Two kinds of body check benefit-spending claims. Statistics regulators step in when a precise, defined figure is stated wrongly, but not when spending is described with mood words. In the UK that work sits with the Office for Statistics Regulation and the Office for Budget Responsibility; in Canada, Statistics Canada and the Parliamentary Budget Officer.
You have heard the line: welfare spending is “spiralling out of control”. It is one of the most repeated claims in the benefits debate, and it shapes how people with disabilities are talked about and treated. So it is worth knowing who, if anyone, is supposed to referee a claim like that, and what happens when the referee decides not to blow the whistle.
In the United Kingdom, that question just got a concrete answer. This piece reports what the regulator decided, then does the more durable thing: it lays out who checks benefit-spending numbers in the UK and in Canada, and how you can check one yourself.
The decision that prompted this
The Office for Statistics Regulation, the regulatory arm of the UK Statistics Authority, had been reviewing a complaint from Disability News Service about repeated claims by senior Conservative figures that welfare spending is “out of control” or “spiralling”. The complaint pointed to figures from the Office for Budget Responsibility, the UK’s independent fiscal forecaster, showing that total spending on social security as a share of GDP is roughly stable, and is forecast to sit at the same level in 2030-31 as in 2026-27, at 11.2 per cent.
That series is worth reading in full, because it shows two things at once. Disability News Service published it alongside the complaint, sourced to the OBR’s Economic and Fiscal Outlook: the share sat at 10.9 per cent in 2025-26, rises to 11.2 per cent in 2026-27, then runs at 11.1, 11.0 and 11.1 per cent before returning to 11.2 per cent in 2030-31. Read from 2026-27, it is flat. Read from 2025-26, it rises, which is how the government’s own published figures describe the period. The House of Lords Library, drawing on the Department for Work and Pensions benefit expenditure and caseload tables of 14 April 2026, records that total UK welfare spending was £314.9 billion in 2024-25, or 10.7 per cent of GDP, and that it is forecast to rise to £333.7 billion in 2025-26 (10.9 per cent of GDP) and to £408.6 billion in 2030-31 (11.2 per cent of GDP). Both readings come from the same numbers. The difference is the year you start from, which is a point this piece comes back to.
The regulator decided not to act. It said the examples it was given “do not meet the threshold for further regulatory intervention”.
Ed Humpherson, the regulator’s head, explained the reasoning in correspondence with Disability News Service, which published his words rather than the regulator issuing a public decision. The statements, he said, “use qualitative characterisations such as ‘spiralling’ and ‘out of control’, rather than making precise factual claims about defined statistical measures”. He drew a contrast with an earlier claim, that “for the first time ever, the total welfare bill is now higher than total receipts from income tax”, which the regulator judged a “precise factual claim about defined statistical measures”. On that one, he said, “we were able to step in, and did so firmly”, and the UK Statistics Authority, acting on the regulator’s behalf, said the party should correct it after the fact-checking charity Full Fact exposed it. Loose rhetoric, in the regulator’s view, was “an expression of political opinion rather than a direct reference to, or misuse of, a specific statistical output”, and so fell outside its remit.
Whatever you make of that line, it is a useful one to understand, because it tells you exactly where the official refereeing stops. A regulator will step in on a precise, checkable number stated wrongly. It will not step in on a mood word. That leaves a large space in which numbers get characterized rather than stated, and in that space the reader is the referee.
Who checks the numbers in the UK
Two bodies do most of the work. The Office for Statistics Regulation sets the standards that official statistics must meet, through its Code of Practice for Statistics, and can challenge publicly when statistics are misused. Anyone can raise a concern with it. The Office for Budget Responsibility is the body that actually produces the forecasts of what the state spends and takes in, including the social-security figures at the centre of this dispute. When you see a claim about the “welfare bill”, the OBR’s published series is usually the primary source the claim is standing on, or departing from.
Who checks the numbers in Canada
Canada has its own version of this division of labour, and it is worth knowing before the same rhetoric arrives in a Canadian election.
Statistics Canada is the country’s central statistical office. It operates under the Statistics Act as an independent agency, and its own mandate describes producing data “using sound and objective methods, free from political or interest group pressure, with the results available to everyone”. If a claim rests on how many people receive a benefit, or how a population is changing, StatCan is generally where the underlying number comes from.
The Parliamentary Budget Officer is the closer Canadian equivalent to the OBR. The PBO provides, in its own words, “independent, authoritative and non-partisan financial and economic analysis” to Parliament, scrutinizing the government’s budget and estimates and costing proposals on request. It is an office of around 40 staff that publishes full reports showing how it reached its findings, and it is accountable to Parliament rather than the government of the day. When a Canadian politician makes a claim about the cost of a benefit or a proposed change, the PBO is often the body that can put an independent number beside it.
Neither StatCan nor the PBO is a rhetoric police. Like the UK regulator, they work on defined measures, not mood words. Which is exactly why the last section matters.
How to check a benefit claim yourself
You do not need to be an economist to pressure-test a “spiralling costs” line. Four questions do most of the work.
First, ask what the measure actually is. A raw dollar or pound figure almost always rises over time, because prices and populations rise. The more honest measure is usually spending as a share of the economy, which is why the UK dispute turned on a percentage of GDP that barely moves, rather than a total that grows. A number can go up and down at the same time, depending on which measure you pick.
Second, go to the primary source. A press release or a headline is a characterisation of a number, not the number. In the UK that means the OBR’s published outlook; in Canada, Statistics Canada’s tables or a PBO report. The bodies above all publish their workings on purpose, so the original series is usually one search away.
Third, notice the difference between a precise claim and a mood word. “Benefit spending rose 4 per cent last year” is checkable. “Benefits are out of control” is not, and as the UK case shows, that is often the point. When a claim cannot be checked, treat it as opinion, not fact.
Fourth, check the timeframe. A claim that picks a single low year as its starting point can make an ordinary rise look like an explosion. Look at the longer series the official bodies publish, not the two points someone chose for you.
Why this matters for people with disabilities
Numbers about “welfare” are rarely just numbers. They set the temperature of a debate that decides whether a person’s support is protected or cut, and how their neighbours regard them for needing it. When a claim about spiralling costs goes unchallenged, the people most likely to pay for the mood it creates are the ones who rely on the support being described.
The refereeing bodies are real, and they are more independent than the debate sometimes suggests. But they referee a narrow thing, the precise misuse of a defined figure, and they say so plainly. The wider claim, the one that does the political work, usually sits just outside their lines. That is not a gap you can fill by waiting for a regulator. It is one you fill by knowing where the primary numbers live and reading them yourself.
For the Canadian figures behind the benefits most readers here actually deal with, our disability funding map and our guide to what the Canada Disability Benefit review is asking for are good next stops, and our piece on why disability costs more in the first place is the context the “cost” debate usually leaves out.
By the Living Unlimited Team
Sources
- Disability News Service, Regulator gives Tories green light to keep saying ‘welfare is spiralling out of control’… even though it’s not, 23 July 2026
- Office for Statistics Regulation, about the regulator and the Code of Practice for Statistics
- Office for Budget Responsibility, social security spending as a share of GDP, series as reproduced by Disability News Service from the Economic and Fiscal Outlook (chart 5.2, October 2024, and table 4.1, March 2026)
- House of Lords Library, Welfare reforms and youth unemployment, 5 June 2026, section 1.1, citing Department for Work and Pensions, Benefit expenditure and caseload tables 2026, 14 April 2026, for the welfare spending totals and GDP shares
- Statistics Canada, Mandate and role: who we are and what we do
- Office of the Parliamentary Budget Officer, The Office of the Parliamentary Budget Officer
Who checks benefit-spending numbers in the United Kingdom?
Two bodies do most of the work. The Office for Statistics Regulation sets the standards that official statistics must meet, through its Code of Practice for Statistics, and can challenge publicly when statistics are misused. The Office for Budget Responsibility produces the forecasts of what the state spends and takes in, including social-security figures. Anyone can raise a concern with the regulator.
Who checks benefit-spending numbers in Canada?
Statistics Canada is the country’s central statistical office, operating under the Statistics Act as an independent agency. The Parliamentary Budget Officer is the closer Canadian equivalent to the OBR, providing independent, authoritative and non-partisan financial and economic analysis to Parliament and costing proposals on request. Neither one is a rhetoric police.
Why did the UK regulator decline to act on claims that welfare is spiralling?
The Office for Statistics Regulation said the examples it was given do not meet the threshold for further regulatory intervention. Its head, Ed Humpherson, said the statements use qualitative characterisations such as spiralling and out of control, rather than making precise factual claims about defined statistical measures, so they sat outside its remit.
Is UK welfare spending actually rising?
Both readings come from the same figures. The House of Lords Library records total UK welfare spending at £314.9 billion in 2024-25, or 10.7 per cent of GDP, forecast to reach £408.6 billion in 2030-31, or 11.2 per cent. Read as a share of GDP, the series starts and ends at 11.2 per cent between 2026-27 and 2030-31, running at 11.1, 11.0 and 11.1 per cent in between. Read from 2025-26, it rises.
How can I check a benefit-spending claim myself?
Four questions do most of the work. Ask what the measure actually is, since spending as a share of the economy is usually more honest than a raw total that grows with prices and population. Go to the primary source rather than a press release. Notice the difference between a precise claim and a mood word. And check the timeframe, because a single low starting year can make an ordinary rise look like an explosion.
