Cash-based transfers represent a powerful weapon through which the governments of low- and middle-income countries can fight poverty and increase the resilience of their citizens. A key question for these programmes is how to prioritize the people that are most in need of assistance. However, this can be difficult in developing countries, where official government registries are often incomplete and out of date. The problem is particularly acute in settings where collecting information on potential beneficiaries is more challenging like in remote rural areas or during an emergency such as the COVID-19 pandemic.
New methods based on non-traditional data sources have expanded the opportunities for identifying the poor in information-scarce settings. But are these new methods a feasible and effective option for targeting social protection programmes at scale? In this webinar, Professor Joshua Blumenstock will present ongoing work that uses recent advances in machine learning, applied to data from satellites and mobile phone networks, to target and deliver cash transfers to individuals and families living in extreme poverty. These new methods have been used to support the governments of Togo and Nigeria in expanding innovative emergency programmes during the pandemic, including to rural areas. Prof. Blumenstock will provide an overview of the new methods, comparing their performance in terms of reaching the poorest to more traditional targeting options. He will also reflect on the practical, social, and ethical challenges involved in the application of these novel methods to social protection programmes at scale.
• Introduction by Marco Knowles, Senior Social Protection Officer, FAO
• Presentation by Joshua Blumenstock, Associate Professor, UC Berkeley School of Information
• Open discussion