When The State Cannot Tell Who Qualifies
On 23 June 2026 the OBR published its tenth Welfare Trends Report. The number that hit the front pages was eight point five per cent. That is the share of Universal Credit spending in 2025–26 that left the Department for Work and Pensions in the wrong pocket — paid to claimants who were not entitled, or in amounts they were not entitled to. Before the pandemic the rate was nine point four per cent. In the worst Covid year of 2021–22 it spiked to fourteen point seven per cent. Among claimants who joined UC in the first twelve months of the pandemic — the so-called "Covid cohort" — the rate hit twenty-five point eight per cent in 2020–21.
This is the cleanest number in British public administration. It is also a confession. The state has spent a decade digitising welfare, hired ten thousand more fraud investigators, and still cannot tell, for one pound in twelve, whether the person on the other end of the payment is entitled to it. The OBR's own conclusion is the opposite of reassuring: the rate has "fallen back to 3.2 per cent, close to its pre-pandemic level" — close, but never quite back. An economist from Vienna in 1920 would have said this was inevitable. So would an economist from London in 1944. So would an economist from anywhere who has read either.
What the Story Claims
The official narrative is operational success against a backdrop of stress. Universal Credit was rolled out fully in 2019–20; it now accounts for ninety-one per cent of combined UC and legacy welfare spending, against thirty-one per cent six years earlier. The pandemic forced the DWP to relax checks — minimum-income-floor tests, self-employed verification, face-to-face appointments — to get money out fast. That decision drove the spike. The subsequent fall, from 14.7 per cent to 8.5 per cent, is presented as the recovery: more staff (4,700 in 2020–21 rising to 15,000 in 2025–26), real-time PAYE earnings data, monthly reassessment, and tighter capital and housing tests.
The OBR is clear that the all-welfare fraud-and-error rate has fallen back to 3.2 per cent, close to the pre-pandemic level of 3.1 per cent, and that the Covid cohort rate has fallen by two-thirds over five years. The older problem remains: the rate never gets below the rate the central planner can compute. Every structural reform has bought improvement in one place by creating new leakage in another. Capital and housing overpayments averaged 0.4 per cent of UC and legacy spending between 2011–12 and 2018–19. They averaged 2.1 per cent from 2019–20 onward. The new system made the housing element more honest about capital, and opened up a fresh fraud-and-error channel worth more than £1.7 billion in 2025–26 (per OBR Table 3.1).
The Austrian Diagnosis
The Austrian school of economics, a tradition going back to Viennese thinkers of the late 1800s and most associated today with writers like Ludwig von Mises (1881–1973), an economist who spent his career showing that centrally planned economies cannot price what central planners cannot value, and Friedrich Hayek (1899–1992), a Nobel-prize-winning economist who argued that no central authority can possess the dispersed knowledge held by millions of individuals, would not be surprised by these numbers. The diagnosis is older than Universal Credit.
The first idea is what Mises called the calculation problem: without market prices set by voluntary exchange, no one can tell whether a given allocation of resources is good or bad. Universal Credit is, in effect, a centrally administered price for the labour of not working — set by regulation, adjusted by algorithm, paid by giro. The price is bad because the office that sets it cannot know each claimant's true entitlement. The claimant knows. The landlord knows. The employer knows. The DWP only sees what the claimant reports. Eight point five per cent is the residue that survives every audit.
The second is what Hayek called the knowledge problem: useful economic knowledge is not held in any single office; it is dispersed across millions of people. Fifteen thousand fraud-and-error staff against a welfare bill that runs to roughly £300 billion a year is the visible cost of compressing that dispersed knowledge into a single ledger. It bought a six-point reduction over four years. A market-based system does not need to buy it at all, because the counterparty on the other side of every payment has the information and the incentive to police the transaction. The market prices fraud in advance. The state only reacts after the fact.
What the Official Number Does Not Show
Frédéric Bastiat (1801–1850), a French economist famous for the line "that which is seen, and that which is not seen," would have asked a simple question about the OBR's headline. What is seen: the welfare bill, the hardship, the public compassion that put the system in place. What is not seen: every marginal job that was not taken because UC made it uneconomic; every self-employed claimant who reported £400 a week because the monthly assessment made anything below that unattractive; every landlord who did not let to a UC recipient because the housing element's 2.1 per cent fraud-and-error rate made the tenancy uninsurable.
The "seen" is the right payment to the right person. The "unseen" is the productive activity that would have happened in the absence of the payment. The OBR is honest about the first. It cannot count the second, because the second never appears in any departmental ledger. The Austrian critique is not that welfare is immoral. It is that welfare administered by a central planner generates two parallel ledgers — a public one, audited and reported, and a private one, distorted and uncounted — where the eight-point-five per cent lives next to the suppressed incentive to work and the inflated housing costs.
Why This Matters for Sound Money
Rails to Freedom makes the deeper Austrian case in Part 1: the problem is not that governments spend too much; it is that they spend without price signals, and a system without prices produces both the inflation that erodes the currency and the leakage that erodes the programme. The argument in Part 3 is that Ethereum, a programmable blockchain whose transactions can be encoded with conditional logic that no fiat rail can match, offers an alternative not because it is cheaper but because it is honest. A smart contract that releases a payment when a verified oracle confirms an event — school attendance, an appointment kept, an employer-confirmed start — needs no fraud-and-error team. The logic is the audit; the audit cannot lie.
Rails to Freedom is clear in Part 4 that the on-chain economy has its own frauds and collapses. The point is narrower. The eight-point-five per cent is what central computation of entitlement looks like at its best. The on-chain alternative is conditional payment with the conditions written into the rail, auditable by anyone, reversible only by the rules the contract was deployed with. It is, in Mises's terms, a price for the transaction rather than a price for the person. Hayek's knowledge problem does not vanish, but the locus of the knowledge moves from a Whitehall office to the people closest to the transaction.
What Markets Are Already Doing
The Ethereum rails that would let a state displace central computation with conditional logic are not hypothetical. Aave, a lending protocol on Ethereum, settles billions in loans every quarter against over-collateralised on-chain positions — no human underwriter, no fraud-and-error line item, no 8.5 per cent leakage. Gitcoin Grants, a quadratic-funding mechanism on Ethereum, has distributed tens of millions of dollars to public-good projects since 2019, weighted by the number of contributors rather than the size of each donation — sybil-resistant by construction. World ID, a proof-of-personhood network, has piloted identity verification for aid distribution so humanitarian payments reach the recipient and not the broker. All three do what the DWP fraud-and-error directorate was hired to do.
This is the Austrian point, applied to a sector the Austrians rarely wrote about. The state cannot tell who qualifies because it does not have the information to tell. It hires more staff; it gets a modest reduction. It builds better rules; it opens new loopholes. The market alternative is not better auditing. It is the displacement of auditing with architecture — a rail in which the conditions of payment are public, the rules of disbursement are immutable, and the cost of fraud is paid by the fraudster and not the taxpayer. Eight point five per cent is what we pay for the privilege of pretending the state can do what only the market can.
Looking Ahead
The OBR will publish its eleventh Welfare Trends Report in 2028. It will almost certainly show a fraud-and-error rate between seven and nine per cent. The DWP will hire another thousand staff. The OBR will, once again, note that the rate has fallen back close to pre-pandemic levels. The taxpayer will absorb the difference, as it has since the welfare state was founded. The Austrian diagnosis is not that this is anyone's fault: central planning of entitlement looks like this at the limit of its information — and the limit is structural, not operational.
The on-chain alternative is not a vote against welfare. It is a vote for welfare that cannot leak — a rail in which the conditions of payment are written in code, the audit is the architecture, and the eight-point-five per cent becomes somebody else's problem. The technology exists. The maths is proven. What is missing is the political permission to try.