The household survey was carried out in the Nile River Basin in Ethiopia. The household sampling frame in Ethiopia was developed to ensure representation for the Nile River Basin at the woreda (district) level regarding level of rainfall patterns in terms of both annual total and variation; the four classes of traditionally defined agro-ecological zones (AEZs) found in the basin; vulnerability of food production systems (through the proxy of frequency of food aid in the past ten years); and irrigation prevalence. All data used for the sample frame is from the Atlas of the Ethiopian Rural Economy (Benson et al., 2006). Each woreda was classified based on : agroecological zone (Kolla, Weynadega, Dega, and Bereha), the percent of cultivated land under irrigation (no data, 0-2%, 2-4%, 4-8%, and 8% or greater), average annual rainfall (0-854mm, 854-1133mm, 1133-1413mm, 1413-1692mm, 1692mm or greater), rainfall
variability (coefficient of variation for annual rainfall), and vulnerability (number of years of food aid received in the past 10 years). Twenty woredas were selected such that across each of the above dimensions the proportion falling into each class for the sample matched as closely as possible the proportions for the entire Ethiopian Nile basin. Peasant associations (administrative units lower than districts) were also purposely selected to include households that irrigate their farms. One peasant association was selected from every woreda for a total of 20 peasant associations. Random sampling was used in selecting 50 households from each peasant administration within the 20 woredas. Thus, the final dataset contains 1,000 observations from 20 woredas in 5 regional states in Ethiopia (Tigray, Amhara, Oromiya, Benishangul Gumuz, and Southern Nations Nationalities and Peoples (SNNP)). The related South Africa Limpopo Basin Climate Change Adaptation dataset is also available from IFPRI’s website at www.ifpri.org/dataset/south-africa-limpopo-basin-climate-change-adaptation-dataset.