Abstract:
Social service providers such as teachers, healthcare providers and homeless shelters receive trillions of dollars each year to help people. Recently, policymakers and other funders have attempted to obtain better outcomes by implementing performance-based incentive schemes that pay more money to higher performing providers. In this paper, I develop a simple model of this incentive design problem with a distinguishing feature – providers compete for users to help by adjusting service quality. I characterize a broad class of incentive schemes that elicit efficient service quality, and I show that popular incentive schemes (including value-added and pay-for-percentile schemes) are generally suboptimal and can have perverse distributional consequences, even when all characteristics of individuals are observed. I discuss implications for performance-based incentives in education and healthcare.