Britain's flat decade: output per hour after 2008
For ten years British workers produced barely more in an hour than they had in 2007. Part of that, it now turns out, was a mistake in counting the hours.
The downturn’s longest shadow outside the United States fell on British productivity. For five decades before 2008, according to the Office for National Statistics, output per hour worked in the United Kingdom rose steadily by around 2 per cent a year. In the decade after the downturn it rose by 0.3 per cent a year, a weakness the ONS described in 2020 as deeper and longer than any since the 1890s.
Small annual numbers compound into large gaps. By the end of 2018 the ONS put output per hour 18.3 per cent below the level its pre-downturn trend would have reached. An ONS blog a year earlier had put the same loss another way: had the old trend held, output per hour would by then have been 21 per cent higher. Andy Haldane, then chief economist at the Bank of England, called it a lost decade and counting.
The explanations
The Bank made the most careful early attempt to share out the blame. Its Quarterly Bulletin of June 2014 took a shortfall of about 16 per cent at the end of 2013 and allotted around 4 points of it to measurement. Between 6 and 9 points went to persistent damage: less investment in machines and in intangible capital such as software and research, and weak firms surviving in unusual numbers instead of releasing their workers and premises to stronger ones. The investment channels alone accounted for 3 to 4 points. Slower growth in financial services added about a point. Labour hoarding, firms keeping staff through the slump in the expectation of better times, was judged uncertain.
Later accounts lengthened the list. The ONS added the long decline of North Sea oil and gas, weak management, the misallocation of workers between firms and the slower spread of ideas. Haldane put weight on that last one: in 2017 he said the rate at which techniques travel from the most productive firms to the laggards had slowed and perhaps stalled, and in 2018 that most firms in the lagging tail were not zombies. They were simply slow.
None of the explanations was decisive, which is why the subject went on being called the productivity puzzle.
The hours
One explanation was dismissed early. The 2014 Bulletin judged that headline data were unlikely to overstate substantially the number of hours worked,.
On 17 September 2026 the ONS questioned it. A new approach to measuring hours found that the hours reported in the Labour Force Survey had been overstated, through recall errors, answers given on someone else’s behalf and gaps filled in by imputation. On the old measure, output per hour grew 2.1 per cent a year before 2008 and 0.7 per cent after. On the new one it grew 2.0 per cent before and 1.3 per cent a year from 2009 to 2019. The slowdown shrinks by about half. The puzzle remains, the ONS wrote, though more muted than before, and the method goes into the headline figures from the November 2026 bulletin.
The ONS is careful not to say more than that. Output per worker, which does not depend on the hours, still slows from 1.7 to 1.0 per cent a year on its figures, and its methodology note says the revision does not overturn the wider evidence of weak growth since the financial crisis. The decade was still slow. It looked flat in part because the hours on the bottom of the fraction were too large, and the puzzle economists spent ten years explaining has turned out to be about half the size they were explaining.