Productivity explained: output per hour versus hours worked
5 min read / 2026-10-17
Productivity is output per worker, and it can slow because people work fewer hours or because each hour produces less. Here is how to tell the two apart and why better data can change the split.
Two parts inside one productivity number
Productivity here means output per worker: how much the economy produces for each person working. That figure depends on two things. One is output per hour, meaning how much is produced in each hour worked. The other is average hours worked, meaning how many hours the typical worker puts in. If either one falls, output per worker can fall, even when the other stays steady.
A shop example with the same drop and different causes
Suppose a hypothetical shop worker produces goods worth ₹800 per hour and works 40 hours a week, so output per worker is ₹32,000. In Case A, hours fall to 35 while output per hour stays at ₹800, giving ₹28,000. In Case B, hours stay at 40 but output per hour falls to ₹700, also giving ₹28,000. The headline drop is identical, but Case A is about time worked and Case B is about output in each hour. These numbers are made up to show the arithmetic.
Why a better split changes the explanation
The ONS says its new component approach combines multiple data sources and international best practice. The data still point to a UK productivity slowdown after the financial crisis. What changes is the split: falling average hours worked now play a larger role, and slower growth in output per hour a smaller one. Since the two parts point to different explanations, a different split gives analysts a different picture to test. The source does not list the causes of either part.
Which data sources moved the estimates
Two examples appear in the ONS update. Linked HM Revenue & Customs Pay As You Earn payroll records showed a stronger jobs recovery after the pandemic, and a smaller, shorter rise in economic inactivity, than the Labour Force Survey estimates alone suggested. Separately, richer data from the Annual Survey of Goods and Services revealed a larger service sector and raised the estimated level of nominal GDP in 2024 by 0.5%. Measuring both output and work more fully changes the numbers that productivity is built from.
What is still to come
The new method is not yet in the headline figures. The ONS plans to incorporate it into headline productivity estimates in November, with regional and industry breakdowns early next year. Until then, figures already published are not fully updated. A reader comparing older productivity numbers with later ones should check whether the same method was used.
Key words
Productivity
Output per worker. It can be split into output per hour and average hours worked.
Output per hour
How much is produced in each hour of work, regardless of how many hours people work in total.
Average hours worked
The typical number of hours each worker puts in over a period such as a week.
Linked payroll data
Employer payroll records from HM Revenue & Customs that are connected to labour market analysis, rather than answers collected in a survey.
Key facts
- 1Productivity here means output per worker, which depends on output per hour and on average hours worked.
- 2The ONS says its new component approach combines multiple data sources and international best practice to help explain a significant part of the UK productivity puzzle.
- 3The data still show a slowdown in UK productivity after the financial crisis, but falling average hours now play a larger role and slower output per hour a smaller one.
- 4Linked HMRC Pay As You Earn data showed a stronger post-pandemic employment recovery and a smaller, less prolonged rise in economic inactivity than the published Labour Force Survey estimates alone.
- 5The new method enters headline productivity estimates in November, with regional and industry breakdowns planned for early next year.
Why it matters
Fewer hours worked and slower output per hour point to different explanations, so the split affects what questions analysts ask about a slowdown.
