Out-of-pocket health spending and population health: evidence from universal coverage expansions.

INTRODUCTION

Reducing out-of-pocket (OOP) health spending is the headline target of universal health coverage (UHC) reforms, but two questions remain actively debated despite a substantial prior literature on each: does cutting OOP improve population health, and—when reform brings OOP down on average—who in the income distribution still pays OOP?

METHODS

We use a panel of 241 countries (2000–2022), 13 UHC expansion episodes and harmonised microdata from four large national surveys (the Behavioural Risk Factor Surveillance System, Pesquisa Nacional de Saúde, Encuesta Nacional de Ingresos y Gastos de los Hogares, National Health and Nutrition Examination Survey). Estimators include two-way fixed effects (TWFE), synthetic difference-in-differences, Callaway-Sant’Anna staggered difference-in-differences (DID) and wild cluster bootstrap inference for the small-group TWFE.

RESULTS

A one percentage point (pp) increase in the OOP share is associated with 0.16 additional infant deaths per 1000 (p=0.003), robust to income-growth, inequality and health-spending controls. TWFE estimates a 5.9 pp average OOP reduction across UHC episodes (p=0.024); synthetic DID yields a more conservative 3.3 pp (p=0.039), with Indonesia (−10.8 pp), Rwanda (−8.7 pp) and China (−5.3 pp) driving the pooled effect while Mexico (−0.3 pp) and Chile (−1.6 pp) show small short-run changes. In the USA and Brazil, the income gradient in poor health is steep and pro-poor in both self-reported and clinically measured biomarkers (hypertension, diabetes, obesity); in Mexico, catastrophic OOP spending displays a ‘reverse gradient’ concentrated among upper-income households who self-finance private care.

CONCLUSION

Universal coverage on paper does not mechanically deliver universal coverage in practice. The institutional design of where residual OOP falls is at least as policy-relevant as the headline magnitude of any OOP reduction.