This article provides experimental statistics covering subregional productivity measures. The productivity measures provided are GVA per hour worked and GVA per filled job and they are provided for NUTS 2 and NUTS 3 subregional geographies.
In January 2009, the National Statistician produced an article examining the measurement of regional economic performance. The article discussed the limitations of GVA per head as a measurement of the economic performance of a region and the income of its residents. The article instead proposed that to understand regional economic performance a suite of indicators should be used. These include the productivity measures GVA per filled job and GVA per hour.
The themes of the National Statistician's article also apply to the analysis of subregional economic performance. At the subregional level it is therefore recommended that GVA per head is not used as a proxy for measuring subregional economic performance but rather that a suite of indicators should be used including measures of productivity.
This article therefore provides two measures of subregional productivity; namely GVA per filled job and GVA per hour worked. Subregional productivity data was first published by ONS in an article in October 2011 and this article updates that publication with an additional year of data. All the subregional productivity data in this article is calculated to be consistent with the regional productivity data provided in the quarterly ONS Labour Productivity Statistical Bulletin.
The subregional productivity data in this article has been compiled to be consistent with the regional productivity data published in the ONS Labour Productivity Statistical Bulletin
Both regional and subregional productivity measures are produced by ONS on a nominal basis only. In other words, there is no separation of volume and price in the final output. As such, different levels of nominal productivity across different subregions will be impacted by any difference in prices between these subregions in addition to differences in production volumes per input.
The methodology used in this article is unchanged from that used in the previous subregional productivity article published in October 2011. The full methodology is outlined below.
The data accompanying this article is based on the NUTS geographical classification that was in use from January 2008 to December 2011. More information on this is available in the background notes section.
Regional productivity data is published by ONS in the ‘Productivity Measures by Region’ table which is updated annually in the Labour Productivity Statistical Bulletin (Q3). This regional table includes two productivity measures; GVA per filled job and GVA per hour worked. The subregional productivity data has been compiled to be consistent with the data in this regional table.
This requires ensuring that the subregional measures of GVA, jobs and hours are all consistent with the regional totals. The methodology is therefore concerned with how best to apportion the regional totals to the subregional areas. The approach taken is as follows:
Regional GVA data is published by ONS as either smoothed (Headline) GVA or unsmoothed GVA. Regional (NUTS 1) productivity calculations use the unsmoothed workplace based GVA at current basic prices series (Table 1.9 in the latest regional GVA release). Therefore the aim in the subregional productivity calculations is to apportion out this NUTS 1 unsmoothed workplace based GVA at current basic prices series to NUTS 2 and NUTS 3 subregions.
Although subregional GVA data is published at NUTS 2 and NUTS 3 geographies in the annual Regional GVA release, it is based on apportioning out the Headline (smoothed) NUTS 1 GVA to the NUTS 2 and NUTS 3 subregions rather than the unsmoothed data. As stated above, to be consistent with the regional productivity calculations it is necessary to apportion out the unsmoothed NUTS 1 GVA.
As such, the subregional GVA data being used in these subregional productivity calculations differs slightly from the GVA data series currently published by ONS for NUTS 2 and NUTS 3 subregions. This is because the published subregional GVA data are all constrained to the Headline NUTS 1 GVA series, whereas the data being used in these productivity calculations have been constrained to the unsmoothed NUTS 1 GVA series.
At the regional level, GVA per filled job is calculated using a ‘Productivity Jobs’ series as the denominator. This is compiled from four components; employee jobs, self employed jobs, government supported trainees (GST) & HM Forces. The regional Productivity Jobs series is benchmarked to the national Productivity Jobs series for consistency on a quarterly basis. To produce annual totals for regional Productivity Jobs, an average of the four quarters in the year are taken.
For subregional geographies, the ‘Total Jobs’ data series is used to apportion regional productivity jobs to subregions. This Total Jobs measure is a workplace based measure of jobs that ONS produces principally for use in calculating job densities at regional and subregional level. Total Jobs comprises employees (from the Business Register Employment Survey), self-employment jobs (from the Annual Population Survey), government-supported trainees (from DfES and DWP) and HM Forces (from MoD).
The Total Jobs series is used to calculate the proportions of regional jobs within each subregion for each year. These results are then used to apportion the regional ‘productivity jobs’ data series to the subregional level.
At the national and regional level, GVA per hour worked is calculated using a ‘Productivity Hours’ series as the denominator. This is calculated by multiplying the jobs series at industry level (which is based on Labour Force Survey (LFS) data allocated by industry) by the average actual hours worked for the industry, also derived from the LFS. Results are then scaled to ensure the whole economy productivity hours equal the appropriate LFS hours total. This data is calculated quarterly and an annual total is constructed as the average of the four quarters in the calendar year.
At subregional level, only annual productivity data is being produced. Therefore, the Annual Population Survey (APS)1 is used rather than the LFS as it has a larger sample size. The process involves calculating total hours for each subregion as the sum of employee hours, self employment hours, hours worked in government training schemes and hours worked by HM Armed Forces.
Employee hours are calculated by using the APS to estimate average hours worked per employee job for each subregion. These totals are then multiplied by the level of employee jobs for each subregion, taken from the Annual Business Inquiry (ABI) to 2008 and the Business Register and Employment Survey (BRES) from 2009, to give a total employee hours total for each subregion. Self employment hours are calculated from the APS. For government training schemes and HM armed forces, the regional totals are allocated to subregions based on each subregions share of regional employee + self employment hours as calculated in the previous stage.
Adding together the sum of employee hours, self employment hours, hours worked in government training schemes and hours worked by HM Armed Forces provides a total hours estimate for each subregion. Once calculated this NUTS 2 and NUTS 3 subregional data is then constrained regionally to the NUTS 1 ‘Productivity Hours’ data to ensure consistency with regional productivity data.
Data in this article has been revised in comparison to the article published in October 2011. The revisions are due mostly to revisions in the regional and subregional GVA estimates that form the numerator of productivity calculations. More details of these revisions are available in the latest Regional GVA release.
A project is underway to develop estimates of real regional and subregional GVA growth using a production approach. The future development of such estimates would lead to an improvement in the quality of the regional and subregional productivity data that ONS is able to produce. This is because it would allow for a separation of volume and price in the final outputs. This is currently not possible when calculating productivity using regional and subregional GVA calculated using the income approach.
A report of progress to date on development of the production approach to regional and subregional GVA will be published in spring 2012.
For 2005, it was not possible to use APS data, therefore LFS data was used – with the average taken of the four LFS surveys carried out in 2005.
There is a strong correlation between the two measures of labour productivity being considered in this article, GVA per hour worked and GVA per filled job. For this reason, in this brief analysis of the results we focus on just one of the measures, GVA per hour worked.
Figure 1 shows the fifteen English NUTS 3 subregions with the highest GVA per hour worked in 2009. The highest productivity levels are found in Inner London with productivity in Inner London West 59 percent above the UK average and Inner London East 29 percent above the UK average.
Outside of London, productivity is highest in Berkshire and Surrey whilst Swindon, Buckinghamshire CC and Luton all have productivity levels more than 10 per cent above the UK average.
It should be noted that one impact of the very high productivity levels in London is that this leads to a skewed distribution of productivity levels across the UK such that relatively few subregions have productivity levels above the UK (mean) average. In fact, just 24 out of the 128 NUTS 3 subregions across England, Scotland and Wales had a GVA per hour above the UK average in 2009. Of these, 22 are English subregions, of which three are located in the North of the country (Halton & Warrington, Cheshire CC, and Leeds), two in the Midlands (Solihull and Derby), and the remaining seventeen in the South of the country (including all five London NUTS 3 subregions).
Given the skewed nature of the distribution it is worth considering the mid-ranking (median) subregion (i.e. the subregion that ranks 47th out of the 93 English subregions) and comparing its performance to the (mean) UK average. For 2009, the mid-ranking subregion in England was Bedfordshire with a GVA per hour that is 92% of the UK average.
Figure 2 shows the fifteen English NUTS 3 subregions with the lowest GVA per hour worked. They all have productivity at least 17 per cent below the UK average. Three subregions have productivity over 25 percent below the UK average.
The subregions included in Figure 2 are varied in location and type. A number are located towards the geographical periphery of England or are coastal resorts. Additionally, a number of largely rural subregions are included in the list such as Northumberland, Herefordshire, Lincolnshire and Staffordshire. Equally, however, there are some mostly urban local authorities such as Blackburn with Darwen, Stoke-on-Trent, Wallsall and Wolverhampton, and Greater Manchester North.
There is a strong correlation between the two measures of labour productivity being considered in this article, GVA per hour worked and GVA per filled job. For this reason, in this brief analysis of the results we focus on just one of the measures, GVA per hour worked. For Northern Ireland, GVA per hour worked data is unavailable for the NUTS 2 and NUTS 3 geographies. Therefore, for Northern Ireland, GVA per filled job data is shown instead.
Figure 3 shows GVA per hour worked data for all NUTS 3 subregions in Scotland. The data shows a number of subregions with high productivity performance. Both City of Edinburgh and Aberdeen City and Aberdeenshire are ranked within the top fifteen UK subregions in terms of productivity performance. Overall, twelve of the twenty three Scotish NUTS 3 subregions have subregional productivity above the median average for Great Britain of 91.3 per cent (i.e. they rank amongst the top 50 per cent of GB subregions).
At the opposite end of the scale a number of Scottish subregions display relatively low levels of productivity, with productivity in six subregions over twenty per cent below the UK average. These are predominantly the more rural and geographically peripheral subregions of Scotland.
Figure 4 displays productivity data for the NUTS 3 subregions of Wales. Cardiff and Vale of Glamorgan has the highest productivity amongst the Welsh subregions. Although its productivity is slightly below the UK mean average, it actually ranks a relatively high 32nd out of the 128 subregions in Great Britain. Two other subregions in Wales are also ranked amongst the top 50 per cent of NUTS 3 subregions in Great Britain in terms of GVA per hour, namely Monmouthshire and Newport, and Bridgend and Neath Port Talbot.
The majority of the subregions in Wales, however, are towards the lower end of the rankings with five subregions having productivity over 20 percent below the UK average. These subregions are generally the more rural subregions of Wales.
As mentioned earlier, there is no GVA per hour worked data available for the subregions of Northern Ireland. Therefore, Figure 5 shows GVA per filled job productivity data for Northern Ireland instead. Productivity is highest in the subregion of Belfast, albeit it is 9 per cent below the UK average. Productivity is lowest in the subregions covering the North, West and South of Northern Ireland where productivity is over 21 per cent below the UK average.
As discussed in the introduction, in January 2009 the National Statistician produced an article examining the measurement of regional economic performance. This article discussed the limitations of GVA per head as a measurement of the economic performance of a region and the income of its residents.
So what is the problem with using GVA per head to measure subregional productivity? As productivity describes the ability to produce outputs, taking into consideration the amount of inputs used to produce them, then at first sight it may appear that GVA per head would make an appropriate indicator of productivity, as it is providing a measure for each subregion that compares output (GVA) in the numerator with an input (population) in the denominator.
The problem, however, stems from the fact that the input measure used in GVA per head (residential population) is not a good measure of the actual labour input involved in the production of the subregion's output (GVA).
There are a number of reasons for this. Firstly, by including all the residential population and not just those who are in employment, the denominator includes residents who are not directly contributing to GVA.
A second key problem with GVA per head is that it is dividing a workplace-based numerator (workplace based GVA) by a residence-based denominator (residential population). This means it does not account for people commuting in and out of a subregion.
For these reasons, GVA per hour worked or GVA per filled job are the most appropriate measures of subregional productivity. These measures only count the input of those who are directly employed in the production process (rather than the population as a whole) and additionally they provide a workplace based labour input denominator to match the workplace based GVA numerator, thus fully accounting for the impacts of commuting.
The differences in the results between measures of subregional productivity and GVA per head are not negligible. To illustrate this, Figure 6 shows GVA per head, GVA per filled job and GVA per hour worked for a small number of selected NUTS 3 subregions.
Take the examples of Nottingham and South Nottinghamshire in Figure 6. Their productivity performance, as measured by GVA per filled job and GVA per hour worked, is very similar. As such, both subregions are achieving a similar level of output per labour input being utilised within their subregions. In other words the productivity measures presented in this article suggest the two subregions have a similar economic performance.
However, if an analysis of economic performance had only been carried out using the GVA per head measure, then it would have appeared that the economic performance of Nottingham were much greater than that of South Nottinghamshire. This is not the case. The large difference in GVA per head between these two subregions is due to the impact of commuting patterns and not the actual economic performance of the two subregions as measured by labour productivity.
The major factor that tends to lead to GVA per head being higher than productivity for a subregion is a high level of in-commuting. Therefore, urban centres often display this pattern. The examples of Nottingham and Glasgow are shown in Figure 6.
The opposite is also true. Therefore in Figure 6, GVA per head is much lower than productivity for Wirral and South-Nottinghamshire. These subregions both display significant out-commuting. This pattern is typical for many rural and suburban subregions.
Other factors that can influence the GVA per head measure include the employment rate in a subregion and the share of population who are aged 16-64. However, the commuting effect tends to be the one that most dominates in diverging GVA per head from productivity across many UK subregions.
The net result is that, as described in the National Statistician's article, ONS recommends the use of GVA per filled job or GVA per hour worked as the appropriate measures of subregional productivity. These productivity measures can be included in a wider basket of indicators alongside income and labour market data to more fully describe subregional economic performance.
The difference between GVA per filled job and GVA per hour worked is that GVA per hour worked takes into consideration the hours worked per job. These may differ across subregions due to different industrial structures and different levels of part-time working. As such GVA per hour worked is considered the preferred measure.
Unsmoothed time series data at small geographies such as NUTS 2 and NUTS 3 tend to show volatility, created by sampling and non-sampling errors. Therefore, a five year weighted average has been used to remove this volatility and produce a smoothed time-series. The data presented in this article is this smoothed subregional productivity data series. It should be noted that when calculating the subregional productivity data, unsmoothed data has been used at all times. The smoothing process has only been applied to the final results. For any users who would like to make use of the unsmoothed results, this data is included in the data section of this publication.
The latest subregional productivity data available is for 2009. The timeliness of the data is determined by the release calendar for subregional GVA data. Subregional GVA data for 2010 will be available in December 2012 and an update to this article will follow shortly afterwards. At the regional level, data is available a year ahead of subregional level data.
Subregional productivity data is produced in this article for the NUTS 2 and NUTS 3 geographies. Nomenclature of Units for Territorial Statistics (NUTS) is a geography developed by the European Union to allow comparison of regional and subregional data across the EU-25 member states. A number of ONS regional and subregional outputs are produced based on the NUTS geography. These include regional and subregional GVA. This GVA data is an input in the calculation of subregional productivity. Typically, NUTS 3 subregion covers the same area as either a single local authority or a combination of two or more adjacent local authorities, although there are some NUTS 3 subregions in Scotland that don't follow administrative boundaries.
The data accompanying this article is based on the NUTS geographical classification that was in use from January 2008 to December 2011. This is because the subregional GVA data used as an input for the productivity calculations was published in December 2011 using the above NUTS classification and is not currently available for the new NUTS boundaries introduced in January 2012. Regional GVA data on the new classification will be available in December 2012 and this will allow for subregional productivity data to be published using the new NUTS classification during 2013.
Compared to the NUTS classification used in this article, the new classification involves changes to the following NUTS 3 subregions, Bedfordshire CC, Cheshire CC, Northamptonshire, Dudley and Sandwell, Walsall and Wolverhampton, Halton and Warrington, East Merseyside, and Calderdale, Kirklees and Wakefield as well as the NUTS 2 subregions of Cheshire and Merseyside.
Regional Productivity data is published in Table 9 of the Labour Productivity Statistical Bulletin.
Contact Richard Prothero, Regional Economic Analysis
telephone: 01329 447825
You may use or re-use this information (not including logos) free of charge in any format or medium, under the terms of the Open Government Licence. You can write to the Information Policy Team, The National Archives, Kew, London TW9 4DU, or email: firstname.lastname@example.org.
Details of the policy governing the release of new data are available by visiting www.statisticsauthority.gov.uk/assessment/code-of-practice/index.html or from the Media Relations Office email: email@example.com