Missing Women in India: Gender-Specific Effects of Early-Life Rainfall Shocks

We link over a century of monthly precipitation data (1911-2011) to the population by gender and age at the district level in the 1991, 2001, and 2011 Indian censuses to study how differential impact of early-life (around birth year) rainfall shocks on women affect cohorts' population sex ratios. Using an approach that generates separate indices for excess (wet) as well as negative (dry) rainfall shocks and adjusts for the level of persistence in monthly shocks, we find that both too much and too little rainfall are associated with lower relative female population. The results are robust to using alternative rainfall shock indices and different sets of fixed effects. Using sex ratio at age 0 in year 2011 and the district-level gender literacy gap in 1991 as proxies correlated with social preference for males, we find that the differential negative effects of birth year rainfall shocks on women are indeed larger in places with stronger male preference, suggesting a significant role for male-biased resource allocation in the face of negative shocks. Population-weighted regressions yield an aggregate estimate of about 1.156 million missing women due to differential effects of early-life rainfall shocks, which is about 3.01% of the estimate of 38.46 million missing women from Bongaarts and Guilmoto (2015).