Advancing the uptake of AI for enhanced social protection in Africa
Though more than 45 countries on the African continent have some social protection program, such systems only account for a small portion of the population [1]. Increasingly machine learning, predictive analytics and other AI tools are being applied to problem solving. From predicting patients’ drug adherence behaviour in healthcare to providing alternative financing systems through data mining, AI is revolutionizing sectors and spaces. Could this tool be what the African continent needs to resolve limited coverage and access to social protection? At a very opportune time, during the COVID-19 pandemic, Togo embraced AI in facilitating social inclusion. Through the use of machine-learning and effective collaborations and partnerships, Togo’s Novissi, a 100% digital cash transfer solution, successfully improved access to social protection to those who need it most.
Togo, a country in West Africa with just over 8 million people has a poverty rate estimated at 45.4% [2]. Over 80% of the poorest individuals are in rural areas. The country has been in the news for its fully digital social protection system, Novissi. Novissi deployed machine learning to predict the consumption pattern of 70% of the population. This enabled the country prioritize social protection payments to the poorest individuals living in the poorest areas in Togo. Novissi was launched in April 2020 by the Ministry of Digital Economy and Digital Transformation (MENTD) of Togo to mitigate the economic impact of COVID-19 on Togo’s informal workers [3].
Novissi has reached 819, 972 beneficiaries, disbursing approximately US$23.9 million (13,308,224,040 FCFA) as at March 2021 [3].
To increase the uptake of machine learning and AI enabled social protection systems, what can African countries learn from Togo’s Novissi program?
Target criteria setting
In rolling out targeted and digitized AI social transfer programs, poverty alleviation should be a priority of the government. The government sets the tone for the importance or non-importance of social protection and to which segment(s) of the population this protection is directed. This is dependent on the population structure and a host of socio-economic factors. Should social protection be for persons with disability who are unable to enhance their livelihoods due to limited work opportunities? Or should it be for women who are primary caregivers? Or the elderly without reliable pensions, who may be unable to return to work or increase their income earning potential? For those in the informal or formal sector? Whatever the criteria is for social protection prioritization, it should be clear to enable implementation with AI.
The target for Novissi are informal workers representing 90.4% of Togolese workers [4], majority of whose professions prevented them from working during COVID-19 pandemic lockdowns. Also, these informal workers are those living majorly in rural areas, areas where poverty in Togo is highest. These two criteria, among others, were considered in initiating Novissi. After the government determines which population segment it prioritizes for social protection, policies to institutionalize social assistance can be adopted.
Information
To broaden access to social protection via digitized channels, information and information systems are required. Information was critical for Novissi. For AI or digital solutions to work, identifying potential beneficiaries through some central registry is required (National Identification system etc.). Togo resorted to using the country’s voter registration database for Novissi [3]. Other sources of data were from beneficiary registration and disbursement monitoring sources. Novissi additionally used AI and geospatial data to aid the identification of the 100 poorest rural cantons in Togo. Information from mobile call records were fed into machine-learning algorithms to predict the consumption patterns of over 5.7 million individuals. These predictive algorithms facilitated the prioritization of beneficiaries [5].
Collaborations and partnerships
In leveraging the benefits of AI and machine learning to improve existing social protection systems, Togo collaborated with internal and external partners. These collaborations were threefold: financial support, technical assistance and disbursement assistance. In delivering the digital initiative, Togo received technical assistance from the World Bank, University of Berkeley and Northwestern University who were instrumental in providing AI and geospatial tech support [6]. For financial support, the World Bank's International Development Association (World Bank IDA) provided $72 million to Togo as part of the $400 million regional IDA operation to help finance, upgrade and improve contactless and integrated social protection payment systems [5]. For disbursement assistance, Give Directly, the philanthropy arm of the collaboration, brought together agencies and other humanitarian organizations to expand the reach of Novissi [7]. Additionally, local Telecom operators were on ground to aid in facilitating mobile money distribution.
Togolese inter-governmental collaborations created an enabling environment for the development of Novissi. From policy to budgeting and roll out, the social protection priorities of the government was a driving force for Novissi. Representatives and policy advisors from the Ministries of Digital Economy and Digital Transformation, Grassroots Development, Economy and Finance, Infrastructure, Informal Sector, among others were part of the inter-ministerial committee that mobilized for the launch of Novissi [3].
What the Novissi program has shown is that possibilities exist for African countries. Possibilities to target, broaden and improve social protection systems using AI and other smart technologies for poverty alleviation can and should be explored. The Novissi program is not without teething problems– using voters registration databases may be limiting. The voters database for instance excludes those who may be in need of social protection but are not registered voters and may include those who have passed away and are still registered voters. Additionally, continued program financing to aid the consistent roll out of transfers, among others, are similar challenges that other African countries would face. These challenges can be overcome with proper planning and decision making to prioritize social protection mechanisms, supported with regular monitoring and evaluation. With these in place, initial teething problems can be quickly resolved.
References
https://blogs.worldbank.org/africacan/covid-19-africa-how-can-social-safety-nets-help-mitigate-social-and-economic-impacts
https://www.worldbank.org/en/publication/macro-poverty-outlook/mpo_ssa
https://www.poverty-action.org/sites/default/files/publications/Togo-Novissi-Cash-Transfer-Brief-August%202021.pdf
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6755594/
https://www.worldbank.org/en/results/2021/04/13/prioritizing-the-poorest-and-most-vulnerable-in-west-africa-togo-s-novissi-platform-for-social-protection-uses-machine-l
https://www.togofirst.com/en/social/0605-7791-novissi-scheme-togo-uses-ai-to-boost-economic-inclusion
https://novissi.gouv.tg/en/givedirect-novissi-2/
Author's bio
Oluwatobi Ogundele's economics experience and interests have been in the areas of competition law and policy, health and migration. She has masters degrees in Economics and in International Public Policy with specialization in International Economic Relations and Global Governance.
Comments