Evaluation of the Burkina Faso Agriculture Development Project: Baseline Report

Publisher: Washington, DC: Mathematica
Apr 24, 2018
Authors
Christopher Ksoll, Seth Morgan, Kristine Bos, and Randall Blair

Key Findings:

  • Persons affected by the project (PAPs) whose land was expropriated to build the Di perimeter lost a substantial fraction of their land and agricultural sales revenue because of the construction of the perimeter (for which they were later compensated through higher-value but smaller plots of land). PAPs had undiversified production at baseline, but they expressed plans to shift to high-value crops.
  • Di Lottery applicants’ use of traction animals, improved seed, fertilizer, and pesticide suggests that they could take advantage of new plots. Male applicants had higher literacy rates, greater ownership of larger areas of cultivable land, and more experience in irrigation and rice production.
  • Di Lottery beneficiary and control applicants and households were balanced across the overwhelming majority of variables, suggesting that the lottery was properly implemented and that treatment–control differences can be used to estimate the causal impact of the lottery. There were, however, a few exceptions. Impact regression models in the interim and final evaluation will control for all variables in which beneficiaries and nonbeneficiaries are not balanced, in particular the number of household members listed on the application, gender of the applicant, and access to land.
  • Farmers who subsequently benefitted from MCC’s farmer training faced suboptimal market access and use of nonmechanized tools, leading to limited baseline production, sales, and income.

Mathematica is evaluating the Millennium Challenge Corporation’s (MCC) Agricultural Development Project in Burkina Faso to understand the project’s effects on improved agricultural outcomes and land tenure security. This baseline report provides an analysis of baseline data sources for three evaluations: (1) the Di Lottery evaluation, (2) the Di perimeter evaluation, and (3) the farmer training evaluation. The report relies on data collected by separate organizations that were contracted to evaluate these activities before Mathematica’s involvement. For each evaluation, we first describe the data collected and—where applicable—limitations of these data. Second, we present descriptive statistics on beneficiaries, with a focus on assessing the extent to which targeted participant groups faced the constraints the project activities were designed to overcome. The goal of this exercise is to assess whether beneficiaries could feasibly benefit from the multiple forms of assistance provided by the project activities, including irrigated land and land tenure assistance, training, free inputs, and rehabilitated markets. Finally, we provide baseline balance tests for the Di Lottery randomized controlled trial.