Selection and Causal Effects in Voluntary Programs: Bundled Payments in Medicare
68 Pages Posted: 1 Aug 2022 Last revised: 21 Sep 2022
Date Written: July 11, 2022
Regulators are testing many payment reforms to improve US healthcare productivity. The vast majority of these reforms allow market participants to opt in rather than mandating participation. Allowing choice can enhance outcomes if firms select contracts based on unobserved treatment gains. However, unobserved selection may bias program evaluation, misleading policymakers. We study this trade-off in the case of a prominent national reform introduced by Medicare to reduce spending on surgeries. Idiosyncratic program rules generated plausibly exogenous variation in potential financial gain across hospitals, which we exploit to instrument for participation. We estimate marginal treatment effects and find considerable treatment effect heterogeneity but no evidence to support sorting on treatment gains. In contrast, there is substantial selection on untreated outcomes, leading the difference in differences estimator to understate the causal effect on spending. Observable factors do not explain more than 30% of this bias, highlighting the challenge of evaluating voluntary programs.
Keywords: payment reform, hospital, marginal treatment effects
JEL Classification: I10,L51
Suggested Citation: Suggested Citation