Weather Shocks, Labor Reallocation, and Structural Transformation: Micro Evidence from Africa

Structural transformation, the reallocation of labor from low-productivity agriculture to higher-productivity non-agricultural sectors, is widely viewed as a key driver of economic development. Yet in many African countries, labor leaving agriculture does not always transition into productive employment. This paper examines how agricultural productivity shocks affect labor reallocation and why similar shocks generate different employment outcomes across countries. Using household panel data from Niger, Malawi, and Ethiopia, combined with geospatial climate data, we construct a continuous measure of weather-driven agricultural productivity changes based on rainfall and temperature using a machine learning approach. Empirical results show that favorable weather productivity shocks reduce agricultural labor shares, but the employment outcomes of workers leaving agriculture vary substantially across countries: labor shifts into non-agricultural activities in Niger, spatial barriers limit access to jobs in Malawi, and institutional constraints lead to surplus labor in Ethiopia. To explain these patterns, we develop and estimate a microfounded model that combines a Lewis-type labor release mechanism with a Harris–Todaro–type job absorption process. Counterfactual policy simulations show that the effectiveness of policies—such as food subsidies, infrastructure improvements, and institutional reforms—depends on both country-specific labor market constraints and climate conditions. These findings highlight the importance of designing development policies that account for institutional context and climate variability when promoting structural transformation.