Have better data solved the UK’s productivity puzzle?

Past productivity trends shape forecasts of future productivity, which drive the fiscal forecasts at the centre of UK policy debate. Unfortunately, the recent revision to measurement of hours worked has barely changed the growth of the measure that matters most for fiscal space: output per worker.

Productivity matters. Labour productivity – defined as the output produced by some measure of labour quantity (workers or hours worked) – is the long-run driver of wages and living standards. Higher output per worker allows firms to pay higher wages without raising prices, and enables governments to fund public services without raising tax rates.

Since the global financial crisis of 2007-09, UK productivity growth, like that of many other advanced economies, has been unusually weak (Fernald and Inklaar, 2025). On the figures that were available until recently, output per hour grew at an annual trend rate of about 2% for several decades prior to the crisis, but at only around 0.7% annually in the subsequent decade.

Economists have long debated how much of this slowdown reflects measurement problems (Goldin et al, 2024). Hours worked – which are measured in the UK by the Labour Force Survey (LFS) – have been a prime suspect: OECD research found that countries relying on unadjusted survey data, including the UK, tended to overstate hours, although the bias seemed to matter more for productivity levels than for growth (Marianna et al, 2018).

A new approach to measuring hours, set out by the Office for National Statistics (ONS) in September 2026 and due to be adopted from November, has re-opened the question, suggesting that the bias affected growth too.

What has changed in UK labour statistics – and why?

UK hours worked have long been measured by the LFS, a household survey that interviews each household for five consecutive quarters. Response rates fell from almost half of households in 2013 to under 15% by mid-2023, a decline that accelerated when first interviews moved from the doorstep to the telephone during the Covid-19 pandemic.

After a particularly poor collection in the summer of 2023, the ONS suspended LFS estimates in October, resuming them in February 2024 after adjusting the weights applied to the surveyed sample. The number of responses has since recovered to near pre-pandemic levels, at around 81,000 individuals a quarter. But the ONS no longer seeks the official quality badge for the LFS, and it is focusing on the planned replacement – the Transformed Labour Force Survey – which households complete mainly online.

The ONS’s new ‘component approach’ builds labour input from separate measures of jobs, hours and workers. Jobs, and from them workers, now come mainly from employer surveys: a quarterly survey of around 37,000 businesses, benchmarked each year to a larger survey of around 87,000.

But the existing LFS remains essential for average hours: it supplies the quarterly movements in hours per job, overtime and most absences, as well as the hours of the self-employed, while the Annual Survey of Hours and Earnings, which employers complete from payroll records for a 1% sample of employee jobs, anchors the level once a year. Hours paid but not worked, such as annual leave, sickness and bank holidays, are then removed. None of this touches output, the numerator of productivity measures: the whole revision runs through hours, the denominator.

Both measures show working hours drifting down before the global financial crisis. After 2008, the old measure has hours edging back up, while the new measure shows the decline continuing, and even accelerating. By 2024, the gap between them had doubled, from just over two hours a week in 2008 to more than four.

Figure 1: Average weekly hours worked per UK employee under the new and old methods

Note: Old method covers main jobs only; new method covers all employee jobs
Source: ONS, component approach to measuring labour productivity, UK: 1997 to 2024 (Figure 12a data

How much do the productivity figures change?

Output per hour looks much stronger after 2008; output per worker barely moves. Under the new method, output per hour in 2024 is 40.7% above its 1997 level, against 34.0% under the old method. Growth over the period from 2009 to 2019 almost doubles – from 0.7% to 1.3% a year – and the ONS estimates that better hours data explain about half of the slowdown following the global financial crisis.

Table 1: UK labour productivity growth under the old and new ONS methods, 1997 to 2024

Average annual growth (%)	Per hour		Per worker
Average annual growth (%)	Old	New	Old	New
1997–2007	2.1	2.0	1.8	1.7
2009–19	0.7	1.3	0.9	1.0
2021–24	−0.1	−0.7	1.0	0.8
Slowdown after the global financial crisis (2009–19 vs 1997–2007)	−1.4	−0.7	−0.9	−0.7

Source: ONS and author’s calculations.

Since 2021, both measures are somewhat weaker under the new method. But these years are distorted by the pandemic, and three years is too short a period from which to infer a trend.

What does the revision tell us?

For the outlook, output per worker matters more than output per hour. The two differ only by average hours per worker. These can drift down for decades, as more workers work part-time and take more leave, but hours cannot fall indefinitely. And a rise in output per hour that comes only from fewer hours adds nothing to output, incomes or tax receipts.

The revision does change how we read the past. Under the old data, average hours per worker were broadly flat after 2008; the new data suggest that they fell once leave is properly counted and hours are checked against employer records, so pay per hour actually worked was higher than thought, even though overall take-home pay has stagnated somewhat.

This opens a wedge between output per hour and output per job. By 2024, output per hour was some 41% above its 1997 level, against 29% for output per job (ONS); the gap reflects the fall in average hours per job.

How much of that fall reflects an ageing workforce? ONS analysis attributes 0.7 of the 1.3-hour fall in average weekly hours between 1998 and 2022, more than half, to the changing make-up of employment: more women and more older workers, both groups that tend to work shorter hours.

These shifts are captured in both the old and new data, so they cannot explain the gap between them directly. They may add to it indirectly: if older workers take more leave, their rising share would increase the amount of leave that the survey misses.

The ONS finds that the largest single part of the widening comes from leave that the survey increasingly failed to record. People on leave are harder to reach, and their answers from a normal working week are carried forward instead, a problem that grows as response rates fall.

The revision also narrows the gap between the UK’s level of productivity and that of its peers. In 2019, output per hour in France, Germany and the United States, adjusted for differences in purchasing power, was around 23% higher than in the UK (Chadha and Samiri, 2026).

The new ONS figures raise the UK’s 2019 level of output per hour by about 6%, closing roughly a quarter of that gap. That is in line with earlier OECD estimates of the effect of correcting UK hours (Marianna et al, 2018).

But the revision does not change the anatomy of the slowdown in output per hour. With fewer hours, capital per hour is revised up just as output per hour is; since neither output nor the capital stock changes, the slowdown shrinks but its split barely moves from that estimated in Chadha and Samiri (2026), with roughly two-thirds still down to weaker total factor productivity and one-third to slower capital deepening.

For output per worker, hours become a component in their own right. Under the old data, slightly rising hours per worker after 2008 cushioned the slowdown in output per hour; under the new data, hours per worker fell by about a quarter of a per cent a year both before and after 2008. That steady fall holds back output per worker throughout, but plays little part in its slowdown, which mainly reflects weaker output per hour.

If anything, the investment puzzle deepens. Falling real interest rates should have made capital cheaper to finance, yet business investment stayed weak. A familiar explanation is that abundant, cheap labour after 2008 led firms to hire workers rather than invest (Pessoa and Van Reenen, 2014). The revision somewhat weakens it: with hours revised down, labour input grew less than thought, and labour was less cheap than it appeared.

But investment is not the whole story. Total factor productivity, which accounts for most of the slowdown, is a residual: it captures everything that raises output beyond what capital and labour can explain, including technological progress, management practices, financial conditions and how firms organise themselves (Chadha and Samiri, 2026).

Faster adoption of new technologies and better management practices – which vary widely across firms and are closely linked to their productivity (Bloom and Van Reenen, 2010) – can therefore be part of the solution.

Conclusion

Better data have changed our diagnosis of the past, but not the outlook. The revision leaves the productivity puzzle smaller but no less real: weak total factor productivity and chronic underinvestment remain at its core, and the steady fall in average hours now joins them as part of the story going forward.

Past productivity trends shape forecasts of future productivity, which drive the fiscal forecasts at the centre of UK policy debate. Unfortunately, the revision has barely changed the growth of the measure that matters most for fiscal space: output per worker. Average hours per worker have been falling for decades, and if an ageing workforce sustains that trend, output per hour will keep outpacing output per worker.

Whether shorter hours mean higher social welfare is unclear: they may reflect more leisure, but also more older people, who tend to work shorter hours, staying in work as the state pension age rises. Either way, they do not raise output per worker or tax receipts.

Tackling underinvestment poses a dilemma. Fiscal space is tight, yet public investment has long been squeezed by current spending. Rebalancing towards well-chosen public investment, and tackling the reasons why UK firms invest so little, offer the best route to faster growth and, in time, more fiscal room. Good national statistics are crucial to tracking that progress, and the ONS revision is a welcome improvement.