Research

Abstract. NASCAR and ARCA races used leaded gasoline through 2006, generating concentrated bursts of airborne lead emissions around active racetracks. I study how these exposures affected infant health using detailed Texas birth records linked to race events at Texas Motor Speedway (TMS) in Fort Worth. Identification relies on within-TMS variation in distance, wind direction, race timing, and race-specific emissions among mothers living within 25 km of the same track. Pre-ban trimester-specific log emissions reduce birthweight, shorten gestation, and increase the probability of low birthweight. A post-ban placebo yields coefficients near zero, suggesting that the estimated effects are not driven by factors correlated with race activity. To translate intermittent race-day exposures into pregnancy-relevant blood lead burdens, I develop a pregnancy-specific physiologically based pharmacokinetic (PBPK) model that simulates daily maternal and fetal blood lead trajectories using each mother's race-event history. A one-standard-deviation increase in second-trimester PBPK blood lead reduces birthweight by approximately 73 grams. A significant post-2007 birthweight decline near a former Houston racetrack, unrelated to lead exposure, suggests the limitations of cross-track comparisons and highlights the importance of exploiting within-track variation. I then examine potential mechanisms and find that lead exposure increases gestational hypertension and shortens gestation, with both effects disappearing after the transition to unleaded fuel. These findings suggest that even relatively small and intermittent sources of airborne lead can have meaningful effects on infant health.

Abstract. Exploiting natural experiments in maternal exposure to tetraethyl-lead arising from the differential phase-out of leaded gasoline across the United States, and from the exogenous migration behavior of mothers during pregnancy, we find consistent evidence that fetal exposure to lead through the maternal blood lead pathway significantly affects infant health. Across an ensemble of tests, we find that an increase in maternal blood lead: 1) decreases infant birthweight; 2) increases the risk of low birthweight; 3) shortens gestation length; and 4) increases the risk of prematurity. These results are robust to various measurement operations of our treatment variable, the relaxation of case inclusion criteria, and the substitution of our blood lead treatment variable for proxy measures of lead exposure. Moreover, we satisfy a divergent validity test by falsifying our causal channel of maternal blood lead with analyses of the lead-independent outcome of infant sex. While the flow of lead into the lived environment has decreased substantially since of the phase-out of leaded gasoline, exposure to legacy lead sources remains a public health problem with measurable economic costs.

Abstract. This paper studies whether emotionally salient events affect physician decision-making and maternal health during childbirth using evidence from NCAA and NFL football games. We document three main findings. First, the probability of a non-cesarean, non-induced birth falls on NCAA football game days and rebounds the following day, consistent with deliveries being shifted around games. Second, exposure to NCAA football game weekends increases the probability of provider-sensitive complications by 0.23 percentage points, a 19 percent increase relative to the baseline rate. Third, these effects are concentrated among mothers with fewer economic resources. Low-SES mothers, uninsured mothers, and WIC recipients experience statistically significant increases in complications after correcting for multiple testing, while we find little evidence of differences by race. Overall, the results suggest that NCAA football games, particularly those played on Saturdays when hospital staffing is typically lower, can influence physician behavior and worsen maternal outcomes.

Abstract. This paper asks whether in-utero exposure to extreme heat worsens birth outcomes in Texas, and whether urban heat islands (UHIs) amplify such effects. Using 2.5 million Texas birth records (2003--2010) linked to tract-level daily temperature and NLCD impervious surface coverage, the preliminary findings suggest that heat has no detectable effect on fetal growth (birth weight or low birth weight) conditional on gestational age, which may reflect selection via early fetal loss, behavioral adaptation, or other channels. The heat--imperviousness interaction is null or economically trivial across outcomes, and splits by AC proxies show no differential effects, tentatively consistent with the possibility that AC access mitigates heat exposure, though housing vintage and income are imperfect proxies. These results are provisional.