← Back to Papers

Health Insurance and Health-Seeking Behaviour: The Distributional Effects of Public Health Insurance in Nepal

Working Paper

The first causal evaluation of Nepal's National Health Insurance Programme. Eligibility raises health-facility visits by 14 percent, but the gains appear only in low-poverty districts, pointing to a supply-side constraint on the empanelled-facility network.

Download PDF
Design
Staggered difference-in-differences (Callaway and Sant'Anna)
Data
District-quarter panel of all health-facility visits, with NHIP registration and claims records
Setting
Nepal, all 77 districts, 2016 to 2022

Key Findings

  • NHIP eligibility raises total health-facility visits by 14 percent, with similar gains across outpatient, emergency, and diagnostic margins.
  • Effects emerge gradually and remain at their peak level once reached.
  • The gains are unequally distributed: utilisation rises only in low-poverty districts, while high-poverty districts show no significant effect.
  • A 2019 facility-reassignment reform points to a supply-side explanation: in poor districts the empanelled-facility network is too sparse for coverage to translate into care.

Abstract

Public health insurance is one of the largest fiscal interventions in low- and middle-income countries, but who actually benefits varies sharply across settings. This paper provides the first causal evaluation of Nepal’s National Health Insurance Programme (NHIP), a voluntary household-level scheme rolled out in stages across all 77 districts between 2016 and 2022. Combining a novel district-quarter panel of all health-facility visits with individual-level registration and claims records, I estimate that NHIP eligibility raises total health-facility visits by 14 percent, with similar gains across outpatient, emergency, and diagnostic margins. Effects emerge gradually and remain at their peak level once reached. The gains are unequally distributed: utilisation rises only in low-poverty districts, while high-poverty districts show no significant effect. Evidence from a 2019 facility-reassignment reform points to a supply-side explanation: in poor districts, the network of empanelled facilities is too sparse for coverage to translate into care.

Dynamic treatment effects on log total clients served, by quarter since NHIP implementation. Effects are flat before rollout and rise steadily afterwards.
Dynamic treatment effects on log total clients served, by quarter since NHIP implementation. Effects are flat before rollout and rise steadily afterwards.
Registration is flat across municipality poverty rates (Panel A), but claiming intensity falls steeply as poverty rises (Panel B). Coverage reaches poor areas; care does not follow.
Registration is flat across municipality poverty rates (Panel A), but claiming intensity falls steeply as poverty rises (Panel B). Coverage reaches poor areas; care does not follow.

Keywords: Health Insurance, NHIP, Poverty

JEL Classification: H51, I12, I13, I14, J16, O15

Presented at

  • Health Economists' Study Group (HESG) Winter Conference
  • Scottish Graduate Programme in Economics (SGPE) Conference
  • Economics PhD Reading Group, University of Glasgow
  • Strathclyde PGR Conference