Myopia and Complex Dynamic Incentives: Evidence from Medicare Part D

Standard Medicare Part D drug insurance provides limited coverage in a ``donut hole'' region, making the purchase problem dynamic. We develop a discontinuity-based test for myopia using enrollees who arrived near the coverage gap early in the year. We find that there are fewer and cheaper purchases immediately after reaching the gap, providing evidence in favor of myopia. We structurally estimate a dynamic drug purchase model and find complete myopia. For a nationally representative sample, "behavioral hazard" increases enrollee spending by 41%. Entirely eliminating the gap would increase insurer spending 31%, compared to 6% for generic-only gap coverage.