Come rain or shine: Precipitation and food insecurity in Brazilian households
Autor: Ricardo Campante Vale, Institute of Applied Economic Research (IPEA)
Áreas Temáticas Sustainable Development
Compartilhe este conteúdo em suas redes
Resumo: This brief uses Brazil’s 2023 Continuous National Household Sample Survey (PNADC), which included a food insecurity module for the first time, combined with CHIRPS satellite precipitation data, to estimate how rainfall levels affect the probability that a household reports moderate or severe food insecurity. A logistic regression model finds a U-shaped relationship: food insecurity probability is highest, at roughly 44 per cent, when rainfall is absent during the growing season; it falls as precipitation rises, reaching a minimum around 814mm; and it climbs again above roughly 1,630mm, eventually exceeding the risk associated with no rainfall. These estimates hold after controlling for income, region, education, race, sex, employment, and rural location, indicating the relationship is not simply a proxy for regional poverty differences. The effect varies by region: in the North, excess rainfall drives food insecurity risk above 50 per cent at the highest precipitation levels; in the North-east, drought conditions are the dominant risk factor, with 500mm of rainfall corresponding to a 34 per cent food insecurity probability compared with 22 per cent in the South. The brief argues that Brazil’s existing food insecurity early warning indicator, CadINSAN, tracks socio-economic variables but omits precipitation, and proposes incorporating climate data into it, following the model of the Famine Early Warning Systems Network (FEWS NET), documented to reach 84 per cent forecasting accuracy. It also proposes linking the resulting estimates to the geographic targeting of food banks and the Cisterns Programme.