This was created in partnership with Environment for Development .
Weather fluctuations tend to be as important as climate change in farmers’ decision making in countries such as Ethiopia that have virtually no weather insurance. This paper assesses the distinct impacts of weather and climate change measures on agricultural productivity of households, measured in terms of crop revenue, in the Amhara region of Ethiopia. Four waves of survey data, which included about 1500 households in each round, combined with interpolated daily temperature and monthly rainfall data from the meteorological stations, are employed in the analysis. The distinction between weather and climate is highlighted by observations in the temperature data, which show that the pattern of temperature for both short-term and long-term values follows a bell-shaped distribution, with the striking feature that the extreme ends of the distribution have fatter tails for the long term values. The analysis employs monthly rainfall and 14 temperature categories related to weather measures and four categories corresponding with the extreme ends of the long-term temperature distribution. The analysis also distinguishes between summer and spring seasons and different crops, in recognition that Ethiopia’s agriculture is multi-cropping and multi-season. The major findings show that temperature effects are distinctly non-linear, but only when the weather measures are combined with the extreme ends of the distribution of the climate measures. In addition, rainfall generally has a less important role to play than temperature, contrary to expectations for rainfed agriculture.