From Sky to Ground: Exploring the Power of Earth Observation in International Development Evaluation

Abstract Information: This multi-paper session will feature three applications of earth observation data for evaluating interventions focused on economic development, disaster risk reduction, and agriculture. By providing real world examples, participants will gain insights into the utility of earth observation data for evaluating a broad range of interventions as well as insights into strengths, limitations, and prerequisites for this evaluation methodology. The first paper presented focuses on an impact evaluation of Mercy Corps’ Managing Risk through Economic Development (MRED) program in Nepal. As communities in the developing world face more frequent and severe environmental risks, many efforts to mitigate damages and build resilience fall prey to collective action challenges. The MRED program is a multi-sectoral disaster risk reduction and economic development activity that includes one component focused on promoting economically sustainable crop production to combat severe flooding risks along flood prone rivers in western Nepal’s plains and middle hills. The presentation will present both the key results of MRED’s impacts on adoption of key crops, soil erosion and flooding, as well as insights from building local staff capacity to support and use geospatial data for evaluation. The second paper presents an impact evaluation of landmine clearance on land use change, economic development, and conflict risk in a setting with ongoing conflict. The prevalence of landmines in many countries leads to thousands of deaths and injuries annually. In addition, landmines may impede the flow of goods and people and block the productive use of contaminated land. This paper studies the effects of clearing more than 15,000 hazardous areas in Afghanistan carried out since 1992. The clearance of hazardous areas leads to changes in land use observed using moderate-resolution imagery, and increases in economic activity reflected in nighttime lights data. Precise clearance data and satellite imagery allows us to observe the shifts occurring even in the small towns and villages that comprise most of our sample. We also find reductions in conflict risks due to the clearance of hazardous areas. The third paper explores gender and measurement bias in agricultural surveys with implications for how evaluators measure agricultural outcomes. Agricultural surveys across the developing world typically interview only one respondent about the characteristics, inputs, and outputs of plots. In households where multiple members make agricultural decisions, this practice may introduce measurement error. While recent work on household assets, income, and decision-making shows substantial disagreement between spouses, the extent to which this disagreement is biased has not been assessed. To address this challenge, this paper presents data from individual interviews conducted with both husbands and wives about all farm plots in 1,243 households in Ghana and an independent observation generated using GPS plot walks. Disagreement between spouses was found on even the most basic aspects of farm plots, including the number of plots and decision-making roles. Moreover, results show significant gender-related bias in reports when matching survey data to independent observations of plot size and distance from houses as measured by GPS, as well as whether a plot is fallowed as observed via satellite imagery. Men are differentially likely to overstate farm production, especially on plots for which they report having decision-making responsibilities. In addition, men are more likely than women to report harvests, planting, and inter-cropping on plots observed as fallow through remote sensing.