Abstract

Nepal made health insurance free for citizens over 70—and their claims fell. Using administrative data on 40 million claims from Nepal’s National Health Insurance Program and a regression discontinuity design, I find that claims per enrollee drop sharply at the eligibility threshold: the share with any claim falls by 12.2 percentage points, the number of claims decreases by 0.86 (33%), and total claimed amounts decline by NPR 1,604 (31%). This counterintuitive pattern reflects adverse selection. When insurance is costly, high-risk individuals are more likely to enroll; when it becomes free, lower-risk individuals join, shifting the pool toward healthier enrollees. To isolate selection from moral hazard, I construct an ex-ante expected healthcare utilization index using prior claims data. This index drops discontinuously at age 70, confirming that the composition of enrollees shifts toward healthier individuals when insurance is free. A placebo test using pre-policy data shows no such discontinuity, validating the research design. These findings demonstrate that adverse selection operates even with household-level bundling and enrollment timing restrictions—two policies often proposed to combat selection—suggesting that premium subsidies alone do not eliminate selection concerns in voluntary health insurance.

Quick Summary