The Medicare Shared Savings Program (MSSP) is a voluntary program that provides incentive payments to healthcare providers that form integrated healthcare organizations, known as Accountable Care Organizations (ACOs). The MSSP rewards ACOs for keeping average per capita spending below a benchmark. I provide reduced-form and model-based evidence that the current benchmark design adversely affects ACOs’ participation and performance in the MSSP. First, the benchmark rebasement, which updates the benchmark over time to reflect observed ACOs’ spending, causes the ratchet effect: ACOs delay spending reductions to avoid lowering future benchmarks. Second, the region-specific adjustment applied to the rebased benchmark induces adverse selection: ACOs that are most likely to enroll are those with initial spending below the benchmark. Moreover, I propose a revised benchmarking methodology to address these inefficiencies. Counterfactual analyses demonstrate that this alternative policy mitigates adverse selection without introducing ratchet effect and significantly increases Medicare savings.

Provider-sponsored health plans (PSHPs) are a form of vertical integration (VI) be- tween health care financing (plans) and health care delivery systems (hospitals). This paper studies the impact of PSHP on 30-day unplanned readmissions risk among Medi- care Advantage beneficiaries and distinguishes between two margins of integration: treatment at an integrated hospital (VI-hospital) and enrollment in an integrated plan (VI-plan). We address both patient selection and supply side selection into VI status. To address patient selection, we adopt an identification strategy the leverages quasi- experimental variation in hospital assignment and plan enrollment. To address hospital selection into VI status, we include hospital and plan fixed effects, and exploit within hospital variation in the VI status over time. We find that in markets where hospital competition is stronger, patients admitted to an integrated hospital-plan setting have 9.5 pp lower readmission risk relative to those admitted to standard hospitals, and have also shorter length of inpatient stay. Identification of the impact of enrollment into a VI plan is weaker since we cannot include plan fixed affect as there is no within plan variation in the VI status. After controlling for the admitting hospital, we do not find evidence that enrollment in a vertically integrated health plan reduces readmission risk. Together, these findings indicate that, in competitive hospital markets, this lower readmission risk operate through hospital-wide care practice changes at an integrated hospital rather than through the enrollment in an integrated plan.

Medicare relies on multiple programs that provide coverage to different groups of beneficiaries. The Medicare Shared Savings Program (MSSP) introduces incentives to reduce unnecessary spending for Traditional Medicare (TM) beneficiaries. Changes induced by the MSSP can spill over to beneficiaries enrolled in Medicare Advantage (MA) and affect MA plans’ outcomes. The MSSP can affect MA through two channels. First, a benchmark channel : because MA plans payments are tied to county-level TM per-capita spending, MSSP-induced changes in county-level TM spending can mechanically affect MA plan payments. Second, a provider channel : because providers treat both TM and MA beneficiaries, MSSP-induced changes in care delivery may directly affect MA enrollees and, in turn, plan expected costs and benefit design. We measure the plan-level exposure to MSSP activity as the enrollment-weighted average of the county MSSP penetration in the counties that are part of the plan service area. We identify the provider channel by controlling for the plan-level payments to net out the mechanical effect that operates through the benchmark channel. We leverage quasi-experimental variation created by the 2019 Pathways to Success reform of the MSSP, which shifted the strength of MSSP incentives. Combining this policy variation with pre-reform plan shares and MSSP providers penetration, we construct a Bartik- style instrument for plan-level exposure to MSSP activity. We find evidence that greater MSSP exposure lowers MA plans’ expected cost of coverage which, in turn, reduces MA payment to plans. In particular, a 1% increase in MSSP exposure lowers plans’ expected costs by 0.34% and lower plans’ payments by about 0.20%. Moreover, part of this lower costs are passed through to enrollees through higher rebates. However, we interpret the estimation on rebates more cautiously as the instrument is correlated with pre-2019 trend in rebates. We also find that plan ownership plays an important role: spillovers are not significant among plans sponsored by health systems that participate in the MSSP.