£55–81bn a year is lost to fraud and error.
£55–81bn of your money lost to fraud and error every year. £6.6bn simply written off. About £481m ever recovered. Nobody sacked, nothing fixed, the auditors' own words, itemised.
29 figures, each with its official source.
All public-sector receipts (mostly taxes). First estimate; subject to revision.
Total managed expenditure. First estimate.
The gap between spending and income; down from £153bn the year before.
Public sector net debt, the highest as a share of GDP since the early 1960s.
Interest on the debt alone, more than the schools and defence budgets combined.
Nominal GDP; the ~£3.0tn level is derived from the published % of GDP ratios.
OUR ESTIMATE (illustrative): £110bn debt interest ÷ ~28.4m UK households ≈ £3,873 each. NOT a bill any household is sent, an equivalence to feel the scale. This is interest on the debt alone, before a penny of actual public services.
In the OBR's higher-migration scenario borrowing is ~£20bn lower and debt 3.1% of GDP lower by 2028-29, because working-age migrants lift labour supply. A scenario, not the central forecast; lower migration worsens the finances symmetrically.
Level of real GDP. But GDP PER PERSON is largely unchanged, migration grows the economy more than it shifts average living standards.
Central assumption, revised down from 295,000. Migrants are overwhelmingly working-age with a slightly higher participation rate than residents.
Net present value per main applicant (excl. Health & Care). ~72% of the gain comes from the top 30% of earners, do not generalise to all migrants.
It depends who, positive for skilled work, negative for low-wage, family & humanitarian routes Family route ≈ −£109k, Health & Care dependants ≈ −£67k, asylum/humanitarian significantly negative. The average across all routes is far below the skilled-worker figure.
Roughly DOUBLE the 2020 estimate for the London–Birmingham phase, even after the northern leg was cancelled, i.e. more money for less railway. DfT itself says the estimate is not yet fully reliable. ~20× the annual asylum cost.
Includes ~£3.8bn at Euston alone. NAO blames scope changes, inefficient delivery and underestimated timescales.
Birmingham–Manchester scrapped (Oct 2023); £100m+ just to wind down The previous government cancelled Phase 2 in 2023; simply stopping work already under way is expected to take ~3 years and cost the taxpayer an estimated £100m+ in remediation.
£37bn was the two-year BUDGET, not the outturn, actual spend was ~£26bn. Even so, PAC called it 'muddled, overstated and eye-wateringly expensive' and found no measurable impact on the pandemic. The £37bn figure alone is ~7½× the yearly asylum bill.
DHSC lost about 75% of the ~£12bn it spent on PPE in year one to inflated prices and unusable kit. ~£4bn of it can never be used in the NHS and is being disposed of, some burnt to generate power. (Note: the viral '£15bn PPE fraud' claim is a stock write-down, not fraud.)
A £5.5bn contract for 589 vehicles; NAO found the MoD 'has not demonstrated value for money' on the £3.2bn spent to Dec 2021. Initial capability arrived Nov 2025, eight years late, and was paused again in Jan 2026.
A record ~£3.7bn on consultants in one year (up ~6%). NAO (Nov 2025) warned government 'lacks a clear picture' of how much it spends, undermining a target to halve it.
17 departmental groups wrote off ~£6.6bn of spending that failed to achieve its purpose in 2024-25 alone, plus £293m+ in special payments. PAC called repeated project cancellations 'egregiously poor value for public money'.
A modelled range, the NAO stresses 'most of the potential loss goes undetected'. Whole-of-government: tax, benefits and wider spend. Each year's loss is broadly the size of the entire defence budget, and 11–16× the asylum system.
Only ~£12bn of the £55–81bn is even found (~£9bn HMRC tax, ~£2bn DWP, ~£1bn elsewhere). A further ~£41bn is estimated-but-undetected and £3–28bn is simply 'unknown'. Detection is not recovery, far less is clawed back.
About £560bn of government spend and income is subject to NO fraud-and-error measurement at all. Outside HMRC and DWP, government largely cannot see its own losses, the PSFA assumes 0.5–5% is lost invisibly (£3–28bn).
Tax owed but not collected, up from a revised 6.0% the year before. Small businesses account for ~62% of it. A single year's tax gap is over 12× the whole asylum bill.
Of which fraud £6.8bn; the rest is claimant/official error. Overwhelmingly benefits paid to UK and settled residents, asylum seekers cannot claim them. Universal Credit is the largest single source.
£10.9bn lost across furlough, Bounce Back Loans, Eat Out to Help Out and more; only ~£1.8bn recovered and much is 'beyond recovery'. Of the ~£47bn Bounce Back scheme, NAO estimates ~£4.9bn was fraudulent.
The Comptroller & Auditor General has refused to sign off DWP's accounts as regular: because of material fraud and error in benefits: every single year since 1988-89. Nearly four decades of the same unresolved failure.
The PSFA's flagship annual result, mostly loss PREVENTED, not cash returned, is roughly 0.7% of the low end of the £55–81bn lost. The scale of recovery is trivial next to the scale of loss.
OUR ESTIMATE (illustrative): £55–81bn lost to fraud & error ÷ ~28.4m UK households ≈ £1,940–£2,850 each per year. NOT a literal bill, an equivalence to feel the scale of what quietly goes missing every year, and of which barely £481m is ever recovered.