Miguel and Kremer randomized deworming by school, not child

Intestinal worms — hookworm, roundworm, whipworm, schistosomiasis — infect more than a quarter of humanity and are most damaging to school-age children, draining them into anaemia and low attendance. The drug is cheap, about half a US dollar per treated child. Yet the credible evidence was thin and discouraging: trials that randomly assigned the treatment to individual children within the same school found small or insignificant effects on health and schooling, apparently confirming that deworming was a marginal investment a donor could skip.

The discouraging trials shared a blind spot. Worms are transmitted person to person through the shared environment of a school; when some children in a room are treated but the rest are not, the treated ones are promptly reinfected by the untreated ones, and the untreated ones are readmitted to the worm pool. Everyone in the same school swims in the same worm reservoir, so an individual-level trial quietly destroys the very benefit it is trying to measure.

Randomized deworming by whole school, not child, so the shared worm pool dropped below reinfection level and the transmission spillover became visible.

Individual trials let treated children get reinfected by untreated ones, erasing the effect.

School-level randomization broke transmission at the boundary, letting the worm pool drop.

This measured both direct and spillover effects, showing deworming's true payoff.

The externality justified full subsidy, changing cost-benefit calculations.

When an intervention spreads through a shared channel, evaluate at the unit that carries the spillover, not the unit that gets the dose.

Deworming cut absenteeism by a quarter; untreated neighbors gained because transmission fell. The externalities justified full subsidy. The 2004 Econometrica paper became the empirical backstop for mass deworming worldwide — WHO school-based deworming programs and GiveWell's top-rated charities rest on it — and a ten-year follow-up in the QJE found long-run impacts on earnings, with a conservative 32% internal rate of return.