stchakwdev/when-care-is-free

Quasi-experimental evidence on user-fee abolition and maternal & child survival in sub-Saharan Africa — LP-DiD, synthetic control, randomization inference, wealth-quintile mechanism analysis # When Care Is Free **Quasi-experimental evidence on user-fee abolition and maternal & child survival in sub-Saharan Africa.** **Live site:** stchakwdev.github.io Between 2001 and 2016, nine African countries abolished point-of-care user fees for maternal and child health services, each in a different year. This project treats that staggered rollout as a natural experiment and asks whether removing fees saved lives. ## Findings - **Under-5 mortality fell ~6%** (95% CI −10.2% to −1.5%) in the eight years after abolition, pooled across nine adopters against 26 never-treated controls. The estimate holds its sign and size across every specification I attacked: timing shifts, broad vs. narrow treatment coding, dropping Ebola countries, dropping any single cohort (ATT range −5.2% to −7.1%, all p < 0.05 cluster-robust). The two inference approaches disagree in strength. Cluster-robust standard errors give p = 0.01; 1,000-permutation randomization inference gives p = 0.05 one-sided. I grade the evidence moderate and report both. - **Sierra Leone's maternal mortality ran ~11% below its synthetic counterfactual on average across 2010–13** (pre-Ebola), and ~23% below averaged across 2010–19, with the yearly gap widening throughout. Placebo rank tests put the 2010–13 gap at p ≈ 0.14, so a single-country result like this one is suggestive, not conclusive. - **The mechanism shows up in the survey data.** DHS/MICS wealth-quintile data (WHO MNCAH database) show poorest-quintile service use rising faster after abolition than in the never-treated controls: all 13 adopter country-service pairs with bracketing surveys beat the control-country median, the best of them ahead of 98% of control survey pairs. Poorest-quintile facility deliveries tripled in Sierra Leone within three years. - **Neonatal mortality barely moved**, consistent with newborn survival depending on quality of care rather than fees at the door. Free doors don't staff operating theatres. …