s:2105:"%T Stochastic impact evaluation of an irrigation development intervention in Northern Ethiopia 
%A Yigzaw, N. 
%A Mburu, J. 
%A Ackello-Ogutu, C. 
%A Whitney, C. 
%A Luedeling, E. 
%X Irrigation plays a significant role in achieving food and nutrition security in dry regions. However, detailed ex-ante appraisals of irrigation development investments are required to efficiently allocate resources and optimize returns on investment. Due to the inherent system complexity and uncertain consequences of irrigation development interventions coupled with limited data availability, deterministic cost-benefit analysis can be ineffective in guiding formal decision-making. Stochastic Impact Evaluation (SIE) helps to overcome the challenges of evaluating investments in such contexts. In this paper, we applied SIE to assess the viability of an irrigation dam construction project in northern Ethiopia. We used expert knowledge elicitation to generate a causal model of the planned intervention's impact pathway, including all identified benefits, costs and risks. Estimates of the input variables were collected from ten subject matter experts. We then applied the SIE tools: Monte Carlo simulation, Partial Least Squares regression, and Value of Information analysis to project prospective impacts of the project and identify critical knowledge gaps. Model results indicate that the proposed irrigation dam project is highly likely to increase the overall benefits and improve food and nutrition status of local farmers. However, the overall value of these benefits is unlikely to exceed the sum of the investment costs and negative externalities involved in the intervention. Simulation results suggest that the planned irrigation dam may improve income, as well as food and nutrition security, but would generate negative environmental effects and high investment costs. The Stochastic Impact Evaluation approach proved effective in this study and is likely to have potential for evaluating other agricultural development interventions that face system complexity, data scarcity and uncertainty. 
";