Lottery-based randomized control trials (RCT) are feasible and effective ways of rigorously measuring the impacts of large-scale irrigation infrastructure on agricultural outcomes, but have some limitations. Significance and balance tests demonstrated that control and treatment groups were similar along observable characteristics. The process was transparent and exceeded the target for female winners. While the significant differences between control and treatment groups are unsurprising, the evaluation’s findings are an important contribution to the slim literature on the impact of irrigation in Sub-Saharan Africa. Furthermore, RCTs alone cannot be used to estimate the project’s economic rate of return, which requires more precise estimates of perimeter-level outcomes. This RCT also did not allow the separation of effects of the project’s different land and agriculture components of the project. Lastly, establishing the lottery takes time and requires considerable and early engagement among staff. MCC is using this learning by considering lottery approaches in other RCTs.
Lesson Learned