Designing for Innovation in Payments

‘I don’t know what you mean by “glory,”’ Alice said. Humpty Dumpty smiled contemptuously. ‘Of course you don’t—till I tell you. I meant “there’s a nice knock-down argument for you!”’

Lewis Carroll, Through the Looking Glass

***

In Lewis Carroll’s famous children’s book Alice and Humpty Dumpty argue about birthday presents: is it better to receive gifts on one’s birthday—or one’s unbirthday? Humpty Dumpty favors the latter, noting that there are 364 more such days. Alice is befuddled. But Humpty Dumpty declares: “There’s glory for you.” Pressed to explain himself, he doesn’t tell us what the word means, but rather what it does; it ends debate. “There’s a nice knock-down argument for you!”

In Carroll’s Victorian world, “glory”—a word saturated with associations of religion, empire, and military sacrifice—serves as a knock-down argument. Today that word might very well be “innovation.” A rallying cry for technology firms, an organizing principle for policymakers, and a talisman for economists, “innovation” now carries the same air of authority without analytical precision that “glory” once held. Like glory, its strength as a knock-down argument is made more potent, rather than less, by the difficulty of pinning down what the term actually means.

In our exploration of digital rails, we encounter invocations to the innovation argument at every turn. Alongside other objectives such as financial inclusion, market competition, and digital sovereignty, innovation is routinely presented as an argument in favor of the “DPI approach” to state-built digital infrastructure. A recent report on digital payments in India and Brazil mentions “innovation” 29 times. Early procedural guidelines published in July 2016, shortly after India’s new payment system’s launch, described “exponential innovation” as one of the technology’s core value propositions. Two years later, at a launch event highlighting a suite of new features, Nandan Nilekani, one of the visionaries behind the India Stack, offered a more concrete gloss: the United Payments Interface (UPI) represents “the ability to combine a core sound payment system regulated by the banking system and having all the security and reliability that you want and have many distributed innovations at the edge – that combination is a very unique thing.” Across these moments, innovation appears in multiple guises: as a policy objective, a descriptive claim, and a justificatory argument.

But what does it actually mean to design an ecosystem for innovation? How do the “DPI principles” of interoperability and modularity advance that goal? Last month in Bangalore, together with Artha Global and the eGov Foundation, we organized a workshop to wrestle with these questions in the context of India’s payments ecosystem. This post distills key takeaways from that conversation and situates them within our broader research agenda.

I. Innovation in Payments

Money, as the traditional definition goes, is a unit of account, a store of value, and a medium of exchange. Money refers to what is exchanged. Payments, by contrast, speak to how exchange occurs—the technologies and institutions that enable value to move between parties. While money defines value, payment systems determine the conditions of speed, cost, and reliability under which value can be transferred.

The distinction is ancient. One of the oldest artifacts cited as evidence of an abstract concept of value in human societies is a baboon femur discovered on the banks of the Congo River in 1950. The so-called Ishango Bone, scored with a series of notches, dates to roughly 18,000 BCE. What those notches represent remains contested. Some archaeologists interpret them as a tally of credits and debits—possibly, as David McWilliams suggests in Money: A Story of Humanity, “the first known example of value.” But the bone itself is not evidence of money in the contemporary sense. It is better understood as a payment mechanism—a Stone Age ledger, the baboon femur analogue to a modern credit card reader. The bone records not what was owed, but the system by which obligations were tracked and settled.

That system has been reinvented many times since. In the 20th century, innovations in payments piggybacked on cascading revolutions in communications technologies: the telegraph enabled the first large-scale interbank settlement systems; mobile telephony gave rise to M-Pesa and its successors. Today, as the Bank of International Settlements observes in a recent report, “innovation in payments is proceeding rapidly.” But what exactly is this innovation?

The economics of innovation is a vast literature with no settled definition at its center. Joseph Schumpeter, writing in the early 20th century, described innovation as “creative destruction”: the process by which new technologies, products, and organizational forms displace existing ones. On this view, innovation is inherently disruptive; it advances the economy by rendering the old obsolete.

Clayton Christensen, a business school professor, complicates this picture in the Innovator’s Dilemma. Not all innovation is disruptive, he argued. Much of what successful firms do is sustaining innovation—incremental improvements to existing products that serve existing customers. Disruption, in Christensen’s account, comes from below: small firms serving overlooked markets with cheaper, simpler products that gradually improve until they overtake incumbents. The distinction matters because it suggests that the most consequential innovations often come not from the most resourced actors, but from the least expected ones—and that established firms are systematically bad at seeing them coming.

Economists working in the growth theory tradition approach innovation differently. In Robert Solow’s foundational model, innovation is what’s left over—the portion of economic growth that cannot be explained by increases in labor or capital. Solow called this residual “technological progress,” but it was essentially a confession of ignorance: we know innovation drives growth, but we cannot fully explain where it comes from. Paul Romer addressed this gap by treating innovation as endogenous—something the economy produces, not something that happens to it. His key insight was that ideas are fundamentally different from physical inputs. A machine can only be used in one place at a time; a recipe can be used by everyone simultaneously. Innovation, in Romer’s framework, is the generation of new recipes and the institutions that support idea generation (universities, intellectual property regimes, R&D investment) are therefore central to growth.

Notably, these theories try to explain innovation in competitive product markets—semiconductors, automobiles, software—where firms compete to sell goods to consumers. Payment systems are different because of the close involvement of the state in regulating an essential infrastructure element of the economy. So what does innovation mean in this context?

We suggest three objectives that give shape to the concept. First, efficiency: reducing the cost, friction, and redundancy involved in moving value from one party to another. Fewer intermediaries, less capital locked up in transit, lower transaction costs for merchants and consumers alike. Speed is one expression of efficiency, but it is an instrumental goal; what matters is that resources are used well, not that they are used fast. Second, safety and trust: ensuring that transactions settle as expected, that the system is resilient to failure, and that users can rely on the infrastructure. This is the dimension that innovation discourse tends to glide over. When policymakers invoke innovation in payments, they almost always mean efficiency gains. They rarely mean improvements in fraud prevention, identity verification, or systemic resilience—even though these are themselves sites of intensive innovation and require enormous ongoing investment. Third, accessibility: extending the system’s reach to populations and use cases previously excluded—the unbanked, small-value transactions, rural merchants. This is the inclusion dimension, the one that gives DPI its political legitimacy and distinguishes public payment infrastructure from private platform competition.

These three objectives do not naturally harmonize. Efficiency and safety can pull in opposite directions: instant settlement is efficient but makes fraud harder to reverse. Accessibility imposes costs that efficiency-minded actors would prefer to avoid. The design challenge is managing all three simultaneously—which is precisely what makes the question of how to design for innovation so much harder than the question of whether to do so.

Notably, innovation in payments is not only productive. It is also extractive. Fraud is innovation too—and arguably, it is where some of the most rapid and creative innovation in digital payments is occurring. As payment systems scale and transaction volumes grow, bad actors innovate relentlessly: new phishing techniques, synthetic identities, social engineering attacks that exploit the very speed and accessibility that make the system valuable. This is not a peripheral concern. In India, reported incidents of UPI fraud have risen alongside transaction volumes, and the tools required to combat fraud—real-time monitoring, machine learning, identity verification—demand significant and ongoing investment. The existence of fraud innovation means that “designing for innovation” cannot simply mean maximizing it. A payment system must simultaneously encourage productive innovation and contain extractive innovation, and the resources required for the latter increasingly shape who can participate in the former.

That last point is what makes the design question urgent. More efficient, more secure, and more accessible for whom—and at what cost? To think clearly about innovation in payments, we need a framework suited to infrastructure—one that asks not just “how much innovation?” but “what kind, where, for whom, and at whose expense?”

We propose a tentative framework that has three dimensions:

Institutional: who enables and governs innovation? In the conventional account, market incentives do the enabling: firms innovate because competition rewards them for doing so. But as Mariana Mazzucato has argued, the state often plays a far more active role than this picture suggests. In sectors from pharmaceuticals to semiconductors to the internet itself, public investment has underwritten the foundational research on which private innovation later builds. The state, in Mazzucato’s account, does not merely correct market failures; it shapes and creates markets. UPI can be seen as a case in point. The Indian state did not simply regulate a market for digital payments; it built the rails on which that market operates. The institutional question is therefore not whether “the state or the market enables innovation?” but “who controls the surface area for experimentation—and on what terms?”

Architecture: where in the system does innovation occur? Annabelle Gawer and Michael Cusumano have drawn a distinction between a platform’s stable core and its innovative periphery. The core provides reliability and interoperability and it is the part of the system that everyone depends on and that changes slowly. The periphery is where experimentation happens and where new applications, services, and business models emerge. The platform leader manages this boundary, deciding how open the core’s interfaces should be and how much freedom peripheral actors enjoy. In private platform ecosystems—Apple’s iOS, Intel’s PC architecture – this boundary is a competitive strategy. In public infrastructure, it is a policy choice with distributional consequences. DPI adopts a version of this architecture: public rails at the core, private innovation at the edges. But the analogy is imperfect. Design decisions about interoperability and modularity determine not just whether innovation happens, but where in the system it occurs and who captures the value it creates.

Market structure: who has the resources and incentives to innovate? This is one of the oldest questions in innovation economics. Kenneth Arrow argued that competition drives innovation: monopolists, already earning rents, have less incentive to invest in something that might cannibalize their existing products. Joseph Schumpeter argued the opposite: that large firms with market power are better positioned to take on the risks and costs of R&D, and that the expectation of temporary monopoly profits is what motivates innovation in the first place. The debate remains unresolved because both logics operate simultaneously; which one dominates depends on the structure of the market in question. In payments, this tension is mediated by pricing. UPI’s zero-fee regime accelerates adoption, but it also compresses margins, raising a basic question for private actors: where are the returns to innovation supposed to come from? When margins are thin or nonexistent, only players with existing scale or adjacent revenue streams can afford to invest, which can reinforce concentration. The market structure question, then, is not simply “how many firms are competing?” but “does the economic architecture of the system give diverse actors the resources and incentives to innovate?”

These three dimensions are not independent. Institutional choices shape architecture; architecture constrains market structure; market structure feeds back into institutional politics. But distinguishing them is analytically useful because each identifies a different site where a critical problem can take hold: lock-in.

Infrastructure systems tend to consolidate over time. Early phases are often marked by experimentation and rapid entry; mature phases, by standardization and concentration. Some consolidation is the natural result of network effects and economies of scale, and is not inherently problematic. But in payment systems, lock-in carries distinctive risks. In private platform markets lock-in can be a competitive strategy and its welfare effects are debated. But, payment infrastructure is built with public purpose. When lock-in emerges in a system designed to be open, funded by public subsidy, the entities capturing rents have not earned them through competitive innovation. The rents flow from their architectural position. And because payment systems have redistributive consequences—determining which merchants bear what costs, which populations get served, how the gains from digitization are distributed across an economy—lock-in in payments is a public finance question as much as a competition question.

The risk is that getting these design choices wrong produces lock-in that concentrates the benefits of innovation among a small set of actors rather than diffusing them broadly. Each dimension identifies a different form of lock-in—institutional capture, architectural rigidity, economic concentration—and each requires a different set of policy responses.

II. Bangalore Workshop on DPI & Payment Innovation

Our workshop in Bangalore brought together senior policymakers, fintech entrepreneurs, and think tank researchers to test these ideas against India’s payments ecosystem. Over the past decade, that ecosystem has been shaped above all by UPI—an interoperable, open API that has transformed everyday payments in India and brought 504 million users into the digital economy. As UPI scales toward its next phase, with an ambition to double its user base and reach one billion users, we encouraged participants—under Chatham House rules—to step back from immediate implementation challenges and examine UPI’s innovation story more holistically. What emerged was a picture of an ecosystem in transition: widely credited with early dynamism, now confronting the constraints that its design choices have produced.

Architecture: Innovation at Core and Edge

We began by asking participants what innovation in UPI actually looks like. The responses clustered at two distinct levels of the system’s architecture.

At the core, participants pointed to improvements in the payment layer itself: the transition from cash to digital payments, gains in reliability and speed, and the layering of new features on top of the original UPI API—UPI Lite for low-value offline transactions, Hello UPI for voice-based payments, credit lines on UPI. The user experience of bank-to-bank payments has become “10x better” since UPI’s introduction, as one participant put it, partly thanks to the “user-centric” approach adopted by UPI’s designers. These are cumulative innovations in the sense described above: they build on existing infrastructure rather than displacing it, and their value lies in making the system more efficient, safer, and more accessible over time.

At the edge, participants described a different kind of innovation—not within payments but enabled by them. Credit, insurance, brokerage, and e-commerce have all been transformed by UPI’s data exhaust and low-cost transaction infrastructure. Two out of three brokerage accounts in the country were opened between 2020 and 2025. Growth in credit and personal loans was described as “only possible because of the low cost of UPI.” Here, the payment system functions less as a site of innovation than as a platform for it—a stable core generating the conditions for experimentation at the periphery.

Yet participants also noted a tension within this architecture. Despite a rapid shift from cash to digital payments, India is “still behind” on digital payment penetration compared to systems like Brazil’s Pix. And the ecosystem, several participants observed, has “become static.” The early years of UPI were marked by experimentation and new entrants; the current phase feels more consolidated. This is the architectural form of lock-in: as the core matures and stabilizes, the boundary between core and periphery hardens, and the space available for peripheral innovation narrows. The question is whether this consolidation is a natural phase of infrastructure maturation—or a design problem that policy can address.

Market Structure: Competition, Pricing, and Incentives

The conversation turned next to the conditions that enable and constrain innovation at the market level. Three features of the UPI ecosystem featured prominently: market concentration, interoperability, and zero-fee pricing. What emerged was a picture of design choices that had functioned as enablers in UPI’s early phase but now cut in both directions.

Concentration. Market concentration emerged more quickly than policymakers anticipated. Proposed interventions such as a 30 percent market share cap have been delayed and may ultimately be abandoned. This prompted a more fundamental question from participants: is maximal competition in payments necessarily conducive to innovation? Would a “balanced ecosystem” require growth to 200 apps, as one participant proposed? Or are hundreds of marginal players competing in the payments space less conducive to innovation than a small number of firms with the scale, margins, and incentives needed to invest? “Do we need to reexamine our attitude toward monopolies?” one participant asked. The Arrow-Schumpeter tension, in other words, is a very live policy debate in India’s payments ecosystem.

Interoperability. Interoperability is often defended instrumentally, as a means of increasing competition by lowering barriers to entry. But if robust competition hasn’t actually materialized, then what is the rationale for the claim that “interoperability drives innovation,” as one participant put it? Some participants emphasized that contestability is valuable in itself; users benefit from the low switching costs that interoperability enables. But if so, the value of interoperability may be connected less to innovation than to consumer welfare or market fairness—a different policy objective altogether.

Zero-fee pricing. UPI’s zero-fee pricing regime sits at the center of ongoing debates over merchant discount rates (MDR) in India. Participants emphasized that pricing cannot be analyzed independently of interoperability. If interoperability functions as a push factor—lowering barriers to entry by opening the network—zero-fee pricing acts as a countervailing pull factor by compressing margins. While zero fees accelerate adoption and reinforce UPI’s status as infrastructure, they raise a basic question for private actors: where are the returns to innovation supposed to materialize?

Think of interoperability and pricing as hot and cold temperature knobs, jointly shaping the “temperature” of market competition. The difficulty is that there is no clear sense of the optimal temperature if innovation is the objective. Moreover, the answer is unlikely to be uniform across the different layers of innovation—at the core, at the edge, in adjacent markets.

The implications of pricing become more acute as UPI enters its next phase. Scaling to one billion users and transaction volumes at three to four times GDP, as one participant described the ambition, will further strain infrastructure already under pressure. Episodes of server downtime have increased. Incidences of fraud have risen. The next decade of UPI will require massive investment in infrastructure and fraud prevention tools. That investment depends on credible pathways to returns. Revisiting the MDR regime therefore becomes not merely an abstract question of fairness, but an urgent one of system sustainability. This is the market structure form of lock-in: when pricing compresses margins to zero, only actors with existing scale or adjacent revenue streams can afford to invest, and concentration deepens.

Institutional: Who Innovates?

A final theme concerned the role of the state itself. Several participants noted that NPCI has become a very profitable actor in India’s fintech ecosystem. This observation raised a deeper question about the division of labor between public and private actors. If the infrastructure institution captures a growing share of value, what space remains for private-sector innovation?

“Public rails, private innovation” is the founding motto of UPI and of digital public infrastructure more broadly. But the workshop suggested that in the Indian case, a more apt description might be “public rails, public innovation, private service delivery.” NPCI does not merely maintain the tracks; it decides which new stations get built, which routes are opened, and which services are permitted. Many of the features that have defined UPI’s evolution—UPI Lite, UPI 123, credit on UPI—were developed or directed by NPCI, not by private actors competing at the edges. Private players, meanwhile, increasingly function as distribution channels rather than innovators in their own right.

This is the institutional form of lock-in. When the entity that governs the platform is also the entity that innovates on the platform, the distinction between infrastructure and application collapses. The ecosystem puts at risk the distributed experimentation that the “public rails, private innovation” model is supposed to enable. And because NPCI’s position is quasi-public—not quite a regulator nor a private firm—the usual tools for addressing this kind of concentration (antitrust enforcement, market entry) do not straightforwardly apply. “The catalytical role of the state” in the payments ecosystem, as one participant described it, may itself be the innovation most in need of examination.

III. The Road Ahead

The conversation in Bangalore helped crystallize a set of questions that point toward a broader research agenda on innovation in payments and digital public infrastructure.

First, innovation and payments beyond UPI. The discussion underscored that payment systems resist the frameworks we typically use to analyze innovation. They are not competitive product markets; they are regulated infrastructures with deep entanglements with state authority, and the existing literature has not fully reckoned with what that difference means. As payments develop atop communication infrastructures, they are often shaped by concerns about stability, inclusion, and national security. Developing an account of innovation that takes these features seriously remains an open task.

Second, rethinking “public rails, private innovation.” An implication of our Bangalore workshop was that the familiar slogan of “public rails, private innovation” may be too simple. In practice, public and quasi-public actors are often innovators in their own right, while private actors may focus less on infrastructure and more on service delivery. This suggests a shift from thinking about innovation primarily as concerning the actions of private actors. What role do state-created features like interoperability and modularity play in building capacity across an ecosystem? And what would a coherent theory of innovation for DPI look like if it took these institutional dynamics seriously?

Like Humpty Dumpty’s “glory,” innovation in payments policy often functions as a word designed to end debate rather than advance it. The conversation in Bangalore suggests a different approach. Innovation in payments is not a single thing to be maximized. It is a layered phenomenon—shaped by institutional arrangements, architectural choices, and market structures that interact in ways policymakers cannot afford to ignore. Treating innovation as an object of analysis rather than an article of faith does not weaken the case for digital public infrastructure. It strengthens it, by making explicit the tradeoffs that are always present but rarely named in the design of systems that increasingly mediate economic life.