Agriculture accounts for 85 percent of employment and 46 percent of GDP in Ethiopia. As a result, development in Ethiopia depends on strengthening rural capacity through extension services and through supporting farmer associations and training centers. However, it is difficult for such development to be equal across gender because women farmers have less access to agricultural technology. Given that women account for about 60 percent of agricultural labor in Ethiopia, it is important to understand how and why they differ from men in Ethiopia's agricultural sector. The Farmer Innovation Fund (FIF) is a component of the Rural Capacity Building Projects (RCBP) which seeks to strengthen the extension system and increase gender equality in extension services. FIF provides funds to farmer groups to implement innovative ideas developed and partially funded by the groups themselves. FIF also plans to decentralize funding from the woreda, or ward, level to the farmer training center level. To evaluate the effectiveness of FIF, an impact evaluation study was conducted in Amhara and Tigray states, where FIF was rolled out as a randomized intervention. The impact evaluation included three surveys: a baseline, conducted in August-October 2010; a midline, carried out in April 2012; and an endline, administered in June 2013. The data collected from the surveys examined how women-only training programs effect women's participation in agricultural and extension services and which kind of training package is the most effective in improving women's economic empowerment. In addition, the impact evaluation studied the effects that participation in training has on intra-household allocation of resources, decision making within households, and domestic violence. Also, variables related to food consumption enabled an analysis of how training programs affect children's nutrition. The midline survey covered 2,492 households, a subset of the original sample of 2,675 from the baseline survey. Within each household, surveys were given to men and women. In addition, a separate survey was given to individuals who were a single head of household. Among the original 2,675 households, 869 were assigned as non-FIF households to serve as a pure control group and on the remaining households a simple lottery design was used to randomly assign 958 of the households to the treatment group and 848 households to the control group. Individuals in treatment households received FIF training, while individuals in the control households did not.
