How early-life conditions shape human capital remains a central question in applied microeconomics. This paper examines a policy-relevant case: fluoride in drinking water, a focus of ongoing public health debates. We estimate the causal effects of childhood fluoride exposure at levels lower than those typically used in artificial fluoridation, exploiting quasi-exogenous variation in naturally occurring concentrations determined by regional geology in Japan. Combining this variation with unique longitudinal survey data, as well as nationally representative medical claims and patient surveys, we provide comprehensive evidence on fluoride’s beneficial impacts: even at low concentrations, it substantially improves dental health in childhood and adolescence. In contrast to concerns from studies of higher exposures, we detect no adverse effects on educational attainment. Importantly, fluoride enhances non-cognitive skills (such as self-esteem) among females but not males. The mechanism appears to operate through oral health’s influence on appearance: fluoride reduces cavities, which in turn lowers the likelihood of malocclusion and other visible dental issues. These improvements in dental aesthetics matter during adolescence, thereby boosting confidence and socio-emotional skills. Our findings underscore the role of oral health in shaping non-cognitive development, especially for girls, and highlight the broader policy relevance of fluoride exposure.
Disentangling Sources of Variation in Cesarean Delivery Rates (with Stefanie Fischer, Heather Royer, and Corey White)
[Draft coming soon]
[Abstract]
Cesarean section rates vary widely across U.S. counties, but it remains unclear how much of this variation reflects demand-side factors (e.g., patient risk factors, preferences, or beliefs) versus supply-side factors (e.g., physician practices or hospital incentives). We develop a new empirical strategy to estimate the effect of supply-side influences on the likelihood of receiving a C-section. We leverage over three decades of hospital obstetric unit closures (1989–2019), which require some mothers to give birth in nearby counties with systematically different C-section rates. Our instrumental variables approach exploits heterogeneity in the induced change in county C-section rates. We find that a mother’s probability of receiving a C-section moves one-for-one with the change in the county C-section rate she faces. The results point to a dominant role for provider behavior and local practice norms in driving geographic variation in C-section use —one of the most common surgical procedures in the U.S. health care system.