India, Digital Economy: The State Built the Rails.

Digital payments scaled; cash grew anyway.

The state builds the rails first.

Estonia, Brazil, and India all arrive at the same conclusion from different starting points. When a country wants a digital economy to work nationwide, the state builds the shared rails first. These are the public pipes that move money between any bank and any app. The market then builds on top of them. It is an institutional design logic, not a technological trick. India is a case that tests that logic rather than simply repeating it. Here, digital payments scaled to record volumes while cash use also grew to a record — the opposite of the falling-cash path Brazil took with Pix. If the pattern only held where cash collapsed, it would be a weak pattern. India is where it must earn its keep.

To see why, start with one night. In November 2016, the Indian government declared its two largest banknotes worthless overnight — they were no longer valid money — voiding roughly 86 percent of the cash in circulation. It is the moment everyone remembers and that most accounts treat as the cause of India’s payments revolution. This blog returns to that night in every section. The more closely you look at it, the less it explains. Consider what each section takes up. The bank-led network could not catch the fallout; the non-profit utility did. The identity layer made the rails usable. The shock forced traffic onto those rails. The cash came roaring back regardless. And the same design accelerated concentration. The shock is real. It is just not the architect.

India replaced a fragmented system wholesale.

A bank-led network already tried this — and it collapsed.

Before 2008, India’s retail payments ran on cash, on card networks built for someone else’s market, and on bank-transfer systems that barely spoke to one another. Fragmentation was the default, not an accident. And it was not for want of trying: the banks themselves had attempted to build a shared, interoperable network and failed.

Swadhan, a bank-led interoperable automated teller machine (ATM) network, launched in 1997 and shut down by 2003. It died of the coordination failure that dogs any voluntary industry consortium. No single bank could get the others to invest, share, and standardize in ways that ran counter to their own competitive interests. The alternatives were no better suited to a mass market. International card networks asked merchants to buy point-of-sale terminals costing $120 to $150 each. On top of that came merchant fees of one to two percent on every swipe — economics that simply do not work for a tea stall. Bank-transfer rails like National Electronic Funds Transfer (NEFT) and Real-Time Gross Settlement (RTGS) ran in silos, demanded full account details, and settled with a delay.

Now run the 2016 shock against that system. Void 86 percent of the cash overnight in a country wired this way, and the displaced volume has nowhere to go. There is no interoperable rail to route it onto, no cheap acceptance device in the merchant’s hand, no shared address to send money to. The contrast is the whole point. Without an answer already in place, the shock everyone credits would have been a catastrophe, not a catalyst. The question is who built that answer, and how.

What drove the shift?

A non-profit utility broke the bank logjam.

The answer was an institution, not an app. In 2008, under the Payment and Settlement Systems Act, the Reserve Bank of India (RBI) and the Indian Banks’ Association jointly created the National Payments Corporation of India (NPCI). NPCI is easy to mistake for something it is not, and the headline of this piece can mislead. NPCI is not a government department. It is a bank-owned non-profit, incorporated under the section of company law reserved for not-for-profit entities. Ten promoter banks initially capitalized it, and it operates under RBI oversight. “The state built the rails” means the state designed the mandate and forced the cooperation, not that a ministry owns the pipes.

Two design choices made the utility self-sustaining.

Two design choices did the work. First, the RBI transferred the existing National Financial Switch — the backbone of the country’s ATM network — to NPCI at book value. That made the new entity self-sustaining from its first day, rather than dependent on years of subsidy. Second, the mandate bound NPCI to run like a public utility. It charges only enough to cover its costs plus a small margin. And it is structurally barred from seeking the kind of profit that would turn a network into a walled garden. A walled garden is a closed system that locks users in and keeps competitors out. That combination is what the bank consortium of the 1990s could never achieve on its own. A neutral utility, owned by all the banks but beholden to none of them, could set a standard and compel everyone to use it.

The result is coordination at a scale the voluntary model never reached: by April 2025, 668 banks were participating on a single NPCI interface. That is the number that matters — not any single app’s user count, but the count of institutions that agreed to interoperate. A non-profit utility, backed by regulators, removed the incentive to defect. Swadhan tried to get banks to cooperate and failed. NPCI changed who owned the problem.

Identity came first, payments second, by design.

Rails are useless if people cannot get onto them. So, the deeper move in the Indian design was to solve identity before payments. Aadhaar, the population-scale digital identity system, collapsed the cost of verifying a customer — electronic know-your-customer checks fell from about $15 per person to roughly seven cents. At that price, opening an account for someone with no banking history stops being a loss-leader. The Jan Dhan accounts program then brought more than 500 million previously unbanked people into the formal system. Cheap identity is what made mass inclusion arithmetically possible. Reliable, cheap connectivity supplied the other half: Reliance Jio’s 4G rollout drove data costs down far enough to put the rail in a poor household’s pocket.

UPI hid the plumbing behind two simple handles.

Only then did NPCI layer payments on top. The Unified Payments Interface (UPI) added two abstractions that hid the plumbing. A virtual payment address (VPA) is a simple, human-readable handle (like a username) that stands in for a bank account number. People can pay each other without exchanging account and routing details. An interoperable quick response (QR) code does the same for merchants. It turns a printed square of paper into an acceptance device and eliminates the $120 terminal entirely.

This is the “hourglass.” Picture a wide base of banks and a wide top of consumer apps, joined by a deliberately narrow waist of open, public standards in the middle. Those standards are the only thing the state mandates; everything above and below is free to compete. Because the waist is open and public, any third-party app can plug into the rail without building its own pipes. That is precisely what unbundled the bank account from the interface a customer touches. The bank still holds the money; an app it does not own now owns the relationship. That unbundling is the engine of contestability. As the next section shows, it is also the source of the design’s deepest tension.

The 2016 shock drove traffic onto existing rail lines.

Now we can put the famous night in its place. That overnight cancellation of the banknotes — demonetization — was real and forceful. Voiding 86 percent of circulating cash forced cash-dependent households and merchants to adopt digital payments just to keep buying and selling. And the trial stuck. A careful study found that the places most dependent on cash before the shock saw the largest and most lasting jump in digital payments afterward. Those habits did not snap back once the government reintroduced cash. COVID-19 reinforced the habit a few years later, when contactless suddenly meant safe. And the zero merchant discount rate (Zero-MDR) mandate set the transaction cost to zero for both sides. That kept the smallest payments economically viable on the rail — which matters in a market where 86 percent of merchant transactions are under 500 rupees.

So, the shock did something. What it did not do was build anything. Look at the timeline and the causal story inverts: the Immediate Payment Service (IMPS), the real-time interbank rail, launched in 2010. RuPay, the domestic card network, arrived in 2012. NPCI designed UPI itself in 2016, before the demonetization announcement, not in response to it. The scaling began before the shock and continued long after it. The shock was an accelerant poured onto rails the state had already laid — and an accelerant needs something to burn. This is the hinge of the whole argument. Strip out the pre-built rails and the identity layer, and the shock has nowhere to send the displaced volume. We already saw what that looks like. It looks like 2003.

What the state’s design produced.

Cash did not die, exposing what the rails did.

Here is where India breaks the pattern. The standard story of a payments revolution ends with cash in retreat. India does not. At the very peak of UPI adoption, the total cash in people’s hands and tills hit a record 41.6 trillion rupees. That was up 11.9 percent in the 2025–26 fiscal year. Cash did not merely survive; it grew, and it grew fast. More tellingly, per-capita cash grew by 9.0 percent, while per-capita gross domestic product (GDP) grew by 9.4 percent. The two track each other almost exactly. Cash demand is rising in step with the economy, not collapsing under digital pressure.

Digital and cash split the market; they did not trade it.

That single correlation dismantles the substitution narrative. Digital payments did not replace cash; they carved out a segment of the transaction market and dominated it. UPI owns the small, everyday payment under about six dollars — the cup of tea, the bus or taxi fare, the vegetable stall. Cash holds the rest: the informal economy, the precautionary store of value, the transactions people prefer to keep off any ledger. The result is a dual-track economy, two systems growing at once because they do different jobs.

But be careful: this is segmentation of the transaction market, not a measured shift of the economy from informal to formal. The cash-tracks-GDP correlation is strong evidence for the former and says nothing direct about the latter. The data show that none of the companion cases share the same finding. Brazil’s cash use fell as Pix rose, which is the tidy story. India’s rose alongside digital, which is the harder and more interesting one. That is why India tests the cross-case pattern instead of merely confirming it. The rails did not kill cash. They built a second economy beside it.

The same design built a Leviathan and an oligopoly.

The costs of this model are not bugs to be patched later. They are intrinsic — they follow from the same design choices that made it work, and an honest account must hold both at once.

Start with the coordinator. NPCI is simultaneously the standard-setter, the scheme owner, and — through its own BHIM app — a competitor in the very market it referees. A body cannot be a neutral umpire and a player at the same time. Then look at who won the customer-facing apps. The same open design that let anyone compete also allowed just two firms to dominate. Google Pay and PhonePe now own the apps that customers open to pay. Open competition in principle produced a two-firm grip in practice. And Zero-MDR, the policy that kept micro-payments alive, did so by erasing the fee banks used to recover their costs. It effectively subsidized Big Tech platforms, which monetize behavioral data and cross-sell rather than transactions. The state spread the cost of the rail across the public. Then it handed the most valuable ground on it to firms that profit from data, not payments.

The deepest cost is a surveillance record without a law to govern it.

Which leads to the deepest cost. A nationwide payment system generates a population-scale record of behavior. India built that record faster than it built the law to govern it. The evidence is blunt about severe, ongoing gaps in comprehensive data-protection legislation. This is the gap that earns the system its “Digital Leviathan” label. It is the capacity for continuous behavioral surveillance, by state and corporation alike, running ahead of any framework to constrain it. At the unbundled edge, customers deal with an app rather than a bank. Advice gaps and “reverse-phishing” scams flourish there precisely because the design removed the security-rich bank relationship. And the biometric and device dependence that lets a laborer prove identity with a fingerprint also excludes the laborer whose fingerprints cannot be read. None of this is incidental. It is the bill for the architecture.

What Latin America can take from this.

The lesson: sequence the foundation, and price the rails honestly.

Return one last time to that night in 2016. It is memorable because it looks like the cause — a single dramatic act, a country pushed onto digital rails overnight. The payoff of reading the case closely is that it was only an accelerant on rails the state had already built across the preceding decade. The transferable lesson is the rails, not the shock. No country can schedule its own demonetization and expect this outcome. What it can do is build the foundation that lets any shock — or no shock at all — convert into durable adoption.

This is also where India earns its place in the series. Estonia and Brazil established the pattern: the state builds the shared infrastructure first, and the market follows. India confirms and extends it. Its dual-track Cash Paradox shows digital and cash growing together, while Brazil’s cash use has fallen. Three independent jurisdictions, three different institutional forms, one underlying logic. The pattern has now been tested, not just asserted, across cases that do not otherwise resemble each other.

That last point carries a warning for Latin America. Institutional coordination was the necessary condition for India’s transformation — but necessary is not sufficient, and it is not a unique recipe. Brazil reached comparable open-rail scale through a different institutional form, a central bank operating the platform directly rather than a bank-owned non-profit. So, the NPCI design is one path to the coordination, not the only path. What does not vary across the successful cases is that the state somehow forced coordination. And that is exactly what most jurisdictions lack:

Four conditions decide whether the model travels.

• Coercive state capacity. The model runs on the raw institutional power to force concentrated, profitable banking oligopolies into a non-profit interoperability mandate. It also requires standing up a trusted, population-scale identity rail. Many states across Latin America and the Caribbean (LAC) lack the first. And civil-society resistance to centralized biometric identity, rooted in real histories of state surveillance, makes the second a live political obstacle rather than a technical one.
• Honest pricing of the rails. Zero-fee buy-adoption by spreading the cost of the infrastructure across the public. But the bill lands on banks, and the design hands the customer-facing apps to data-monetizing Big Tech. The evidence shows this dynamic harming small and regional banks and credit unions in both India and Costa Rica. Regulators must choose a deliberate cost-recovery and concentration-cap design at the outset. Retrofitting one once two apps already dominate the market is far harder.
• Sequencing. Identity and inclusion accounts must come before payment rails. Build the roof before the foundation, and you get fragmented, low-trust adoption — which is the failure mode, not the exception.
• The environmental and sovereignty bill. A 24/7, high-throughput rail forces migration to energy-intensive cloud infrastructure. For LAC states with grid instability or no sovereign data centers, this raises digital-sovereignty and environmental trade-offs. India, with its scale and resources, could absorb them more easily than smaller economies can.

Get those four issues right, and the rails carry the traffic. Get them wrong, and the open technical standards alone replicate nothing. That is the real lesson India offers a region looking for a shortcut. There isn’t one. There is only the foundation, built in the right order, and paid for honestly.