Estimating Health Expenditure Elasticities Using Nonlinear Reimbursement

Recent health care initiatives attempt to stem rising costs by increasing patients ’ cost shar-ing. These initiatives include high deductible plans, the Medicare Drug Plan “doughnut hole”, and Health Savings Accounts (HSAs). The success of such initiatives depends on how health expenditures change as patients ’ reimbursement decreases. Estimating this elastic-ity is complicated by selection bias, as high expenditure patients can self-select into high reimbursement plans. Additionally, nonlinear reimbursement is prevalent in U.S. insurance contracts and the aforementioned initiatives. Nonlinearities introduce bias when using pre-vious estimation methods by simultaneously determining expenditure and reimbursement rate. This paper develops an elasticity estimation method that controls for selection bias by taking advantage of nonlinear reimbursement rates. Discontinuous reimbursement rates induced by a nonlinearity are used to isolate patients ’ expenditure choices. Using detailed claims-level data of employer-sponsored health insurance, I find a tight range of elasticities between-0.25 and-0.33 in the range of average U.S. spending. I then use these estimates in a policy experiment measuring moral hazard and calculate the resulting welfare effects. This paper’s estimation method may be used on many policies with nonlinear reimburse-ment which previous tools could not address