Global DPI models: Lessons from India, Brazil, and beyond

Brazil Digital Currencies Digital Policy Economy & Business India International Markets

Issue Brief October 25, 2024

Global DPI models: Lessons from India, Brazil, and beyond

By Barbara Kotschwar and Colin Colter

The concept of digital public infrastructure (DPI), while relatively new, has rapidly gained traction among policymakers. Countries around the world have long attempted to digitize government service delivery. Some wealthier nations in the Global North build atop legacy systems that include public and private actors that offer essential goods and services, such as identification, payments, civil registration and vital statistics (CRVS), and data exchange.

In contrast, low- and middle-income countries of the Global South have built novel indigenous systems with new technologies and best practices, leapfrogging the Global North’s digital government systems. Models such as India’s highlight DPI’s potential as a tool for financial inclusion and economic development. Because of its initial success, DPI has gained traction as other Global South countries embark on their own DPI projects or adopt technology from counterparts such as India and Brazil, which offer open-architecture access. In contrast to some legacy systems, these new digital goods aim to employ open standards and protocols, be interoperable and non-excludable, use federated architecture, engage privacy by design, and offer the digital equivalent of physical infrastructure in providing access to each country’s overall digital economy. In building these sui generis digital systems, Global South countries are rethinking how to balance public and private-sector involvement, regulations for interoperability, the appropriate role and limits of markets, how to create trust in institutions, and how to build consequentially inclusive digital government goods and services.

Two key recent events have accelerated the interest in digital public infrastructure: the COVID-19 pandemic and India’s presidency of the Group of Twenty (G20) Summit. The COVID-19 pandemic exposed vulnerabilities and the urgent need for scalable, digital services, accelerating countries’ investments in offering goods and services digitally.
¹ India’s leadership at the G20 further galvanized this movement, positioning DPI as a crucial global conversation. International and multilateral groups from the Group of Seven (G7), the Quad, and now the United Nations (UN) Global Digital Compact have all been working to address and define DPI.
² The European Parliament is now holding conferences on a “Euro Stack” as it explores new ways to assert digital sovereignty and create equitable access to the digital economy.³

In the context of these developments, the Atlantic Council’s South Asia Center assembled working groups to research and discuss the definition of digital public infrastructure, what makes it “new,” learnings from historical examples, and key questions going forward. These working groups comprised digital government, trade, innovation, payments, foreign policy, industrial policy, and internet experts. They kindly shared their time and insights in a series of meetings and panels. The working groups also conducted structured interviews with DPI experts around the world. We graciously thank all participants and commentators for their openness and time given to the project.

This issue brief establishes the background of DPI development, discusses existing examples of DPI, and provides the policy recommendations essential for the next stage of DPI exploration, implementation and deployment. This brief is followed by two papers on cybersecurity and financial inclusion, addressing the fundamental issues affecting the development of DPI.

DPI: Scope and definition

The nomenclature “digital public infrastructure” is a new entrant in the government technology literature, but projects bearing elements of what we now consider DPI have existed for decades.⁴ As technologies have changed, so have the definitions of what is public and what is infrastructure.⁵ Although often credited to the Nobel Prize-winning economist Paul Samuelson, the concept of public goods extends back to John Stuart Mill, Italian writer Ugo Mazzola, and Swedish economist Knut Wicksell.⁶ Samuelson extended this definition to include non-rivalry (i.e., one person’s use of a good does not diminish another person’s use) and non-excludability (i.e., everyone has equal access to that good, such as air).⁷ Digital public goods (DPGs) are open-source software packages meant for governments to build digital tools that broadly fit these two criteria. ⁸ Not all DPI projects use DPGs, and using open-source software is neither necessary nor sufficient for a government DPI project to be truly open protocol, transparent, interoperable, and reusable, as many definitions of DPI require. The definition of DPI remains a topic of a vibrant and ongoing debate. For the purposes of this paper, we use the Global DPI Repository definition: “interoperable, open, and inclusive systems supported by technology to provide essential, society-wide public and private services.”

What is new about DPI?

DPI assumes that citizens have a right to access basic features of a country’s digital economy. These features typically include identification, civil registration and vital statistics, payments, and data exchange. Proponents of DPI argue that markets have not, or have not properly, provided these products and services to all members of the digital economy. Therefore, they argue, the government should step in to provide these basic goods and services in ways that allow societal reach.

The DPI movement asks what the role of the state should be in the digital economy. Has the private sector failed in delivering goods and services to the world’s poorest and most underserved?

Government involvement in payments systems provides the clearest example of digital domains historically run by the private sector, but there are also emerging attempts to deploy DPI models for open commerce, ride hailing, and even decentralized compute.

India stack

India’s leadership of the 2023 G20 catapulted the conversation about these digital government offerings to the international stage. The New Delhi Leaders’ Declaration defined DPI as “a set of shared digital systems that are secure and interoperable, built on open technologies, to deliver equitable access to public and/or private services at a societal scale.”⁹ Over the past decade, India has digitally and financially included millions of people and built a system of digital goods around a core group of IDs, payments, and data exchange. This set of government-led digital products, known as India Stack, enables access to the Indian domestic digital economy.¹⁰

As a result, India is often seen as the model for DPI implementation due to the successful launches of its Aadhaar digital identity platform, Unified Payments Interface (UPI) instant payments system, and civil registration services. Its ability to scale DPI is attributed to its large population, technological expertise, low mobile data costs, and supportive political and economic conditions.

For New Delhi, the development of the stack has been a project for more than a decade. In 2010, the Indian government launched Aadhaar, a biometric digital identity system. Enrollment centers across the country collected face and retina scans, fingerprints, and demographic data. The government of India has tied a variety of government benefits and account setups to Aadhaar, such as setting up a bank account. Although legally optional, having an Aadhaar card remains imperative for access to the digital economy in India.¹¹ In less than a decade, more than 1.3 billion people, nearly 90 percent of the population, have joined Aadhaar.

Digital payments form the second layer in the stack. The Central Bank’s Pradhan Mantri Jan Dhan Yojana (PMJDY), a project to bring a bank account to all Indian households, opened accounts for 166 million people in its first year. This grew to 510 million by the end of 2023, according to India’s Ministry of Finance. This allowed the introduction of the UPI, a new layer of the retail payments systems that allowed banks to exchange messages with each other and with non-bank firms, capitalizing on the financial technology (fintech) innovators that had developed tools to cheaply and easily store and transfer funds.¹²

Anyone with access to the system—including consumers and small merchants who previously found it difficult to make and receive payments—could now send or receive payments for goods and services through a digital app. A crucial feature of this system was its intra-system interoperability; users were able to transact with all actors on the UPI rails. The government accomplished this interoperability by establishing a single application programming interface (API)-based rail along which all payments providers had to transact.

Data verification and consented exchange represent the third layer of the stack. The Data Empowerment and Protection Architecture, launched in 2020, aims to facilitate the seamless and consent-based exchange of personal data. An Account Aggregator framework aims to enable atomized control over one’s personal data, whereby each individual user can track and consent to different digital actors accessing said data.¹³

India Stack has experienced several challenges and controversies since its rollout. The growth of account ownership hit a lull with the pandemic, even declining slightly from 80 to 78 percent by 2021. Some doubts have also been raised regarding the universal utility of the accounts, as India has one of the world’s highest percentages of inactive accounts. Additionally, several high-profile data breaches have raised concerns about the security of the Aadhaar system.¹⁴ Experts have flagged the potential security risks of centralizing such large quantities of identity information and the possibility that saved fingerprints could be used improperly.¹⁵ Others have worried that governments could use such large databases as tools to track and surveil citizens.¹⁶

Brazil’s digital payments system

Brazil’s new payment system, Pix, is another example of a DPI. Launched in November 2020 by the Central Bank of Brazil (BCB), the Pix electronic payment system aimed to reduce reliance on cash, increase financial inclusion, strengthen competition, and reduce the cost and ease the acceptance for merchants. Several features have ensured the success of Pix. First, the BCB made participation by banks and payments providers mandatory, allowing peer-to-peer usage to increase over time. Second, Pix payments settle in three seconds on average, faster than credit or debit cards.¹⁷ Third, the BCB set zero-fee transaction costs for individuals, with a cost to the merchant of 0.33 percent of the transaction amount. These dynamics led the largest banks operating in Brazil to work together to develop the network in a way that mandated interoperability. The BCB also established a Pix Forum, in which users and stakeholders can have a dialogue through its implementation cycle.¹⁸

Pix has had impressive results since its rollout. In its first two years, 140 million Brazilians—nearly 80 percent of the adult population—used Pix. By the end of 2022, more than 3 billion transactions took place on Pix per month, five times more than credit and debit cards. The head of Brazil’s central bank, Roberto Campos Neto, famously declared that the Pix system would result in “credit cards ceasing to exist at some point soon.”¹⁹ Pix has led to the growth of non-bank payments fintechs and a decrease in the price of payments.²⁰

In terms of its governance, Pix is more centralized than India Stack. The BCB both owns and operates the payment scheme, as well as the user address database that contains user identification. The BCB is also the regulator for the overall payments system and, thus, the regulator for Pix. The central bank has added several features to Pix since its inception, including payment scheduling, access to third-party payment providers, and the ability to withdraw cash from automated teller machines. Pix has also been the subject of controversy. As volumes and online account ownership have increased, so have instances of cybercrime and fraud.²¹

Estonia’s digital highway: The X-Road

More than half of the Estonian population voted in the 2023 election from their home computers. Estonia’s digitized government system—which allows access to government services, e-health records, and secure digital identity—made this feat possible.

Estonia’s DPI project began in the 1990s with a decision to rebuild the country’s economy on a digital basis. Through the so-called “Tiigerhüpe,” or tiger leap program, the government used public-private partnerships to invest in network infrastructure to modernize Estonia’s post-Soviet infrastructure, including providing internet to all Estonian schools and government agencies. An electronic identification (e-ID) program followed suit. The electronic governance platform also includes digital voting, an e-file system for access to the judicial system, and the government cloud, which, through partnerships with private companies, has put 99 percent of public services online.

The X-Road system represents the key infrastructure behind Estonia’s digital government. A secure data-exchange platform that connects more than 450 public and private-sector organizations, X-Road enables more than three thousand digital services. The Nordic Institute for Interoperability Solutions—a nonprofit organization created by the governments of Estonia, Iceland, and Finland—now manages the X-Road platform and its international adoption projects. More than twenty countries have adapted or plan to adapt the X-Road program through open-source access.

Working with the private sector proved essential for the success of the Estonian experiment. While the government ideated the digital ID card early in the 1990s, the first digital IDs were “only good for scratching the ice off the windshield of a car,” according to one of their developers.²² Working with banks to improve the user experience and creating incentives to use the cards proved essential to the system’s ultimate success.²³

In contrast to the payment systems in Brazil or India, X-Road has no single point of failure. X-Road’s peer-to-peer architecture is focused primarily on resiliency.²⁴ Because X-Road allows Estonia’s public agencies to share data securely with each other, every ministry manages access to its own database, which ensures data are not stored in a common pool that could become a single point of failure.

As with India and Brazil, Estonia has faced cyber threats to its DPI system. In 2007, the country faced a weeks-long attack by cyber criminals in which all government services were taken down. This led Estonia to develop a data embassy, which created a backup of critical data and services stored in a remote location.²⁵

The role of payment systems in DPI

India, Brazil, and Estonia offer distinct yet instructive models for implementing DPI. Their unique experiences reflect the different regulatory, technological, and governance choices that countries can make. India’s society-level approach, Brazil’s emphasis on speed and accessibility, and Estonia’s integration of security and privacy into digital services each provide lessons for how digital infrastructure can be developed to meet local needs. The recurring challenges of cybersecurity and privacy standards across these examples illustrate the need for secure and resilient digital architecture. These examples set the stage for a deeper exploration of a key component of DPI: payment systems.

Why study payments?

The term “payments” means moving value between actors across businesses, consumers, and governments. It is the process or service of exchanging units of value and was historically led by the private sector (e.g., by banks or merchants). Money is a discrete unit of value and governments historically play the role of enforcing that it has been spent only once at a time. Moving value digitally incurs transaction costs.²⁶ Someone must facilitate, clear, settle, and assume risk in the movement of that value from one account to another. Running a cash-based system also incurs costs to both the operator and the users of cash.²⁷ The advent of government-offered payment rails, such as India’s UPI and Brazil’s PIX, has raised new questions about the lines between the state and its citizens, the definition of “public” in public goods, and the long-term direction of financial exchange in the digital economy.²⁸

Types of DPI payments systems

For the purposes of this brief, we identified four major types of DPI payments systems. These different forms come from the different payment instruments they support and the participants among which they can transact payment instruments.

All the below DPI payments systems involve actors moving value between them and either charging to do so (via interchange) or being funded by some other mechanism (e.g., government funding/subsidies or value-add services such as telecommunications subscription services). Each of these methods is either:

a low-cost system linking a handful of banks or other institutions to do account-to-account (A2A) payments;
government-led facilitation like public rails (e.g., UPI or a central bank digital currency (CBDC)); or
involving entities that do not directly monetize the payment flow itself because they monetize another aspect of customer interaction (e.g., M-PESA).

In each of these examples, either government funding plays a role in facilitating the transaction or retail banks cover the cost and the issuing bank charges for the service. In all, some alternate source of funding (whether government subsidy or cross-subsidy) maintains the rails.

Cross-domain payment systems

The first payments system is the interoperable, or cross-domain, system. Cross-domain systems allow all accredited financial actors to transact payment instruments in near real time and on equal or progressive cost footing. They ideally allow for all-to-all switching, clearing, and exchange of instruments within one system between banks, microfinance institutions (MFIs), mobile money operators (MMOs), savings and credit cooperatives, and a government’s central bank. Cross-domain payment systems represent the aspirational goal of many DPI programs because they allow the most payments interoperability in an economy.

Bank instant payment systems

A bank instant payment system (IPS) allows for the instant messaging and transaction of payments instruments between member banks. Thus, this system only allows for transactions involving instruments associated with bank accounts (e.g., debit or credit electronic funds transfers). To facilitate instant payments between other parties, those parties would need to partner with a bank that is a member of the bank IPS.

Interoperable mobile money payments systems

Interoperable mobile money operator (I-MMO) payment systems allow for the messaging and transaction of payments instruments between or within mobile operators. These systems typically work in e-money instruments and are often run by the private sector. In contrast to centralized DPI payment systems, these I-MMO payments rely on a series of multilateral and bilateral agreements between MMOs to facilitate the transfer of funds between them. Sometimes the MMOs act as indirect participants in the instant payments system via a bank that is a direct participant in the settlement infrastructure (e.g., PesaLink in Kenya). Note that this type of interoperable payment system contrasts with closed-loop payment systems such as Venmo, in which a customer can only transact with other in-network participants. The ability of MMOs to move these e-money instruments depends on the legal architecture of the country in which they operate and whether it facilitates such private-to-private exchanges between non-banks.²⁹

Central bank digital currencies

Retail CBDCs are a way for governments to issue fiat as digital legal tender. The advent of blockchain and cryptocurrencies has increased interest in CBDCs as this kind of ledgered cryptography can increase the security of storing fiat digitally. CBDCs can be seen as a type of instant payments DPI. Both retail CBDCs and traditional DPI payments software systems can use cryptography and APIs to ensure security and accessibility. Because they both reduce transaction costs, they can enable the creation of private-sector entrants and increased competition. In contrast to other digital payments systems, CBDCs represent a claim on the central bank, not on the intermediaries.

Cross-border DPI systems

In a regional DPI system, various countries group together to allow for instant payments transactions across borders and, sometimes, between different currencies. In the case of regional DPI payments, clearing either occurs through an agreed-upon central bank or through a third-party hub. Each participant (MMOs, MFIs, commercial banks) connects to the hub either directly or through a national switch.

For example, two of the three regional IPS in Africa—Pan-African Payment and Settlement System (PAPSS) and Groupement Interbancaire Monétique de l’Afrique Centrale (GIMACPAY)—use hub arrangements, while Transactions Cleared on an Immediate Basis (TCIB) follows a hub-switch arrangement.

The public-private divide

As some countries look to use open-source software to build indigenous payments systems, they must make sure to use the correct tools to scale, meet their goals, and continue to innovate as consumer needs change. However, the political systems and civil society underpinning software design and implementation arguably play a larger role in determining that DPI system’s success than the technology itself. This working group emphasizes that a diversity of institutions and balanced trade-offs can create long-term sustainable payments systems that both include and serve their end customers.

Working group policy recommendations

Governance

While financial inclusion remains important, sound internal governance and oversight of DPI projects are paramount for their long-term success.³⁰ Governance will determine the success of DPI projects in serving all communities and replicating success globally. The working group members with experience studying industrial policy and trade protection raised questions about the role of the central bank and its mandate in a particular jurisdiction, and how to resolve potential conflicts of interest when governments act as both regulator and operator. The group felt that governance should be designed to enhance public-private collaboration to encourage competition and innovation, as well as to safeguard against government favoring specific technologies (that is, technology neutrality) and to prevent crowding out of private-sector solutions.

Cost

As excellent research from the World Bank’s Project FASTT group shows, cash-based systems also incur costs (on cash providers as well as users). It is, therefore, essential that governments and private-sector players are aware of the costs of upgrading and digitization, as well as the costs of opting out of these efforts. Research on pricing, interchange, and consumer elasticity in financial products can help illuminate the conversation on free or low-cost, instant, push-payments systems.

To ensure the success of DPI, particularly in the realm of financial inclusion, it is essential to enable digital readiness by investing in key infrastructure like internet access and cellular networks while also rigorously evaluating a country’s preparedness for digital transformation. This digital readiness should be complemented by strong privacy and cybersecurity frameworks that ensure user trust, safety, and resilience. By implementing internationally recognized standards for data privacy and cybersecurity (such as the National Institute of Standards and Technology or International Organization for Standardization frameworks), countries can safeguard user data, promote transparency, and ensure that financial inclusion efforts are secure, inclusive, and sustainable for the long term.

Design for users

DPI should be inclusive, affordable, and able to address digital divides. It should prioritize users and their needs like literacy, accessibility, and fraud protection.³¹ Consumer preferences and design should be at the center of DPI improvement, which will require continuous monitoring and evaluation even as these technologies are deployed.

Transparency, citizen involvement, and accountability are keys to a successful implementation. Sharing scheme rules and uptake data builds trust and establishes independent progress evaluation. Much of the leading research on DPI and instant payment systems comes from stakeholder interviews and not from public-access websites.³²

In conclusion, the exploration of digital public infrastructure (DPI) across various national models highlights the transformative potential of these systems in addressing key societal needs such as financial inclusion and service delivery. Countries from both the Global North and South are shaping DPI to suit their respective context, with developing nations often trying innovative indigenous systems. As India’s leadership at the G20 and Brazil’s Pix system show, DPI offers a critical tool for digital governance, enabling broad and public and private access to essential services. This working group’s findings underscore the need for continued international collaboration, robust governance, and strong privacy-oriented cybersecurity frameworks to ensure longevity and inclusivity in DPI. Policymakers and stakeholders ought to focus on building resilient, and interoperable systems to fully harness the benefits of DPI.

About the authors

Authors & working group co-chairs

Barbara Kotschwar, Georgetown University
Colin Colter, Atlantic Council

Working group members

Rob Atkinson, ITIF
Ravi Shankar Chaturvedi, Tufts University
Dan Chenok, IBM Center for The Business of Government
David Eaves, University College London
Arya Goel, ASG
Jeff Lande, The Lande Group & Atlantic Council
Mark Linscott, Atlantic Council
Srujan Palkar, Atlantic Council
Aparna Pande, Hudson Institute
Anand Raghuraman, Mastercard
Susan Ritchie
Kati Suominen, Nextrade Group
Atman M Trivedi, ASG & Atlantic Council
Tiffany Wong, ASG

Share

Issue briefs and reports

Fast Thinking

New Atlanticist

Inflection Points

UkraineAlert

MENASource

Econographics

Global DPI models: Lessons from India, Brazil, and beyond