← Back to Papers

Adverse Selection in (Un)Subsidised Health Insurance: Evidence from Nepal’s Age-70 Threshold

Job Market Paper · Working Paper

Nepal waives health insurance premiums at age 70. I use that statutory waiver in a regression discontinuity design to produce the first quasi-experimental estimate of adverse selection in a national insurance scheme in a low- or middle-income country.

Download PDF
Design
Regression discontinuity in age at enrolment
Data
Administrative claims records from Nepal's National Health Insurance Programme
Setting
Nepal

Key Findings

  • Coverage nearly doubles at the age-70 premium waiver, from 21.9 to 42.3 percent of the age-eligible population, and the waiver extends coverage into places that are poorer and farther from care.
  • Cost per enrollee falls across every claims margin at the threshold.
  • The marginal cost of the enrollees the zero price draws in sits significantly below the average cost of the paying pool, which is adverse selection into the paid programme.
  • Selection persisted while household bundling and restricted enrolment windows, the two remedies usually proposed against it, were both in force.

Abstract

A growing number of governments in low- and middle-income countries are building universal health coverage on contributory health insurance schemes. The mandate behind them is rarely enforced, so enrolment is voluntary in practice and enrollees sort on the premium. How strongly that price sorts them on risk determines what the pool costs to insure. This paper provides the first quasi-experimental estimate of adverse selection in a national scheme in a low- or middle-income country, using a statutory premium waiver at age 70 in Nepal’s National Health Insurance Programme in a regression discontinuity in age at enrolment. Coverage nearly doubles at the threshold, while the waiver extends it into places that are poorer and farther from care. Cost per enrollee falls across every claims margin. Because enrolment itself responds to the price, these compare pools of differing composition rather than a fixed population’s use of care. The marginal cost of the enrollees the zero price draws in is significantly below the average cost of the paying pool, which is adverse selection into the paid programme. The pool was adversely selected while both remedies proposed against selection in these markets, household bundling and restricted enrolment windows, were in force. These findings show that selection in these markets runs through the price, so remedies that leave the price alone cannot remove it.

Claims fall discontinuously at the age-70 threshold. Panel A shows the probability of any claim; Panel B shows total amount claimed, in NPR, against days from the 70th birthday.
Claims fall discontinuously at the age-70 threshold. Panel A shows the probability of any claim; Panel B shows total amount claimed, in NPR, against days from the 70th birthday.
The marginal cost of the enrollees drawn in by the zero price lies below the average cost of the paying pool, across the coverage margin induced at age 70.
The marginal cost of the enrollees drawn in by the zero price lies below the average cost of the paying pool, across the coverage margin induced at age 70.

Keywords: Adverse Selection, Health Insurance, Regression Discontinuity, Nepal

JEL Classification: D82, G22, H51, I13, I18, O15

Presented at

  • European Health Economics Association (EuHEA) Conference
  • Global Health Seminar Series, Centre for Health Economics, University of York
  • Health Economists' Study Group (HESG) Summer Conference
  • Early Career Economics Seminar Series, Heriot-Watt University
  • Strathclyde PGR Conference