Digital Public Infrastructure in Low-Income Countries: What the DPI Framework Actually Delivers

Digital public infrastructure — the term of art for the shared, interoperable digital systems that governments build or enable at national scale — has become one of the organizing concepts of international digital development policy over the past several years. The three-layer stack most commonly referenced consists of digital identity systems, digital payment rails, and data-exchange platforms enabling interoperability across government services. This “DPI stack,” as it is sometimes called, is promoted by the World Bank, the G20, the Bill and Melinda Gates Foundation, UNDP, and a growing set of bilateral donors as a foundational layer for digital government, financial inclusion, and social protection delivery.

The concept has real substance. India’s experience with Aadhaar, UPI, and DigiLocker — often cited as the reference case for DPI-at-scale — demonstrates that large-scale shared digital infrastructure can change how government services reach citizens, how financial transactions are processed, and how data moves across previously siloed public institutions. The question that the development-informatics literature is now beginning to examine seriously is: what happens when DPI frameworks designed in high-income or middle-income contexts are applied as a template in low-income country settings with different governance conditions, different state capacity, and different infrastructure constraints?

What “DPI” Actually Encompasses

The DPI concept covers heterogeneous systems that are often grouped together in policy documents but differ substantially in their technical requirements, governance demands, and implementation risks.

Digital identity systems assign unique identifiers to individuals and allow those identifiers to be verified electronically across services. They range from relatively simple biometric ID programs — recording fingerprints and iris scans in a national registry — to sophisticated federated identity architectures allowing citizens to authenticate their identity across multiple government and private-sector platforms. The governance implications differ sharply across this range.

Payment rails are the shared infrastructure for digital financial transactions: real-time payment systems, interbank clearing, mobile money interoperability frameworks. Some of the most successful DPI examples are in this category: UPI in India, Pix in Brazil, TIPS in Tanzania. These systems require central bank governance, regulatory frameworks for participating financial institutions, and technical standards for interoperability.

Data exchange platforms — sometimes called “consent layers” or “data empowerment” architectures — allow individuals to authorize the sharing of their data across institutions. These are the least mature DPI layer globally and require the most sophisticated governance infrastructure to function as intended.

The Evidence From Low-Income Country Deployments

The World Bank’s Global Digital Public Infrastructure Program, which supports over 80 countries, published a results summary in May 2026 documenting outcomes across several country programs. The evidence across low-income country deployments is mixed in instructive ways.

Ethiopia’s Fayda digital ID system enrolled over 36.6 million residents as of the reporting period, with a notable finding on gender: the program explicitly addressed a 15-percentage-point gender gap in registration. Complementary programs supporting digital payment transitions under a government-to-person payment initiative shifted 1.3 million households from cash to digital payment channels.

Nigeria’s National Identification Number program passed 100 million enrollments, explicitly including persons with disabilities in coverage metrics. The National Social Safety Net program places cash-out points within 60 minutes’ travel of enrolled households — a practical operationalization of financial access that is more meaningful than raw enrollment numbers.

The Philippines’ PhilSys identity program reported 80 million enrolled individuals (87 percent of the eligible population), 60 million IDs issued, 8 million new bank accounts opened by previously unbanked individuals, benefit processing time cut from four days to one, and business permit processing reduced by 80 percent.

These are meaningful numbers. But several methodological cautions apply. These results come from program monitoring data, not from independent impact evaluations with appropriate counterfactuals. Attribution — separating the effect of the DPI from contemporaneous changes in economic conditions, other government programs, or the private sector — is not established in the results summaries. Self-reported program results from implementing agencies have structural incentives for optimism.

What the DPI Framework Gets Right — and What It Elides

The DPI framing succeeds in several respects that prior e-government frameworks did not. It explicitly centers interoperability: the goal is not a government portal but shared infrastructure that multiple services can use. It explicitly centers inclusion: enrolling previously excluded populations — women, people with disabilities, rural residents, refugees — is a stated design goal, not an afterthought. And it has achieved genuine policy traction: the G20 framework on DPI, endorsed at the 2023 and 2024 summits, has created political attention and financing flows that prior digital development frameworks lacked.

Where the DPI framework is weaker — and where the development-informatics research community has the most useful contributions to make — is on the governance conditions required for DPI to deliver inclusion rather than surveillance.

The UCL Institute for Innovation and Public Purpose’s 2025 global DPI landscape assessment noted that progress has been “global but uneven,” with low-income countries consistently underperforming relative to the framework’s aspirations. The IIPP report identified the same constraints that have characterized digital development challenges for decades: limited technical capacity in implementing agencies, weak regulatory frameworks, insufficient investment in cybersecurity, and governance structures that allow DPI to be captured by ruling-party interests rather than serving citizens’ interests broadly.

The cybersecurity gap is particularly striking. By the World Bank’s own assessment in 2025, only 20 percent of low-income countries have a fully functional Computer Security Incident Response Team — the basic institutional infrastructure for responding to security incidents in national digital systems. DPI built on this foundation carries substantial risk: a large-scale identity or payment system breach in a country without the institutional capacity to detect and respond to it can cause serious and lasting harm to the people enrolled in the system.

The Governance Question the DPI Framework Underweights

There is a structural tension in how DPI is promoted internationally that the development-informatics literature needs to engage more directly.

The DPI framework is heavily promoted by international donors and multilateral organizations whose primary interest is in getting digital infrastructure built and achieving enrollment numbers. The governance conditions that determine whether that infrastructure serves citizens or concentrates state power over them are harder to measure, take longer to develop, and are more politically sensitive than the technical build. They therefore receive less attention in program design and evaluation.

Research from across the e-government literature — including work published at IDIA conferences and in journals such as Information Technology for Development and the Journal of Information Technology — documents repeated examples of digital government systems that achieved high enrollment and technical functionality while failing to deliver their intended inclusion benefits, or actively harming the people they were supposed to serve. Aadhaar-linked exclusions from India’s food subsidy programs (documented by Jean Drèze and colleagues at J-PAL) are the most cited example, but similar dynamics have been observed in social protection digitization in Kenya, Ghana, and several other countries.

The DPI framework’s language of “open,” “interoperable,” and “inclusive” infrastructure does not guarantee these governance outcomes. Whether the infrastructure actually serves the people enrolled in it depends on the legal frameworks governing data access and use, the accountability mechanisms available when the system produces errors, and the political relationships between the state, donors, and citizens that shape how the infrastructure is actually operated.

Further reading from authoritative sources:

• World Bank Global Digital Public Infrastructure Program — results summary covering Ethiopia, Nigeria, Philippines, and other DPI program countries.
• UNDP Digital Public Infrastructure — UNDP’s DPI playbook and approach for low- and middle-income countries, including an inclusion and rights-based framework.