Counting hours worked from several sources, rather than one household survey, lifts estimated output-per-hour growth for 2009 to 2019 from 0.7% a year to 1.3%. The figures are indicative and not yet official, and the ONS plans to adopt the method from November 2026.
The Office for National Statistics (ONS) currently measures labour input mostly from the Labour Force Survey, a household survey. Its new component approach builds the same total from three separate parts, hours, filled jobs and workers, each taken from the source judged strongest, such as business surveys, employer-reported pay data and administrative records.1
Before the financial crisis the two methods agree. Output per hour grew by 2.1% a year from 1997 to 2007 on the current measure and 2.0% on the new one. From 2009 to 2019 they diverge: 0.7% a year on the current measure, 1.3% on the new one.1 The drop from the pre-crisis rate is therefore 1.4 percentage points on the current measure and 0.7 points on the new one, exactly half. The ONS says improved hours measurement explains roughly half of the slowdown as previously measured.1
Output is the same under both methods, so every change comes from the denominator. On the new method total hours worked in 2024 were 11.1% above the 1997 to 2007 average, against 18.5% on the current one. The number of jobs is almost identical, up about 19.9% against 19.7%. The difference is in hours per job: the new method has average hours per job 7.4% below the base period, the current one only 1.1% below.1
That is why output per job and per worker barely change between methods. For 2009 to 2019 output per job grows 1.0% a year on both, and output per worker 0.9% on the current measure against 1.0% on the new.1 Only the measure that divides by hours moves.
The ONS compares both methods with an administrative measure built from PAYE Real Time Information. Output per worker in 2024 was about 2.4% above its 2019 level on the current measure, 1.9% on the new one and 1.5% on the payroll data. The new method is closer to the administrative figure, but still above it.1 On both ONS measures, output per job is above its 2019 level but below its 2022 peak.1
The ONS labels the component-approach figures indicative and not official statistics. It plans to adopt the method for whole-economy figures in November 2026, with industry detail to follow in February 2027, and the numbers could change before then.1 This piece does not show industry detail, where the effects of the change may differ. Nor does it say why hours per job fell more on the new method than the survey suggested, beyond the ONS's own explanation of better source data.