Advancing the measurement of favorable selection in Medicare Advantage: evidence from encounter data.

Introduction: Debate persists over how favorable selection in Medicare Advantage (MA) affects federal spending relative to Traditional Medicare (TM). Prior studies, including the Medicare Payment Advisory Commission's (MedPAC's) March 2025 Report, focus on TM-to-MA switchers-roughly one-third of MA enrollees-but extrapolating these findings to the broader MA population is sensitive to identifying assumptions because switchers' cost patterns may differ from non-switchers. This study uses MA encounter data to compare switchers and non-switchers and assess an alternative extrapolation approach to estimating favorable selection. Methods: We estimated favorable selection among 1.4 million MA enrollees from 2020 to 2022, generating separate estimates for switchers and non-switchers. Using a 5% national Medicare sample, we attempted to replicate MedPAC's switcher-based analysis, then modified its extrapolation by comparing risk-adjusted costs (imputed with TM claims) between switchers and non-switchers using MA encounter data. Results: Favorable selection ranged from 4.9% to 5.8% between 2020 and 2022-32% to 40% lower than our attempted replication of MedPAC's estimate (8.2%-8.8%). Conclusion: Using an alternative extrapolation approach lowered favorable selection, underscoring its sensitivity to methods and the importance of further refinement for MA payment policy.