We study a 2008 policy reform in which Medicare revised its hospital payment system to better reflect patients’ severity of illness. We construct a simulated instrument that predicts a hospital’s policy-induced change in reimbursement using pre-reform patients and postreform rules. The reform led to large persistent changes in Medicare payment rates across hospitals. Hospitals that faced larger gains in Medicare reimbursement increased the volume of Medicare patients they treated. The estimates imply a volume elasticity of 1.2. To accommodate greater volume, hospitals increased nurse employment, but also lowered length of stay.

We study the effect of public health insurance on labor supply by exploiting a large public health insurance disenrollment. In 2005, approximately 170,000 Tennessee residents abruptly lost Medicaid coverage. Using both across- and within-state variation in exposure to the disenrollment, we estimate large increases in labor supply, primarily along the extensive margin. The increased employment is concentrated among individuals working at least 20 hours a week and receiving private, employer-provided health insurance. We explore the dynamic effects of the disenrollment and find an immediate increase in job search behavior and a steady rise in both employment and health insurance coverage following the disenrollment. Our results are consistent with a significant degree of “employment lock”—workers who are employed primarily to secure private health insurance coverage.

Health insurance affects the rate at which individuals visit hospitals and emergency departments (EDs). We identify the causal effect of losing health insurance using a regression discontinuity design. We compare individuals just before and after their twenty third birthday, which insurers have used as a cutoff after which students are no longer eligible for their parents' health insurance: 1.5% of young adults lose their health insurance upon turning 23, and this transition leads to a 1.6% decrease in ED visits and a 0.8% decrease in hospital stays. We discuss why these estimates are larger than those observed among teenage populations.