Pay-for-Performance and Selective Referral in Long-Term Care

We examine how pay-for-performance (P4P) affects long-term care (LTC), exploiting a natural experiment in Japan. Matched user/care manager/provider data are used to observe care managers' referral decisions. Care managers/providers can vertically integrate, and P4P creates new incentives for selective referrals. Overall, we found no robust evidence that P4P improves LTC outcomes. However, after P4P, LTC outcomes improved more when care managers referred users to affiliated providers than to non-affiliated providers. Moreover, care managers referred users whose care levels were more likely to improve to affiliated providers. Selective referrals are apparently explained by vertical integration and a lack of risk adjustment