Does service-level spending show evidence of selection across health plan types?

We provide an explanation for the widespread finding that capitated managed care plans attract comparatively healthy, low cost enrollees relative to traditional unmanaged plans. Using disaggregated commercial insurance claims from the Thomson-Reuters MarketScan database, we show that managed care plans spend proportionally less on those types of services that are predicted to be more profitable to ration tightly using a selection index developed by Ellis and McGuire that captures the derivative of profits with respect to reduced spending on disaggregated services. Conventional diagnosis-based risk adjusted premiums reduce selection incentives by about 50% relative to premiums that are not risk-adjusted.