Dynamic Incentive Design in the Medicare Shared Savings Program

Abstract

This thesis designs the Medicare Shared Savings Program (MSSP) in a dynamic context. The MSSP is the largest healthcare spending-reduction program in the U.S. The program incentivizes providers to improve efficiency by paying bonuses if they have reduced their spending. Providers create an Accountable Care Organization (ACO) to participate in the program, for cost accountability and care coordination. The MSSP has been plagued by modest progress and selective participation since its inception. Static contract design ignores ACOs’ evolving capabilities, which determine their sustainable efficiencies, can be improved by continuous investments, but are ACOs’ private information. This thesis considers ACO’s private information (adverse selection), unobservable investments, and uncertain spending (moral hazard) in a dynamic environment. Overall, it sheds light on the scenarios in which the current shared-savings payment format performs well and guides the policy refinement.

We model the ACO’s capability evolving process, which is stochastic and involves costly investments. When ACOs’ heterogeneity only exists in their capabilities but not in their investment costs, the optimal policy takes a shared-savings format. Theoretically, we unravel the fundamental roles of shared-savings payment parameters in providing incentives. The optimal policy (i) elicits capability information through ACOs’ self-selection of tracks and proper benchmarks, which adapt to ACOs’ capabilities and prevent selective participation; (ii) sets proper sharing rates to incentivize continuous investments, so that high-capability ACOs are also willing to invest for higher future efficiency gain. Importantly, the benchmark rule resolves the infamous ratchet effect caused by benchmark rebasing, which inhibits the program’s effectiveness.

We use data released by CMS to calibrate our model through inverse optimization and maximum likelihood estimation. We further investigate whether the optimality of the shared-savings policy is robust with more complex ACO heterogeneity. When ACOs differ in both their capabilities and their investment costs, moral hazard cannot be eliminated by the linear shared-savings policy, and the optimal policy requires a large quadratic penalty on the ACO’s spending deviated from the spending target.

Date of Award	16 Aug 2024
Original language	English
Awarding Institution	City University of Hong Kong
Supervisor	Youhua Frank CHEN (Supervisor)