Physician Responses to Nonlinear Contracts

Policymakers have increasingly focused on the design of provider contracts to reduce health care costs and increase care quality. Many of these contracts provide bonus payments to providers contingent on meeting externally set performance threshold levels. Using data from a large insurer in Hawaii, this paper estimates physician responsiveness to two features of these contracts: 1) threshold level and 2) bonus amount. This data is unique in its size: it encompasses half of Hawaii’s population and almost 70% of its commercial market. I estimate provider performance response for individual measures using a large discrete change in threshold level and bonus amount during the sample period. I also estimate a pooled provider performance response across all measures using two instrumental variables. I find that a one percentage point increase in threshold location leads to a 0.3 to 0.5 percentage point increase in performance the subsequent quarter. I do not detect an average response to bonus size. I find heterogeneous responses based on prior performance: low performing physicians were more responsive to threshold level, and high performing physicians were responsive to bonus amount. My results demonstrate that the bonus amount has little effect on provider effort and incentivizes already high-performing physicians. Small increases in threshold levels improves performance without increasing cost. These results have implications for innovations in physician payment models and contract designs.