Wednesday, 29 September 2010. Tembhli, a tribal village in Maharashtra’s Nandurbar district. Enrolment for the number takes a photograph, ten fingerprints and two iris scans.
The Prime Minister, Manmohan Singh, has come to launch a number. Beside him are Sonia Gandhi, who chairs the National Advisory Council, the Governor of Maharashtra, K. Sankaranarayanan, and the Chief Minister, Ashok Chavan. The first number goes to Ranjana Sonawane, a housewife from the village.
Almost thirty-two years earlier, the farmers of Xiaogang had pressed their thumbs into red ink and hidden the paper from the state. Replay of an earlier scene: Xiaogang, 24 November 1978
One thumbprint was hidden from the state. The other was given to it.
An illustration, not a facsimile. Aadhaar enrolment records a photograph, all ten fingerprints and both irises; the twelve-digit number is a placeholder.
Source: UIDAI, Aadhaar (Enrolment and Update) Regulations 2016; PTI via Governance Now
The two acts are not the same. One was a defiance and the other a registration. But each was a poor person’s mark, and each began something that reached a very large number of people. The village that hid its prints was asking the state to look away. The village that gave its prints was asking the state to see it. This chapter is about what happened after the state looked.
An authority with one job
The authority had been set up twenty months earlier, on 28 January 2009, by an executive notification under the Planning Commission, with Nandan Nilekani as its first chairman. Nilekani was a co-founder of Infosys, and his move is the last link in a thread this book has followed since the lift: he left the firm to run a public project with cabinet rank, building for people the lift never reached. The budget of 26 February 2010 gave it Rs 1,900 crore as it entered its operational phase. In April 2010 it acquired a name, Aadhaar, and a logo.
It was meant to be boring. The job was to answer one question, whether a person is who they say they are, for anyone who asked. Its first customers were meant to be people who held no proof of identity at all.
Notice what came first and what came last. The authority was running on a notification before Parliament had passed anything. In December 2011 the Standing Committee on Finance, chaired by Yashwant Sinha, considered the bill that would have given it a statute, rejected it in that form on grounds of security and privacy, and asked for a privacy and data-protection law first. The system carried on, on its executive footing. When a statute finally came, the Lok Sabha passed it on 11 March 2016 as a Money Bill, a category the Rajya Sabha cannot reject. The Leader of the Opposition in the Rajya Sabha, Ghulam Nabi Azad, was reported to have written that the route would make the house redundant. The argument that followed went to a courtroom.
Whatever one thinks of the system, the order is plain: scale first, a statute six years after the first number, and a ruling on privacy a year after that.
The rails
The identity was one layer. The next was a bank account to attach it to, and a way to move money between accounts. The Economic Survey of 2014–15 named the combination of Jan Dhan accounts, Aadhaar and mobile phones the JAM trinity, and set it out as the route to targeted cash transfers. Direct benefit transfer itself began in January 2013, and began lamely: in Andhra Pradesh it started in one district instead of the five announced, because accounts linked to Aadhaar lagged. The lesson was cheap: lay the rail before you promise the train.
Its first large test was the Andhra trial in the Registrar’s folder: biometric smartcards rolled out at random across 157 sub-districts, with payments faster, more predictable and less corrupt, and no loss of access. Hold on to that result, because the same economists found a different one elsewhere.
The government’s own site counts 59.59 crore beneficiaries banked under Jan Dhan, which is about 590 million people, though the page gives no date for the count.
The cleverest rail came last. On 11 April 2016 the UPI pilot opened with 21 banks. UPI is not an app. It is a set of public rules, developed by the National Payments Corporation of India, that lets a customer of any bank pay a customer of any other, through any app, in seconds. According to the government’s own factsheet, UPI handled about 2 crore transactions in the financial year 2016–17 and over 24,162 crore in 2025–26, a rise of roughly twelve thousand times. In September 2026 it carried 24.07 billion payments worth Rs 29.37 lakh crore, 23 percent more by number than a year before, or about 802 million a day. Per the same government factsheet, the IMF puts UPI at about 49 percent of the world’s real-time payment volume in 2024.
Linear scale.
All others (19%) is not a reported figure: it is what remains once the five named shares (49 + 14 + 8 + 7 + 3 = 81) are taken from 100.
In ten years UPI went from 17.8 million payments a year to 242 billion
Each bar is one Indian financial year, April to March. The first year is too small to see on a linear scale; switch to the ratio scale to compare growth. The strip is ACI Worldwide's count of real-time payments in 2023; 'all others' is the remainder of the five named shares.
Source: NPCI and government figures via Business Today, YourStory and others; ACI Worldwide, Prime Time for Real-Time 2024
See the numbers
| Financial year (Apr–Mar) | UPI transactions (billion) | As reported (crore) |
|---|---|---|
| 2016-17 | 0.0178 | 1.78 |
| 2017-18 | 0.915 | 91.5 |
| 2018-19 | 5.35 | 535 |
| 2019-20 | 12.52 | 1252 |
| 2020-21 | 22.33 | 2233 |
| 2021-22 | 45.97 | 4597 |
| 2022-23 | 83.75 | 8375 |
| 2023-24 | 131.16 | 13116 |
| 2024-25 | 185.87 | 18587 |
| 2025-26 | 241.62 | 24162 |
The payments authority announced a merchant fee of 0.4 percent on some person-to-merchant payments above Rs 2,000, to start on 15 October, and then deferred it until after Diwali. It is the first test of whether a rail that has been free to use can be paid for without driving people off it. A study of the universe of consumer loans found that UPI deepened finance, that fintechs extended credit to borrowers new to credit, and that there was no sign of higher default. That is a good result for a payments rule.
Who was inside
The rails changed who was inside the banking system. By the World Bank’s Global Findex, 35 percent of Indian adults had an account in 2011, against 64 percent of Chinese adults. By 2024 the shares were 89.0 and 89.4. Among women India went from 26.5 percent to 89.2. It is one of the few charts in this book in which the gold line climbs to meet the red.
India caught up with China on bank accounts, and Indian women caught up fastest
Share of adults with an account at a bank or a mobile-money provider. The World Bank's survey runs every three to four years; the lines join survey years and imply nothing in between.
Source: World Bank, Global Findex via World Development Indicators (FX.OWN.TOTL.ZS, FX.OWN.TOTL.FE.ZS), fetched 21 October 2026
See the numbers
| Survey year | India, all adults (%) | China, all adults (%) | India, women (%) | China, women (%) |
|---|---|---|---|---|
| 2011 | 35.2 | 63.8 | 26.5 | 60.0 |
| 2014 | 53.1 | 79.0 | 43.1 | 76.4 |
| 2017 | 79.9 | 79.5 | 76.6 | 75.7 |
| 2021 | 77.5 | 88.7 | 77.6 | 87.3 |
| 2024 | 89.0 | 89.4 | 89.2 | 89.2 |
Not every account is used. In 2024, 16 percent of Indian account holders had not paid in or taken out a rupee for a year. And the book’s earlier claim about the lift has to be held beside this chart. An account is not a job and a payment rail cannot teach a girl to read. The chart measures that India reached the ground floor with a bank; it does not measure what the ground floor can do once it is there.
The year the cash vanished
On the evening of 8 November 2016 the Prime Minister announced that the Rs 500 and Rs 1,000 notes, 86 percent of the cash in circulation by value, would cease to be legal tender at midnight. Minutes of the Reserve Bank’s board, released years later under the right to information, showed that the government and the bank had discussed the idea for six months, and that directors had warned of a short-term hit to growth and that black money is mostly held in gold and property rather than cash.
By the end of June 2017 about 99 percent of the banned notes were back in the banking system, and the Reserve Bank’s annual report later put the final share at 99.3 percent. The Indian Express’s reconstruction of the following months shows the stated purpose shifting as the numbers came in, from catching hoarded money to a less-cash society.
What did it cost? Chodorow-Reich, Gopinath, Mishra and Narayanan compared districts that had less cash with districts that had more. In the districts that lost more they find a fall of about two percentage points in employment and in nightlights-based output, and about two points in bank credit, in the last quarter of 2016, which they call a lower bound and which faded within a few months. They also find faster uptake of alternative payments in the districts that were hit hardest. A measured, temporary hit, then, and a push toward the very rails this chapter describes. The book records the push and does not claim that anyone planned it so.
2017New DelhiWhat the court said
On 24 August 2017 a bench of nine judges of the Supreme Court held, unanimously, that privacy is a fundamental right under Article 21. The case had been brought by a retired High Court judge named Puttaswamy against Aadhaar. The bench decided only the principle, and said that any limit must rest on a law, pursue a legitimate aim and be proportionate.
The ruling on Aadhaar itself came on 26 September 2018, by four judges to one. The majority upheld the Act and called the number a document of empowerment, but confined it to welfare delivery. It struck down the mandatory linking of Aadhaar to mobile phones and to bank accounts, struck down the section that let private companies use it to verify identity, and said schools could not make it a condition of admission. Justice D. Y. Chandrachud, dissenting, questioned the Money Bill route and pointed to biometric failures that had kept people from their benefits.
Note what the court did not do. It did not shut the system, and it did not wave it through. It named the places where the state could not reach: private companies, bank accounts, phone connections, a child’s school place. Those were the places where a person has the least say in whether to hand over a fingerprint, because refusing means losing the account or the connection or the seat. The Court also wrote the test for the future: a law, a legitimate aim and proportionality.
A democracy argued with its own database in open court, and the database lost some of the arguments. China’s argument with its payment giants took a different form, as the next cut shows.
China got to digital payments first, by another road. In early 2014 Tencent’s WeChat introduced digital red envelopes, a version of the Lunar New Year cash gift, and its rival Alibaba added a similar function to its payment service. According to reported figures, over 32 billion virtual envelopes were sent over the 2016 holiday, ten times the 2015 number. A leap, then, on private platforms.
2020ShanghaiOn 24 October 2020 Jack Ma, the founder of Alibaba and controlling owner of Ant Group, told a finance conference in Shanghai that China’s banks behaved like pawnshops. On Monday 2 November he and two Ant executives were summoned by four financial regulators. On Tuesday 3 November the Shanghai Stock Exchange suspended Ant’s listing, two days before the share sale was due. It would have raised an estimated $37 billion and valued Ant above $300 billion.
China’s mass payments grew on private platforms and the state later intervened. In India the core rules are set by a public body and private apps ride them. Whether that design can be summoned in the same way has not so far been shown, and what it has done is be argued over in court.
The cost, in a village in Simdega
In late September 2017 an eleven-year-old girl, Santoshi Kumari, died in Jharkhand’s Simdega district. Her family’s ration card had been cancelled on 22 July because it was not linked to Aadhaar, according to the state’s food minister, and the family said they had received no rice from the public distribution system for several months. Family testimony put her last stretch without food at eight days. The state government said she had died of malaria and suspended a nurse-midwife for not reporting her illness. The family denied it.
This book does not settle a cause of death. It records what is documented: a card was cancelled for want of a link, and a family went without grain for months.
The economists Karthik Muralidharan, Paul Niehaus and Sandip Sukhtankar studied what happens when a welfare programme tightens its identity checks. They found that corruption fell, at a substantial cost to legitimate beneficiaries: an estimated 1.5 to 2 million lost access at some point during the reforms, mainly, the authors judged, because of how the transition was managed rather than because of the technology. In Andhra Pradesh, where the rollout had reached 19 million people and was tested before it was tightened, they had found no such loss.
Where the savings went
The government says direct benefit transfer has saved Rs 3.48 lakh crore by reducing leakage. An earlier claim of Rs 90,000 crore a year, put forward in 2018, was shown by FactChecker.in to be a cumulative figure to March 2018. The claim is not necessarily wrong. It is a claim whose method nobody has published, made by the party that benefits from making it.
Ranjana Sonawane’s number has had a rougher journey than the system’s headline figures. In April 2025 the Nandurbar district collector, Mittali Sethi, intervened after it emerged that Sonawane’s monthly payment under Maharashtra’s Ladki Bahin scheme had been going to an IDBI Bank account in Andheri, to which her Aadhaar had been linked, and not to her own. Sethi said her Aadhaar and mobile number had been linked to a different bank account. Sonawane was paid Rs 13,500 by cheque at her home.
2026In September 2026 the same rails carried 24.07 billion payments in a month. The village’s first number and the system’s latest monthly total belong to the same history, and neither cancels the other.
Where the lift did and did not reach
Here India did leap, over cheques and cards, and put nearly nine in ten adults inside a bank, which the lift never did. The same rails carried the relief of 2020: Rs 500 a month for three months to 20.4 crore women Jan Dhan account holders, announced on 26 March 2020, in the lockdown’s first week. The state paid them by account number; chapter 25 follows the people who were on the road when it did.
But rails are not a staircase. A payment network cannot teach a girl to read, or change the law that kept a workshop at ninety-nine. What it proves is narrower. The Indian state, when it chooses, can build one boring thing for nearly everyone and keep it running for sixteen years, through two governments and a court that told it where to stop. That is not nothing. It is also not the same as building the ground floor.
In the same twelve months the Indian state built a second boring thing, and its hard part was not a database. It was a table at which the Centre could not win alone.
Door D5 · opened