The Services Turnover Survey (STS) is a survey of business-to-business turnover by service product, conducted every five years to support the reweighing of the Services Producer Price Index (SPPI). An updated version of the STS has been carried out to collect turnover for 2010. This is the first time results from the STS have been published and all data associated with this article are to be treated as experimental statistics. This article summarises the approach used to collect and produce 2010 STS estimates and publishes the detailed turnover estimates in an accompanying reference table.
The Services Turnover Survey (STS) is a survey of business-to-business turnover at a detailed service level, conducted by the Office for National Statistics (ONS) every five years to support the re-weighting of the Services Producer Price Index (SPPI).
The SPPI, formerly known as the Corporate Services Price Index (CSPI), was renamed in 2006 for consistency with the more established Producer Prices Indices (PPIs). In turn, the Corporate Services Prices Index Survey into Turnover by Type of Service was renamed as the STS during development in 2010.
The STS is primarily conducted to collect a detailed level of turnover data required to support the SPPI. This year, aggregate estimates from the STS are also being published in their own right.
The STS methodology for deriving a dataset suitable for re-weighting purposes (and publication) has been improved upon, specifically estimation, imputation and disclosure rules.
Improvements have been made to the STS sample design, with the survey having a sample size of 8,000 business across 41 different service sector industries.
The results of the STS are published as experimental statistics and subject to various caveats. The detailed 2010 STS estimates can be accessed via the accompanying reference table (384 Kb Excel sheet) that is published alongside this article.
The SPPI is an index compiled quarterly to measure the changes in price of a range of services provided by UK businesses to other UK businesses. Between 2006 (the last time the STS was conducted) and 2011 the SPPI continued to expand its service sector coverage, and was accredited as a National Statistic in 2011.
Further details on the SPPI can be found on the ONS website (154.1 Kb Pdf) .
The historical purpose of the STS has been to assist with the quinquennial re-weighting of the SPPI by collecting estimates of business-to-business turnover for those service sectors covered by the SPPI, broken down into individual detailed services referred to as 'service products'. These service products are the building blocks of the SPPI and are the lowest level for which price data is collected. The service products are obtained by arranging the activities of a service industry into a hierarchal structure of product groups, sub-products and items. Each item is assigned an index number and these indices are given a percentage weight calculated from the turnover information obtained by the STS.
Full details on this weighting process can be found on the ONS website (260.4 Kb Pdf) .
The percentage weight of a product represents its market share within that industry, so while the SPPI collects the price charged for each service the STS collects the total turnover received for each of these services during a base reference year, in this case 2010. These low level indices form the basis that all upper-level indices are derived from within the SPPI, so the STS plays a vital part in supporting the SPPI and its continued evolution.
No other ONS survey collects turnover at this detailed level for the service sector, so the estimates obtained via the STS are unique and therefore it has been decided that the results produced should be published. Due to the expansion of the SPPI in recent years the STS has, in turn, needed to expand its scope and methodology to meet the quality demands of the SPPI, and it is for these reasons that publication was initially considered. With a sample size of just over 8,000 businesses across 41 different service industries, the STS was now deemed to be of sufficient size, scope and interest to justify the publication of top level results. To an extent, the STS is constrained by those service sector industries covered by the SPPI, however, a number of additional industries have been included in the 2010 STS that fall outside the scope of SPPI. These additional industries represent those service industries where the SPPI may, in future years, expand its coverage and begin collecting price data so a detailed source of service product turnover has been collected in anticipation. It is to be noted that neither the SPPI nor the STS cover the whole service sector.
To collect the STS data, a separate questionnaire was created for each of the 41 industries, due to the industry specific nature of the business-to-business turnover questions asked. As over half of the industries were represented in 2006 then this mainly involved updating the design of the questionnaire to meet current ONS standards, however, further work was carried out to determine whether the detailed service product questions being asked were still relevant to their industry. Typically, when attempting to define the set of service products for inclusion in the SPPI or STS, a classification system is used as a starting point. The classification system used by ONS is either the 2007 Standard Industrial Classification (SIC2007) or the 2008 Classification of Products by Activity (CPA2008). A draft questionnaire is developed using this initial classification and the questionnaire is then developed further with businesses to determine if the classification structure is relevant. This also allows ONS to further improve the level of detail so that the final industry classification system is broken down into a detailed set of 'service products' that are representative of the industry in question. During this phase of development, it is also important to consider any difficulties which businesses may have in providing the required turnover at this detailed level. Therefore, in some cases there needs to be a balance between the level of detail for which data is requested and the availability of turnover from businesses. Each questionnaire is set out in a similar format to collect the required data as follows:
the first section of the questionnaire asks for turnover generated from business-to-business services to UK businesses only, broken down by the detailed service products as defined above,
the second section of the questionnaire asks for turnover generated from business-to-business services to UK businesses only, for services not included above,
the third section of the questionnaire asks for all other turnover generated in the reference period, not included above (so this could include turnover generated from services offered to the household sector, exports, manufactured goods etc.),
the final section asks for total turnover in the reference period (effectively this is a sum of the above sections).
As the STS has previously been used only for internal purposes, the survey itself inherently contains several limitations. Most prevalent of these is that the survey is principally a tool for re-weighting the SPPI, so the industry breakdowns are to an extent restricted to matching what is required for the SPPI. Also, all those businesses providing data to the SPPI (approximately 3,000) were automatically included in the STS sample, whereas normally a unique random sample would be created for such a survey. With 8,000 businesses selected the sample was far greater than in 2006, but meant that on average, with over 40 industries to be covered, under 200 businesses were selected per industry. Although industries were proportionally represented according to their size based on total employment as held on the Interdepartmental Business Register (IDBR), the individual industry samples were fairly low and, while this is sufficient for the purposes of SPPI re-weighting, a published survey of this type would ideally have a much larger sample to produce more accurate estimates of turnover.
As already mentioned, it was decided that the results from the 2010 STS would be published as experimental statistics alongside their main use for re-weighting the SPPI. Therefore, to ensure the quality of the estimates produced were acceptable for publication the methodology that underpinned previous versions of the STS was reviewed and improved. The 2010 STS sample consisted of approximately 8,000 businesses, over a quarter of which were automatically included as a result of their current participation in the SPPI. This meant that the design had to take into account the SPPI sample when producing an overall sample where each industry was accurately represented according to their universe size (where universe represents all those businesses in the industry which fall within scope of the survey). The sample was stratified by industry and size of business, as defined by employment. A number of exclusions were also made, to keep the survey in line with previous iterations and/or to match that used by the SPPI. For example, partnerships/sole proprietors and local authorities/Government agencies were excluded, as were businesses in Northern Ireland.
To arrive at a suitable estimation method for the STS that allowed aggregated estimates of turnover for each industry to be published, research was carried out on the properties of various estimators including a variant of the estimator used on the PRODCOM survey (140.2 Kb Pdf) . The research indicated that for this particular application the standard Horvitz-Thompson estimator generally produced estimates with smaller standard errors. In simple terms, the Horvitz-Thompson estimator calculates for each industry/size stratum the average turnover of the sampled businesses within that stratum. The estimated total turnover for that stratum is then calculated as the product of this average and the total number of businesses that meet the industry/size condition. The final estimate is calculated by summing the estimated total turnover for each stratum.
Publishable estimates were required for each of the questions asked in the STS, in order to present turnover data which would represent each industry based on the data provided by businesses within each industry sample. For each estimate a Coefficient of Variation(CV) was also calculated and is published alongside the turnover estimate. The CV is the ratio between the standard error of the estimate and the estimate itself. It is usually expressed as a percentage and can be interpreted as a measure of sampling error. If another sample were taken, due to the random nature of sampling, it is likely that a different estimate would be produced. If this process were repeated then approximately two thirds of the estimates would be within the CV percentage of the estimate. For instance if an estimate was 200,000 and the CV was 10 per cent then roughly two thirds of the estimates would be within 10% of the estimate (that is, between 180,000 and 220,000). Another way of expressing this uncertainty about the estimate is that we can be approximately 66 per cent confident that the true estimate lies between 180,000 and 220,000.
The 2010 STS was conducted during the period July to November 2011 and finished with an overall response rate of 83.9 per cent. Of this response, the STS had a clearance rate of 96.2 per cent (where the clearance rate represents data returned that has successfully passed the validation rules defined for the survey). Both of these rates exceeded the internal quality requirements of the survey agreed during the planning stage. Overall, 36 of the 41 industries included in the STS finished with a response rate of above 80 per cent. The validation rules defined for the STS were the same for each industry, in order to provide ease of use in system development. Where these predefined rules were not met by data being returned, the business would be contacted to further quality assure the turnover provided and to request corrected data where an obvious error had been identified.
Clearly the definition of a service can be very subjective, in particular due to the wide ranging and fast moving nature of the service sector.
See the article on the ONS website (154.1 Kb Pdf)
This in turn led to some difficulties in a number of industries regarding how turnover for specific services is generated and how it should be reported for STS. Two industries where the majority of issues were raised were recycling and publishing services. With recycling, the main problem was that some respondents, including those who clearly recycled based on the name of their business, did not consider themselves to be service based. Whilst the aim was to try to identify turnover generated from the service offered (that is, the actual process of recycling itself), some respondents instead decided that they were essentially manufacturers, generating their turnover from the sale of recycled goods and products.
A decision was made where if the respondent produced a recycled product to be used as a basis for further manufacturing/processing then they were correctly classified as a recycling business and their turnover could be apportioned between the specific service questions on the questionnaire. However, if the respondent was producing a finished end product from recycling raw material (for direct retail) then they were deemed as manufacturers, and their turnover would therefore be generated from the sale of the finished products. An example used was that of a company that recycled gold. If the business produced recycled gold to produce ingots for the purposes of selling on to other businesses (who would then use this to make gold rings etc.) then they were offering a recycling service. Whereas, if the company were recycling gold to produce the actual rings themselves (for retail) then this would be classified as manufacturing. This approach to error clearance and quality adjustment was unique to the STS as other, more established business surveys use their own approach to such problems due to the differing requirements and composition of each survey.
A similar issue was experienced in the publishing industry, where respondents did not consider themselves to offer a 'publishing' service, but instead classified themselves as manufacturers in that they were creating a physical good for sale at retail. There were many varied examples given, ranging from newspapers and magazines to chequebooks and statements on behalf of banks. It became apparent that there were broadly two types of business - publishers and printers. Again, a decision had to be made, and in this case if the business owned the content copyright and arranged to make this available to the public then they were publishers, whereas if they did not own the copyright and did not disseminate the printed content to the public then they were printers.
Other industries included in the STS also experienced difficulties due to the nature of the services they produced and how the turnover generated should be reported. Notably these other industries included those that are technology based such as telecommunications and computer services. The telecommunications industry was particularly difficult for respondents to complete and the issues with data collection are still being resolved. As such, the 2010 STS estimates for this industry are not of sufficient quality to be published at this time, whilst further quality assurance is carried out. Additionally, strict disclosure rules have been applied to the STS estimates to ensure the confidentiality of respondents who provided data. Where the final STS estimates for a question have been identified as disclosive, the data has been suppressed. In the case of the rail freight and sewerage services industries, this has unfortunately meant removing the whole set of industrial data.
As detailed above, the main purpose of the STS is to collect a detailed source of service sector turnover that is required to re-weight the SPPI. Therefore, to an extent the STS is constrained by the requirements of the data required for SPPI and there are certain limitations that need to be taken into account when analysing the data. The results of the STS are also classified as experimental statistics and therefore they have not met the rigorous quality standards of National Statistics. The following limitations should be taken in mind when using the STS results:
The coverage of the STS sample is constrained by the sample defined for the SPPI. This means that partnerships/sole proprietors and Government agencies were excluded from the STS sample and estimation. This may impact on the estimates produced in those industries where the above type of businesses are prevalent (for example the legal industry is likely to have a large amount of partnerships and these would be excluded from the STS). Care should therefore be taken when comparing the STS estimates to other sources of turnover data due to differences in coverage and sample size/allocation. As a benchmark, total 2010 turnover as published in the ONS Annual Business Survey (ABS) has been included alongside the STS estimate for each industry (at an aggregate industry level only), however, due to the issues detailed above, direct comparisons are not recommended.
Whilst every effort has been made to accurately define the detailed services for which business-to-business turnover is requested, there is still likely to be an element of subjectivity when completing the questionnaire, whereby a respondent will need to make an informed judgement regarding which detailed services are applicable to their business. In some cases, respondents are not always able to report the turnover in the level of detail required for the STS and therefore an informed judgement has been made to disaggregate turnover to the level of detail required.
Businesses were selected for inclusion in the STS based on their industrial classification (as defined by SIC2007) as held on the IDBR. This classification represents the main area of trade for a business. The industrial version of the STS questionnaire a business received depended on this classification and no consideration was made regarding any turnover generated from services offered outside the main classification of a business. For example, a business classified to the legal industry on the IDBR received a legal version of the STS questionnaire. This business may have generated significant turnover from activities classified to another industry (such as accountancy). The business did not receive an additional questionnaire to collect detailed turnover for this further activity (accountancy services in this example), however, the aggregate turnover for these additional services would be reported in the second section of the STS questionnaire.
An accompanying reference table (384 Kb Excel sheet) has been published alongside this article which provides estimates of the turnover generated for each industry included in the 2010 STS (where the industry has been deemed suitable for publication). For each industry the following information is published (all data for 2010):
total business-to-business turnover for detailed services in the relevant industry,
total business-to-business turnover for services within an industry not classified above,
total turnover not elsewhere classified (such as from exports, manufacturing etc.),
total industry turnover,
accompanying quality measure for each of the above estimates (in the form of CVs - see methodology section above),
total 2010 industry turnover as published in the ABS (for benchmarking purposes).
Further details on any aspect of the 2010 STS are available by contacting the authors of this article.
Quality and Methodology Information
A Quality and Methodology Information (QMI) paper for both the PPI (109.5 Kb Pdf) and SPPI (139.3 Kb Pdf) describes in detail the intended uses of the statistics presented, their general quality and the methods used to produce them.
Follow us onTwitter and Facebook.
View the latest ONS podcasts on YouTube.
National Statistics are produced to high professional standards set out in the Code of Practice for Official Statistics. They undergo regular quality assurance reviews to ensure that they meet customer needs. They are produced free from any political interference.
Tel: Luke Croydon 0845 604 1858
Emergency on-call 07867 906553
Tel: Chris Jenkins +44 (0)1633 455474
PPI/SPPI General Enquiries:
Tel +44 (0)1633 456628 or +44 (0)1633 455941
© Crown copyright 2012
Under the terms of the Open Government Licence and UK Government Licensing Framework, anyone wishing to use or re-use ONS material, whether commercially or privately, may do so freely without a specific application for a licence, subject to the conditions of the OGL and the Framework.
For further information, contact the Office of Public Sector Information, Crown Copyright Licensing and Public Sector Information, Kew, Richmond, Surrey, TW9 4DU.
Tel: +44 (0)20 8876 3444
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