ABSTRACT Payment to providers and plans in Medicare has far-reaching implications for the health of the elderly and disabled. The fee-for-service basis for payment has been rightly criticized for incentivizing excessive and potentially harmful provision of care and for interfering with the production of health by linking reimbursement to specific inputs. To establish greater flexibility and accountability, Medicare has increasingly transferred financial risk to entities that can manage care—first to private plans in the Medicare Advantage (MA) program and more recently to providers directly in alternative payment models (APMs) in traditional Medicare (TM). In both models, additional payment adjustments are tied to performance on quality measures. These approaches to integrating the financing and delivery of care create opportunities to improve patient care, but empirical study has been challenging. Evidence on the merits of the MA program and the relative performance of MA and APMs has been largely observational and limited in analysis of patient experiences and health outcomes. In addition, risk contracting relies heavily on risk adjustment to profile plan or provider performance and to align resources with the needs of patient populations, but the adequacy of risk adjustment has not been well described. This project will make substantial contributions by leveraging natural experiments and new data to understand plan and provider responses to payment incentives and their implications for patients and payment system design. The project includes three aims. First, we will use natural experiments to compare performance between MA and TM and different variants of each system. Sources of quasi-randomization to MA vs. TM will include differences across state and county borders in MA exposure and transitions from Medicaid to dual eligibility at age 65 in areas of high or low MA exposure. We will explore effect heterogeneity based on geographic variation in the configuration of MA (e.g., by plan type) and TM (e.g., by APM activity). Second, we will assess the marginal value of post-acute care in skilled nursing facilities (SNFs) and effects of incentives to alter post-acute length of stay on patient outcomes in different payment models, including TM, MA, and APMs. Post-acute care is an area of intense activity in MA and APMs that offers insights into provider agency. Approaches will make use of discontinuities in SNF reimbursement and coinsurance related to length of stay and natural randomization of patients to physical therapists in hospitals. Third, we will determine the extent to which risk-adjusted MA-TM differences in outcomes reflect performance differences vs. residual risk selection and assess the adequacy of risk-adjustment methods. We will use strategies from our first aim to compare MA- TM performance differences in self-sorted vs. quasi-randomized populations. Findings will characterize the prospects for value-enhancing competition or unint...