- UK Labour Productivity as measured by output per hour grew by 0.5% from the second to the third calendar quarter of 2015 to the highest level ever recorded for this series, albeit still some 13% below an extrapolation based on its pre-downturn trend.
- By contrast, output per worker and output per job decreased by 0.2% and 0.1%, respectively, as a result of average hours decreasing in Q3.
- Output per hour in services grew strongly in Q3 to a record high, but manufacturing output per hour grew by only 0.1%, continuing the exceptionally weak trend for this series since the economic downturn.
- Whole economy unit labour costs were 2% higher in Q3 than the same quarter last year, as earnings and other labour costs have outpaced productivity. Unit wage costs in manufacturing grew by 4.6% over this period.
- This edition contains new estimates of regional labour productivity to 2014. These estimates show that productivity is above the UK average in London and the South East region, and well below the UK average in Wales and Northern Ireland.
This release reports labour productivity estimates for the third quarter (July to September) of 2015 for the whole economy and a range of sub-industries, together with selected estimates of unit labour costs. Labour productivity measures the amount of real (inflation-adjusted) economic output that is produced by a unit of labour input (measured in this release in terms of workers, jobs and hours worked) and is an important indicator of economic performance.
Labour costs make up around two-thirds of the overall cost of production of UK economic output. Unit labour costs are therefore a closely watched indicator of inflationary pressures in the economy.
Output statistics in this release are consistent with the latest Quarterly National Accounts published on 23 December 2015. Labour input measures are consistent with the latest Labour Market Statistics as described further in the 'General commentary' and 'Notes on sources' sections below.Back to table of contents
Whole economy output (measured by gross value added - GVA) increased by 0.4% in the third quarter of 2015, while the Labour Force Survey (LFS) shows that the number of workers and jobs increased by 0.6% and 0.5% respectively but hours worked decreased by 0.1% over this period.This combination of movements in outputs and labour inputs implies that labour productivity across the whole economy increased by 0.5% in terms of output per hour, but output per worker and output per job decreased by 0.2% and 0.1% respectively.
Differences between growth of output per worker and output per job reflect changes in the ratio of jobs to workers. This ratio decreased very slightly in Q3. Differences between these measures and output per hour reflect movements in average hours per job and per worker which, though typically not large from quarter to quarter, can be material over a period of time. For example, between Q2 and Q3 average hours per worker fell from 32.1 to 31.9 hours per week, partly reflecting faster growth of part-time employment than full-time employment. For this reason, output per hour is a more comprehensive indicator of labour productivity and is the main focus of the commentary in this release.
This equation explains how Labour Productivity is calculated and how it can be derived using growth rates for GVA and labour inputs.
Unit labour costs (ULCs) reflect the full labour costs, including social security and employers’ pension contributions, incurred in the production of a unit of economic output, while unit wage costs (UWCs) are a narrower measure, excluding non-wage labour costs. Growth of ULCs can be decomposed as:
This equation explains how ULCs are calculated and how it can be derived from growth of labour costs per unit of labour (such as labour costs per hour worked) and growth of labour productivity.
In the third quarter, whole economy output per hour grew by 0.5% and ULCs grew by 0.3%. Plugging these values into the ULC equation and re-arranging yields an implied increase of approximately 0.8% in labour costs per hour. This implied movement differs from other ONS information on labour remuneration such as Average Weekly Earnings (AWE) and Indices of Labour Costs per Hour (ILCH), partly because the labour cost component includes estimated remuneration of self-employed labour, which is not included in AWE and ILCH.
Following a review of 17 component industries within the Index of Services, we changed the status of these series from Experimental to Official Statistics in the Q2 2015 release. For further information, see ‘Improvements to the output approach to measure UK GDP, 2015’ published on 30 September 2015. Accordingly, we have also changed the status of output per job and output per hour estimates that use these series as numerators. This means that none of the series in this release are Experimental Statistics.Back to table of contents
Productivity estimates in this release are derived from estimates of output of goods and services and of labour inputs, the latter measured in terms of workers, jobs (‘Productivity Jobs’) and hours worked (‘Productivity Hours’). In general, estimates of output and of labour inputs are measured independently of one another, with labour productivity calculated as the ratio of the two estimates. However there are some activities where, in the absence of direct measures of output, labour inputs are used as a proxy, with productivity either assumed to be unchanged over time (as in public administration and defence) or assumed to move in line with the productivity trend in a measurable equivalent activity (as in a few small components of the index of services).
Total hours worked fell in Q3 for the second consecutive quarter. This reflected a fall in average hours per job to the lowest level since Q4 2012; the number of workers and jobs both grew in Q3. The fall in average hours was fairly broad-based across services and the production industries.
On the other hand, output registered its 11 consecutive quarterly increase. Since Q4 2012 output has grown at an average annualised rate of 2.6%. Much of the growth in output is accounted for by services, and particularly non-financial, non-government services.
In consequence, whole economy output per hour was the highest ever recorded in Q3, albeit less than 1% higher than its pre-downturn peak in Q2 2008, and some 13% below an extrapolation based on the trend prior to the economic downturn. Output per hour in services was also the highest on record in the latest quarter, and about 3% above its pre-downturn level. By contrast, output per hour in the production industries was more than 6% below its pre-downturn level in Q3.
Prior to the economic downturn, productivity growth was slower in services (2.1% per annum) than production (3.5%) and manufacturing (4.0%). Productivity growth has slowed across the board since the downturn, but it has slowed much more in production and manufacturing than in services, such that this pattern has reversed. Output per hour in services has grown at an annualised average rate of 0.9% since the trough in output in Q2 2009. This is slightly ahead of the equivalent figure for manufacturing (0.8%) and well ahead of production (-0.4%). The disparity between manufacturing and production reflects a halving of productivity in mining and quarrying, although there have been some tentative signs of recovery in output and productivity in this industry in recent quarters.
Some of the slowdown in productivity can be attributed to slower output growth, particularly in services. But this does not fully explain the change in ranking, not least because output growth in manufacturing has, if anything, been a little higher post downturn than over the pre-downturn period.
Another factor is a striking transformation in trends of labour inputs. Across the service industries, growth of hours worked has been fairly similar, on average, over the post- and pre-downturn periods. Hence the slowdown in productivity growth in services mainly reflects the slower output growth noted above. This has been focussed in a few service industries, especially financial services.
By contrast, hours worked in manufacturing have gone from a long term trend decline of around 3.5% per annum prior to the economic downturn to positive growth since 2009, albeit still below the whole economy average (so the manufacturing sector has continued to shrink as a share of total hours). In manufacturing, therefore, the collapse in productivity growth primarily reflects unprecedented growth of labour inputs.Figure 1 provides a high level summary of movements in output per hour since Q1 2008 in terms of cumulative quarterly changes. Whole economy output per hour fell sharply in 2008 before staging a recovery up to mid 2011. Productivity then fell again through the second half of 2011 and through 2012, initially reflecting sluggish output growth and then reflecting strong growth in hours worked. Output per hour has grown by about 2.7 percentage points since Q4 2012, albeit with a period of broadly flat productivity between Q2 2013 and Q4 2014.
Over the period since Q1 2008, movements in whole economy output per hour have been dominated by positive contributions from ‘Other services’ (that is, excluding financial services) and negative contributions from industries ABDE (non-manufacturing production and agriculture). Focussing on the period since Q4 2012, the net contribution of ABDE has been to add 0.4 percentage points to overall productivity. The contribution of ‘Other services’ has added about 3.5 percentage points over this period, while the combined contributions of the remaining industries has been close to zero. Additionally there has been a negative contribution of about 1 percentage point due to shifts in resources from relatively high productivity industries to industries with lower productivity. (There is some non-additivity in contributions over this period because GVA weights are fixed from 2012).
In this case, the negative allocation effect partly represents the impact of lower oil prices, which reduces the value of output of ABDE relative to other industries, and partly reflects a reduction in the share of manufacturing in total output.Figure 2 shows annual changes in ULCs since Q1 2008, with the bars representing the decomposition of ULC changes into changes in labour costs per hour and changes in output per hour. The latter have been reversed in sign, so a negative bar represents positive productivity growth.
Downward contributions to ULC growth from generally positive productivity growth since the end of 2012 were initially accompanied by low or negative contributions from growth in labour costs per hour, allowing ULC growth to fall to -2.0% in Q2 2014. Since then, however, growth of labour costs per hour has accelerated, reaching 3.3% in Q3. This upward trend is also apparent in the Index of Labour Costs per Hour, which shows labour costs per hour excluding bonuses and arrears growing 3.5% in the year to Q3, and also, though to a lesser extent, in average weekly earnings growth.
Analysis of ULC growth by industry (available in Reference Table LPRODSULC (266.5 Kb Excel sheet) ) shows ULC growth across the service sector of around 2% pa over the last two quarters, with wide variation between service industries. These experimental statistics show similar rates of ULC growth for the market sector and for the production industries, where falling ULCs in non-manufacturing production are offsetting above-average ULC growth in manufacturing.Back to table of contents
Output per hour continued on an upward trend that began in late 2012. This measure grew in quarter 3, 2015 by 1.3% on the same quarter a year ago, the same as Q2 2015, and on the quarter it saw a rise of 0.5% (see Figure 3).
Output per hour growth in the latest quarter was driven by a 0.4% increase in GVA, aided by a slight fall in hours. Figure 4 shows that hours worked grew slightly faster than jobs from the start of 2011, suggesting that average hours per job has increased, but more recently this trend has reversed.
The flattening of hours relative to GVA and jobs has led to a faster increase in output per hour as opposed to output per job. In the past year, hours have grown by 0.7%, versus 1.3% for jobs. GVA growth of 2% over the same period has led to the productivity increase of 1.3% and 0.7% in output per hour and output per job, respectively.In general terms, Market Sector output per hour growth continues to track that of the whole economy after the two measures converged following the financial crisis (see Figure 5). Back to table of contents
Figure 6 shows output per hour in manufacturing in terms of cumulative quarterly annual changes and decomposed into broad component industries. Here the allocation element captures the effect of changes in output shares and relative prices within manufacturing, and has knocked about 1.5 percentage points off manufacturing productivity since Q1 2012. This primarily reflects relative price movements such as the effect of falling oil prices on chemical prices. Falling productivity in industries CA-CD+CM has knocked another 2.3 percentage points off manufacturing output per hour since Q1 2012, but has shown little overall change in contribution to output per hour since Q1 2008. Manufacturing GVA, jobs and hours have all fluctuated within a narrow range since 2010. Since Q3 2014, manufacturing GVA fell by 0.9% and hours increased by 1.1%, which explains the weakening productivity performance of manufacturing since this time last year.
More information on the labour productivity of sub-divisions of manufacturing is available in Reference Table LPROD01 (358.5 Kb Excel sheet) (Tables 3 and 4), and in the tables at the end of the pdf version of this statistical bulletin. Care should be taken in interpreting quarter on quarter movements in productivity estimates for individual sub-divisions, as small sample sizes of the source data can cause volatility.
Tables 3 and 4 include annual estimates for the level of productivity in current price terms for the National Accounts base year of 2012. These are estimates of GVA per unit of labour input and are not necessarily related to pay rates. Output per job (Table 3) varied from approximately £44,100 in Textiles wearing apparel and leather (divisions 13-15) to £135,900 in Chemicals and pharmaceuticals (divisions 20-21). The average for the whole of manufacturing was £59,700 and the average for the whole economy was £48,200 in 2012.
Chemicals and pharmaceuticals were also top of the distribution for output per hour in 2012 (£73.2 per hour), with Basic metals and metal products (divisions 24-25) at the bottom of the distribution (£25.0). On this basis the average for manufacturing as a whole was £32.3 and the average for the whole economy was £30.2 per hour.Back to table of contents
Figure 8 deconstructs the growth of output per hour in total services (G-U) since Q1 2008 into the cumulative contributions from different service industries. It suggests that much of the recent strengthening of output per hour in services has been driven by a partial recovery in transport and storage services and in retail, reflected in GHI, as well as stronger contributions from MN. The Real Estate industry (L) also stands out because its contributions to total services output per hour have been positive nearly every quarter since Q1 2008. However, Real Estate productivity is affected by the National Accounts concept of output from owner-occupied housing, which adds to the numerator but without a corresponding component in the denominator. As such, users should approach productivity estimates for the industry L with some caution. From Q1 2014 K and OPQ have had a negative impact on output per hour. Figure 8 also shows in the most recent quarter a negative allocation which captures either the impact of resources flowing from higher to lower productivity industries or movements in relative prices. Overall, output per hour in services is now on a steady upward trend after a couple of periods of moving sideways between Q1 2009 and Q1 2011, then again Q3 2011 and Q2 2012.
Services output and hours have both risen since late 2012 (see Figure 9), although more recent data for hours and jobs are less robust. 2015 has demonstrated a mild reduction in productivity hours after reaching a peak in Q1 2015. Coupled with positive GVA estimates, this has driven the rise in output per hour for services in Q3.
More information on labour productivity of services industries is available in Tables 5 and 6 of Reference Table PROD01 (358.5 Kb Excel sheet) and in the tables at the end of the PDF version of this statistical bulletin.
Excluding industry L, output per job in 2012 varied from approximately £20,900 in Accommodation and food services (section I) to £100,300 in Finance and insurance (section K). These industries were also at the bottom and top of the productivity distribution in terms of output per hour (Table 6).Back to table of contents
This section uses regional estimates of nominal GVA (NGVA) up to 2014, consistent with the ONS regional GVA release on 9 December 2015. The estimates in Table 9 of Reference Table PROD01 (358.5 Kb Excel sheet) and in the tables at the end of the PDF version of this statistical bulletin indicate the relative value of economic output per job and per hour across the NUTS1 regions, indexed to the UK=100. In interpreting these statistics it should be borne in mind that they do not take account of price differences across regions (e.g. in housing costs) and should not therefore be interpreted as measures of relative living standards.
There are revisions to regional GVA, jobs and hours worked compared with the previous estimates published on 24 December 2014. However, the overall pattern of regional productivity is broadly unchanged from previous estimates.
Figure 10 shows that, of the NUTS1 regions, only in London and the South East is NGVA per job above the UK average, although this is sufficient for England as a whole to be fractionally above average. NGVA per job was close to the UK average in 2014 in the East of England and in Scotland. In all other regions, NGVA per job was below than the UK average.
A broadly similar pattern obtains for NGVA per hour (Figure 11), although the margin by which London exceeds the UK average is significantly smaller, and Wales and Northern Ireland exchange places at the bottom of the productivity distribution.
Differences between Figures 10 and 11 reflect differences in average hours worked across the regions. In 2014, average hours worked in London were more than 6% above the UK average, and more than 5% above average hours worked in the next highest region (West Midlands). Average hours worked were lowest in Wales, around 3% below the UK average in 2014.
Regional productivity differentials have been fairly stable over time, although there is some faint evidence that London's productivity advantage is narrowing a little, and that productivity in Scotland is closing in on the UK average.Back to table of contents
There are revisions to both the numerator and the denominator for Labour Productivity estimates.
Growth rate revisions to the main productivity variables reflect, in part, a labour market benchmarking exercise. Workforce Jobs (WFJ) data, components of which are used to determine industry splits for productivity jobs estimates, were benchmarked to the annual Business Register and Employment Survey, as detailed in this revisions article (260.6 Kb Pdf). Productivity hours estimates, which are the product of jobs and average hours by industry, are also affected by the changes.
While whole economy estimates for both productivity jobs and productivity hours remain unchanged, the benchmarking changes have affected different industries in different ways from Q4 2008.
Industries that have seen an overall increase in labour input include mining and quarrying, manufacturing, construction, information and communications, and financial services. The increase in labour input has acted as a drag on productivity estimates for these industries.
Industries that saw an overall decrease in labour input, and thus a relative boost to their productivity estimates, include agriculture, utilities, transportation and storage, and other services.
The aggregate effect has resulted in relatively large upward revisions to manufacturing labour input and thus downward revisions to productivity estimates. The impact on services is much smaller.
Revisions on the output side were slightly negative since Q1 2014. Details of these revisions are available from the Quarterly National Accounts release. Financial services saw relatively large downward revisions due to benchmarking of insurance industry GVA to regulatory data and imputed data for banking GVA being replaced by real data.
Changes to unit labour cost growth rates also reflect revisions to mixed income estimates. These changes are due to revised estimates of Average Weekly Earnings (AWE - a component of mixed income) for employers with less than 20 staff, leading to re-evaluation of income against comparable labour market estimates. Details of the AWE revisions are available on the Labour Markets webpage.
The effect has increased mixed income estimates since Q1 2014, while Gross Operating Surplus estimates for the same period have fallen, largely due to the financial services revisions noted above.
Table A below summarises differences between first published estimates for each of the statistics in the first column with the estimates for the same statistics published three years later. This summary is based on five years of data, that is, for first estimates of quarters between Q4 2007 and Q3 2012, which is the last quarter for which a three-year revision history is available. The averages of these differences with and without regard to sign are shown in the right hand columns of the table. These can be compared with the estimated values in the latest quarter (Q3 2015) shown in the second column. Additional information on revisions to these and other statistics published in this release is available in the Revisions Triangles (3.42 Mb Excel sheet) component of this release.
Table A: Revisions analysis
|Change on quarter a year ago||Value in latest period (%)||Average over 5 years (bias)||Average over 5 years without regard to sign (average absolute revision)|
|Output per worker||0.6||0.2||1.0|
|Output per job||0.7||0.0||0.0|
|Output per hour||1.3||0.1||0.8|
|Unit labour costs||2.0||-0.3||1.2|
|Unit wage costs||2.7||-0.6||1.2|
|Source: Office for National Statistics|
|1. Revisions between first publication and estimates five years later (Relating to Period - Q4 2007 - Q3 2015).|
Download this table.xls
A research note, ‘sources of revisions to labour productivity estimates (145.4 Kb Pdf)’ is available on the ONS website.Back to table of contents
The measure of output used in these statistics is the chain volume (real) measure of Gross Value Added (GVA) at basic prices, with the exception of the regional analysis in Table 9, where the output measure is nominal GVA (NGVA). These measures differ because NGVA is not adjusted to account for price changes; this means that if prices were to rise more quickly in one region than the others, then this would be reflected in apparent improved measured productivity performance in that region relative to the others. At the whole economy level, real GVA is balanced to other estimates of economic activity, primarily from the expenditure approach. Below the whole economy level, real GVA is generally estimated by deflating measures of turnover; these estimates are not balanced through the supply-use framework and the deflation method is likely to produce biased estimates. This should be borne in mind in interpreting labour productivity estimates below the whole economy level.
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 the ONS Labour Market Statistics Bulletin, in three ways:
to achieve consistency with the measurement of GVA, the employee component of productivity jobs is derived on a reporting unit (RU) basis, whereas the employee component of the WFJ estimates is on a local unit (LU) basis. This is explained further below
productivity jobs are scaled so industries sum to total 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 then seasonally adjusted.
Industry estimates of average hours derived in this process differ from published estimates (found in Table HOUR03 in the Labour Market Statistics 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.
Whole economy unit labour costs are calculated as the ratio of total labour costs (that is, the product of labour input and costs per unit of labour) to GVA. Further detail on the methodology can be found in Revised methodology for unit wage costs and unit labour costs: explanation and impact.
Manufacturing unit wage costs are calculated as the ratio of manufacturing average weekly earnings (AWE) to manufacturing output per filled job. On 28 November 2012 ONS published Productivity Measures: Sectional Unit Labour Costs describing new measures of unit labour costs below the whole economy level, and proposing to replace the currently published series for manufacturing unit wage costs with a broader and more consistent measure of unit labour costs.
What is a reporting unit?
The term 'enterprise' is used by ONS to describe the structure of a company. Individual workplaces are known as 'local units' and a group of local units under common ownership is called the 'enterprise'. In ONS business surveys, reporting units are the parts of enterprises that return data to ONS. While the majority of reporting units and enterprises are the same, larger enterprises have been split into reporting units to make the reporting easier.
For most business surveys run by ONS, forms are sent to the reporting unit rather than local units, in other words, to the head office rather than individual workplaces. This enables ONS to gather information on a greater proportion of total business activity than would be possible by sending forms to a selection of local units. But it has the disadvantage that it is difficult to make regional estimates – for instance all the employment of, say, a chain of shops would be reported as being concentrated at the site of the head office.
Further differences between reporting unit and local unit data can be seen in the industry coding. Take, for example, a reporting unit with three cake shops and one bakery, each employing five people. The local unit analysis would put 15 employees in the retail industry and five employees in the manufacturing industry. But the reporting unit series puts all 20 people into the industry with the majority activity, in this case, retailing. Detailed industry figures compiled using the local unit approach will therefore be different from industry figures using the reporting unit approach, although the totals will be the same at the whole economy level.Back to table of contents
Contact details for this Statistical bulletin
Telephone: +44 (0)1633 456282