It has been a tough few years for labour market statistics. The COVID-19 pandemic made the collection and interpretation of data on jobs and pay extremely challenging. Many surveys found it harder to reach respondents. Those that relied on face-to-face interviewing had to switch to telephone or online modes. Different sorts of people or firms seemed to respond to surveys when compared with pre-pandemic times, making it hard to be sure whether the trends these data sources were picking up reflected actual changes in the labour market. More concerningly still, these problems do not appear to have entirely dissipated as the pandemic has passed.

The most publicised consequence of these problems has been for the quality of the Labour Force Survey (LFS), which affects the reliability of official statistics on employment, unemployment and inactivity. Here, the pandemic-related complications above combined with an already declining trend in the response rate to the survey. This led the Office for National Statistics (ONS) to temporarily suspend official statistics based on the LFS. It is developing a new ‘Transformed LFS’, which should redress some of the problems.

It has not gone unnoticed that our key data sources also fail to paint a clear picture of what has happened to earnings levels, or the earnings distribution, since 2019. They are not all in accord. As things stand, it is unclear what we should think has happened. This is a particularly salient issue in the context of weak income growth in the decade leading up to the pandemic (Cribb and Waters, 2024) and the subsequent cost-of-living crisis. There is also the risk – particularly around election time – that very different statistics, plucked from conflicting data sources, get bandied around. 

Here we compare earnings trends measured from key data sources. From this, we reach some conclusions about which data are most and least reliable, and hence about what has most likely happened since 2019. 

Key findings

Comparing trends in UK earnings since 2019 as measured by the ONS’ Average Weekly Earnings (AWE) series, HMRC Pay As You Earn (PAYE) data, the Annual Survey of Hours and Earnings (ASHE) and the Family Resources Survey (FRS), the ASHE emerges as the clear outlier. The other data are all in quite close accord. The discrepancy between the ASHE and other data is driven largely by the ASHE painting a far more pessimistic picture of trends in earnings for high earners since 2019. As a result, it seems to significantly understate earnings growth on average, and to significantly overstate falls in earnings inequality, since the pandemic. We argue that this is likely related to the large drop in the rate of response to the ASHE since the pandemic.

The best evidence (i.e. disregarding the ASHE) suggests that inflation-adjusted earnings are on average now almost 4% higher at the mean and 6% higher at the median than in April 2019. This largely reflects three distinct phases: earnings growth after the loosening of COVID-19 restrictions in 2021; the cost-of-living crisis beginning at the end of 2021, which eroded the value of most of that earnings growth; and the welcome return of real earnings growth since Autumn 2023. Indeed, the last few months account for most of the earnings growth seen overall since 2019 (although only the PAYE data and, at the mean, the AWE data are timely enough to tell us about this latter period).

Earnings inequality appears no higher, and probably a little lower, than in 2019. At both the 10th and 90th percentiles of earnings, we again see the three recent phases of earnings trends since the pandemic: growth immediately after the worst of the pandemic, which was largely eroded by the cost-of-living crisis, before returning in recent months as inflation has subsided. That uptick in earnings growth over the past few months has, according to PAYE data, been more pronounced at the bottom of the distribution, leaving 10th percentile earnings in March 2024 about 4% above April 2019 levels in real terms, and 90th percentile earnings about 2% above April 2019 levels – both of which are a little lower than the 6% seen for the median earner. 

Further reading