The Private Provision of Public Services : Evidence from Random Assignment in Medicaid

Nearly all prior work on government outsourcing has focused on the contracting firm's incentives. This paper shows how strong incentive contracts may be insufficient to generate spending reductions (or other desired outcomes) in the presence of a binding technological or managerial constraint. We study this outsourcing problem in the context of Medicaid. In healthcare, plans may have special capacity to interdict care provision at the pharmacy, where real-time adjudication enables insurers to deny services before they are rendered. Exploiting a large natural experiment in which Medicaid beneficiaries were randomly assigned between a state-administered FFS system and private managed care, we document how Medicaid outsourcing impacted spending, utilization, and consumer satisfaction. We find that spending was 5%-10% lower for enrollees assigned to managed care. These effects were concentrated in prescription drugs, where spending declined by about one quarter. Using administrative records that include information on real-time denials, we show that utilization management by plans caused efficiently-targeted reductions and substitutions at the pharmacy. Our results indicate that private managed care plans may have sharp tools for managing pharmacy benefits but blunter tools for managing medical benefits, where real-time claims adjudication is less feasible