The truth will out

Key findings

The Claimant Count measure of unemployment is currently overstating the level of and changes in those genuinely unemployed and claiming unemployment-related benefits, for two reasons. First, the crisis occurred in the middle of the roll-out of Universal Credit (UC), and has dramatically increased the pace of that roll-out. The replacement of legacy benefits by UC leads to more people being captured in the Claimant Count, including those who would have previously only claimed Child Tax Credits and Housing Benefit, and claimants awaiting a health assessment.
Second, and more importantly, the easing of the usual work-search conditions and contact between claimants and work coaches between March and June means that many new UC recipients have not had their work status accurately updated as quickly as they otherwise would. This is particularly important in light of the Coronavirus Job Retention Scheme (JRS), with some furloughed workers making an out-of-work UC claim that would have placed them in the Claimant Count, before they knew they were furloughed.
Analysis of timely longitudinal survey data suggests that at least 27 per cent (400,000), and likely many more, of the 1.6 million Claimant Count rise between March and May is accounted for by those still working, furloughed workers, or Self-Employment Income Support Scheme (SEISS) grant recipients. While some of these people will ultimately be thought of as unemployed, it is far from clear that all should be right now. We estimate that only 45 per cent of the recent Claimant Count rise (700,000 claimants) relates to those newly out of work and not receiving other government support via the JRS or SEISS.
In contrast to the overestimate provided by the Claimant Count, the International Labour Organisation measure of unemployment in April will have underestimated any increase in unemployment. This is because the hard lockdown period, and the collapse in vacancies it caused, meant that many people who did not have a job during April, or who lost self-employed work, did not make an effort to look for new work. This is reflected in the fact that while unemployment did not budge, economic inactivity rose by 425,000 in the month to April, driven by those inactive for reasons other than the usual ones, and those who say they want a job.
It would therefore be more sensible to use the employment rate as the key indicator of the state of the labour market. But key employment measures are also failing to provide a complete picture of the state of the labour market and the amount of productive work being done. The headline employment measure includes people who are temporarily not working (the proportion of employees in this group has risen from around 7 per cent before lockdown to just under three-in-ten in the five weeks after lockdown began), while Pay As You Earn (PAYE) data misses out the self-employed, and includes employees who are not working but are being paid (including furloughed workers).
To gain a complete insight into the amount of productive work actually being done, we need to turn to additional indicators published by the ONS. These include the average and total hours worked in a week by those who are in employment (average hours fell by 23 per cent between early March and the last week in April), and the proportion of workers employed and not temporarily away from work.

Recommendations

Our recommendation to users of these labour market statistics is to recognise the misleading nature of both common ways of measuring unemployment at present, in terms of both levels and changes, and focus on other measures wherever possible. That said, given that we expect the Claimant Count to be ‘corrected’ downwards in the coming months, a further Claimant Count rise in the autumn should be a cause for concern, perhaps indicating a second wave of job losses as the JRS ends.
The change in the level of productive employment is a more useful labour market barometer at this time, but neither the employment rate in the Labour Force Survey (LFS) nor the number of paid employees in the PAYE data provides a complete picture of it (although the PAYE data, being timely, will be particularly useful for understanding the impact of the phase-out of the JRS). On this basis, we recommend that users monitor some of the additional indicators published by the ONS, such as the average and total hours actually worked by those in employment.
We also recommend that the ONS goes to more effort to explain the apparently contradictory findings in its own labour market releases, and provides a more detailed commentary discussing the implications, and reliability, of the different statistics contained within it (including those produced by others). And in the short run, the ONS should split estimates of total employment from the LFS into those who did and did not do any work in the reference week.

Search site

Main themes

Timely topics

Search site

Keywords

Content type

Keywords*

Content type

Key findings

Recommendations

Further reading

Mailing list

Sign up below

I would like to receive:

I consent to my data being used in line with the privacy policy