Project Summary Payment and delivery system reform receives significant and increasing attention from health care payers, providers, and the government. In the pursuit of lowering spending growth and improving quality of care, new payment models are expanding and incentivizing changes in care delivery. Population-based risk-sharing models, or Accountable Care Organizations (ACO), are one of these major reforms and brings groups of providers together to share financial and clinical responsibility for patient populations. While intended to incentivize higher quality and more efficient health care delivery, these payment models also create new opportunities for strategic behavior that serves providers’ financial self-interest but may not support these goals. This study hopes to shed light on how participating providers and organizations may be responding to incentives in unintended ways and whether they are able to leverage information on the quality and spending of their own and surrounding providers. Insight into such sophisticated organizational behavior may have important implications for the success of future payment reform and competition in provider markets, which aligns with Agency for Healthcare Research and Quality priorities. Focusing on the Medicare Shared Savings Program (MSSP), one of the largest ACO models in the country, this study investigates whether ACOs respond strategically to certain MSSP payment rules and will have three aims: (1) study health system ACOs and referral patterns, (2) study selective ACO participation and reconfiguration favoring lower spending providers, and (3) study selective ACO reconfiguration favoring higher quality providers. The study will primarily rely on MSSP ACO Provider and Beneficiary Research Identifiable Files, administrative Medicare data, a random 20% sample of Medicare claims, and a new, comprehensive, national-level dataset on U.S. health systems. To study Aim 1, I will employ an event study interrupted timeseries analysis comparing referral patterns before and after non-system practices begin ACO contracts with health systems. I then compare this single group trend to several different control groups. To study Aims 2 and 3, I use non-ACO counterfactuals in an evaluation-based approach that estimates ACO spending and quality effects with and without fixing ACO’s provider practice configuration to determine the extent to which practice selection affects estimated effects. I follow this approach with a separate flow analysis that compares practices exiting or entering ACOs with those that remain to determine whether baseline spending and quality is associated with the likelihood of eventual exit from or entry into an ACO.