different agents in the health system. The flexible and transparent response to preferential ambiguity is one of the strongest justifications for a DEA-approach. DEA yardstick regimes have been successfully implemented in other sectors (electricity distribution) and we present an operalization of the power-parameter p in an pseudo-competitive setting that both limits the informational rents and incites the truthful revelation of information. Recent work (Agrell and Bogetoft, 2002) on strategic implementation of DEA yardsticks is commented in the healthcare context, where social priorities change the tradeoff between the motivation and coordination functions of the yardstick. The paper is closed with policy recommendations and some areas of further work.