Are out-of-work benefit claims at a record high?

The official statistics are wildly misleading - I've now made some new estimates that show just how wrong they are

“Record 6.5 million Britons on jobless benefits” ran the Telegraph headline a few weeks ago, an apparent 80% rise since 2018 - a claim that is now doing the rounds among journalists and politicians, including Nigel Farage (as you can hear on the BBC Radio 4 More or Less programme that came out this morning, where I was talking about all this).

I’ve explained before about what’s wrong with these figures (across parts 1, 2 and 3) - but those posts are nearly two years old, and in the light of a new, seemingly huge rise in out-of-work benefit claims, I want to revisit this again. And rather than just explaining what’s wrong with the figures, I want to go further - for the first time presenting a detailed alternative picture of the trend in out-of-work benefits.

But what’s wrong with the original claims?

To be clear - I’m not accusing these journalists/politicians (or Fraser Nelson, whose work they’re drawing on) of deliberately distorting the facts. Fraser Nelson actually is impressively clear about what his claims are based on, and in this case, the numbers come from official statistics, without distortion.

The problem is that the raw DWP statistics¹ are wildly misleading - they’re just not counting things in the same way over time, so you can’t tell how much of the trend comes from real changes, and how much comes from changes in how things are counted. There’s lots of ways that the counting has changed, but to give you a flavour of it, let me recap three issues:²

1. Sometimes a couple was previously counted as 1 claim, but is now counted as 2 claims. The DWP has recently estimated that this doesn’t make a huge difference to claim numbers (at least not for the move from means-tested ESA→UC Health), but still: this is a breathtakingly basic bit of inconsistent counting.

2. The State Pension Age has risen, so some people who would have been counted as ‘pensioners’ in the past are now counted as ‘working-age’. This includes more than 500k women aged 60-64 and men/women aged 65 who were out-of-work claimants in 2025 but wouldn’t have been in 2010. What’s even more frustrating is this saves the Government large sums of money (it’s cheaper to pay 0.5m people benefits than to pay the the state pension to all 2.7m people of these ages),³ but is being presented as welfare running out-of-control.

3. Some people in the past were out-of-work and receiving benefits, but were not counted as ‘out-of-work benefit claimants’ - whereas now they are. In particular, people just claiming Child Tax Credits and/or Housing Benefit were not counted as ‘out-of-work claimants’ in the legacy system, but under UC they suddenly become ‘out-of-work claimants’. This is a huge effect, and I say more about this just below.

So what is the real trend in out-of-work benefit claims?

It is possible to create a more accurate picture of the trend in out-of-work benefit claims - but it requires a number of assumptions. I’ve tried to be as transparent as possible, and you must remember that some of things that follow are subject to much uncertainty. It’s also a new estimate, so if you spot something that doesn’t look right, then either drop me an email or say something in the comments below. But still: if you don’t produce an improved number, then the old, flawed, caveated number rules the roost.

Let me firstly show you what my attempt at a consistent series looks like:

In other words - by my estimates, if we count consistently over time, then the current level of out-of-work claims is NOT any kind of record; it’s similar to 2014/15 levels, and noticeably lower than 2013.

There’s been a rise since 2018, but it’s only 22% (from 7.4 percent to 9.6% of the working-age population), not the 80% that the DWP’s figures seem to suggest.

To be clear - I’m not saying that this recent rise isn’t important. But the figures being bandied around at the moment wildly overstate what’s been going on.

But how do you estimate this, and can we trust it?

Let’s get into the detail: to get this more consistent series, I’ve made five adjustments to the official data.⁴

Firstly, I’ve accounted for the changing definition of ‘work’ in 2023 for something be an ‘out-of-work’ benefit (this uses the DWP’s estimate of the impact of the changing Administrative Earnings Threshold, as I described in my earlier post). This only makes a small-to-medium-sized difference to the figures (235k).

Secondly, I’ve accounted for the rising State Pension Age by excluding all claims made by women aged 60-64 and men & women aged 65 - these are people who would have been pensioners in 2010, but are ‘working-age’ by 2025.⁵ The oldest year I go back to us 2013, when there were already 90k women aged 60-64 claiming benefits; by 2025 in total there are 525k people (385k women aged 60-64, 140k men/women aged 65).

These are the estimates that we can have the most confidence in - there’s very little chance that these are wrong. But everything else is about the move from legacy benefits/tax credits to Universal Credit (UC), so has varying amounts of uncertainty.

Third, I’ve used the official estimates of how the transition to UC affects the claimant count. These are people who were claiming Tax Credits and/or Housing Benefit in the past, but who would have been counted as unemployment benefit claimants today (i.e. the UC Searching for Work group). Again, I explain this more in my earlier blog posts, which show just how massive a difference this makes - it reduces the count by 775k people relative to 2013.⁶

Fourth, I’ve used some DWP estimates of the effect of managed migration on out-of-work claims. These data aren’t part of the main statistical releases, but are buried in a bunch of papers that Stephen Timms released earlier this year⁷ - so fair play to DWP for the commitment to transparency here. That report shows by Dec 2024, of the 580k people who had moved from Tax Credits only → UC, 130k were in the ‘No work requirements’ group (so are categorised as ‘Out-of-work benefits’); a further 75k moved to this group from Housing Benefit only (or HB+TCs). I’ve extended this slightly to get a picture for Feb 2024.⁸ This reduces the level of out-of-work claims by 250k in 2025 relative to 2023.

Finally, and most tentatively, I’ve accounted for the wider migration of people from Tax Credits to UC. The previous step only looks at ‘managed migration’ - people who were still on Tax Credits/Housing Benefit, who were sent a letter (‘migration notice’) telling them to move onto UC. But most of the transition to UC has involved ‘voluntary migration’ (people who decided to claim UC) or ‘natural migration’ (new claimants who would have been on Tax Credits before, but now have to claim UC).

There’s no easy way of estimating the size of this level of migration, but a crude way of doing it is compare the number of people claiming Tax Credits in Dec 2017 (before the full national rollout of UC) vs. Dec 2021 (before national managed migration started). This is a lot of people - 3.3m, of which I estimate 2.9m were claiming just Tax Credits or TCs + Housing Benefit.⁹ If we assume that 26% of these people would have been ‘out-of-work benefit claimants’ - which is the proportion that the DWP report shows for people subject to managed migration¹⁰ - then we estimate that there are 650k people claiming out-of-work benefits now who wouldn’t have been counted this way before.

A few things to note: I’ve never seen anyway try to estimate this before - so if you think there’s a better way of doing it (or that I’ve made a mistake), please do let me know. It’s also subject to several uncertainties that I’ve tried to describe transparently. But these figures shouldn’t be a surprise - the official Tax Credit statistics consistently show that 25-30% of families claiming Tax Credits were out-of-work.

You can see the difference that each of these steps makes on the figure below. This clearly shows that differences in how UC counts things (compared to legacy benefits) accounts for most of the revisions that I’ve made.

Final thoughts

I don’t blame any journalists for failing to make the adjustments above - they’re really complex, they’ve taken me ages, and it will probably take a few weeks of crowdsourced discussion to finesse these into their most robust state.

But as it stands, under very plausible assumptions, headlines about ‘record’ levels of out-of-work benefit claims aren’t true - there’s been a recent rise, but this is 22% not 80%, and we’re now back at about the level of 2014/2015.

Alongside this, claims that welfare spending is out-of-control are also wildly misleading - I’ll put up some updated figures next week, but for the meantime, this post is still basically accurate. And the picture on employment overall is basically still as I described here.

There are real issues around welfare and the labour market that we need to tackle - but these are not helped by wildly overstating the scale of the problem. And the root of the problem here are the figures produced by the DWP - if they can’t find a way of presenting them clearly, then they just need to be pulled, with a clear explanation of why.

1. You can find the DWP statistics in Stat-Xplore, under the ‘Benefit Combinations’ section, where one of the categories is ‘out-of-work benefits’ - you can read the DWP’s explanation of this category here.

2. Aside from all the issues that I describe here, there’s also some further issues described in a DWP report that’s come out since I last looked a this - it’s really helpful (and shows that the issue about people with partners is relatively minor), but doesn’t cover the big issues that I describe in this post.

3. On More or Less, I referred to this as ‘300,000 people’ - this refers to the rise in the number of people of those ages claiming out-of-work benefits since 2018.

4. Figures refer to February in each year, as this is the latest month for the official out-of-work benefits series.

5. These figures are easily available from the Stat-Xplore Benefit Combinations series that underlines the headline figures on out-of-work benefit claims (and also Fraser Nelson’s work).

6. Note that the official ‘alternative claimant count’ increases the historic claimant count as if the UC system was in place, whereas I reduce the current out-of-work benefit count as if the legacy systems was in place - so mine is a mirror image of the alternative claimant count. So for example, the alternative claimant count increases the 2013 count by 775k, whereas I reduce the 2025 count by 775k. Note also that the alternative claimant count stops in 2022, so I’ve assumed a linear trend from the 2022 figure to an assumed zero in 2025 (when the migration of the affected benefits had been completed). Month-to-month totals in recent years are therefore estimates, but the 2022 and 2025 figures match official statistics.

7. The document is the ‘Move to UC Data and Insights Pack’ from Feb 2025, released by the Minister as part of a large volume of deposits from the Universal Credit Programme Board in April 2025 - slide #10 is the source for the estimates here. Note that this is a visualisation that doesn’t give the exact figures, so I used https://plotdigitizer.com/app to scrape the figures from it; they’re therefore more approximate than they might be.

8. Note that the Stat-Xplore figures on Move to UC refer to the date that the migration notices were sent, so don’t quite match up with the DWP insights report figures that are about migrations completed by Dec 2024 - but there’s no reason to think that the proportions will be any different. Note also that I’m excluding the people who have gone from Tax Credits only → UC Searching for Work, as these are presumably included in the Alternative Claimant Count.

9. Data comes from Child and working tax credits statistics: provisional awards, December 2024. The % of all Tax Credit claims that are TC only or TC+HB only comes from the Move to UC statistics on Stat-Xplore, which show that 89% of all those on Tax Credits who were invited to move to UC were in these groups.

10. The 26% figure comes from combining (i) 22% of TC only claimants during managed migration ended up in UC Out-of-work; (ii) 42% of TC+HB only claimants during managed migration ended up in UC Out-of-work; and (iii) the share of TC+HB only (relative to TC only + TC+HB) in the managed migration data is 21%.