What’s going on with the unemployment data?

Blog posts

23 Jun 2020

Tony Wilson

Tony Wilson, Institute Director

Tuesday’s labour market statistics contained two very different stories on unemployment.  According to the ‘claimant count’, unemployment has shot up to 2.8 million – a rise of 125% in just two months, the fastest rate of growth in one hundred years (as we said in our briefing on the day).  However on the official ‘ILO’ measure, unemployment continues to glide along at close to 1.3 million – still at its lowest since 1974 and showing no signs yet of any ill effects.

In the past there have often been big differences between the claimant and ILO measures of unemployment, but rarely have these been so great.  And this matters more than ever now, as policymakers grapple with trying to understand whether we’re in the middle of an unemployment crisis or merely on the eve of one, when the Job Retention Scheme (JRS) starts to unwind.

So what’s going on?  Overall, there are four main differences between the two measures.

First, the two measures describe different periods.  For claimant unemployment, the numbers refer to claimants on a specific ‘count date’ – with the last three being 12 March, 9 April and 14 May.  So these are point-in-time estimates, and handily we can compare what the situation was about a week before the crisis (12 March) with how things were two months later (14 May).

For the ILO measure, the headline figure is a three-month average of survey responses between early February and late April 2020.  This means that two months pre-date the crisis, while one month (April) is since the crisis began.  The ONS also produces single month estimates, however while these show a drop of 320 thousand in the number of people in employment in April – which is almost entirely explained by fewer people self-employed – they do not show any significant change in unemployment (with unemployment just 40 thousand higher in April, which is within the margin of error). 

Instead, the single-month figures show a steep rise in ‘economic inactivity’ – which is the measure of those who are out of work but are not looking and/ or available for work.  And this brings us on to the second big difference between claimant and ILO unemployment: the claimant count measures those who are required to look/ be available for work as a condition of benefit, while the ILO measure is those who say that they actually are actively seeking and available for work.  So all told, the Labour Force Survey is recording a single-month increase in the number of people out of work of 330 thousand, but nearly 290 thousand of these people are not looking for work – however if they are on benefit, then they will be counted as claimant unemployed.  As noted this appears to be mainly people previously self-employed, and as we set out in our briefing if you dig deeper into the data it appears that all of this rise (and more) is explained by people who want to work.

This figure of 330 thousand however is still well below the claimant count increase from March to April, of 1.0 million.  So the third factor that could explain the difference is around the recording of people in work.  In the claimant count, people with earned income can be counted as claimant unemployed if their earnings in the reference month are below a set threshold (£338 per month for a single person, or £541 per month for a couple).  Before Universal Credit (UC), short-hours working was penalised and so these numbers were generally low.  However UC incentivises short-hours work, and so we’ve seen a growth in recent years in the number of people treated as being unemployed but who have some earnings (and the way that earned income is recorded is even more complicated than this, but I won’t go into that here).  The detailed data for UC suggests that 190 thousand of the 1.0 million increase between March and April was accounted for by working claimants – so around one fifth of the rise.

So by now we’ve accounted for about half of the increase in the claimant count for April – with around 300 thousand explained by higher economic inactivity (and maybe a marginal increase in unemployment) and around 200 thousand by working claimants.  But what appears to explain the other half is another quirk in how we record whether people are ‘in work’ – namely that in the Labour Force Survey, you are recorded as in employment even if you have not done any work that week but ‘have a job or business that you were away from… (and that you expect to return to)”.  Obviously this category of workers ‘away’ from work now captures about nine million people furloughed under the Job Retention Scheme (JRS) who are continuing to earn, but it also includes people who consider themselves to be employees or self-employed but who have no earnings.  And we have one crucial piece of further evidence that suggests that this may be mainly employees – as ‘real time’ Pay As You Earn data shows that the number of paid employees fell by 450 thousand between March and April.

So could there be 450 thousand more people describing themselves as having a job but not having earnings?  I think it’s possible, if these are people who were due to start a job in March or April but have been told that their job isn’t available yet (and about half a million people start a new job each month, as I’ve written about before).  There may also be people who had very few or irregular hours before the JRS was introduced and for whom employers have not submitted a JRS claim.  With so few other jobs available, it is plausible that people in these circumstances are describing themselves as being workers with a job that they are away from, rather than as being actively seeking a new job.

Finally, if this explains the rest of the ‘gap’ between the claimant and ILO unemployment measures, the fourth difference between the two measures could mean that the claimant count is under-stating the growth in worklessness in the last two months.  This is because the claimant count, as its name suggests, only counts those who claim benefit.  We know that many unemployed people do not claim, and particularly young people (usually due to eligibility).  Under Universal Credit, there have been on average 450 thousand more ILO unemployed young people than claimant unemployed and even if that gap narrows in the crisis, as tends to happen in recessions, it’s well possible that ILO youth unemployment will be more than 200 thousand higher than the claimant measure.

So putting this all together – timing differences, economic inactivity, treatment of work and benefit take-up/ eligibility – are we any closer to understanding what is going on?!

I think that we are, and that the data suggests that:

  • There has already been a record fall between March and April in the number of people working and not being supported by the JRS – I think that this is plausibly at least a million, i.e. in line with the rise in claimant unemployment;
  • The 450 thousand fall in paid employees between March and April may well have been almost entirely driven by people who reported that they had a job to go to, but that they were not being paid for;
  • A further 300 thousand people previously self-employed are now out of work, while many more have likely claimed Universal Credit either because they have stopped trading or have very low earnings (and are either not eligible for, or not yet been paid, income under the Self-Employment Income Support Scheme); and
  • Virtually none of the decline in the numbers ‘working’ had translated into higher unemployment in April – but this would appear to be very likely to start coming through if people do not get back into work quickly.

Looking ahead, the growth in claims for UC is slowing and so the growth in the claimant count should start to slow as well.  I think that it will have increased by around 300 thousand on the next count date (11 June), meaning that claimant unemployment is probably right now the highest that it has ever been.  However it may not increase by much more until the JRS starts to unwind.  Given that most of these people will not have earnings, nor jobs to go back to, and will want and need to work, I still take the view that the jobs crisis is happening now, and that we need an urgent and significant response to meet it now.

Finally – having either helped or confused things over these 1,500 words, I’ll just end with a plea.  I spend a large part of every ‘stats day’ trying to help people in local government, charities and the media to understand what are incredibly confusing, complicated and often contradictory data.  So my plea is for government and the ONS to produce analyses each month that explain and reconcile these different data sources.  This is not to criticise ONS, who I think have done great work during this crisis.  But with employment falling and worklessness rising faster than at any time before, and changes that have taken eight months or more in previous recessions now happening monthly, helping to explain these changes is more important now than ever.

Subscribe to blog posts

Any views expressed are those of the author and not necessarily those of the Institute as a whole.