Weather Risk and Crop Switching in Perennial Agriculture: Evidence from Ethiopia

Perennial agriculture poses a distinct adaptation problem because planting decisions are long-lived, costly to reverse, and often adjusted only at replanting. This paper studies whether Ethiopian smallholders shift land away from coffee when weather exposure worsens coffee’s risk-return profile relative to khat, an alternative perennial crop. I develop a portfolio-choice framework in which farmers allocate replanting land between coffee and khat based on expected profitability and profit risk. Using Ethiopia LSMS-ISA household panel data matched to ERA5 weather data, I first estimate reduced-form effects of weather exposure during the February–April budding and growing window on subsequent coffee allocation. I then estimate crop-specific weather-profit functions and use them to construct weather-induced measures of expected profitability and predicted profit risk. Extreme heat reduces the coffee share of coffee-khat land, with the clearest effects among households that cultivated coffee in the previous round. Coffee profits are more sensitive to extreme heat than khat profits, and higher weather-induced coffee profit risk is associated with lower next-round coffee allocation. The current estimates provide weaker evidence that expected profitability independently predicts adjustment. These findings suggest that, in perennial systems, weather risk affects adaptation not only through current profits but also through the crop choices farmers make when replanting becomes possible. 2026 Annual Meeting, July 26 - 28, 2026, Kansas City, Missouri