Too hot to give birth? Temperature shocks and general fertility rates in Japan

This paper studies how alternative measures of ambient temperature change the estimated short-run relationship between heat and fertility. I combine monthly vital statistics for all 47 Japanese prefectures from 2012 through 2019 with daily weather reconstructed from raw station records. The outcome is the monthly general fertility rate, defined as live births per 1,000 women ages 15–49. Models include complete temperature-bin vectors, lagged precipitation and missing-weather controls, year-month fixed effects, prefecture-by-calendar-month fixed effects, and prefecture-by-year fixed effects. A day with a minimum temperature of at least 25 ∘ C is associated with 0.39 and 0.25 percent lower fertility nine and ten months later. A day with a mean temperature of at least 30 ∘ C is associated with 0.50 percent lower fertility at lag ten. In contrast, the maximum-temperature result is borderline at lag nine, insignificant at lag ten, and disappears when complete mean-temperature bins enter jointly. The evidence therefore points more consistently to sustained heat and limited nighttime cooling than to a brief daytime peak. Dynamic analysis and the recorded-fetal-death check do not identify whether the fertility estimates reflect conception timing, early pregnancy loss, gestational length, or another mechanism. The birth-rate estimates capture short-run reductions or displacement around the gestational horizon, not changes in completed fertility.