- Labour productivity for Quarter 4 (Oct to Dec) 2019, as measured by output per hour, saw a small rise of 0.3% compared with the same quarter a year ago.
- This rise was caused by gross value added (GVA) growing at 1.1% compared with the same quarter a year ago, while hours worked grew by 0.8%.
- The 0.3% growth in output per hour was largely caused by a strong performance from construction, while manufacturing made the largest negative contribution to whole-economy productivity growth.
- Over the decade, whole-economy growth in output per hour was led by productivity improvements in non-financial services, while financial services saw a decline in output per hour.
Productivity is the main cause of economic growth and largely determines the long-term economic health of a nation. It helps define both the scope for raising living standards and the competitiveness of an economy, increasingly informing government policy.
Labour productivity measures the volume of gross value added (GVA) produced per unit of labour input, with hours worked as the preferred labour input. It has demonstrated weak growth since the 2008 economic downturn, while in the previous 10 years it was close to historical long-term average growth rates of 2.0% per year. This sustained period of minimal labour productivity growth has been labelled the UK’s “productivity puzzle”, and it is arguably the defining economic question of our age.
In December 2019, the Royal Statistical Society named the estimated average annual increase in UK productivity in the decade or so since the financial crisis the “Statistic of the Decade”, reflecting the significance of the unusual weakness observed since the 2008 economic downturn.
In Quarter 4 (Oct to Dec) 2019, output per hour was 0.3% higher when compared with the quarter a year ago (this is called "quarter on year" growth). After four consecutive quarters of zero or negative growth, the last two quarters of positive growth are an improvement. However, in historical terms 0.3% is still a very low productivity growth rate, significantly below the post-2008 economic downturn median and less than a seventh of the pre-downturn median growth rate. The takeaway story is the continued weakness of the UK’s productivity growth since the economic downturn.
Figure 1 shows the log growth rate of output per hour compared with the same quarter a year ago, noting the 25th, 50th and 75th percentiles of growth. Labour productivity growth since the downturn has been consistently far weaker than in the decade before it. The median productivity growth of the post-downturn period is less than one-quarter of what it was during the pre-downturn period (starting with Quarter 1 (Jan to Mar) 1998).
Furthermore, productivity growth has still not recovered to anything like pre-downturn levels. Since the downturn, more than one in four quarter on year growth rates were negative, while Quarter 4 2004 saw the only negative growth rate in the decade before the downturn.
Figure 1: Output per hour has risen by 0.3% from the same quarter a year ago, the second consecutive quarter
Output per hour, quarter on year log growth rates, seasonally adjusted, UK, Quarter 1 (Jan to Mar) 1998 to Quarter 4 (Oct to Dec) 2019
Percentiles are measurements that indicate the percentage of observations beneath a specified point. The 25th percentile is the value below which 25% of the observations reside.
Percent log growth used in the chart will differ slightly from percent growth in published datasets.
Figures for output per job, an alternate measure of labour productivity, are included in the LPROD01 dataset. In Quarter 4 2019, output per job contracted by 0.1% compared with the the same quarter a year ago. Over a longer time period, output per job exhibits a similar pattern to output per hour, with the post-downturn median roughly a third of the pre-downturn growth rate.
Output per hour grew by 0.3% compared with the previous quarter, the same as its quarter on year growth rate.
Gross value added and hours worked
Output per hour is calculated as gross value added (GVA) divided by the number of hours worked. This means that using all logged growth rates, the change in output per hour can be expressed as the change in GVA minus the change in hours. Compared with the same quarter a year ago, GVA grew by 1.1% while hours worked grew by 0.8%, resulting in the 0.3% growth in labour productivity.
Figure 2 shows the quarter on year log growth rate for output per hour, decomposed into the growth rates of GVA and hours. Because an increase in hours causes productivity to fall, holding GVA constant, hours growth is “reversed” in Figure 2 to show it as a negative contribution.
Since the downturn, annual log growth in GVA averaged 1.8%, somewhat higher than the average log growth rate of hours worked of 1.3%. As a result, productivity log growth has averaged 0.6%, with brief instances of higher growth, most noticeably in the initial recovery from the 2008 economic downturn. Subsequent years experienced a slump in productivity, with consecutive periods of negative growth between Quarter 2 (Apr to June) 2012 and Quarter 1 (Jan to Mar) 2013. Since then, productivity growth has remained weak, despite a somewhat stronger performance from mid 2016 to early 2018.
Nôl i'r tabl cynnwys
Another way of breaking down the headline growth figure of 0.3% is using the “Generalised Exactly Additive Decomposition” (GEAD) methodology to decompose whole-economy productivity growth into the direct contributions from different industry sections.
Figure 3 shows that the UK’s whole-economy growth in output per hour was caused by the construction industry. With all growth rates in logs, output per hour in construction grew by 7.0% compared with the same quarter a year ago, because its gross value added (GVA) grew by 1.6% while its hours declined by 5.4%. This contributed 0.4% to whole-economy productivity growth, greater than the 0.3% the whole economy actually saw.
It is worth noting that productivity growth in the construction sector has been historically volatile. The largest negative contribution was made by manufacturing, which declined by 1.6% in productivity and contributed negative 0.2% to whole-economy productivity growth.
While several industry sections saw large increases or decreases in productivity, the positive and negative contributions mostly cancelled out, resulting in the small whole-economy increase of 0.3%.
Contributions to productivity growth over 10 years
At the end of the decade, it is now possible to compare the most recent labour productivity data with the decade as a whole. In this analysis, we focus on which sectors of the economy have contributed to productivity growth.
In the 10 years to Quarter 4 (Oct to Dec) 2019, output per hour grew by log 5.2%, indicating an annual growth rate of log 0.5%. This growth is not evenly distributed throughout the economy. Instead, it was overwhelmingly led by the non-financial services sector, notably through strong growth in the wholesale and retail trade and administration and support services industries. Apart from non-financial services, the rest of the economy saw in aggregage virtually no growth in output per hour over the 10 years up to Quarter 4 2019.
Strikingly, financial services has seen a net negative contribution to output per hour growth, reflecting that its productivity has actually declined over the decade. This contrasts with the years leading up to the 2008 economic down, where financial services saw extremely strong productivity growth, so the post-downturn decline may offer evidence this earlier growth was unsustainable.
Figure 4 compares the quarter on year industry contributions to whole-economy productivity growth with the same industry’s contributions over a 10-year period ending with Quarter 4 2019. Generally, there is little link between the quarter on year contributions and the decade’s average contributions.
In contrast to the quarter on year contributions, over the 10-year period, construction’s contribution to productivity growth was relatively insignificant, showing the most recent figures are unusual for construction. Despite the non-financial services sector’s recent negative contribution, over the last decade it was the main cause of productivity growth.
Over the 10-year period, the allocation effect (the effect of changes in the relative size of industries on productivity) was somewhat negative, implying a transfer of labour inputs from more productive to less productive industry sections. However, this may not present the full picture because of the effect of imputed rental. Imputed rental represents the value of the accommodation services that owner-occupiers provide to themselves. It makes up about 10% of real UK GVA and the majority of the GVA of the real estate industry. There is a strong case for excluding imputed rental from GVA for the purposes of calculating labour productivity; see, for example, Section 11.16 of the European System of Accounts 2010 (ESA 2010).
This is relevant because research suggests that excluding imputed rental causes significant changes to the allocation effect because of the low labour input relative to the outputs. In particular, if imputed rental is excluded from GVA, the average direct effect over the period 1998 to 2008 increases by 0.14 percentage points, while the average allocation effect decreases by 0.54 percentage points.
Over the decade 2008 to 2018 (the latest period for which this analysis is available), the overall growth differential is much smaller. Excluding imputed rent makes no difference to the average direct contribution to output per hour growth in the 10 years to 2018, while the allocation effect decreases by 0.12 percentage points per year on average.Nôl i'r tabl cynnwys
Labour Productivity Tables 1 to 8 and R1 (LPROD01)
Dataset | Released 7 April 2020
Estimates of main productivity metrics, corresponding to tables from the PDF version of the statistical bulletin
Productivity jobs, productivity hours, market sector workers, market sector hours (LPROD02)
Dataset | Released 7 April 2020
Underlying labour inputs behind the labour productivity estimates by industry and industrial sector as defined by the Standard Industrial Classification (SIC). Contains statistics on productivity jobs, productivity hours and market sector workers. These statistics are the main intermediates in producing output per worker and output per hour statistics.
Breakdown of contributions, whole economy and sectors
Dataset | Released 7 April 2020
Provides estimates of contributions to labour productivity (measured as output per hour) using the “Generalised Exactly Additive Decomposition” (GEAD) methodology as described in Tang and Wang (2004), UK. Contains data on total worked hours, gross value added (GVA) estimates, output per hour series and prices deflators. Includes data disaggregated by sector. Also contains quarter on quarter, quarter-on-same-quarter a year ago and annual formats for selected outputs.
Labour productivity by industry division
Dataset | Released 7 April 2020
Contains statistics on productivity hours, output per hour and output per hour at current prices. Productivity hours measures the whole economy and sectoral hours worked. Output per hour is GVA divided by productivity hours. Output per hour at current prices is displayed in British pounds. These are experimental statistics for the UK.
Labour productivity: revisions triangles (LPRODREV)
Dataset | Released 7 April 2020
Revisions triangles for the main labour productivity variables. Data present the first estimates of chosen statistics used in the labour productivity publication against later revised estimates. Includes output per worker, output per job and output per hour, first estimates and revisions.
Labour productivity time series (PRDY)
Dataset | Released 7 April 2020
Quarterly output per hour, output per job and output per worker for the whole UK economy and a range of industries.
Quarterly regional productivity hours and jobs (NUTS1)
Dataset | Released 7 April 2020
Quarterly UK productivity hours and jobs for the Nomenclature of Units for Territorial Statistics: NUTS1 regions. Seasonally adjusted and non-seasonally adjusted experimental statistics.
Labour productivity is calculated by dividing output by labour input.
Labour inputs in this release are measured in terms of workers, jobs (“productivity jobs”) and hours worked (“productivity hours”).
Output refers to gross value added (GVA), which is an estimate of the volume of goods and services produced by an industry, and in aggregate for the UK.
The measure of output used in these statistics is the chained volume (real) measure of gross value added (GVA) at basic prices.
Labour input measures used in this bulletin are known as “productivity jobs” and “productivity hours”. Productivity jobs differ from the workforce jobs (WFJ) estimates, published in Table 6 of our Labour market overview, in three ways:
- to achieve consistency with the measurement of GVA, the employee component of productivity jobs is derived on a reporting unit basis, whereas the employee component of the WFJ estimates is on a local unit basis
- productivity jobs are scaled so industries sum to total Labour Force Survey (LFS) jobs – note that this constraint is applied in non-seasonally adjusted terms; the nature of the seasonal adjustment process means that the sum of seasonally adjusted productivity jobs and hours by industry can differ slightly from the seasonally adjusted LFS totals
- productivity jobs are calendar quarter average estimates, whereas WFJ estimates are provided for the last month of each quarter
Productivity hours are derived by multiplying employee and self-employed jobs at an industry level (before seasonal adjustment) by average actual hours worked from the LFS at an industry level. Results are scaled so industries sum to total unadjusted LFS hours and are then seasonally adjusted.
Industry estimates of average hours derived in this process differ from published estimates (found in Table HOUR03 in the Labour market overview release), as the HOUR03 estimates are calculated by allocating all hours worked to the industry of main employment, whereas the productivity hours system takes account of hours worked in first and second jobs by industry.
Labour productivity is then derived using growth rates for GVA and labour inputs in line with the following equation:
Presentation of growth rates in log percentage changes
In this release, charts and associated text measure growth in terms of percentage log changes, and we will continue to use this presentation in future releases. The datasets will still contain percentage growth rates and these statistics hold the National Statistics status.
For typical rates of change for labour productivity and labour inputs, this change will not make much difference to the result. For example, a 2.0% percentage change translates to a 1.98% log change. We are adopting the approach because a log change between two observations has the same numerical value regardless of which observation is the starting point. This is not true for a percentage change. For illustrative purposes, in the following example log changes are substantially different from percentage changes.
Suppose a series starts at 7, doubles to 14, then halves back to 7. The log change from 7 to 14 is 69%, and the log change from 14 to 7 is negative 69%. But the percentage change from 7 to 14 is 100%, while the percentage change from 14 to 7 is negative 50%. The log change reflects the fact that the second change reverses the first (and so has the same value) while the percentage change series appears to be very different in the first period compared with the second.
This approach is the same as that used by the Office for National Statistics (ONS) to compile multi-factor productivity.
This release reflects revisions to GVA resulting from quarterly national accounts, affecting time periods since the start of 2019. Revisions to the current data also reflect revisions to jobs data for Quarter 3 (July to Sept) 2019. Revisions resulting from seasonal adjustment affect all periods.
This research note provides further information on the sources of revisions to labour productivity estimates.
Quality and methodology
More quality and methodology information on strengths, limitations, appropriate uses, and how the data were created is available in the Labour productivity QMI.Nôl i'r tabl cynnwys
This release reports labour productivity estimates for Quarter 4 (Oct to Dec) 2019 for the whole economy. Productivity is important as it is considered to be a cause of long-run changes in average living standards.
This edition forms part of our quarterly productivity bulletin, which also includes unit labour costs, quarterly estimates of public service productivity, quarterly estimates of multi-factor productivity, and a productivity economic commentary.
Comparability and consistency
The output statistics in this release are consistent with the latest Quarterly national accounts, released 31 March 2020. Note that productivity in this release does not refer to gross domestic product (GDP) per person, which is a measure that includes people who are not in employment.
The labour input measures used in this release are consistent with the latest labour market statistics,released 17 March 2020.
In October 2018, the Office for National Statistics (ONS) informed users in a notice that we will no longer be publishing estimates on international comparisons of productivity, owing to an ongoing review of the methodology. In December 2018, the Organisation for Economic Co-operation and Development (OECD) published a working paper, “International productivity gaps: Are labour input measures comparable?”, which showed the methodologies, data sources and adjustments used to estimate labour inputs varied significantly across countries. The ONS published an article exploring these differences and the impact they had on our international comparison of productivity statistics.
The GEAD methodology
The analysis of industrial contributions to whole-economy changes in labour productivity in this article uses the “Generalised Exactly Additive Decomposition” (GEAD) algorithm described in Tang and Wang (2004). This is the same algorithm as the one used to derive the contributions published in the dataset, “Breakdown of contributions, whole economy and sectors”, which is part of the labour productivity release.
The GEAD algorithm has wide acceptance in the economics community. However, users should be aware that some commentators (for example, Reinsdorf 2015) have raised concerns about the interpretability of the GEAD contributions. In particular, the direct effects reported in the charts in this release are weighted by nominal GVA, not real GVA. This is necessary to ensure additivity (because real GVA is not additive) but means that the GEAD direct effects reflect the impact of numerically equivalent changes in real GVA on labour productivity differently depending on whether they arise from price changes or not.
Unless otherwise stated, all figures are seasonally adjusted.
More information on the strengths and limitations of the data, as well as the quality and accuracy of the data, is available in the Labour productivity QMI.Nôl i'r tabl cynnwys
Manylion cyswllt ar gyfer y Bwletin ystadegol
Ffôn: +44 (0)1633 455086