October Labour Market Statistics: Comment from the Institute for Employment Studies – no news is not good news
24 Oct 2023
Today was meant to see publication of the latest labour market statistics from the Labour Force Survey. Unfortunately however, issues with the reliability of the survey mean that instead only three figures were published: experimental estimates of employment, unemployment and economic inactivity which have been derived instead from administrative data from the tax and benefits system. Given the unavailability of any underlying data on labour force activity for different groups, we will therefore not be producing our usual briefing and analysis.
The issues with the reliability of the Labour Force Survey have been building for some time, with response rates to the survey having declined significantly in recent years – from an average of nearly 50% of surveyed households a decade ago to under 15% in the most recent data. This accelerated with the pandemic, as the first wave of the survey moved from being face-to-face to telephone-based. Falling response rates in themselves are not necessarily a problem, as long as samples are big enough and we know enough about the people who do not reply (so that responses can be weighted properly). However it seems that the scale of recent falls and the impact of the shift to phone-based interviewing has made it impossible to produce reliable, weighted estimates based on the dwindling numbers of people who do complete the survey.
The ONS deserves credit for the work that they have done to try to address these issues in recent years, with extensive changes to how data is weighted and also a doubling of the number of households contacted in order to boost sample sizes. However, it appears that these efforts ultimately have not been successful. As a result we are left in the very unsatisfactory position of not being able to use Labour Force Survey data at all, at least for the time being.
As noted, the figures that have been published today appear to be based on the most recent LFS data, then rolled forward and adjusted based on changes in estimates of employee jobs from HMRC data and changes in unemployed benefit claimants from Universal Credit and Jobseeker’s Allowance systems. All of these data sources have had their own issues in the recent past, but in as far as they can tell us anything, they suggest that the overall labour market picture is broadly unchanged on recent months: with employment still a bit below where it was before the pandemic, unemployment and economic inactivity both a bit higher, and signs that things may have weakened slightly in the last few months.
Looking forward, it is not yet clear whether LFS data will be published again. If it is not, then we may not need to wait too long for a decent replacement, as thankfully the ONS has also been working on a replacement, ‘Transformed’ LFS which uses a combination of online, telephone and face-to-face collection. The ONS intends to start publishing findings from this in early 2024. However given the importance of having a full understanding of what is happening in the labour market, we would hope that they will be exploring the scope to bring these plans forward – even on an experimental basis.
Finally, it is important to note that other data on the labour market continues to be published as normal, including on earnings and on vacancies – on which Tony Wilson published a short thread last week – and data from HMRC and Universal Credit systems. We will review in the next month whether to publish a revised monthly briefing drawing only on these sources.