Biometrics – such as fingerprints, voices, faces or iris patterns – can be used to establish the unique identify of an individual person. The need to provide people with identification is a huge issue given estimates that nearly one in seven people in the world have no way to prove their identity, an issue the World Bank hopes to address with the launch of a new High Level Advisory Council to lead its initiative to provide IDs for everybody.

Currently, a number of individual social protection schemes are collecting fingerprints – through sub-contracted payment service providers – as part of the enrolment of beneficiaries onto programmes. They store these biometrics on their databases and verify beneficiaries – in other words, checking those presenting themselves for payment against this list – by using fingerprint-enabled point-of-sale devices each payment day. But I would argue that this is a costly, ineffective and unsustainable means of implementing an electronic payment mechanism for social protection schemes.

When social protection programmes themselves capture and store biometrics, this cannot prevent people from registering more than once. Individual programmes are unlikely to be able to eliminate any duplicate records. This process, ‘de-duplication,’ requires the use of expensive proprietary algorithms protected by large software vendors which is rarely accessed by social protection programmes.

A more robust solution would be for payment service providers of social protection programmes to collaborate with national ID agencies. Such agencies have the capability to build databases that can identify each citizen using multiple forms of biometrics, while also performing de-duplication. After building such a database, the task remaining would be to verify that those presenting themselves for payment are the unique, registered individuals.

Links can be established between ID databases and social protection databases to allow lists of registered beneficiaries to be compared and checked against an existing national list of unique individuals. When real-time links cannot be established, social protection programmes could request biometric data from ID agencies so they can make offline cross-checks of their lists.

A recent study by Consult Hyperion and FSD Africa – Biometrics in Digital Financial Services: An Overview – has concluded that payment service providers predominantly use biometric technology for the latter task of verification. This means that they do not need the expensive, proprietary software that is necessary to determine whether or not potential beneficiaries have registered more than once, but can take advantage of this service through national ID databases. While fingerprint readers and iris scanners have come down significantly in price in recent years, the software provided by largescale vendors has not. Therefore, it is not necessary for social protection programmes to embark on the expensive process of collecting fingerprints.

It would be more sensible is to obtain fingerprints from ID agencies, and then personalise them on a smart-card. A fingerprint reader at a point-of-sale system can then be used to determine whether fingerprints belong to a certain individual or not. As the Consult Hyperion and FSD Africa report underscores, it is practicable “to use biometrics for verification, to help authenticate that a service user is who they claim to be”.

Richard Chirchir has designed and built Management Information Systems for social protection programmes across Africa and Asia including the Social Assistance Grant for Empowerment MIS in Uganda, the MIS for Kenya’s four main SP programmes, and providing advice on Pakistan’s Benazir Income Support Programme.

Consult Hyperion and FSD Africa make a further important point: “There is no form of biometrics that can provide 100% accuracy, even the much-touted ‘finger vein technology.’ This is often proposed as an alternative to overcome the limitations of collecting fingerprints presented by sweaty, dry, or aged fingers.

India’s Aadhaar national identity scheme has responded to this challenge by capturing ten fingerprints and two irises when registering people to provide them with identification. Social protection programmes should also respond to the small risk that people might not be accurately matched against an existing record and captured biometric.  Best practice verification is to match a person with a stored unique record using two methods, and not just biometrics. Kenya’s Hunger Safety Net Programme has, for example, proposed the use of personal identification numbers as a backup to the fingerprint verification mechanism.

Electronic payment systems, requiring two methods by which somebody proves who their identity, ensure the robust, secure delivery of social protection systems, one vital part of the management information system solution for effective, equitable programmes.