In the midst of Africa's mining boom, communities downstream from industrial mines face increased exposure to toxic waste, yet the causal impact of mining-induced water pollution on local health remains unquantified at the continental scale. We construct a novel dataset of opening dates for 2,016 industrial mines and combine it with child health outcomes from the Demographic and Health Surveys spanning 1986 to 2018 across 26 African countries. Using a staggered difference-in-differences design comparing villages located downstream and upstream—based on hydrological networks—of mines before and after their opening, we find a 25% increase in 24-month mortality downstream after mine opening, relative to upstream areas where mortality rates remain stable. The effect is concentrated among children who were no longer breastfed after six months, consistent with the protective role of breastfeeding against heavy metal ingestion through contaminated water and with exposure to heavy metals rather than microbial contamination. Effects are stronger during active mine operation, at high international mineral prices, and in densely mined regions, and decrease with distance from the mine, consistent with higher mining intensity generating higher pollution levels. We rule out alternative mechanisms including changes in fertility behaviors, access to healthcare, in-migration, conflict, and income effects. These findings underscore the significant local health impacts of industrial mining and highlight the need for stronger environmental regulation and water quality monitoring in resource-rich developing countries.

Since 2000, Kenya has experienced an increase in the frequency of droughts, significantly affecting agriculture and driving labor force migration. This paper investigates strategic migration patterns among farmers and pastoralists in response to repetitive droughts. I use fine-grained data that capture short-distance migration and heterogeneity, combining satellite-based daily rainfall data from CHIRPS with exhaustive censuses from 1989, 1999, and 2009. I use a two-way fixed-effects model to exploit spatial variation in drought frequency across 2,518 sub-locations, comparing their demographic growth according to the number of dry rainy seasons over each decade. First, I show that increased drought frequency triggers out-migration: one additional drought decreases demographic growth by 1.7 percentage points, equivalent to a 1% population decline. This result holds within the 15–65 age group, excluding other demographic effects and confirming migration as the driving factor. The main contribution of this paper is the identification of different migration strategies across livelihoods. Rural areas dominated by pastoral activities experience significant out-migration, generating a rural-rural shift from pastoral to agriculture-oriented regions. Herders’ migration displays little heterogeneity, suggesting the migration of entire households and consistent with migration as a last resort. Agricultural areas are less vulnerable to drought and display significant heterogeneity. The results show the migration of the most educated working-age individuals, while uneducated individuals remain trapped in affected areas. This paper highlights the importance of detailed data for understanding diverse migration strategies and designing effective policies.