Jonathan Portes explains why the ONS updated its productivity estimates for the period between 2009 and 2019 from 0.7 per cent to 1.3 per cent and what this change actually means for the UK economy. 

Last week, the Office for National Statistics published indicative estimates suggesting that British productivity, measured by output per hour, grew almost twice as fast in the decade after the financial crisis as previously estimated. This is a very large statistical revision. It changes our understanding of Britain’s economic performance since 2008—but it does not mean that the economy actually performed better, or that the “productivity puzzle” has disappeared.

Productivity—how much output we produce for each hour worked—is ultimately what drives wages and living standards. Before the financial crisis, UK output per hour grew by about 2 per cent a year. On the old figures, that fell to just 0.7 per cent between 2009 and 2019. On the new ONS estimates, it was 1.3 per cent.

So the post-crisis slowdown, while substantial, was perhaps half as large as we previously thought. By 2019, output per hour was roughly 6–7 per cent higher than suggested by the old statistics.

What happened? Productivity is simply output divided by labour input. Sadly, the ONS has not discovered that we were making lots more stuff than we thought: the productivity-method change does not alter estimates of GDP. Rather, it has concluded that we simply were not working as many hours as the previous figures suggested.

Previously, the ONS largely derived total hours worked directly from the Labour Force Survey. Respondents were asked how many hours they had worked in a particular week. But this is surprisingly difficult to measure accurately. People may report their usual or contracted hours rather than their actual hours. One household member may answer on behalf of another. And when respondents drop out temporarily, the survey may carry forward their previous answer—even if the reason they did not respond was that they were on holiday.

These problems became more serious as response rates fell. In effect, the survey increasingly imputed that people were working as usual when some were not working at all, or were working fewer hours. This created a growing upward bias in measured hours and, correspondingly, a downward bias in measured output per hour.

The ONS’s new “component method”, already used by many other countries, builds the estimate up from several sources. It starts with paid usual weekly hours from the Annual Survey of Hours and Earnings, then deducts annual leave, bank holidays, sickness and other absences, while adjusting for overtime. It combines these figures with estimates of jobs drawn largely from business surveys. The result is falling average hours per job and much slower growth in total hours after 2008.

The good news, then, is that the UK’s productivity record no longer looks quite so bad. Productivity slowed across the advanced economies after the financial crisis, reflecting weaker investment, the fading of the information-technology boom and other common factors. Previous figures suggested the UK was a particularly poor performer. The revised figures appear to put us much closer to the rest of the G7, although precise international comparisons are difficult and the UK remains well behind the United States, with a much weaker record than before 2008.

That should modify some of the more apocalyptic accounts of the UK economy. There was a productivity crisis, but not one of quite the extraordinary magnitude we thought. Some explanations specifically designed to account for Britain’s uniquely dreadful performance may therefore be solutions to a statistical problem that did not, in fact, exist.

But the revision does not vindicate British economic policy since 2008. An average growth rate of 1.3 per cent remains far below the pre-crisis norm. The cumulative consequences for wages, public services and the public finances are still enormous. Nor does a statistical revision put any extra money in anyone’s pocket. GDP per person and real household incomes have not been revised upwards as a result.

And there is a flipside. If the productivity disaster was less disastrous, the supposed “jobs miracle” of the 2010s looks somewhat less miraculous. The familiar story was that Britain responded to the financial crisis with remarkably strong employment growth but almost no productivity growth: lots more workers and hours, producing relatively little extra output. We now think some of the recorded growth in total labour input—specifically hours worked, rather than employment itself—was illusory. We were producing the same amount as we thought, but we were working fewer hours to do it.

The UK’s post-crisis employment record remains good. The employment rate rose sharply and unemployment fell to historically low levels. Indeed, the new estimates make little difference to output per worker or output per job. But the intensive margin—how many hours those in work actually worked—was weaker than recorded. The apparent trade-off between employment growth and hourly productivity was therefore less stark.

What does all this tell us about the economy now? Less than some of the coverage implied. The revision is primarily about the period before the pandemic. It does not transform our understanding of the past few years, where the record remains bleak, although it supports the move towards the new methodology.

Administrative and business-based estimates had already suggested that productivity has recently been doing somewhat better than the survey figures imply. The Resolution Foundation estimates, for example, suggest that output per hour grew by about 1.1 per cent annually over the two years to mid-2026, rather than declining slightly, as the LFS suggests.

That may represent a genuine improvement. But it is much too early to declare a productivity revival. Much of the arithmetic reflects weak or flat growth in employment and hours, rather than strong GDP growth. It is not yet clear whether underlying efficiency has improved, let alone whether AI is beginning to deliver measurable economy-wide gains.

Nor is this historical revision, by itself, likely to change the Office for Budget Responsibility’s forecasts significantly. Forecasts concern future productivity and labour-force growth, not our revised estimate of precisely how badly we performed after 2008. The OBR’s medium-term assumption for productivity growth remains around 1 per cent—better than stagnation, but not enough to deliver rapidly rising living standards or significantly change the government’s fiscal arithmetic.

So, the right conclusion is neither that everything we thought was wrong nor that nothing has changed. Britain’s post-2008 productivity slowdown remains real and economically damaging. But it was considerably smaller, and less uniquely British, than the official data told us.

The broader lesson is that economic statistics are not facts handed down on tablets of stone. They are estimates constructed from imperfect data. That points to the urgency of improving our statistical system, especially when it comes to measuring the labour market. Better measurement cannot make Britain richer. But without it, we may diagnose the wrong problems—and prescribe the wrong cures.

By Professor Jonathan Portes, Professor of Economics and Public Policy, Department of Political Economy, King’s College London and Chair of the Office of National Statistics Stakeholder Advisory Panel for Labour Market Statistics. Please note, his views here are personal and do not represent the views of the ONS.