It’s the Ecosystem, Stupid – Exploring the ‘Digital Poverty Stack,’ Part 1
It’s a familiar refrain these days – the failure of the development sector to leverage the potential of digital technologies.
While things like mobile phones and internet connectivity have spread rapidly in much of the world, donor investments to leverage technology have largely failed to achieve their imagined potential. Instead, these funds have produced a plethora of sub-scale pilot projects which are simplistic, bespoke, unsalable and often do not solve a real problem.
Instead, we argue that donors should adopt what we call “stack thinking” and focus more energy on building a “digital poverty stack,” a set of inter-operable and reusable digital tools that can be built into a large number of applications. This would enable innovation by the private sector, while also involving governments and NGOs that can use the resulting infrastructure in the process of delivering services.
To illustrate the concept, we look at two examples of projects that take a solution stack approach to the use of digital technology for development: the Level One Project financial services platform developed by the Bill & Melinda Gates Foundation, and the Indian government’s Aadhaar Universal ID system. Both of these projects invested significant thought and research to specify core infrastructure that would fill a fundamental need and thus be beneficial to a wide range of projects and actors. Both also focus on the core enabling digital infrastructure rather than on innovation at the edge. In a subsequent post we will lay out some of the basic tenets of stack thinking, and how they can be applied elsewhere in the sphere of international development.
The Level One Project (L1P)
Based on a number of studies, the Financial Services for the Poor (FSP) team at the Gates Foundation came to believe that financial inclusion would not be solved by simply expanding microfinance, nor by encouraging the down-market movement of traditional banks. More fundamental change is needed, requiring a redesign of the existing payments infrastructure such that it reduces the restraints which eliminate profitability for most commercial players as they try to expand to low-income customers.
The Level One Project is a model for a country-level digital financial services (DFS) system designed to dramatically reduce the cost of transactions, adjusting the economics to encourage aggressive attempts to bring the poor into the formal economy. FSP developed a working prototype for a new, lost-cost payment system infrastructure that is enabled by shared, open and standards-based digital components. The project aims to bring together an ecosystem around the technology, business models and policy recommendations that underpin this digital infrastructure. The underlying design principles of the L1P include:
- A push payment model with immediate funds transfer and same-day settlement
- Open-loop interoperability between providers
- Adherence to well-defined and adopted international standards
- Adequate system-wide shared fraud and security protection
- Efficient and proportional identity and know-your-customer requirements
- Meeting or exceeding the convenience, cost and utility of cash
These principles are critical to achieving a robust, low-cost, modular system based on interoperability and openness – one that fits within the existing payments system but substantially augments it to support more universal, high-volume/low-value, mobile use cases relevant to low-income consumers. Schematically, the L1P consists of two shared core layers:
- The Interoperability Service for Transfers (IST), which is a switch that passes payment messages from one participating digital financial services provider to another. The IST can handle net settlements among participants on a same-day basis and will operate a directory that can recognize one DFS provider from another. The IST will ensure that all transactions are covered under a common set of system rules.
- The Fraud and Risk Management Service (FRMS), which is a collaborative fraud control system that will require participating digital financial services providers to contribute data about their fraudulent and non-fraudulent accounts and transactions. These data will help create an FRMS database that will underpin a series of services, including fraud detection algorithms and scoring. The FRMS will also help participants manage compliance for anti-fraud regulations.
In addition, the system design contemplates a number of other generic components which would be built and operated by DFS providers or other third-party providers. These parts are not shared, but instead are built and deployed on a competitive basis. These components include:
- Account Opening Services (AOS): AOS is part of the digital money platforms operated by DFS providers. The AOS interacts with the FRMS to screen for fraudulent account applications.
- Consumer and Agent Management Services: These are important digital service capabilities, most commonly offered by the DFS provider. These services track the balances in individual accounts, and submit transactions to the IST for switching.
- Merchant Acquiring Service: This is a specialized type of account management system that supports the needs of merchants, including individual sellers and smallholder farmers. These services make it easier for a merchant’s customers to transact at the point of sale, and electronically provide merchants with additional tools to manage their businesses.
- Open APIs: DFS providers that expose an application programming interface (API) to third-party service developers will enable open innovation. These service providers will be attracted by the ability to accept payments from low-income consumers using a hardened, secure, reliable and available retail payments system.
This sort of stack solution for digital financial services would do much to encourage and enable DFS innovation and competition around the shared core layers. Those service providers looking to run businesses characterized by a high volume of low-value transactions could maintain reasonable costs while serving under- and unbanked customers. Competition should revolve then around the differentiation of digital financial services with an advantage given to those who can provide niche applications that add value to clients (rather than the advantage going to those who monopolize the core transactional layers, as mobile money providers currently do).
The L1P is not currently deployed in market, but the Financial Services for the Poor team is working in partnership with a number of governments to develop and deploy commercial versions of the software. It will be exciting to see how this plays out in the coming two-three years.
Meanwhile, India has seen full government participation in developing and implementing a country-level solution stack around identity and digital financial services.
Aadhaar Unique ID Number
In India, the Aadhaar digital ID number is the basis for its own stack of interconnected digital services termed the “India Stack,” which aims to improve the lives of its residents and increase the efficiency of government programs. Aadhaar was set up in January 2009 by the Unique Identification Authority of India under the leadership of Nandan Nilekani. At the core of the stack is Aadhaar, a national biometric ID platform. Registration requires the submission of demographic and biometric data, including all 10 fingerprints and scans of both irises. The information is then attached to a 12-digit individual identification number. Eventually, the government envisions using the Aadhaar identity platform as the foundation for a digital India, providing Aadhaar-enabled applications for electronic signatures, digital banking, credit registries, government services, mobile commerce and more.
The system itself is built on open-source components and standards to ensure that the ecosystem can access Aadhaar’s APIs. It offers authentication servers that can conduct over 100 million authentications per day and can serve applications deployed on any device form-factor, operating system, programming language or network. The government itself has already begun development around the core Aadhaar layer and has created a stack of services. The India Stack comprises four layers:
- A presence-less identity layer which consists of the core Aadhaar biometric ID system. Once registered on Aadhaar, citizens can remotely identify and authenticate themselves by inputting their fingerprint or iris information on a biometric reader. As of June 2016, 1.03 billion citizens had enrolled in Aadhaar. Its root functionality is the basic validation service which takes as inputs the biometric and identity numbers and validates that they match in the database. This simple function is the heart of the system on which all else is built.
- A paperless layer which enables customers to register for bank accounts, SIM cards and other services without having to provide paper forms and pen-and-ink signatures. This layer includes three key components:
- Electronic “Know-Your-Customer”: A system that allows customers to use their Aadhaar numbers and biometric information to auto-populate a bank’s core systems, converting what was a 30-day paper-based process into a digital, 30-second process.
- Digital Locker: A database that enables citizens to digitally store and retrieve loan documents, land titles, diplomas, training certificates and other documents in a single place.
- e-Sign: A digital security process that enables customers to request a secure digital signature from a central government server to remotely sign documents.
- A cashless layer which is an interoperable payments infrastructure which makes it easier for people to transact digitally without having to take cash out of the system. This layer includes four components:
- Immediate Payments Service (IMPS): A payments switch which enables mobile account holders to transfer funds to and from any account in the country.
- Aadhaar Payments Bridge (APB): A national database which maps each individual’s Aadhaar number to their bank’s routing code. Payers send payments to a recipient’s Aadhaar number and the APB routes the transaction to her bank.
- The National Unified USSD Platform: A system which enables customers to enter a USSD short-code (*99#) on any handset on any mobile network to launch their bank’s mobile banking menu.
- Unified Payment Interface (UPI): An interface which enables customers to send and receive money using a range of identifiers, including Aadhaar, mobile or bank account numbers. To enable UPI, the customer first has to map her account details via USSD, internet or Java.
- A consent layer which enables customers to digitally authorize the sharing of their personal information with service providers.
Each of these layers is interoperable and can be accessed by any application or service provider with the consent of the customer. In parallel with the aims of the L1P, the India Stack aims to provide a level playing field for new entrants looking to serve low-income customers. Examples of the leverage provided by the stack include:
- The IMPS lowering barriers to entry in India’s banking sector as new entrants can instantly tap into the network effects across India’s entire banking sector, avoiding locked-in incumbents like we see with M-PESA in Kenya.
- The National Unified USSD Platform ensuring that a customer can access the mobile banking menu of any bank, regardless of their SIM provider.
- The ABP making account portability much easier, and boosting competition among new banks vying to capture some of India’s sizable government to person (G2P) payment flows.
Additionally, much like the L1P, the India Stack is an open API stack, with the key layers built and managed at cost as a public good by the Unique Identification Authority and the National Payments Corporation of India. The leadership and support of the government has been vital to the transformation of Aadhaar from a biometric ID database to the core layer in a stack of public services helping to boost digital and financial inclusion and other priorities in the country.
We believe that donors and governments should focus more on these kinds of approaches in the future – creating public goods that enable the broader ecosystem rather than focusing the bulk of their energies on bespoke innovation in service delivery at the edge, as they often do currently.
This shift will not be easy. It took considerable leadership and financial support from organizations like the Indian government and the Gates Foundation to see these projects to completion, because digital infrastructure is risky, expensive, slow and requires a high level of standardization across different stakeholders. Thus, a shift in this direction will require new thinking and approaches.
In a subsequent post, we’ll explore some early progress in this direction by donors, including the Principles for Digital Development as well as the DIAL initiative – a major new initiative by the Bill and Melinda Gates Foundation, USAID and the Swedish International Development Cooperation Agency, which will help the field shift more toward the solution stack approach in their funding.
Photos: Screenshots from the Level One Project’s website and video. Credit: Level One Project.