Annual Survey of Goods and Services QMI

This quality and methodology report contains information on the quality characteristics of the data (including the five European Statistical System dimensions of quality) as well as the methods used to create it.

The information in this report will help you to:

• understand the strengths and limitations of the data

• learn about existing uses and users of the data

• reduce the risk of misusing data

• help you to decide suitable uses for the data

• understand the methods used to create the data

• The Annual Survey of Goods and Services (ASGS) presents annual statistics on the value of turnover from goods and services provided from the UK service sector, including a breakdown of the turnover generated from both UK and overseas customers.

• The ASGS collects data from UK businesses in the service sector using the UK Standard Industrial Classification 2007: SIC 2007 – excluding construction, public administration and defence, compulsory social security, and banking. The public sector, non-profit making organisations, and government are also excluded from this survey.

• The ASGS was introduced in 2017 (reference period 2016) to replace the Services Turnover Survey (STS). The first ASGS report was published in August 2018 containing data for 2016.

• The ASGS is sent to a sample of approximately 40,000 businesses.

• The ASGS data are collected through an electronic questionnaire, rather than a paper questionnaire, improving respondent experience and data accuracy.

• There are background validation checks while the respondent is completing the electronic questionnaire. Further validation and editing are carried out once the online questionnaire has been submitted.

• Imputation methods are applied to compensate for non-response, and a non-linear estimation method is used to estimate for non-sampled businesses. Outlier methodology is applied to identify anomalous survey responses that are not representative of other like-sized businesses. Further details on the editing, imputation and estimation methods are provided in the Methods used to produce ASGS data section.

• ASGS data provide estimates of the value for each combination of Industry class (four-digit SIC) and Product subcategory (six-digit Statistical Classification of Products by Activity). Alongside the estimates of domestic services turnover, exported services are also reported. The ASGS Annex provides more detail of the services product breakdown.

• ASGS data are used by the Prices and Supply and Use teams in the Office for National Statistics (ONS).

Overview

The Annual Survey of Goods and Services (ASGS) was launched by the Office for National Statistics (ONS) in summer 2017 in response to one of the main recommendations from Professor Sir Charles Bean’s independent review of economic statistics. This recommendation was to provide more comprehensive and detailed statistics of the services industries for the UK, which were not captured by existing surveys. The ASGS replaced an existing ONS survey, the Services Turnover Survey (STS), which was last conducted in 2015. The STS only covered approximately 40% of total service industry turnover.

The main purpose of the ASGS is to measure turnover, broken down into the individual services and goods that a business provides. As businesses are becoming more diverse, there is a need to measure secondary activities outside their main industrial classification as well as activities in their main classification.

The ASGS helps to improve the quality of product breakdowns in the supply and use tables, and in turn the quality of balancing and Gross Domestic Product (GDP), a main economic indicator. Additionally, ASGS data will inform the Services Producer Price Index (SPPI), which provides a measure of inflation for the UK services sector, and are used as deflators in the Index of Services (IoS) and the output measure of GDP.

History of the Annual Survey of Goods and Services

The Annual Survey of Goods and Services (ASGS) was launched in 2017, replacing the Services Turnover Survey (STS), which was last run in 2015. The STS was conducted biennially and collected turnover data for various services provided by 42 different services industries, with a sample size of 20,000. The STS only covered approximately 40% of total service industry turnover.

The ASGS was therefore developed to broaden service industries coverage. Consistent with the STS, the ASGS uses a Statistical Classification of Products by Activity (CPA code) harmonised with the Standard Industrial Classification (SIC), collecting information about products a business produces inside and outside its main industrial classification or activity. This helped to meet recommendations made in Professor Sir Charles Bean’s independent review of economic statistics. The use of a harmonised product classification also enables comparisons with other national statistics institutes. Therefore, to fully capture services industries turnover, the ASGS was developed to collect a detailed breakdown of the services provided by businesses and whether they were provided to customers inside or outside the UK.

Within the UK, the services industries represent a large proportion of Gross Domestic Product (GDP) (80%), in comparison with the manufacturing industries (10%). However, while a detailed survey exists for the manufacturing industries (UK Manufacturers’ Sales by Product, also known as ProdCom), the services industries were not being measured in a sufficient level of detail. An existing ONS survey, the International Trade in Services (ITIS) survey, measures UK services industries exports and imports; however, it does not measure the domestic supply of services.

Uses and users

There are a range of users for the ASGS data. Users include those from the government, both internal within the ONS and external in other government departments, including:

• government departments including the Department for Business, Energy and Industrial Strategy (BEIS)

• devolved administrations including the Northern Ireland Statistics and Research Agency (NISRA) and Scottish Government

• the Prices team within the ONS

• the Supply and Use team within the ONS

• academics and research agencies

The data collected from the ASGS will feed into the Services Producer Price Index (SPPI) and the national accounts supply and use tables, which are a central component of the National Accounts balancing process, setting the annual level of nominal Gross Domestic Product (GDP). ASGS data have improved the quality of balancing and GDP.

Data delivered to the Prices division ensure that legislative Eurostat requirements are met.

Future ASGS data will be delivered to the Blue Book, which is an important annual publication of national accounts statistics and the essential data source for anyone concerned with economic policies and studies.

Strengths and limitations

The main strengths of the survey include:

• comprehensive coverage of services provided at a more detailed level which is not available from other ONS surveys

• the questions are industry specific, which reduces the burden on sampled businesses

• the electronic questionnaire ensures filling in the ASGS is quicker and easier to complete than if it were a paper version

• validation is carried out in real time with the respondent at the point of data entry into the electronic questionnaire

The main limitations of the survey include:

• businesses classified outside of the service industry are excluded, even though providing a service may be an important part of their income

• turnover is collected for service activities, which can be at a more specific level than respondents accounts, therefore estimates are accepted

• the ASGS is a survey, not a census or an administrative data source

• some published tables contain high levels of suppressed data as a result of disclosure methodology being applied

• no geographical information is available (the sample is not stratified by region)

Recent improvements

The electronic questionnaire has undergone improvements since the first year of dispatch based on respondent feedback and user testing. Improvements include updating the key words in the search function, ensuring searches are more refined, and removing the option for respondents to select government product codes.

Products within wholesale, retail and waste management industries were reduced to a less detailed level following feedback from respondents after the first year. The wholesale industry services to select from the online questionnaire were reduced to three-digit level, retail was reduced to five-digit level and waste management was reduced to five-digit level. This modification was deemed a success and it should become the default level going forward.

In the first two years of the ASGS, key responders were defined as companies that have more than three hundred employees or companies that have the highest contributing turnover per four-digit SIC, despite their employment size. A new rule was added when collecting data for the 2018 reference period that defines which companies are selected as key responders. In addition, a new rule introduced in the 2018 reference period was that if 20% or more of a total product (six-digit CPA) was generated by a business, regardless of their employment size, this business will be a “key responder”. Key responders are prioritised when response chasing.

This report provides a range of information that describes the quality and characteristics of the data and identified issues that should be noted when using the output.

We have developed Guidelines for measuring statistical quality based upon the five European Statistical System (ESS) quality dimensions. This report addresses the quality dimensions and important quality characteristics, which are:

• relevance

• accuracy and reliability

• coherence and comparability

• accessibility and clarity

• timeliness and punctuality

• output quality

• geography

• concepts and definitions

• why you can trust our data

More information is provided about these quality dimensions in the following sections.

Relevance

The ASGS outputs will be used as part of the national accounts , an integral part of the measurement of Gross Domestic Product (GDP). GDP is the primary measure of the overall state of a country’s economy; it is extensively reported in the media to track the UK’s economic performance.

The data will also be used to improve the weighting and sample used for the Service Producer Prices Index (SPPI). The SPPI provides a measure of inflation for the UK service sector. The main users of the SPPI include other areas of the ONS, for instance the National Accounts, where SPPIs are used as deflators in the Index of Services (IoS) and the output measure of Gross Domestic Product (GDP). It is also used by external organisations for purposes such as aiding escalation clauses in contracts, monitoring inflation within the services sector and informing economic policy decisions.

The delivery of ASGS data was required in 2018 to meet Eurostat specifications. The Framework Regulation Integrating Business Statistics (FRIBS) requires that UK turnover for the services industries is broken down by domestic and exported turnover and using the Statistical Classification of Products by Activity (CPA) version 2.1 instead of the previous bespoke SPPI product classification.

Finalised ASGS data are timetabled to be included in the Blue Book from 2020 onwards. The Blue Book is an important annual ONS publication of national accounts statistics. The Blue Book records and describes economic activity in the UK and is used to support the formulation and monitoring of economic and social policies.

Accuracy and reliability

This section explains statistical concepts used in the processing of the ASGS data.

Total error

The total error in a survey estimate is the difference between the estimate derived from the data collected and the true (unknown) value for the population. The total error consists of two main elements: sampling error and non-sampling error.

Sampling error

The error that arises because the estimate is based on a survey rather than a census of the population. The results obtained for any single sample may vary from the true value for the population, but the statistical theory is well developed and estimates of accuracy can be produced, such as the standard error.

Non-sampling error

Non-sampling errors cover all errors unrelated to sampling methodology. These can be difficult to quantify and relate to errors in coverage, measurement, processing and non-response.

Coverage Errors

The full coverage of the ASGS is impossible to assess, because it is currently not possible to identify businesses that provide services as a secondary activity. A general assessment of coverage must therefore be viewed with caution.

Coefficient of variation (CV)

This is estimated by the standard error of a variable divided by the survey estimate. This is used to compare the relative precision across surveys or variables. The closer the CV is to zero, the less uncertainty there is in the estimate. To minimise the burden on businesses and to complete the survey within allocated resources, the sample size is limited, which can lead to large CVs and some uncertainty around detailed product estimates.

Response rate

The response rate gives an indication of the number of respondents who completed the survey divided by the number of respondents in the sample.

Response accuracy

It is difficult to accurately quantify the effect of response inaccuracy. Questionnaires are tailored to industrial sector, so that businesses are only asked to respond to questions relevant to their industry. This helps to reduce inaccuracy. In addition, there are regular reviews to improve the survey questions and supporting notes, helping respondents complete the survey more accurately.

Validation procedures are implemented to minimise errors, including built-in checks throughout the electronic questionnaire where an error message will show if the respondent has entered a value that does not pass one of the logic rules (for example the sum of turnover allocated to each product does not equate to the total turnover figure given). Further validation is conducted by manual quality assurance checks involving desk research, such as year-on-year validation, and as a last resort, respondents are recontacted to clarify erroneous responses.

Industry classification on the Inter-Departmental Business Register (IDBR)

Industry reclassification (moving from one Standard Industrial Classification (SIC) class to another) can occur because of a relatively small change to the nature of its operation but can have a significant effect on estimates.

During the ASGS data collection period, misclassifications of businesses are corrected on the sample and on the business population from the IDBR. Other surveys that extract their sample from the IDBR benefit from these corrections.

When a survey does not cover the whole business population, reclassification can lead to units moving in and out of scope of the sample, even if the business is still offering services as a secondary activity.

Coherence and comparability

Coherence with the Annual Business Survey

As businesses are becoming more diverse, there is a need to measure secondary activities outside their main industrial classification. The ASGS collects more comprehensive and detailed statistics of the services industries for the UK, which were not captured by any existing surveys.

The Annual Business Survey is a well-established survey providing information on turnover, among other variables, to help inform and feed into supply and use tables (SUTs) and national accounts. However, the ASGS was developed to collect the detailed breakdown of this turnover between customers inside and outside the UK, as a requirement of our internal SPPI team and the Supply and Use team.

Despite the ABS sample being larger, sampling approximately 73,000 businesses, the ABS does not collect information at a service product level. The ABS questions total turnover and total exports, therefore the totals are comparable between surveys. However, the ASGS also questions a breakdown of total turnover against the services selected, generated by customers based in the UK and overseas.

The industry coverage overlaps at four-digit SIC, but differences could arise due to the estimation methods. The ABS uses the expansion estimator whereas the ASGS uses the Chamber-Cruddas estimation method. This method is composed of two parts. The first estimates the probability that a business provides a service, and the second conditionally models the service turnover as proportional to register employment.

Coherence across time series

Minimal changes are made in data collection and methods resulting in more comparable estimates between each reference period. ASGS data are collected in accordance with guidance issued to European Union member states which are used in the production of ASGS estimates; therefore, figures published by the UK should be comparable with other member state countries.

Accessibility and clarity

The format for accessible content is a combination of HTML webpages for narrative, charts and graphs, with data being provided in usable formats such as CSV, XML and Excel. The ONS website also offers users the option to download the narrative in PDF format. In some instances, other software may be used or may be available on request. Available formats for content published on our website, but not produced by us, or references on our website but stored elsewhere, may vary.

For information regarding the conditions of access to data, please refer to these links:

• terms and conditions (for data on the website)

• freedom of information

• accessibility

Timeliness and punctuality

We are currently reviewing the timeliness of the release of ASGS data and our aim as the survey develops is to reduce the time taken to publish the data. However, the time lag between collection and publication reflects the complexity of the survey. The aim is that publications will feed into the timetable for Blue Book publications. It is not possible to collect this large quantity of data more frequently, as this would place an unacceptable level of burden on businesses and would require a prohibitively large resource to process results.

For more details on related releases, the release calendar is available online and provides 12 months’ notice of release dates. If there are any changes to the pre-announced release schedule, public attention will be drawn to the change alongside the full explanation of the reasoning behind it, as set out in the Code of Practice for Statistics. This itself has been recently updated, with a greater focus on statistical context and recommended usage.

Output quality

Initially during the first year with no previous ASGS data, 2016 data were quality assured against the total turnover reported on the Annual Business Survey (ABS), a well-established survey that also collects business turnover. 2016 data were also revised against 2017 data, once validated. At the time of the 2016 publication, the percentage of validated returned questionnaires was 62%.

As the ASGS is now in its third year (reference period 2018), an additional year-on-year validation check has been implemented to ensure the highest quality data possible. We are currently working towards achieving National Statistics status. Part of this work will be to establish and publish a full publication schedule, comparable to those of other annual ONS outputs.

Geography

The ASGS covers businesses in the UK (England, Scotland, Wales, and Northern Ireland, excluding the Channel Islands and the Isle of Man). Turnover is collected for both inside (domestic) and outside (exported) the UK according to where the customer is based, rather than the location of where the service is delivered. No geographical information is collected.

Concepts and definitions

Turnover is defined as sales to any individual, business entity or other parts of an organisation. It is based on an accruals accounting basis (invoices raised), rather than a cash accounting basis.

The following is excluded from turnover:

• turnover from overseas branches, subsidiaries and agencies

• VAT

• capital receipts from disposal of assets

• interest and dividends

• grants and subsidies

Turnover data are collected at the Statistical Classification of Products by Activity (CPA code) three-digit level for wholesale trade except of motor vehicles and motorcycles; five-digit level for retail trade except of motor vehicles and motorcycles; five-digit level for waste management; and six-digit level for all other sampled industries.

Why you can trust our data

The ONS is the UK’s largest independent producer of statistics, and its national statistics institute. The Data Policies and Information Charter, available on the ONS website, details how data are collected, secured and used in the publication of statistics. We treat the data that we hold with respect, keeping it secure and confidential. We use statistical methods that are professional, ethical and transparent. You can find more about our data policies on our website.

The Annual Survey of Goods and Services data are currently experimental but we are working towards achieving National Statistics status, designated by the UK Statistics Authority (UKSA) in accordance with the Statistics and Registration Service Act 2007. This designation signifies compliance with the Code of Practice for Statistics, which has recently been updated and focuses on trustworthiness of data in greater depth.

We are keen to get your views on the methods and work to date to help inform and make improvements to the ASGS. If you have any comments, please email us at ASGS@ons.gov.uk

How we collect the data, main data sources and accuracy

Sample frame

The Annual Survey of Goods and Services (ASGS) sample frame is the Inter-Departmental Business Register (IDBR). The IDBR covers businesses in all parts of the economy, except those that are not registered for Value Added Tax (VAT) or Pay As You Earn (PAYE), for example, very small businesses, the self-employed, those without employees and those with low turnover. Some non-profit making organisations are also not registered on the IDBR. The IDBR has details of approximately 2.6 million businesses and covers approximately 99% of UK economic activity. The IDBR is used by government departments, including the Office for National Statistics, as the sampling frame for most business surveys.

The ASGS draws its sample from the 1.8 million businesses classified in the service sectors, which are in the scope of this survey.

Sample design

The ASGS uses a stratified random sample design. This is grouped by:

• five employment size bands: 0 to 9, 10 to 49, 50 to 99, 100 to 299 and 300 plus

• industry class: four-digit UK Standard Industrial Classification 2007: SIC 2007

All businesses with three hundred employees or more will automatically be included in the sample to ensure that the businesses generating the highest turnover are selected. For the businesses below this threshold, a simple random sampling method based on Permanent Random Numbers (PRN) is used for selection.

The ASGS covers a large proportion of the service industry but excludes banks, public sector and non-profit making organisations. Some service sectors were not included, for example some parts of financial, because these data are collected by the Bank of England or other ONS surveys. Inclusions and exclusions in the sample are listed below:

Industry inclusions:

• Electricity, gas, steam and air conditioning supply (section D)

• Water supply; sewerage, waste management and remediation activities (section E)

• Wholesale and retail trade; repair of motor vehicles and motor cycles (section G)

• Transport and storage (section H)

• Accommodation and food service activities (section I)

• Information and communication (section J)

• Financial and insurance activities (section K)

• Real estate activities (section L)

• Professional, scientific and technical activities (section M)

• Administrative and support service activities (section N)

• Education (section P)

• Human health and social work activities (section Q)

• Arts, entertainment and recreation (section R)

• Other service activities (section S)

Industry exclusions:

• Agriculture, forestry and fishing (section A)

• Mining and quarrying (section B)

• Manufacturing (section C)

• Construction (section F)

• administration and defence; compulsory social security (section O)

Sample size

The ASGS has a sample size of approximately 40,000 businesses.

Data collection

Data are collected from businesses primarily through industry-specific electronic questionnaires, to reduce respondent burden. Respondents can select any other services the business provides outside of their industry classification using the search function if applicable. An example of this can be found in the “Electronic Questionnaire” section of the article: Development of the Annual Survey of Goods and Services.

Reminders are dispatched for non-responses. In addition, a telephone exercise is conducted by a dedicated respondent relations team at IFF Research who recontact respondents to correct any erroneous data.

Measured variables

The primary aim of the ASGS is to more appropriately measure the turnover generated in the services sector and provide a much more detailed coverage of the UK service sector, including off-diagonal products (products generating turnover outside the business’ main industry classification). The important variables for the ASGS output are turnover at a three-, five- or six-digit CPA level split by customers inside and outside the UK.

In addition to service products, there is also the option for respondents to report turnover from production.

How we process

Once the data are collected, the data are validated by recontacting businesses which have failed at least one of the fourteen validation rules. The data feed into the ONS results system which contains modules to preprocess the data, impute values, estimate results and apply disclosure rules.

Preprocessing

This module checks that validation rules are applied correctly.

Imputation

Imputation techniques are used to replace missing data as a result of unit non-response (no information available on the current period) or item non-response (partial information available: total turnover but not product breakdown). Imputation is carried out only for businesses with more than nine employees. All the non-responses not dealt with in imputation are accounted for in the estimation process.

Item non-response

Imputation classes are created for each industry level (two digit) and employment size-band. Product patterns based on services provided within the same imputation class are derived. In effect, the total turnover for a case to be imputed would be distributed among the services of a selected pattern; the amounts assigned to each service would depend on the mean proportions of turnover calculated from the complete cases.

Unit non-response

When a business does not respond to the survey in the current year but has responded in the previous year, the returns from the previous year are then brought forward to the current year and multiplied by a growth factor/imputation link. This growth factor has been calculated using companies that have responded in the previous and current year. One growth factor is calculated for each industry-employment size-band combination. This imputation method (unit non-response) has also been applied retrospectively further enhancing the quality of the first-year estimates (2016 reference period). The estimation method deals with businesses that have never responded.

Estimation

Estimates are produced using a non-linear estimator (Chamber-Cruddas). The estimation procedure is based on the employment of the non-sampled units (obtained from the IDBR), along with the sales per head (SPH) by respondent industry and product propensity, both of which are obtained from the sampled units. These are defined as follows:

• SPH: sales value of product or employment count

• Product Propensity (PP): proportion of businesses in strata carrying out the service or number of businesses in the strata; this is the probability that a business within a particular stratum will carry out this service

Businesses with extreme SPHs are removed (see outliers below) and a SPH for the industry-product combination is calculated.

An estimate of the sales for the non-sampled units, including those that have never responded, is produced by multiplying the PP by the employment of the non-sampled business, which is then multiplied by the SPH for the industry.

The total UK services sales for the product is calculated by summing the total returned sales of the sampled units (including imputed values) and the estimated sales of businesses that have never returned their questionnaire or were not in the sample (non-sampled units).

Outliers

Removing atypical responses is undertaken according to sales per head (SPH). Values within the top and bottom 5% of the average SPH are removed only when there are more than ten responses in each industry-product combination.

How we quality assure the data

Data are quality assured throughout the data collection, processing and analysis processes through regular consistency checks, investigation of anomalies, ensuring disclosure procedures and reviewing data sources. These checks are presented at regular quality assurance meetings, where important internal stakeholders are able to interrogate the data and explore any anomalies or interesting findings. External stakeholders, for example other government departments, also have regular opportunities to analyse the data and share feedback with ONS colleagues.

How we disseminate the data

The Annual Survey of Goods and Services data are disseminated primarily through the publication of statistical bulletins. The publication schedule is detailed under the Timeliness and punctuality section. The Annual Survey of Goods and Services publishes regular statistical bulletins, with releases increasingly promoted through the use of ONS social media accounts.

Statistical disclosure

We employ consistent disclosure control procedures ahead of publication. The Government Statistical Service defines statistical disclosure control as follows:

“Statistical disclosure control (SDC) is the term used to cover the many methods of safeguarding the confidentiality of the information about individuals and businesses. Information obtained from surveys or administrative data is usually given in confidence. SDC is applied to ensure that individuals, businesses or other statistical units cannot be identified from published data, whether record-level data or tables. This will involve modifying data so that the risk of identification is reduced to an acceptable level.”

The Annual Survey of Goods and Services is conducted under the Statistics of Trade Act 1947. This act imposes restrictions on the way that data collected during the survey may be used, to ensure that information that can be attributed to an individual organisation is not disclosed in any publication.

The Code of Practice for Statistics set out practices for how we protect data from being disclosed. More information can be found on the ONS Disclosure Control Methodology webpage.

Disclosure is a particularly sensitive issue in business surveys, given the commercial confidentiality of the data collected. The following rules are applied to ASGS estimates (except where the respondent is willing to allow the data to be published):

• minimum threshold rule: this rule states that there must be at least n reporting units (businesses) in a cell

• P% rule: the total contribution of the m largest contributors to the cell aggregated total must be less than p% of the total in that cell; sometimes this rule can be applied multiple times, with different values of m and p

• dominance rule: the estimated total of all the units in the cell (including those not included in the sample), without the d largest units from the sample, must be at least s% if the largest b units in the sample

Businesses whose data have been suppressed, because of the level of disclosure, may be contacted following publication of the data to request permission to publish disclosed data.

How we review the data

Previous data are used to ensure the quality of the time series. For example, at the aggregate level (industry and product), checks are undertaken to ensure the latest year’s data are comparable as part of the time series in comparison with the previous year’s values. If not, data queries are sent to understand drivers behind movements in the data. The latest year’s data are also used to inform and revise previous data in circumstances where respondents have clarified that the latest data are correct during validation call backs.

Development of the Annual Survey of Goods and Services can be found on our website.

A technical report, describing detailed information about the methodological and technical procedures used to produce the Annual Survey of Goods and Services estimates, will be available shortly.

For further information, please contact the Annual Survey of Goods and Services team via email at ASGS@ons.gov.uk.