GiveDirectly and Center for Effective Global Action – Using Data Science to Target Cash Transfers for COVID-19 Relief

CEGA used satellite imagery and call detail records (CDR) data to develop hyper-localized wealth estimates at the canton-level for individual cell phone users in rural Togo. The research team then conducted a phone-based survey of a representative sample of 8,915 cell phone subscribers across the poorest rural cantons, and used this data to train and validate a machine learning algorithm to predict an individual’s wealth from their mobile use. The government of Togo then used the algorithm to target individuals living in extreme poverty nationwide. By combining this innovative targeting approach with remote USSD enrollment and instant payments, the Togolese government provided unconditional cash transfers contactlessly to 138,585 of its citizens in most need between December 2020 and April 2021.

Researchers from CEGA and the Innovations for Poverty Action (IPA) published an article about this work in Nature, demonstrating that the phone-based targeting methodology helped the government of Togo target COVID aid more effectively than traditional targeting methods through enabling more accurate and rapid identification of the most vulnerable. The phone-based approach improved the precision of the social assistance program targeting by 42 percent relative to a naive geographic targeting of the 100 poorest cantons in Togo. GiveDirectly and CEGA published the targeting methodology behind MobileAid as an open-source codebase so that governments and NGOs can use this new methodology to improve the targeting and delivery of aid.

Using lessons learned from the work in Togo, GiveDirectly launched pilots for remote cash transfer programs (called MobileAid pilots) with additional external funding of $18.75 million in three new countries: the Democratic Republic of Congo (DRC), Malawi, and Nigeria.