Every Rupee, But Only for Grain: Two Governments Define Programmable Money

Two governments have now written down what programmable money is, and they reached opposite conclusions about whether a state should be permitted to issue it. The European Commission's draft regulation on the digital euro, COM(2023) 369, disposes of the question in eight words at Article 24(2): "The digital euro shall not be programmable money." The proposal defines what it is forbidding: units that, "due to intrinsically defined spending conditions", can be used only to buy particular goods or services, or that expire. On 15 February 2026, at Mahatma Mandir in Gandhinagar, the Government of India launched a public distribution system built on precisely that. The Press Information Bureau's own release describes tokens "redeemable exclusively for purchase of entitled foodgrains", producing what it calls "purpose-bound usage of subsidy". It names the achievement without euphemism: a "programmable sovereign digital payment layer" inside the ration system. Part II of these papers argued that the ledger was the state's first instrument of centralisation: that legibility precedes control, and that the register of who owns what gets built before the power to take it. Programmable money collapses those two steps into one object. The record and the permission become the same artefact. On this series' reading that is not an incremental change in payment technology; it is the first time the instrument of account has also been the instrument of enforcement, at the level of the individual transaction, at national scale. What makes the European text worth reading closely is not the prohibition but the exception sitting beside it. The same regulation permits conditional payments, and defines them in Article 2 as transactions instructed automatically upon conditions "agreed by the payer and by the payee". Recital 55 lists what that covers: standing orders, automated top-ups, machines paying for their own spare parts, machines buying electricity at favourable market conditions. Europe is not banning clever money. It is drawing a line between conditions two parties negotiate and conditions the issuer imposes on the holder. Payer and payee may bind themselves; the central bank may not bind them. That is a decentralisation principle stated in legislative drafting, and it is sharper than most of what the crypto industry has produced on the subject in fifteen years. Now the part the coverage of India's pilot did not reach. The scale-up path announced from the Gandhinagar stage runs through Puducherry, Chandigarh, and Dadra & Nagar Haveli. Those three are not arbitrary. They are the only jurisdictions in India that have implemented the National Food Security Act in cash mode, since 2015 and 2016, under the Cash Transfer of Food Subsidy Rules, 2015: Rules whose own definition is explicit about why the money is cash: it exists "to enable purchase of the entitled quantity of foodgrains from the open market". The government confirmed the arrangement in a Rajya Sabha written reply on 5 February 2021. And the standfirst of the Puducherry launch release, on 26 February 2026, states what is replacing it: food subsidy paid into CBDC wallets "in place of conventional bank accounts". Read those documents together and the reform inverts. For a household in Puducherry this is not the digitisation of a ration queue. It is the conversion of money back into a voucher. Eleven years ago the state decided these families could be trusted with rupees and an open market. This year it decided they could be trusted with tokens and an approved merchant list. The programme's slogan is Har Dana, Har Rupiya, Har Adhikar: every grain, every rupee, every entitlement. The rupee in question can only buy grain. The strongest objection is serious and it deserves its own paragraph. Leakage from the public distribution system was never a talking point; it was catastrophic. Estimates built from consumption survey data put diversion at roughly 42 per cent in 2011-12, which is to say that two-fifths of what the state bought for the poor did not reach them. Purpose-binding works against that, mechanically and immediately, and it solves a second problem too: a Maharashtra state economist told Reuters in April 2026 that programmable disbursement lets farmers take an irrigation subsidy without fronting the money themselves, which is a real gain for people with no working capital. A token that cannot be diverted is worth something to someone who has been on the losing end of diversion for thirty years. Anyone arguing against programmable welfare owes that person an alternative, not a principle. But the arithmetic has moved, and it has moved away from the case being made. Work by Prabhu Pingali and Raghav Puri published in August 2025 tracks leakage from 41.7 per cent in 2011-12 to 24.1 per cent in 2022-23, and then to an implausible-looking 8.8 per cent in 2023-24: a figure the authors distrust themselves, noting that five states report negative leakage and putting the true range at 10 to 20 per cent. Take the pessimistic end of their own estimate. The collapse from two-fifths to one-fifth was delivered by digitised ration cards, e-POS terminals, Aadhaar authentication and One Nation One Ration Card, none of which are programmable money. The residual that programmability is being asked to close is a fraction of the problem the rhetoric invokes, and the price of closing it is charged entirely to the discretion of the poorest households in the country. Scale is why this cannot wait for the pilot to report. The Home Minister said from the Gandhinagar stage that the system would be implemented across the country within three to four years, into a distribution network covering 80 crore people, and framed it as "Minimum Government, Maximum Governance". It is hard to think of a mechanism that puts more government inside a transaction than one in which the state specifies, per token, what the token may buy. Neha Narula of the MIT Media Lab put the general case to Reuters in nine words: "This is a really dangerous road to go down." Two caveats, because this register cannot retract. Europe's prohibition is not law. Parliament's economics committee adopted its position on 23 June 2026 and the file entered trilogue after a plenary mandate on 9 July; Article 24(2) stands in the Commission's text but has not yet survived a negotiation, and a prohibition is only as good as the last compromise amendment. On the other side, India's programme is being pushed uphill rather than pulled: press accounts of the Reserve Bank's 2025-26 annual report put e-rupee in circulation at Rs 771.7 crore on 31 March 2026, down roughly a quarter year on year [UNVERIFIED: figure not read from the Bank's own report]. Voluntary retail adoption falling while compulsory welfare deployment expands tells you something about which users this currency is actually acquiring, and on what terms. The house test asks whether a function, decision or asset moves from a central authority to the edge, and what that does to the people underneath. Here the movement is legible and it runs the wrong way. The decision being relocated is small, what a household buys with its own entitlement this month, and that is exactly why it matters, because centralisation at that grain is new. A rupee in a bank account is a claim its holder directs. A token in a CBDC wallet is a permission its issuer wrote. Europe has put that difference into a statute. India has put it into production.