This article is the second in a series examining both the geographical concentration of industries and the industrial specialisation of local authority areas. This article investigates the patterns of spatial concentration of industries across Great Britain. Two measures of spatial concentration, the Moran’s I statistic and location quotients, are used to help describe these spatial patterns. Data tools and maps of the data are provided for users to explore the data. The data used has been derived from the Business Register and Employment Survey.
A previous article, 'The geographical concentration of industries' examined the degree to which employee jobs in different industries are concentrated within a small number of local authorities or spread more evenly across many. This article explores further those industries that display a concentration of employee jobs by examining the different spatial patterns of this concentration.
The results show that:
The information and communication sector and the professional, scientific and technical activities sector are examples of sectors where activity is spatially concentrated across neighbouring local authorities. In both these cases, employee jobs are concentrated across a large area of the south of England, particularly London and the South East region.
The finance and insurance activities sector is an example of a sector concentrated in smaller clusters of local authorities in which employee jobs are very highly concentrated. The highest concentration being within Central London.
The degree of spatial clustering varies amongst the sub-industries of the manufacturing sector. Some manufacturing industries show a high tendency for employee jobs to cluster across neighbouring local authorities but this is not the case for all manufacturing industries.
Utility industries are an example of an industry in which employee jobs are concentrated in a small number of local authorities, but these local authorities are dispersed across the country rather than being spatially concentrated.
There are also many industries where there is not spatial concentration into certain areas of the country because the industries act to serve local populations and as a result tend to be dispersed across the country relatively evenly. Retail is an example.
The article also discusses why spatial concentration occurs and why it matters:
If spatial concentration occurs in an industry, it reflects the fact that individual firms must see the benefits of such co-location as greater than the costs. This applies whether the spatial concentration is in the form of an industry cluster or an urban agglomeration.
The benefits of spatial concentration occur through increased productivity, due to a number of spillover effects usually labelled ‘agglomeration economies’. There are also costs associated with spatial concentration, particularly costs of congestion and increased land prices.
Because of this balance of benefits and costs, it is not always clear what the optimal level of spatial concentration is, either within a specific industry or over the whole economy. The balance will vary by industry and location.
To an individual spatial concentration matters because the spatial pattern of employment across industries influences the economic opportunities available in the area in which they live.
To a policy maker, spatial concentration is of interest because of the positive productivity benefits to the economy that spatial concentration can produce. There may also be an interest in any impacts on spatial inequalities across the economy.
The article also includes data and tools that allow the user to more fully investigate the spatial concentrations for different industries. These include:
An excel tool (1.93 Mb Excel sheet) that allows the user, for each of 257 industries, to obtain a list of those local authorities, and regions, which have the highest shares of their overall employee jobs within that industry (via calculated location quotients).
Google maps that map the location quotients for all the industries with more than 25,000 employee jobs in Great Britain.
Reference data tables with calculated Gini coefficients and Moran’s I indices (226.5 Kb Excel sheet) for each industry and location quotients (1.65 Mb Excel sheet) for each industry, local authority and region.
A complementary article, 'Industrial Specialisation in Local Areas', has also been published. It analyses data on an area by area basis to investigate each area’s industrial specialisations and it also includes a number of further tools to allow easy examination of the data.
There are many industries in which employee jobs are not evenly spread across the country. Instead, they are concentrated in just certain areas1. This article focuses on such industries and investigates the different spatial patterns that occur.
For example, the local authorities that specialise in a particular industry may be located disparately, in an apparently random pattern across the country; or at the other extreme they may be located together in adjacent local authorities forming a spatial cluster of activity for that industry.
The results sections of this article examine these possibilities and provide examples. The discussion starts in this section and the next with those industries that tend to cluster together spatially, before moving on to industries that show a mixed picture and ending with those industries that tend to be more randomly spread across the country.
This section focuses on a number of industries that are mostly concentrated in one single large cluster of neighbouring local authorities. Within Great Britain this is particularly observed in the case of some of the knowledge intensive service industries such as information and communication and professional, scientific and technical activities. These industries tend to display a spatial pattern illustrating a clustering of activity in the Greater South East area of England.
Note, all the results sections include analysis of location quotients. Therefore, it is helpful to understand what location quotients are. Briefly, the location quotients compare for each geographic unit, its share of employee jobs in a specific industry with its local share of total employee jobs.
A location quotient of 1.0 indicates that the local share of employee jobs in an industry is equal to the local share of total employee jobs. A location quotient greater than 1.0 therefore indicates a relative concentration of the industry in the geographic area. More detail is available in the 'Data and Methodology' and 'Interpreting the results' sections.
The data for the information and communication sector shows that most of the local authorities (23 out of 27) with high shares of their employees in information and communication activities are clustered in London and in the South East. These two regions combined account for more than 50 per cent of the employee jobs in the industry, compared with their combined share of national employee jobs across all industries of 30 per cent.
This is not a sector that only one or two local authorities dominate. Rather, employment is concentrated in a wider area around London and the area to the West of London. This can be seen in Map 1.
The information and communication sector is Section J of the 2007 UK Standard Industrial Classification. Divisions within this sector include media activities, such as publishing and television, as well as computer programming and consultancy; each of which shows a similar tendency to cluster within or close to London. The telecommunications sector, however, is less tied to the South East, with the North East region also having a relatively high share of employment in this industry.
The data for the information and communication sector overall, together with the different divisions (two digit SIC codes) and groups (three digit SIC codes) that make up the sector, can be examined in detail via the Google maps and also the concentration tool (1.93 Mb Excel sheet) which can show which regions and which local authorities have a relatively high share of their total employee jobs within the industry being examined (see the 'Interpreting the results' section for more detail).
Employee jobs in professional, scientific and technical activities are also mainly concentrated in one big cluster of local authorities. The analysis of location quotients at local level show most of the local authorities with proportionally high shares of employee jobs in the industry are located in London and in the South East, as shown in map 2.
The analysis of the location quotients at regional level confirm that only London and the South East have a more than proportional share of employee jobs in professional, scientific and technical activities with London having a location quotient of 1.7 and the South East region of 1.1. All other regions and countries in Great Britain have a location quotient below 1.0 with Wales having the lowest at 0.6.
This pattern is confirmed further by looking at the location quotient for local authorities. Table 1 shows the ten local authorities in Great Britain with the highest location quotient in this sector. Six of them are located within London with the other four in either the South East or the East of England regions. Expanding this list to the top thirty local authorities shows that 26 of the 30 are located in these three regions. The exceptions are Aberdeen City in Scotland, Manchester and Cheshire East in the North West of England and Stratford-on-Avon in the West Midlands.
|Region||Local authority||Location quotient|
|London||City of London||3.3|
|South East||South Oxfordshire||2.7|
|South East||Bracknell Forest||2.3|
|East of England||South Cambridgeshire||2.2|
|South East||Mole Valley||2.2|
|London||Richmond upon Thames||2.2|
A value greater than 1.0 means that the local authority share of employee jobs within professional, scientific and technical activities is greater than the local share of total employee jobs.
The table does not include any local authorities for which the number of employee jobs in this industry, derived from the Business Register and Employment Survey (BRES), is disclosive.
The professional, scientific and technical activities sector is Section M of the 2007 UK Standard Industrial Classification. It includes activities that require a high degree of training, and make specialised knowledge and skills available to users. Amongst the industries included within this sector are legal and accounting activities, management consultancy, architectural and engineering activities, advertising and the activities of head offices. Details of these can be found in the accompanying maps and tables.
Results presented in the previous article ‘The geographical concentration of industries’ were used as a starting point for this analysis. The previous article highlighted those industries in which employment is concentrated.
In contrast with the previous groupings, in which industries that showed a high degree of geographical concentration were spatially concentrated across a relatively wide area of the country, there are some cases in which the high concentration of employee jobs is concentrated within only a few neighbouring local authorities, forming small clusters of very high concentration of employee jobs. This is the case, for example, with the finance sector, which is strongly concentrated in London, and the mining sector, which is strongly concentrated around Aberdeen.
The majority of local authorities have a relatively small share of employee jobs in finance and insurance activities, compared with their local share of national employee jobs. However, four local authorities clustered in London (City of London, Tower Hamlets, Islington and Westminster) together account for more than 25 per cent of Great Britain’s employee jobs in the industry, which is five times as much as their local share of Great Britain’s overall employee jobs (5 per cent).
Aside from this high concentration in Central London, the finance and insurance sector is also highly represented in a few other local authorities spread across the country. Table 2 shows local authorities with the highest location quotients for the divisions that make up the finance and insurance sector.
It should be noted that this table, and others showing location quotients, are not showing the largest employers in aggregate terms. Rather, they show those local authorities that have the highest share of employee jobs in the specified industry relative to their local share of total GB employee jobs.
The finance and insurance sector is Section K of the 2007 UK Standard Industrial Classification. It is composed of financial service activities, which are often concentrated in major urban centres or in local authorities nearby; activities auxiliary to financial services, which are more concentrated in London; and insurance activities which display a lesser tendency to spatially cluster within the South East region of England. This can be seen, for example, by the fact that Scotland and Wales both have location quotients in the insurance activities division above 1.2.
|Region||Local Authority||Location quotients|
|London||City of London||10.6|
|South East||Reigate and Banstead||3.2|
|Scotland||Edinburgh, City of||3.0|
|Yorkshire and The Humber||Calderdale||2.9|
|East of England||Ipswich||2.3|
|South West||Bristol, City of||2.2|
|Yorkshire and The Humber||Craven||2.2|
|East of England||Brentwood||2.1|
A value greater than 1.0 means that the local authority share of employee jobs within finance and insurance is greater than the local share of total employee jobs.
The table does not include local authorities for which the number of employee jobs derived from the Business Register and Employment Survey (BRES) is disclosive.
The mining and quarrying industry also tends to be highly concentrated in a relatively small number of local authorities. Employee jobs in mining and quarrying are concentrated in a relatively small number of big firms, which constrains employment to a relatively small number of areas. The location of those firms usually depends on the location of the natural resources.
Scotland shows the highest concentration of employee jobs in mining and quarrying. Those employee jobs are mainly concentrated in two neighbouring local authorities, Aberdeen City and Aberdeenshire. These two local authorities, who specialise in the oil industry, account for almost half of the employee jobs in the GB mining and quarrying industry (compared with a combined local share of national employee jobs of 1 per cent).
The remaining employee jobs in the sector are spread around a number of other locations across Great Britain with analysis of the location quotients showing that there are local authorities with relatively high shares of employee jobs in the industry in all regions and countries of Great Britain, except in London.
The previous section dealt with industries where there are high levels of spatial clustering of employee jobs. In addition, the patterns tended to be similar for both the aggregated sector being examined and also the subsectors (divisions and groups) that make up the sector. However, this is not always the case.
There are some sectors in which the sub-industries within it display very different spatial behaviour to the sector itself or else display a wide variety of different spatial patterns. This section of the article deals with such industries, concentrating particularly on the manufacturing sector and the transport and communications sector.
Figure 1 displays location quotients for the manufacturing sector as a whole. It shows that the East Midlands and the West Midlands are the regions with the highest share of manufacturing employee jobs relative to their share of GB employee jobs. Manufacturing employee jobs are also highly represented in Wales and in the northern regions of England but virtually absent from London.
Analysis of location quotients at local authority level shows that there are 38 local authorities with a location quotient for manufacturing above 2.0, of which eleven are located in the East Midlands, nine in the North West and six in Wales. Three local authorities, Pendle, Flintshire and Fylde have a location quotient for manufacturing above 3.0.
The aggregated manufacturing sector disaggregates into 23 divisions (two-digit SIC codes) and when the data is examined at this more disaggregated level, the analysis of the spatial patterns shows that the various two-digit code industries within manufacturing can have quite different spatial patterns of employment.
In some industries employment is highly concentrated in a small number of local authorities, but these are not spatially concentrated; rather, they are spread across the country. A similar pattern to that of utility industries. This is the case for the manufacture of basic pharmaceutical products and pharmaceutical preparations, manufacture of basic metals, manufacture of motor vehicles, trailers and semi-trailers and manufacture of other transport equipment.
It is worth noticing that these industries typically show very high levels of industrial concentration1. This means that employee jobs are concentrated in some local authorities due to the presence of one or a few very big firms. However, evidence shows that in these cases there is no spatial autocorrelation between local authorities in which the industry has high location quotients; in other words the large firms in these industries do not tend to locate close to each other.
Most manufacturing industries are, by contrast, spatially concentrated. Manufacture of textiles; manufacture of wood and of products of wood and cork, except furniture; and manufacture of food products show the highest tendency for local authorities with high location quotients in the industry to be clustered. As an example, Map 3 shows the spatial concentration of high and low location quotients in the manufacture of wood and wood products industry.
Another industry that displays some clustering of local authorities with high location quotients in the industry is the manufacture of computer, electronic and optical products, as shown in map 4. This is unique amongst the 36 (two digit SIC) manufacturing divisions in that its employee jobs are predominantly located in the South East of England. For most other manufacturing industries, this part of Great Britain tends to have relatively low employee job levels.
Other industries also showing a relatively high tendency for spatial concentration of local authorities with high location quotients in the manufacturing sector include manufacture of furniture, manufacture of paper and paper products, manufacture of chemicals and chemical products, manufacture of rubber and plastic products, manufacture of other non-metallic mineral products, and manufacture of machinery and equipment.
Similarly to the manufacturing industry, the transport and storage industry, as defined in the Standard Industrial Classification 2007, include several industries that show not only different levels of geographical concentration but also different patterns of spatial distribution.
At the more aggregated industry level, the transport and storage sector does not display much spatial clustering of employee jobs. However, the degree of concentration and the patterns of spatial distribution of the various activities within the transport and storage industry differ greatly.
Within the transport industries, land transport activities account for the largest number of employee jobs. Employee jobs in this industry are relatively spread across many local authorities, i.e., the degree of concentration in each local authority is relatively low. However, as shown in map 5, the concentration often tends to sprawl over neighbouring local authorities.
Employee jobs in either the air or water transport industries, on the contrary, are highly concentrated in a relatively small number of local authorities. These local authorities are not clustered together; rather they are dispersed across various different locations of the country.
Warehousing and support activities for transportation, meanwhile, are mainly concentrated in a large number of local authorities located close to each other. As shown in map 6, warehousing and support activities for transportation are mainly based in central area of England, particularly within the regions of Yorkshire and The Humber, the West Midlands and the East Midlands. There is also a clustering of activity to the East of London.
For more information on industrial concentration see the article ‘The geographical concentration of industries’.
This article focuses on those sectors where there is a high level of geographical concentration such that employee jobs are concentrated in just a small number of local authorities. Some of these industries do not display any spatial clustering. In other words, the local authorities with high location quotients in the industry tend to be dispersed across the country. Air transport was one example discussed in the previous section. Another particular example of this are utility industries.
In the electricity sector1, there is a relatively high number of local authorities with high shares of employee jobs in the industry compared with their local share of national employee jobs, i.e. with a location quotient value greater than 2. Evidence shows, however, that those local authorities tend to be randomly distributed across the country.
In 12 per cent of the local authorities in the country, the share of employee jobs in the electricity industry is at least twice as much as their local share of national employee jobs; and it can be as high as thirteen times bigger. These local authorities combined account for 57 per cent of employee jobs in the electricity industry, while their share of national employee jobs is not greater than 14 per cent.
It is worth noticing at this point that, given the high level of industrial concentration2 of the electricity industry, the high levels of concentration of employee jobs in some local authorities are mainly due to the presence of large single plants or firms rather than to the clustering of several small or medium sized firms.
In terms of spatial distribution, the local authorities exhibiting high location quotients in the electricity industry are very much spread across all regions, as shown in Table 3.
|Region||Number of local authorities||Number of local authorities with location quotients > 2 in the electricity industry|
|Yorkshire and The Humber||21||2|
|East of England||47||3|
'Electricity sector' is used here to refer to section D : Electricity, gas, steam and air conditioning supply, of the 2007 UK Standard Industrial Classification.
The degree of industrial concentration, measured by the Herfindhal index was discussed in the previous article ‘The geographical concentration of industries’.
The preceding sections have focused on industries that display a high level of geographical concentration such that employee jobs are found to be concentrated in a relatively small number of local authorities. The different results sections have investigated how the spatial patterns of concentration can vary amongst this group of industries.
There is, however, another group of industries in which there is very little geographical concentration because employee jobs are spread relatively evenly across the country. These industries tend not to display spatial clustering across local authorities, they also tend not to display geographical concentration (i.e. employment is not concentrated in just a small number of local authorities).
Sectors such as retail or leisure, or public services such as health and education fall into this category. Table 4 gives an example of such an industry and shows that in the retail trade the location quotients for each of the regions and countries of Great Britain is very close to 1.0.
|Yorkshire and The Humber||1.0|
Even amongst these more dispersed industries1, however, it is sometimes possible to identify some patterns of spatial distribution of employee jobs. Although, in these industries, the degree of concentration in each local authority is relatively low, some industries do show some tendency for clustering.
This is more the case for industries that are providing services to businesses rather than directly to consumers. For example, map 7 shows the case of the administrative and support service activities sector. Real estate activities and other service activities also show low level spatial concentration.
The industries with relatively low Gini coefficients include agriculture; construction; wholesale and retail trade; accommodation and food service activities; administrative and support service activities; education; health and social work activities; arts, entertainment and recreation; and other service activities. For more detailed information please see the article 'The geographical concentration of industries'.
The previous article described why concentration occurs. It noted that it is the cumulative result of numerous individual location decisions by businesses and that each of these businesses must have viewed the benefits of such co-location as being greater than the costs. The article noted that such concentration is more likely to occur in certain industries.
This article has taken this further by examining the spatial pattern of concentration across different industries. It has shown that there is more than one spatial pattern to such concentration. Additionally, whilst the data has been examined on an industry by industry basis, it has also become clear that some areas of the country are home to concentrations of many different industries, more so than other parts of the country. (For more information on the local specialisations in any particular area, please consult the third paper in this series, 'Industrial Specialisations in Local Areas’).
The existence of spatial concentration is important to many different fields of academic study as well as to policy decisions made by government. For example, academic study on industrial location theory, geographical economics, trade, economic growth theories, urban economics etc are all concerned with the propensity of production to spatially concentrate and either the causes or effects of this.
Similarly, for policy makers, decisions around industrial clusters, urban agglomerations, and redistribution are all concerned with or influenced by the patterns of spatial concentration in the economy and the benefits or costs of such patterns.
Overall, the benefits arising to firms from geographical concentration can be loosely described as productivity benefits occurring through the three channels first described by Marshall in 1890. These are 1) the existence of specialised providers of industry inputs; 2) a thick (as in large) local market for specialised labour skills and 3) the existence of information spillovers. These benefits are often grouped under the term ‘agglomeration economies’.
These types of benefit can occur for both a single industry cluster and alternatively across different industries within a large urban agglomeration. Those that occur within a single industry concentration are known as ‘localization economies’. Here, knowledge spillovers are seen to be predominantly industry-specific and these act to encourage clusters of particular industries.
Knowledge may also spillover between complementary industries producing ‘urbanisation economies’. The exchange of complementary knowledge across diverse firms and economic agents facilitates search and experimentation in innovation. As such, the diversified local production structure leads to increasing returns such that the increasing size of the city leads to a reduction in individual firm’s costs via improved matching, sharing and learning.
As mentioned in the previous article, a particularly important aspect of concentration is that it is seen as having an impact on the amount of economic growth in the economy. There is ample evidence in the research literature to suggest productivity is (or can be) higher in areas where economic activity is geographically concentrated1. As such, there are economic efficiency gains for an economy from spatial concentration occurring2.
What is not always clear, however, is what the optimal level of spatial concentration is, either within an industry or over the economy overall. There is a balance between the economic benefits of spatial concentration, that can lead to a spatial convergence of firms and economic activity, and the additional costs such as increased land prices or congestion costs that can encourage a spatial divergence of firms and activity.
This balance will vary by industry and location. The situation is further complicated by the fact that many of the costs and benefits are externalities, i.e., they are not transmitted through the price mechanism and are incurred by a party who was not involved as either a buyer or seller of the goods or services causing the cost or benefit.
The terms cluster or agglomeration are often used to describe spatial concentrations of industrial activity. Definitions of clusters and of agglomeration can vary depending on the source being consulted. Sometimes they are used interchangeably to describe any spatial concentration of one or more industries in an area.
An alternative way of looking at the difference between a cluster and an agglomeration is to consider it as a matter of degree with a cluster describing a localised concentration of a single industry and an agglomeration describing an urban grouping of a much larger part of economic activity.
A cluster is usually thought of as an industry or sector level concept. As such, visible evidence of a cluster would be from evidence of spatial concentration of an industry or sector. It is this evidence of spatial concentration in individual industries which this article has sought to illustrate. However, within this paper the smallest geographical unit that has been used is local authority level. Some clusters will be more localised than this.
This definition of clusters can be slightly expanded to also include the spatial concentration of a small number of related industries. These related industries may be part of a horizontal cluster where there is co-location of industries producing similar products, or they may be part of a vertical cluster, where firms involved in various stages of the supply chain for an end-product co-locate within an area.
Agglomeration meanwhile can be seen as the localised clustering of a much larger part of economic activity. In other words, it occurs where many different industries are seen clustering in the same location. An agglomeration will therefore usually be describing an urban area, which is the most likely to see such patterns of activity.
This article has shown a snapshot of the current picture in terms of the spatial concentration of individual industries. As such it, and the accompanying articles, can help provide a starting point for looking into these areas. A fuller examination of any particular subject would require further work in addition. For example, the data in this article can be used to show that a certain area has an apparent clustering of employment in a certain industry.
However, there is much potential detail beneath this such as what factors have influenced the creation of that cluster, or whether there are co-linkages with other industries that can’t be answered from this data alone. Similarly, other extensions to this analysis would be to investigate smaller geographical areas or alternatively to investigate different industrial sectors that are cross-cutting across the Standard Industrial Classification such as ‘green industries’ or ‘creative industries’.
Spatial concentration can work as a channel for increased standards of living. There is evidence that productivity of individual firms rise with the overall amount of activity in other nearby firms, or with the number of nearby workers and consumers. High productive firms are able to pay higher wages. So places where high productive firms are located will tend to see earnings and income grow.
It should be noted that to achieve economic growth across the economy overall, there ideally needs to be productivity growth occurring across all industries, whether or not they spatially concentrate. The fact that some industries achieve higher productivity levels if they spatially concentrate can be relevant to achieving productivity growth in the economy, but it is only one of many differing factors within the economy that influence productivity levels and productivity growth rates.
In this analysis we have used the number of employees derived from the Business Register and Employment Survey (BRES) from 2011. This means that self-employed jobs, HM Forces and Government supported trainees are not included in the data.
The term ‘employee job’ is used throughout this article. It is used to help stress that all the data in this article is focused on the workplace of an employee. This contrasts to some data sources that focus on the residential status of an employee. However, it should be noted that strictly speaking the BRES data being used in this article is not a count of the number of jobs filled by employees in an area, but rather a count of the number of employees who work in an area.
BRES data is used because it is the primary source of employee estimates at a detailed geographical and industrial level . BRES contains information on the numbers of employees by industry down to five-digit industrial classification as defined in the UK Standard Industrial Classification and by geographic area down to local authority.
It is important to note that estimates are subject to sampling error, which increases as geographic areas become smaller and industry classification become more detailed. To limit this issue, four and five digit SIC industries are not studied in this article which instead focuses on analysing to local authority level data and to the 272 industry split, i.e., the three-digit code industry level from the 2007 UK Standard Industrial Classification.
The Standard Industrial Classification (SIC) provides a framework for the collection, tabulation, presentation and analysis of data, and its use promotes uniformity. The UK SIC is a hierarchical five-digit system. UK SIC (2007) is divided into 21 sections, each denoted by a single letter from A to U. The letters of the sections can be uniquely defined by the next breakdown, the divisions (denoted by two digits). The divisions are then broken down into groups (three digits), then into classes (four digits) and, in several cases, again into subclasses (five digits). There are 21 sections, 88 divisions, 272 groups, 615 classes and 191 subclasses. In this article, results are shown for the industry sections, divisions and groups only. The full SIC breakdown can be found in the UK SIC structure and explanatory notes document.
The methodology followed in this analysis is based on three measures:
The locational Gini that indicates which industries are geographically concentrated.
The Moran’s I statistic that determines the degree of spatial autocorrelation.
The location quotients, which are a local measure of concentration.
The locational Gini was presented in the previous article 'The geographical concentration of industries' and its results are used as the starting point of this analysis. Since the first article was published more recent data on employee jobs has became available. We have therefore recalculated the locational Gini using the most recent data derived from the Business Register and Employment Survey (BRES).
The Moran’s I statistic is a global measure that complements the locational Gini by proving a means to assess whether local authorities with high concentration of a specific industry are clustered in space or are randomly dispersed over the country.
The location quotients are a local measure of concentration that allow identifying in which local authorities a specific industry is highly represented by comparing the local share of employment in each industry with the local share of national employment. When represented in a map, location quotients provide a useful visual tool of the distribution of industries across the territory.
The location quotients are a simple and very common measure used to assess both geographical concentration of industries and industrial specialisation of regions. Used in the analysis of geographical concentration, the location quotients compare for each geographic unit, the share of industry-specific employee jobs with the share of total employee jobs. The location quotients expression for geographical concentration is given by:
where Eij is the number of employee jobs in industry i in region j, Ei is the total employee jobs in industry I, Ej is total employee jobs in region j and E is the national total of employee jobs. A location quotient of 1 indicates that the local share of employee jobs in an industry is equal to the local share of total employee jobs. If the local share of employee jobs in industry i (Eij/Ei) is greater than the local share of total employee jobs (Ej/E), than the location quotient takes values greater than 1, indicating a relative concentration of industry i in region j.
For detail on how to interpret the location quotient results, please see the next section 'Interpreting the results: how to use the data'.
The second step was to compute a global measure that indicates whether local authorities with high concentration of a specific industry are clustered in space. The Moran’s I statistic uses the location quotients computed for each local authority to measure the level of spatial autocorrelation. The test statistic for industry i is given by:
where k and j are indices for local authorities, Ux is the mean of X, Xk is the location quotient for industry i in region k and Wkj is one element of the weights matrix W, which indicates whether area (local authority) k is spatially connected to area (local authority) j. Under the null hypothesis of no spatial correlation, the expected value of Moran’s I statistics is E(I) = -1/(n-1). Values of I greater than the expected value indicate a positive spatial autocorrelation in the distribution of industry employment. Values range from -1, indicating perfect dispersion to 1, indicating perfect correlation. A zero value indicates a random spatial pattern.
It should be noted that Moran’s I statistic measures spatial clustering of areas with similar location quotients rather than clustering of firms. It is therefore very important to bear in mind when interpreting the results which geographic unit is being used in the computation of the statistic. In this analysis, we have chosen to use local authority as the main geographic unit.
Moran’s I statistic takes high values for industries in which employee jobs are concentrated in several neighbouring local authorities. This statistic fails, however, to identify high levels of spatial clustering if employment in a specific industry is highly concentrated in a single local authority or in a small number of local authorities that do not share a common geographic border. However, in those cases, we can identify the concentration of activity from the fact that there will be a small number of local authorities with very high location quotients.
The results presented in this article focus mainly on the 21 industry split, the highest industry aggregation level of the UK Standard Industrial Classification (SIC 2007). However, results of the analysis for all industries down to three-digit code level are available in the reference tables and maps accompanying this article. This section aims to provide guidance for users wishing to examine in more detail the full underlying results from the available data.
All measures were computed using employee jobs estimates at local authority level. For further information on the data used and the measures presented please see the sections ‘Data & Methodology’.
The ‘ Concentration indices table (226.5 Kb Excel sheet) ’ provides the values of the locational Gini and Moran’s I statistic for all industries down to three-digit code of the UK Standard Industrial Classification (SIC 2007).
The value of the locational Gini indicates the degree of geographical concentration of employee jobs in a specific industry. If an industry is highly concentrated in a small number of local authorities, the locational Gini will take on a high value; whereas, if employee jobs in an industry are spread relatively evenly across many local authorities, the index will take on a low value.
The first article of this series provided a detailed analysis of the results of the locational Gini (as well as of other global measures of concentration) and guidance on how to interpret the results. A key point is that there are a couple of issues related to interpreting the results that need to be taken into consideration when making comparisons across industries.
Firstly, the more disaggregated is the industry level, the greater the values of the indicator tend to be. This happens because these measures, when calculated for more aggregated industry levels, do not entail any information regarding the distribution patterns of sub-industries within it.
Secondly, the locational Gini does not correct for the size distribution of the industries, therefore it does not distinguish between industries that are highly concentrated because employee jobs are concentrated in a small number of very large plants and industries that are highly concentrated as a cumulative results of many firms’ decisions to locate in the same areas. This issue was, however, covered in the preceeding article so for more detailed information please refer to the sections ‘Methodology’ and ‘Interpreting the results’ of the article ‘The geographical concentration of industries’.
The Moran’s I statistic is used as a complement of the locational Gini giving information on the pattern of spatial distribution of employee jobs. From the Gini we can observe when employee jobs are concentrated in a relatively small number of local authorities. What the Moran’s I statistic tells is whether these particular local authorities tend to be located adjacent to each other or alternatively whether they are scattered across the country in an apparently random way.
In the context of this article and analysis the Moran’s I statistic is only really of interest in the cases where there is a high degree of geographical concentration as shown by the Gini coefficient. If the degree of concentration of an industry is very low then it means that the employee jobs are spread across many local authorities. The chances of witnessing spatial clustering are therefore relatively low and the Moran’s I statistic is not of much importance in such a case.
It is very important to keep in mind what the Moran’s I statistic measures and what geographic units are used in its computation. In this analysis the Moran’s statistic was computed using the location quotients at local authority level.
Therefore, if an industry has a high Gini value and a high Moran’s I statistic it means that employee jobs are typically concentrated in a relatively small number of neighbouring local authorities. The higher the Moran’s I statistic the higher the spatial autocorrelation, so the greater the tendency for clustering of local authorities with high (or low) levels of concentration of the industry.
If, on the other hand, the industry has a high Gini but a low Moran’s statistic, this means that the employment in that industry is concentrated in a relatively small number of local authorities that do not share a common border. It does not mean however that the industry is less concentrated. On the contrary, employee jobs in these cases can be highly concentrated within the borders of one or a very small number of disparately located local authorities.
In terms of geographic units, therefore, the key point is that the smallest geographic unit used in this article and analysis was local authority level. A possible extension to this article would be to utilise a smaller unit of geography such that it would be possible to do additional analysis looking in finer detail at the existence or otherwise of spatial clustering within a local authority, rather than just between them.
The analysis of the global measures of concentration is followed by the analysis of local measures which allow identification of in how many, and in which, local authorities a specific industry is concentrated.
The location quotient table (1.65 Mb Excel sheet) and the concentration tool (1.93 Mb Excel sheet) accompanying this article shows the values of the location quotients of all local authorities and regions of Great Britain for all the industries down to three-digit code. The location quotients are local measures of concentration and therefore indicate in which particular local authorities there is a relatively high share of employee jobs to be found within a specific industry.
An industry is considered to have a high share of employee jobs in a specific area if that area has a higher share of employee jobs in that industry than its local share of national employee jobs. For instance, examine Table 5. This shows how Tower Hamlets is responsible for 7.1 per cent of all employee jobs in the finance and insurance industry in Great Britain but only 0.9 per cent of total (all industry) employee jobs in Great Britain. As a result, Tower Hamlets has a location quotient for financial and insurance activities of (7.1/0.9 = 8.1).
When interpreting the location quotients a key result is an LQ of 1.0. This shows that the local share of employee jobs in an industry is equal to the local share of total GB employee jobs. If the local share of employee jobs in a specific industry is greater than the local share of total employee jobs, than the location quotient takes values greater than 1, indicating a relative concentration of the industry in the local area.
Note, there has to be some judgment on how large a location quotient needs to be to represent concentration. In this article, a cut-off value of 2 for local authorities and a cut value of 1.2 for regions has been used. In conducting analysis of location quotients results it may be helpful to calculate some descriptive statistics, such as the smallest and largest location quotient values, the average, percentiles and counts of local authorities with location quotients in the intervals 0 to 0.5, 0.5 to 0.8, 0.8 to 1.2, 1.2 to 2 and above 2.
Analysis of such measures complements the information provided by the global measures of concentration by providing a means to compare the amplitude of the relative concentration in each local authority and the geographic location of the areas with highest levels of concentration as well.
|Region||Local authority||Share industry employment||Share national employment||Location quotient|
|London||City of London||14.8||1.4||10.7|
|South East||Reigate and Banstead||0.7||0.2||3.2|
|Scotland||Edinburgh, City of||3.3||1.1||3.0|
|Yorkshire and The Humber||Calderdale||1.0||0.3||2.9|
|East of England||Ipswich||0.6||0.2||2.3|
|South West||Bristol, City of||2.0||0.9||2.3|
|Yorkshire and The Humber||Craven||0.2||0.1||2.2|
|East of England||Brentwood||0.2||0.1||2.1|
In this article, an industry is considered to be highly represented in a local authority if the location quotient is 2.0 or above.
The table does not include local authorities for which the number of employee jobs derived from the Business Register and Employment Survey (BRES) is disclosive.
The location quotient table (1.65 Mb Excel sheet) provides a full set of location quotient data for all local authorities of Great Britain and for the industry sections, divisions and groups of the Standard Industrial Classification. The user can investigate which local authorities of any particular region have a high location quotient by expanding the group of cells collapsed under the ‘London’ or ‘North East’ column etc. The user can also investigate specific industries of this sector in a region and in its local authorities by selecting to show the two- and the three digit code industries.
To further help examine the location quotients by industry, a concentration tool (1.93 Mb Excel sheet) is provided. This tool allows the user to select the industry of interest and see all regions and local authorities ordered by the value of their location quotient. This will enable the user to easily see in which regions and local authorities the industry of interest is highly overrepresented and underrepresented.
To fully understand the use and limits of a location quotients analysis there are a few points to bear in mind. Firstly, the size of location quotients can vary greatly depending on the level of aggregation. Analysing location quotients at large geographical or industry levels of aggregation may hide important information that would become apparent based on analysis of a more disaggregated dataset.
In other words, there will be a wider variety of location quotient results when investigating disaggregated three-digit industries at a local authority level in comparison to investigating aggregated industry sectors at a regional level.
Secondly, it should be noted that they are a relative measure. If an area has a high location quotient in one industry, this must be offset by a less than proportional share of employee jobs in other industries. In other words, the location quotients do not help to inform as to whether an area has a very high level of absolute employee jobs (relative to its population).
As such a location quotient may not be the correct indicator to, for example, determine if an area has too high or too low an amount of health or education employee jobs. All it informs on, is the relative share of the local area’s employee jobs that are associated with a particular industry, and whether this relative share is higher or lower than elsewhere.
Similarly, it should be noted that a high location quotient does not necessarily equate to a high absolute level of employee jobs. If the industry overall is small in Great Britain then it is possible for an area to have a relatively high share of these employee jobs (i.e. a high location quotient) but still a low number of absolute employee jobs given the industries small overall size.
Similarly, a location quotient of 1.0 for an area in an industry that has a high level of absolute employee jobs in Great Britain would mean quite a high level of employee jobs in this industry in this area despite its middling location quotient.
For example in Table 5, Manchester has a higher share of employee jobs in the finance and insurance sector (2.4) than Reigate and Banstead (0.7), but Reigate and Banstead has a higher location quotient.
The location quotient of Reigate and Banstead is higher because its share of total GB employee jobs (0.2) is smaller than that of Manchester (1.2); i.e, whilst in Reigate and Banstead the share of employee jobs in finance and insurance is three times as much as the its share of total GB employee jobs and in Manchester it is only twice as much.
To help visualise the spatial distribution of employee jobs across the country, this article is also accompanied by maps showing the location quotients at local authority level for all industries down to three-digit code with more than 25,000 employee jobs. Users can choose to view only one area and they can embed the current view – the chosen area and industry – in a website. The specific selection can also be shared with other users by email, twitter and facebook.
Is it important to keep in mind when looking at data on maps that areas that are large geographically have a greater visual weight on the map, although this often does not correspond to that area having a large population or number of employee jobs.
Some local authorities are geographically small and difficult to observe on a map of Great Britain unless the user zooms in and yet they account for a high share of national employee jobs (as it is the case of Westminster, Camden, City of London, Tower Hamlets) Other local authorities are much larger in terms of area and highly visible on the map but account for a much smaller share of employee jobs (as it is the case of Scottish Borders, Powys, Dumfries & Galloway).
The author is grateful to Robert Fry for his help and for the mapping tool accompanying the article.
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: firstname.lastname@example.org