A Comment on "Does the Effect of Pollution on Infant Mortality Differ Between Developing and Developed Countries? Evidence from Mexico City"

Arceo et al. (2016) (AHO) examine the effects of air pollution on neonatal and infant mortality in Mexico City (CDMX), Mexico. The authors estimate these effects using administrative data on infant mortality across CDMX municipalities in conjunction with measured hourly pollution concentrations, using both two-way fixed effects and an IV approach that exploits thermal inversions. The IV results show that a 1µg/m3 increase in average maximum daily coarse particulate matter (PM10) raises weekly infant mortality by 0.23 deaths per 100,000 births and that a 1 ppb increase in carbon monoxide (CO) raises infant mortality by 0.0046 deaths per 100,000 births; both effects are significant at the 1% level. First, we successfully reproduce all results in the main paper from the provided analysis data; some minor differences emerge between our reproduced results and the reported coefficients, but they do not affect the qualitative findings. Second, we attempt to reconstruct components of AHO's analysis sample using publicly available data. While differences in data availability and processing limit direct comparability, the reconstructed data preserve the direction of the original results and retain statistical significance in some specifications. The degree of agreement varies across reconstruction exercises: estimates are often less precise, losing statistical significance in several specifications-particularly for CO- and frequently fall outside the authors' original confidence intervals. Finally, we verify the authors' results are robust to alternative choices for sample selection and standard error choices.