Reading · Anti-Capitalist Musings

Thanks for reading Anti-Capitalist Musings. It is a small operation, and I hope it offers something worth your time. There will be no premium subscriber content here: everything published will remain free to read. If you value these pieces and want to support the writing, buying me a coffee helps fund media subscriptions and the books that keep the analysis grounded. Every contribution, however modest, is genuinely appreciated.

Britain really did have a productivity crisis. It still does. After the 2008 financial crash, one of the basic mechanisms through which capitalist economies had delivered rising living standards seemed to seize up. The Office for Budget Responsibility calculates that output per hour grew by an average of 2.1 per cent a year between 1998 and 2007, then by only 0.6 per cent during the 2010s. Between 2020 and 2024 it managed 0.4 per cent. Productivity was not the only reason wages stagnated and public finances became tighter, but it sat underneath much of what went wrong. [1]

What now looks doubtful is the next part of the story. We were told that Britain had stumbled out of the pandemic only to suffer another bout of productivity stagnation. The economy was employing more people without producing much more, which suggested that the productivity problem had become even worse. Yet some of those additional workers may never have existed, at least not in the numbers recorded by the Labour Force Survey. The supposed economic deterioration was partly happening inside a damaged statistical instrument.

Productivity is a ratio. Take the amount produced by the economy and divide it by the labour required to produce it. If you overestimate the number of people working, productivity appears weaker even though the amount being produced has not changed. For several years the Labour Force Survey, one of the main ways Britain measures employment, has suffered from serious response problems. Since January 2024 the Office for National Statistics has been trying to improve it. That has brought more employed people back into the sample, but it has also created a peculiar statistical effect: measured employment has risen partly because the survey itself has got better at finding workers. [2]

The ONS now says as much. Its August productivity release warns that recent Labour Force Survey employment growth is likely to be temporarily overstated because improvements to the survey are being mixed together with genuine changes in the labour market. The corresponding measure of productivity is therefore likely to be understated. More significantly, the ONS now recommends its alternative measure based primarily on HMRC’s PAYE Real Time Information as the best guide to changes in labour productivity. [3]

The difference is not trivial. According to the PAYE-based measure, output per hour in the second quarter of 2026 was 0.7 per cent higher than a year earlier and output per worker was up 1.4 per cent. The Labour Force Survey version says output per hour fell by 0.2 per cent and output per worker increased by only 0.4 per cent. These are not rival interpretations of the same statistic. They describe rather different economies. [4]

Niki Barbas, Anna Valero and John Van Reenen at the LSE’s Centre for Economic Performance found an even more revealing discrepancy when they looked at the period from the third quarter of 2024 to the first quarter of 2026. Using administrative employment data, output per worker rose by 2.4 per cent. Using the Labour Force Survey it rose by just 0.3 per cent. The difference comes overwhelmingly from the denominator. The Labour Force Survey recorded an increase of 377,000 employees while PAYE records showed a fall of 133,000. That is a gap of 510,000 workers in eighteen months. [5]

There are differences in coverage between the two datasets, and nobody should pretend that PAYE records constitute a perfect census of work. The self-employed remain difficult to measure and PAYE does not tell us how many hours people actually work. The alternative productivity series therefore still has to borrow average-hours information from the Labour Force Survey. The Resolution Foundation also has to carry forward the latest available self-employment figure because complete tax-return data arrive with a lag. These are important qualifications. They are not, however, large enough to explain away what has happened. The ONS itself has concluded that administrative data currently provide the more reliable measure of employee numbers. [6]

The Resolution Foundation has now extended the analysis through to the second quarter of this year. Its estimate is that output per hour grew by an average of 1.1 per cent a year over the two years to Q2 2026. During the previous two years it had fallen by 0.7 per cent annually. The Labour Force Survey, by contrast, records productivity falling by an average of 0.2 per cent a year during the latest period. The Foundation also runs the calculation using the ONS Workforce Jobs series and gets almost exactly the same result as the administrative-data approach: productivity growth of around 1.1 per cent. [7]

These are UK-wide figures and should not be read as evidence that the recovery has been evenly distributed between England, Scotland, Wales and Northern Ireland; the available regional productivity data do not yet allow that conclusion.

All of this does not restore Britain to the world before 2008. It is not even close. The Resolution Foundation estimates that output per hour remains about 5 per cent below the level implied by the OBR’s final pre-Covid forecast, a shortfall equivalent to roughly £150 billion of annual output, or £4,500 for every worker. The productivity crisis did happen. The provocative part of the argument is narrower: the latest stage of that crisis, when productivity supposedly began falling again, increasingly looks as though it did not happen in the way we were told. [8]

Nor does the recovery seem to be a statistical trick caused by restaurants and shops laying off low-paid workers. That would have been an easy explanation. Remove enough workers from low-productivity industries and average productivity rises even if nobody remaining becomes any better at producing anything. The Resolution Foundation finds almost no evidence of this. Twelve of nineteen broad sectors contributed positively to productivity growth and almost all of the improvement came from higher productivity within industries rather than workers moving between them. Hospitality accounted for 6.6 per cent of employee jobs in the second quarter of 2026, exactly the same share as during 2016-19. [9]

The occupational evidence is similarly awkward for the idea that employers simply removed their least productive workers. In the year to April 2025, employment in occupations paying below the median grew by 3.4 per cent, compared with 1.4 per cent in occupations above median pay. The composition of employment actually shifted very slightly towards lower-paid work. Whatever produced the productivity improvement, it cannot easily be reduced to companies clearing low-wage workers from their payrolls. [10]

Investment does not provide an obvious answer either. Britain’s chronic failure to invest is one of the main explanations for its long productivity problem. Workers equipped with better machinery, software and infrastructure can produce more in an hour. Yet gross fixed capital formation remained around 19 per cent of GDP in both 2023 and 2025. There has not been anything resembling the investment surge required to explain such a sudden improvement in productivity. The Resolution Foundation is therefore left with the economists’ residual category, total factor productivity: we appear to be combining labour and capital more effectively without really knowing why. [11]

Artificial intelligence will inevitably be offered as the answer, and perhaps eventually it will be part of one. The evidence is not there yet. The proportion of UK businesses with ten or more employees reporting some use of AI has risen sharply, reaching 46 per cent by June this year, but only 4.6 per cent of all businesses said they were using it extensively. Information and communications has made a strong contribution to productivity growth, but it was already one of Britain’s better-performing sectors before generative AI arrived. The recovery is too broad to pin comfortably on ChatGPT and its competitors. [12]

There is a temptation at this point to turn the numbers into another argument about whether Britain is secretly doing rather well. GDP rose by 0.6 per cent in the first quarter of 2026 and 0.4 per cent in the second, while GDP per head increased by 1 per cent over the year to Q2. Those figures are better than the prevailing political mood would suggest. They do not amount to a return to rapid growth, and weak productivity over the previous fifteen years has left an enormous cumulative hole that two better years cannot fill. [13]

There is a more interesting question anyway. If productivity really has been recovering, why has so little of it been visible in people’s lives?

Productivity is often discussed as if it automatically becomes prosperity. It does not. Higher productivity means that an hour of labour produces more value. What happens to the additional value is a social question before it is an accounting one. It can appear as higher wages, increased profits, lower prices, greater tax revenues or some mixture of them. The distribution depends on ownership, bargaining power and the institutional balance between capital and labour. Productivity tells us that more can be distributed. It does not decide who gets it.

The latest wages data make the distinction difficult to ignore. In the three months to June, regular pay was growing by just 0.5 per cent a year after CPIH inflation. Private-sector regular pay increased by only 2.8 per cent in cash terms, barely ahead of inflation, and the Resolution Foundation calculates that real regular private-sector pay in June was lower than it had been the previous October. This is occurring during the period in which the revised figures tell us workers have been becoming more productive. [14]

Even the OBR does not assume productivity growth will simply flow into wages. Its March 2026 forecast expects productivity growth eventually to reach around 1 per cent a year, while real hourly earnings rise by only about half that rate. The OBR explicitly assumes that firms will use part of the gap to rebuild their rate of return on capital. This is presented as a forecasting judgement, which it is, but it also reveals something that tends to disappear from the political discussion. Higher productivity creates a larger surplus. The struggle over that surplus comes afterwards. [15]

For years workers have been told that substantial real wage growth was impossible without productivity growth. There was an economic truth buried inside the argument. An economy cannot permanently raise everybody’s consumption faster than its capacity to produce. But the formulation also performed useful ideological work. Stagnant wages could be presented as the unfortunate consequence of a national productivity problem rather than something shaped by the weakening of trade unions, insecure labour markets and decisions about how the proceeds of production were divided.

Now comes an awkward test. If the revised figures survive, workers have been increasing output per hour noticeably faster since 2024. Yet real regular wage growth is currently scraping along at around half a per cent. The old answer, that wages cannot rise because productivity is not rising, becomes less satisfactory. The question moves from production towards distribution. [16]

There are consequences for economic policy too, although these need handling carefully. The OBR cut its medium-term productivity assumption from 1.3 to 1 per cent in November 2025. Productivity matters enormously to its fiscal calculations. In the OBR’s own scenarios, sustained productivity growth of only 0.5 per cent would leave the current budget £6 billion in deficit in 2029-30, while its higher-productivity scenario produces a £59 billion surplus. Small changes in assumptions about what each worker can produce become tens of billions of pounds once projected across an economy and through the tax system. [17]

It would nevertheless be too easy to conclude that the government’s fiscal difficulties were simply created by faulty statistics. The new administrative-data estimate of 1.1 per cent productivity growth is actually close to the OBR’s downgraded medium-term assumption of 1 per cent. The OBR also builds its forecasts from much more than a single employment series. What the episode shows is something more uncomfortable about governing an economy through numbers that are always provisional but are treated politically as though they were facts carved into stone. A faulty denominator can migrate from a household survey into estimates of productive capacity and from there into arguments about how much a government can spend.

There is another warning here for those eager to announce Britain’s economic renaissance. The recent increase in productivity has mainly come because hours worked have stopped growing, not because GDP has suddenly taken off. Over the two years to Q2 2026, the Resolution Foundation calculates that GDP expanded by an average of 1.3 per cent annually while hours worked were almost flat. Seventy-seven per cent of the improvement in productivity compared with the preceding two years reflects slower growth in hours. That is better than employing ever more labour to produce very little additional output, but it is not the same thing as a booming productive economy. [18]

Something has changed nonetheless. Britain appears to be squeezing more output from broadly the same workers doing broadly the same kinds of work. We do not yet know why. We do know that the gains have not produced anything resembling the improvement in living standards that the language of productivity normally promises.

That may turn out to be the more important discovery. The productivity crisis after 2008 was real, and Britain remains a low-investment economy carrying the damage of fifteen years of stagnation. But if the latest figures are right, the recent productivity recovery was hidden by bad measurement while the benefits of that recovery were hidden from workers by something else. One problem belongs to statistics. The other belongs to political economy.

References

1. Office for Budget Responsibility, Economic and Fiscal Outlook, November 2025, productivity growth and historical productivity performance.

2–4. Office for National Statistics, UK productivity: April to June 2026 and January to March 2026, 18 August 2026. Covers the Labour Force Survey problems, the recommendation to use the RTI-based measure and the Q2 productivity comparisons.

5. Niki Barbas, Anna Valero and John Van Reenen, Centre for Economic Performance, LSE, analysis of administrative employment data and UK productivity, 2026. Covers the 2.4 versus 0.3 per cent productivity comparison and the 510,000 discrepancy in employee estimates.

6–12. Resolution Foundation, The Macroeconomic Policy Outlook, Q3 2026, August 2026. Covers its alternative employment and productivity series, self-employment methodology, sectoral composition, occupational changes, investment and AI.

13. Office for National Statistics, GDP first quarterly estimate, UK: April to June 2026. Covers Q1 and Q2 GDP growth and GDP per head.

14. Office for National Statistics, Average weekly earnings in Great Britain: August 2026, together with Resolution Foundation analysis of recent real private-sector pay.

15. Office for Budget Responsibility, Economic and Fiscal Outlook, March 2026, productivity, real hourly earnings and returns to capital.

16. Same wage and productivity evidence as references 7 and 14.

17. Office for Budget Responsibility, Economic and Fiscal Outlook, November 2025. Covers the reduction in the medium-term productivity assumption and fiscal sensitivity to alternative productivity growth.

18. Resolution Foundation, The Macroeconomic Policy Outlook, Q3 2026. Covers GDP, hours worked and the decomposition of recent productivity improvement.

The Productivity Crisis That Wasn’t

Comments