Adverse Selection in Medicaid

Recent expansions of Medicaid eligibility have come with increased experimentation with enrollee cost-sharing. In this paper, we exploit a discontinuous premium increase at the federal poverty level in Michigan’s Medicaid expansion program to test low-income individuals’ sensitivity to premiums using linked enrollment and claims data. At the cutoff, average premiums increase by $3.15 and the probability of disenrollment increases by 2.3 percentage points. Increased disenrollment occurs among those with fewer documented medical needs at baseline, but not among those with greater medical needs. These results suggest that healthier low-income individuals may be sensitive to even modest health insurance premiums, and that premiums may induce adverse selection in Medicaid plans.