There are strong regional pushes (both external and internal) for the adoption of ‘Good ID’ across the continent. Fuelled by inspiration from the African Free Trade Area, there are numerous global initiatives supporting ‘Legal Identity for All’. And, in support of the Sustainable Development Goal (SDG) call for universal identity (ID), various multilateral agencies are providing technical support as well as funding to countries and regional bodies to help roll out biometric ID systems across Sub-Saharan Africa over the next decade.
As a spin off, Good ID proponents are calling for practical collaborations on components of ID, such as the facilitation (and easing) of verification challenges that might see people struggling to be included within such systems, which, as a direct example, could involve relaxing “know your customer” (KYC) requirements by supplanting the requirement with the provision of universal ID instead.
When phrased as an issue of financial inclusion, as they most frequently are, it is easy to see why these imperatives are stated with such urgency. But when we consider ID system issues from the perspective of data subjects first and the lived experience of Africans, it raises some public interest red flags.
Central to the African context is digital inequality. Our nationally representative survey across ten countries demonstrated that Internet penetration only averaged at around 28% in 2018. Several of the least developed countries were well below 15%, including Mozambique, Rwanda, Tanzania and Uganda.
These penetration rates are marked by extreme inequality, reflected in the urban-rural and gender divides. But, aggregated descriptive statistics of this kind mask the causes of these inequalities – education and the associated factor of income. Lower access and use of the Internet determined by figures for women reflect that they make up the majority of the poor and uneducated in countries with the highest gender disparities in the study, such as Rwanda and Mozambique.
And even once online, the experience of the digital realm, where the lack of affordable data means that usage is generally extreme low and infrequent, means that most people are using services passively and not in the high-speed, always-on environment where studies of causality in relation to penetration and economic growth have been done.
An important component for understanding why this inequality matters, in the context of digital identity, is in relation to digital literacy: being incorporated into digital identity systems is marked by passivity for those not ordinarily connected and this passivity seriously challenges notions of consent for the engagement in such processes. How will we engage this as the first step toward implementing Good ID?
Having this background in the consideration of African data subjects also helps us reassess not just that digital risk in relation to engagement and incorporation, but also purported digital dividends. A caveat to listed digital dividends for ID is that financial inclusion need not always be synonymous with ‘banking the unbanked’. Banking is a form of inclusion. It facilitates access to credit and so on – but credit is only a component of financial inclusion, with potentially negative impacts on households in terms of indebtedness.
Take the South African example, where the privatised facilitation of social grants led to increased access to financial products from the service providers’ subsidiary companies and partner bank. Civil society groups had warned that commercialisation of the delivery in this way financially overburdened many of South Africa’s poorest.
And, if we consider financial inclusion as being created through a variety of data subject benefits, cheap and efficient remittance can be facilitated outside of the banking systems evidenced in MPesa, which also offers micro credit and other mobile money services. In fact, since offering micro credit facilities, crippling indebtedness had become a major issue for East African families.
The Indian example is most obviously used as a reference point for considering digital identity. The largest biometric civil registration exercise in the world was initiated under the Aadhar system in India in 2009, which provides Indian citizen with a unique identity number. The Aadhar demonstrates that the facilitation of ID does not assist in dealing with the more central challenge to financial inclusion often referred, in infrastructure and network economies, as the “last mile problem”. This is important to note because, though it is practical to caution against assuming banking will always facilitate financial inclusion, financial exclusion in Africa is still significantly related to geography. RIA’s After Access survey showed that, though 71% of Africans surveyed did not have access to formal financial services, that figure rose to 81% of those in rural areas. And African residents who reside in urban areas (57%) are more likely to be financially included than those who live in rural areas (38%).
Yet perhaps the most important caution for contextualising data subjects needs within the realm of ID and financial inclusion is this: conflating ID utilities leads us to solutions that may contradict minimisation requirements. Why are these important? Simplified, ID data can essentially serve three functions: identity at its most direct (this is you), authentication (this is you and you permit this action), and classification (this is who you are within a particular system).
These three different purposes have very distinct data needs. KYC, as an investigation of connections between Aadhar and financial inclusion showed, deals with identification only at the point of opening an account (and later verifying changes). Unlike India where banking is nationalised, this KYC component is financially borne by the private sector. While it might be challenging to do, facilitating individual identity directly need not be through a universal and national ID, necessarily.
Compare this for instance with ID systems used to facilitate social grants. These systems aren’t just about identifying people, but also about classifying them. The scope of data required is far broader, typically including biometrics and other components. A national ID system need not, for instance, require a physical address (legal challenges in South Africa around the voters roll and physical addresses bear reference here). Yet KYC typically legally requires it. If we propose too readily that one ID ‘does all’ through a centralised system, we are proposing that an ID programme should collect all data first, and then ask questions of it later. We are assuming universal ID programs as the solution and in this case, justifying it through financial inclusion imperatives that are, in fact, not met by such a solution.
Citing the Indian example of Aadhar is problematic for trying to understand the African context in all circumstances. Aadhar was being implemented as a universal ID anyway and thus, the utilisation of Aadhar as a proxy for KYC was occurring in a context where a decision on universality has already been taken (and in the context of a nationalised banking system, which shifts where the costs and burdens lie).
And that example, in fact, demonstrated it did not practically assist in financial inclusion. That is a very different situation from using a secondary function as a justification for universality. A core component of Good ID is considering users and contexts, not universality. This doesn’t mean that ID programs with universal functions won’t be incorporated into financial solutions, it simply means we may need to reconsider justifications with the African user placed more centrally in solutions. Other fixes to KYC, such as removing the physical address requirement, are more important first steps – particularly within our regional context.
