Novissi, Togo’s digital emergency cash program is leveraging geospatial analytics to reach vulnerable populations
At the onset of the COVID-19 pandemic, many governments around the world launched digital emergency cash transfer programs to support the most vulnerable populations whose livelihoods had been severely disrupted by lockdowns. Among these efforts, the government of Togo led the way with the launch in 2020 of Novissi, an advanced digital cash transfer program that utilized AI-powered geospatial analytics. In collaboration with GiveDirectly, Center for Effective Global Action (CEGA), and Innovations for Poverty Action (IPA), Togo’s digital emergency cash program, Novissi, leveraged machine learning along with satellite imagery and cell phone data to identify those living in extreme poverty and enable aid transfer efforts.
Togo’s Minister of Digital Economy and Transformation highlighted that the small nation successfully distributed US$34 million as of 2022 to over 920,000 people, representing 25% of its adult population. Novissi was designed to operate without an internet connection and was compatible with 2G phones. It determined eligibility based on individuals’ professions and locations. To ensure accessibility for its targeted population, including users with 2G phones – which constitute 30% of the Togolese population – Unstructured Supplementary Service Data (UUSD) technology was employed so that users would receive mobile money payments.
Novissi’s main breakthrough was the utilization of satellite imagery and artificial intelligence to create a poverty map that ranked Togo’s districts from lowest to highest income. The program also employed call data records to generate phone numbers of individuals earning less than US$ 1.25 per day. The Togolese government sought the assistance of researchers from the University of California, Berkeley, who first leveraged machine learning to analyze information from high-resolution satellite images, focusing on indicators like roof material density and land plot size, that provided insights into local infrastructure. Furthermore, the researchers examined machine learning algorithms on phone data records, including patterns of international calls, internet usage, and mobile money balances, to estimate individual wealth and determine eligibility for COVID-19 cash transfers. Finally, GiveDirectly, a nonprofit organization that provides unconditional cash transfers to people living in extreme poverty, facilitated the distribution of funds from the Togolese government.
The success of the Novissi program in Togo provides policymakers with a new tool to effectively deliver humanitarian aid during crises, even in contexts where traditional data (e.g., national demographic survey) are unavailable. A local need-driven initiative, Novissi exemplifies how collaboration between international organizations, academics, and government can lead to effective and functional social protection programs globally. Building on the success of Novissi, the government is establishing a new ‘Interoperable Social Information System’ that will aim to lay the foundation for future social protection programs with a prioritization on data protection and cybersecurity. By harnessing new digital data sources and advanced machine learning technology, this AI-driven program offers an innovative approach to crisis response and establishes a new standard for humanitarian aid, which can be scaled across low and middle-income countries.
