Evaluating the impact of poverty reduction programmes with machine learning and satellite images

Rigorously evaluating the impact of poverty reduction programmes is key to increase their effectiveness. Traditionally, impact evaluations of social programmes are based on field household surveys that measure the changes in wellbeing among those who receive the programme. However, household surveys are costly, time-consuming, and often logistically challenging. This is especially the case in rural areas of developing countries. How can we measure the impact of poverty reduction programmes when it is not possible to field household surveys?

In this webinar, Luna Yue Huang will show how the impact of anti-poverty programmes can be measured using only satellite images and deep learning techniques. She will present evidence from a recent poverty reduction programme in rural Kenya, comparing the performance of this new method against traditional impact evaluations based on household surveys. She will also discuss the practical advantages and the main limitations of assessing the impact of poverty reduction programmes in this new way.

SPEAKER:

Ms Luna Yue Huang recently obtained a Ph.D. in development economics from UC Berkeley, where she used to be a doctoral fellow at the Global Policy Lab. Her Ph.D. dissertation focused on how new technologies such as high-resolution satellite imagery and machine learning can help fighting key social challenges such as poverty, climate change, air pollution, and COVID-19.

Webcast