Estimating Heterogeneous Treatment Effects of Medicaid Expansions on Take-up and Crowd-out

Economists have devoted considerable resources to estimating local average treatment effects of expansions in Medicaid eligibility for children. In this paper we use random coefficients linear probability models and switching probit models to estimate a more complete range of effects of Medicaid expansion on Medicaid take-up and crowd-out of private insurance. We demonstrate how to estimate, for Medicaid expansions, the average effect among all of those eligible, the average effect for a randomly chosen person, the effect for a marginally eligible child, and the average effect for those affected by a nonmarginal counterfactual policy change. We then estimate the average effect of Medicaid expansions among all eligible children and the average effect for those affected by a nonmarginal counterfactual Medicaid expansion since these are likely to be the most useful for policy analysis. Estimated take-up rates among average eligible children are substantially larger than take-up rates for those made eligible by a counterfactual Medicaid expansion, moreover both of these effects vary widely across demographic groups. In terms of crowd-out, we find statistically significant, though small, effects for all eligible children, but not for those affected by a counterfactual policy change