Climate variability and its Effects on Agricultural Yields in Sub-Saharan Africa

Abstract

This research project examines the relationship between short-run climate variability in the form of temperature and rainfall anomalies and agricultural yields. Utilizing data from longitudinal surveys conducted by the World Bank (LSMS-ISA), I investigate how short-run temperature and rainfall anomalies affect plot-level yields in Ethiopia, Mali, Nigeria, Malawi, and Tanzania. Previous literature provides strong evidence that climate variability affects agricultural productivity. A study indicates that negative rainfall shocks (drought like conditions) reduce household consumption and agricultural yields (Amare et al. 2018). Additional evidence indicates that temperature has non-linear effects on the yields of staple crops (Schlenker and Lobell 2010). I hypothesize that negative rainfall anomalies and extreme temperature shocks decrease yields, particularly in regions that are highly dependent on rainfall. To test this, I constructed standardized rainfall and temperature anomaly measures and estimated ordinary least squares, household fixed effects regressions, and a series of extreme shock models that estimated droughts, heat shocks, flood shocks and frost shocks along with positive and negative rainfall and temperature shocks. Extreme shocks are defined as being ±1.5 standard deviations from the mean and the positive and negative rainfall shocks are defined as being ±0.5 standard deviations from the mean. The results are consistent for rainfall shocks, an increase in rainfall across the specifications is associated with higher yields, ranging from 5% to 9% (around 7% in the household- fixed effects model), while temperature effects range from 9% to 19%. Temperature effects decline when enumeration area (EA) fixed effects are included; this indicates that cross location differences are driving the temperature coefficients up. These findings remain robust under the household fixed effects specifications. Quadratic forms of the rainfall and temperature z-scores were found to be insignificant across the specifications. Additional specifications indicate that a negative rainfall shock leads to a 7% reduction in yields, while positive rainfall shocks display positive effects, but lose their significance when EA fixed effects are included.