The *real* trend in out-of-work benefit claims

The official statistics are misleading - so what has really been going on?

In my last post, I explained why you can’t trust the statistics around trends in out-of-work benefit claims. Strangely the numbers on ‘out-of-work benefit claims’ include people who aren’t out-of-work and who aren’t really ‘claiming’ (in the sense of receiving any money from the state) - and the problems don’t stop there. But it’s never enough to say ‘these statistics are wrong’ - we need to have a better idea of what is *actually* going on. In this past, I make a first attempt to do just that.

The pieces of the jigsaw

There are several parts of the jigsaw where we have some data, which we can go through piece-by-piece. Just like last week, this is a work-in-progress, so please let me know if you can add to it! [NOTE: with this in mind, I edited this on 23rd Feb, see footnote for details.¹] For a recap of the issues look in last week’s post - I’m here focusing on how many people are affected:

The changing definition of ‘work’: people can do more work than they used to and still be counted as ‘out-of-work’ (which is set by the Administrative Earnings Threshold, ‘AET’). According to official estimates, recent changes to the AET have increased the number of out-of-work claims by 235k.². But going back further, it’s hard to know how things have changed. The few JSA/ESA claimants working more than 16hrs/wk or so would have been classified as ‘out-of-work’ (because if they consistently did more than this, they would have to leave the benefit). We have data on the share of people on the ‘alternative claimant count’ who have done any work in the last month, and this is pretty similar across the last decade³ - and the data stops just before the 2022 rise in the AET just above. So a plausible guess is that this was stable, but then artificially inflated out-of-work claimants by 200-250k since early 2023 (compared to early 2022).

The changing working-age population: this is much easier to figure out, because Stat-Xplore data tells us the number of out-of-work benefit claims split by age. In May 2023, there were about 335k women aged 60-64 claiming out-of-work benefits, together with 125k people aged 65 - none of whom would have been considered ‘working age’ back in 2010. So that’s 465k extra people counted as ‘out-of-work claimants’, who wouldn’t have been included in the past.

In the main comparisons below, we’re comparing the current picture to Feb 2013, when the rise in women’s pension age had already started. There were 90k 60-61 year-old women claiming out-of-work benefits by Aug 2013, so that’s 360k extra people counted as ‘out-of-work claimants’ in 2013, who wouldn’t have been included in Feb 2013.

Claimants’ partners and counting Housing Benefit/Child Tax Credit: this is the hardest bit to figure out, because it is simply impossible to calculate ourselves using public data. However, the DWP recognised that the ‘claimant count’ series was incomparable before vs. after UC, so put some effort into creating an ‘alternative claimant count’ series. It’s a complex process that’s best described here, but basically it estimates the number of ‘additional claimants’ - extra people counted under UC but not under JSA - in the claimant count series, namely:

• Partners of income-based JSA/ESA/IS claimants who would be in the ‘intensive work search’ conditionality group in UC.⁴

• People who were claiming benefits and were out-of-work, but were not claiming ‘out-of-work benefits’ (this means most people claiming just Housing Benefit and/or Child Tax Credits)

This completely changes the number of people on the claimant count (though it’s not possible to pick apart which of these two reasons plays the bigger role). As shown below, the claimant count in 2013 goes from 1.5m to 2.3m when this issue is dealt with consistently, a 52% increase (!). As the financial crisis-fuelled claimant count went down over time, the difference that the additional claimants make is even larger - in early 2015, the claimant count rises by 80% (that is, it almost doubles) when counted consistently.⁵

The difference that consistent counting makes to the ‘claimant count’

Source: official Alternative Claimant Count, final release (emphasis in red added by me)

What’s more, this underestimates the impact on out-of-work benefits because it just focuses on the ‘claimant count’ (basically, unemployment) - it doesn’t look at additional claimants with work-limiting health problems or young children or other caring responsibilities, who are the majority of people claiming benefits. If the proportional increase in other out-of-work (ESA/IB/IS/PC) claims was the same as for unemployment - there’s no reason to think this, it’s just a way of showing the point - then we would have a further 1.6-2.4m people on out-of-work benefits in 2013! To repeat, I’m not saying this should be treated as an accurate estimate - it’s very crude, and we have no way of knowing what the real figure is. But this at least shows that the real out-of-work benefits estimate for 2013 could be completely different…

Putting the jigsaw together

So what is the real trend in out-of-work benefit claims? It’s very difficult to say, because the DWP haven’t used their internal data to produce something that’s genuinely consistent over time. But putting together the bits of information above, we can at least get a ballpark figure for how 2013 compares to today. I’ve tried to show this as clearly as I can in the figure below.

The real trend in out-of-work benefit claims: if we account for methodological changes, there are fewer claims in 2023 than 2013

Source: Fraser Nelson chart from previous post (figures for Feb 2023), with revisions as described in the text.

Let’s talk through this figure bit-by-bit (with one further caveat)⁶:

• The grey bars show the figures from Fraser Nelson’s chart last week (based on the official DWP data) - this shows that out-of-work claims in 2023 were 5.5m, compared to 4.6m back in 2013.

• However, we know that the true figure for Feb 2013 is at least 700k greater than this, because of the the changed way of counting people.

• We also know that the true figure for 2023 should be 600,000 people lower, because of the changed definition of ‘work’ and the changed definition of ‘working’age’.

• If we put this together, then on a consistent way of counting, there were 5.4m out-of-work claimants in 2013, which is more than the 4.9m claimants in 2023. In other words, out-of-work benefit claimants are not higher than a decade ago, if we count this consistently.

• The true figure might show that the fall in claimant numbers is even greater than this. That’s because this doesn’t capture how the changed way of counting people affects incapacity or lone parent or other carer claims. My very, VERY rough estimates suggest this could be well above 1m further claims, but we don’t know at all - I just wanted to flag that the true figure could be even higher (in blue).

Some final thoughts for today

Put simply, the real trend in out-of-work benefit claims over the past decade is the opposite of the conventional wisdom - we have fewer claimants than we did a decade ago (and possibly quite a lot fewer).

But there’s a couple of things to be careful of here. Firstly, I haven’t said anything about the recent trends - the data are just too terrible over this period to say anything useful, so it might be that there was a fall in claims in the mid/late 2010s followed by a more recent rise. Secondly, we have a wide range of uncertainty about the true trend, because DWP/ONS haven’t tried to create a consistent series (unlike for the ‘claimant count’).

It might be possible to use survey data to provide a clearer picture - but this is both time-consuming, and has uncertainties of its own. Still, I think this would be worth doing, and if no-one else does it (though please do if you have extensive FRS experience, and let me know!), I’ll try to do the data analyses myself, so that we can see what’s been happening more clearly…

1. As the original post said, this is a work-in-progress - it’s a complex issue, and I’m keen to keep improving these figures (so please get in touch if you have any suggestions). A list of edits as of 23rd Feb:

• Correction to the changes from the changed working-age population - in the original post I noted the number of additional claimants who are counted as ‘working-age’ because of rises to the State Pension Age since 2010. However, the main comparison at the bottom is vs. Feb 2013, by which time the pension age had already started to rise (there were 90k 60-61 year old women claiming out-of-work benefits by Aug 2013) - so I’ve added the comparison to Feb 2013 into the original bullet point, and amended the final chart accordingly.

• Correction to changes from the Alternative Claimant Count (ACC). Technically the ACC series stopped in Aug 2022 because it had converged to the main Claimant Count. However, there are a still a small group of ‘additional claimants’ in the ACC who should really be added to keep a consistent series (the reason that this has converged to the main Claimant Count is because these are outweighed by the people pre-WCA who would previously been claiming ESA not JSA). There’s no estimate of this for 2023, but we could roughly estimate this as 130k in Feb 2023 and 110k in Aug 2023 assuming a slight linear downward trend since the last published data (which isn’t unreasonable based on the trend from Mar/2021 to Aug/2022).

• Correction to the numbers/chart in the final comparison because of the two issues above- the new (old) figures are: (i) true figure in 2023 is at least 500k (700k) lower; (ii) when consistently counted, the total number of claimants in Feb 2023 was 4.9m (4.7m).

2. This comes from rises in the AET affecting 114k (Sep 2022) and 120k (Jan 2023), after the AET was left untouched 2015-22 (and therefore fell compared to the minimum wage). It’s possible that the combined total is slightly lower than 234k, because some of the people affected by the 1st rise in the AET worked slightly more to avoid it, but are then double-counted as affected by the 2nd rise - but this seems unlikely to change things much. There were also proposals in the Spring Budget 2023 to reclassify 190k more people as ‘out-of-work’ (80k + 110k), but it’s not clear if/when this will finally happen.

3. The level has been 170-250k for most of this period - it surged to 450k in April 2020, but has since fallen, and is about 200k in mid-2022. For more on the alternative claimant count, see later in this post.

4. This excludes those assumed to be exempt due to disability (which here means those claiming DLA/PIP), those assumed to be exempt due to caring (= those claiming Carer’s Allowance or IS/Child Tax Credit claimants with a child under 3), or those assumed to be exempt due to working (WTC claimants or HB claimants earning more than the AET)

5. Note that the figure (taken from the Alternative Claimant Count release) removes the number of people with sick notes on UC waiting for a WCA (who are in the UC ‘Searching for Work’ group, but would previously have been on ESA rather than JSA). The number of additional claimants is therefore even higher than the orange line in the chart.

6. One minor caveat to this: we don’t know how much the different issues above (changed ‘work’, changed ‘working age’, and changed counting) overlap with each other. It’s likely that this isn’t very much, but I just wanted to flag that I couldn’t account for it.