Benefit fraud and unpaid tax: how big are the losses?
Benefit fraud: £6.8bn a year (2025/26). The total tax gap, HMRC estimate: £59.2bn (2024/25). Both are real, both are estimates, and neither is a pot that could all be recovered — here are the big numbers on one scale, with which ones sit inside which.
The question
How do benefit fraud, benefit errors and unpaid tax compare—and what can these amounts tell us?
What is true
Benefit fraud is real money, not a rounding error: £6.8bn a year, part of £9.9bn in total overpayments (fraud, claimant error and official error combined) — a genuine, worthwhile thing to scrutinise.
Put it in scale
Does the coverage match the scale of the losses?
Benefit fraud has received sustained coverage. The research below describes earlier periods; it does not tell us whether benefit fraud receives more attention than unpaid tax today.
What researchers counted
A study of 2008–2017 examined benefit-fraud reporting across newspapers, television, radio and BBC online. It found extensive coverage across these outlets. It does not provide a current, like-for-like comparison with reporting on unpaid tax.
Survey-based evidence · British Social Attitudes · 2016
What people remembered seeing
People were asked how much benefit-fraud coverage they remembered from the previous six months. This measures recalled exposure, not the number of stories published. It cannot establish how much tax reporting appeared.
Sources and how a current comparison would work
- Neil T. Gavin, Below the radar (2021): coverage during 2008–2017.
- British Social Attitudes 34: tax avoidance and benefit manipulation, using the 2016 survey. Historical survey of adults in Britain; question wording and research limitations are in the chapter.
A current study needs the same outlets and dates for both subjects, agreed search terms, duplicate removal and checks of relevance. Tax avoidance, evasion and the total tax gap must be recorded separately. Story counts, prominence and audience reach should not be combined into one unsupported measure. No current comparison has been completed here.
What we know / don’t know
Which of these losses receives the most media and political attention, relative to its actual size?
Historical studies document benefit-fraud coverage, but do not establish a current like-for-like count against tax-gap coverage. Article counts, audience reach and remembered coverage answer different questions.
The verdict
Benefit fraud is real, but far smaller than the tax gap and several other lines on this chart.
Every number on this card
| Figure | What it counts | Geography · period | Status | Source |
|---|---|---|---|---|
| £146.1bn | State PensionContributory pension expenditure, Great Britain, 2025/26 | Great Britain · 2025/26 | KNOWN | DWP, Benefit expenditure and caseload tables |
| £145bn | Working-age and children’s benefitsIncluding in-work support, disability, housing and children | Great Britain · 2025/26 | KNOWN | DWP, Benefit expenditure and caseload tables |
| £59.2bn | Total tax gapTax theoretically due but not collected, 2024/25 | United Kingdom · 2024/25 | KNOWN | HMRC, Measuring tax gaps 2026 edition |
| £21bn | Corporation Tax gapPart of the total tax gap above — do not add | United Kingdom · 2024/25 | KNOWN | HMRC, Measuring tax gaps 2026 edition |
| £9.9bn | Benefit overpayments (fraud and error)Fraud, claimant error and official error combined | Great Britain · 2025/26 | KNOWN | DWP, Fraud and error in the benefit system, FYE 2026 |
| £6.8bn | Benefit fraudThe deliberate-fraud component of the line above — do not add | Great Britain · 2025/26 | KNOWN | DWP, Fraud and error in the benefit system, FYE 2026 |
| £1.2bn | Benefit underpaymentsEntitlement not correctly paid to claimants | Great Britain · 2025/26 | KNOWN | DWP, Fraud and error in the benefit system, FYE 2026 |
| £0.8bn | Tax avoidanceHMRC’s narrow avoidance category only, within the tax gap | United Kingdom · 2024/25 | KNOWN | HMRC, Measuring tax gaps 2026 edition |
- Several of these sit inside one another: the Corporation Tax gap and avoidance are parts of the total tax gap, and benefit fraud is part of total overpayments. Adding them double-counts.
- None of these is money that could simply be collected. The tax gap is an estimate of theoretical liability, not funds waiting in an account, and no tax authority in the world collects 100%. Benefit fraud estimates are sample-based, not detected cases.
Verified against primary sources: 2026-08-31.