Reference metadata describe statistical concepts and methodologies used for the collection and generation of data. They provide information on data quality and, since they are strongly content-oriented, assist users in interpreting the data. Reference metadata, unlike structural metadata, can be decoupled from the data.
Government Buildings Cardiff Road Newport South Wales NP10 8XG
1.6. Contact email address
Confidential because of GDPR
1.7. Contact phone number
Confidential because of GDPR
1.8. Contact fax number
Confidential because of GDPR
2.1. Metadata last certified
27 November 2018
2.2. Metadata last posted
27 November 2018
2.3. Metadata last update
27 November 2018
3.1. Data description
Name of indicator/source
The retail sales index is compiled from the results of the monthly business survey- retail sales index. The purpose of the indicator is to measure the short term movements of retail sales.
3.2. Classification system
NACE Rev. 2.
3.3. Coverage - sector
Activities covered
Division 47 of NACE Rev. 2.
Size classes covered
There is no size class threshold.
3.4. Statistical concepts and definitions
List and definition of variables
On a monthly basis the questionnaire collects information on retail turnover including VAT. It is also broken down by broad product group for 52 enterprises.
Planned changes in information collected
None
Accounting conventions
Turnover data is recorded for a fixed 4 or 5 week period.
3.5. Statistical unit
Reporting unit
Enterprise
Observation unit(s)
Enterprise
3.6. Statistical population
The survey measures value and volume of retail sales in Great Britain.
3.7. Reference area
Geographical area covered
Great Britain- excluding Northern Ireland.
3.8. Coverage - Time
Date of first use as a source
This index has been compiled since the 1950s.
3.9. Base period
Base (reference) year
The reference year is 2016. Annual chain-linking is used.
Indices (2016=100). Growth rates in %
Reference year of this report
2016 Monthly.
6.1. Institutional Mandate - legal acts and other agreements
Legal basis
Statistics of Trade Act 1947.
Obligation on units to provide data
The survey is compulsory. Fines for refusing to provide information are laid down by the Criminal Justice Act 1991.
Planned changes in legal basis, obligation to respond and frame used
None
6.2. Institutional Mandate - data sharing
Eurostat
Scottish Government receive quarterly regional data.
7.1. Confidentiality - policy
Dissemination of terms and conditions under which official statistics are produced, including confidentiality of individual responses
All necessary steps are taken to protect the confidentiality of data collected from retail businesses. This includes statistical disclosure controls to ensure that individual businesses are not identified in the published statistics.
When data are shared with other bodies, for example Eurostat and the Scottish Government it is done so under legislation and using secure electronic file transfer methods.
8.1. Release calendar
Advance dissemination of release calendar
Release dates for all ONS publications are available via the National Statistics website using the following link;
All electronic methods of dissemination are automatically updated at 9.30am on the day of release. See link for National Statistics code of practice. Code of Practice for Official Statistics
Identification of internal government access to data before release
Not applicable
Transmission to Eurostat and further use of the statistics
Data are transmitted to Eurostat shortly after national publication.
Monthly
10.1. Dissemination format - News release
Identification of ministerial commentary on the occasion of statistical releases
No ministerial commentary attached to the release.
The retail sales summary outlines the key messages within each month’s data. This includes the key drivers for movements within particular sectors for both value and volume indices.
The Retail Sales Statistical Bulletin presents the key messages within the data, analysis of raw data including analysis by size of business, a detailed sector summary focusing on the four main aggregates: predominantly food, predominantly non-food, non-store retailing and automotive fuel.
Dissemination of documentation on methodology and sources used in preparing statistics
ONS's recommended format for accessible content is a combination of HTML web pages for narrative, charts and graphs, with data being provided in usable formats such as CSV 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 the ONS website but not produced by the ONS, or referenced on the ONS website but stored elsewhere, may vary. For further information please refer to the contact details at the beginning of this document.See link: https://www.ons.gov.uk/businessindustryandtrade/retailindustry/methodologies/retailsalesindexrsiqmi
Description of standard tables produced
Results are published as average weekly sales for each reference period in an index form. The results cover Great Britain and are inclusive of VAT. The current base year is set at 2016=100. Volume and value data are published which are also seasonally and not seasonally adjusted. The retail sales index is broken down by NACE Rev. 2. A summary of growth rates, such as year on year growths are included.
Summary description of quality criteria calculated for national purposes
A statistical test is calculated to help determine whether there is any bias in revisions to first published results. The results are published in the release.
To meet our user needs, the Retail Sales branch is part of the Short-Term Output Indicators Stakeholder Group (STOISG). Key stakeholders and the Office for National Statistics (ONS) meet every three months to discuss published data, planned and ongoing developments and statistical communication.
A report on RSI standard errors and response rates titled Improving quality information for the Retail Sales Index was published in January 2015. The report shows that the median year-on-year standard error is estimated at 0.9% and the month-on-month standard error is estimated at 0.5%.
The RSI is first published using a 61% response rate, this equates with approximately 88% of sampled turnover. The sample of 5,000 from a population of nearly 200,000 may seem small but together these 5,000 retailers cover approximately 95% of all retailing turnover captured on the IDBR.
Sampling error occurs because a sample, rather than the entire population, is surveyed. It is the difference between the true value for the population and the estimated value. One way of measuring this difference is through standard errors.
13.3. Non-sampling error
Response rate
Response rate in terms of questionnaires is around 60% at the time of the first estimate of the index and rises to around 90% two months later.
Actions to speed up or increase the rate of response
Non-sampling error arises from inaccuracies in collecting, recording and processing the data. The most significant of these errors are: misreporting of data items, deficiencies in coverage, non-response, and processing errors. Every effort is made to minimise reporting error by the careful design of questionnaires, intensive training and supervision of editing and validation staff, efficient data processing procedures and response chasing.
One form of measurement error arises from respondents reporting turnover in pounds (thousands) (£’000s), when actual pounds are required. Where this is the case, the figure is corrected and respondents are informed as to which figures they need to provide.
Coverage error is introduced where businesses are not classified to retail or are not registered for VAT or PAYE and are therefore not listed on the IDBR.
Processing errors can be introduced where data are inputted manually by data analysts in £’000s rather than actual pounds. Training is provided to minimise this processing error.
14.1. Timeliness
Timeliness
The retail sales index is disseminated around 19 days after the end of the reference period.
Timetable of data collection
Questionnaires are sent out three days before the end of the reference period. Closedown for the provisional month is ten working days after the end of the reference period. Late forms are included in subsequent result runs.
14.2. Punctuality
For more details on related releases, the release calendar is available online from the GOV.UK website and provides 12 months’ advance notice of release dates. If there are any changes to the pre-announced release schedule, public attention will be drawn to the change and the reasons for the change will be explained fully at the same time, as set out in the Code of Practice for Official Statistics.
15.1. Comparability - geographical
ONS follow the A (ideal) method for the compilation of these statistics to as described by the Short Term Statistics Regulation - no other European country follows this method.
15.2. Comparability - over time
Comparable time series are available going back to 1988 for the following headline aggregates and their sub-sector series:
• All retailing excluding automotive fuel;
• Predominantly food stores;
• Predominantly non-food stores; and
• Non-store retailing.
For the following aggregates, comparable time series are available going back to 1996:
• All retailing including automotive fuel; and
• Automotive fuel.
15.3. Coherence - cross domain
Dissemination of information that support statistical cross-checks and provide assurance of reasonableness
The Annual Business Inquiry (retail) results are available and can be used to compare annual growth rates.
Confrontation with other data sets
Other statistics on retail sales are available. ONS also produces retail sales statistics as part of the Annual Business Survey (ABS). There are differences between the RSI and ABS retail figures for the following reasons:
• the ABS collects and produces statistics on the United Kingdom whereas RSI covers Great Britain, resulting in differences in total sales values;
• reporting periods can vary, the ABS allows respondents to provide information on either a calendar year or financial calendar year. In comparison the RSI asks respondents to provide information based on a 4 - 4 - 5 week reporting period;
• for the food sector, the ABS includes automotive fuel sales by supermarkets, for the RSI these are extracted and used for the calculation of the automotive fuel series; and
• for chemists the ABS includes National Health Service receipts, these are excluded from the RSI.
Non-official statistics on retail sales are available from the British Retail Consortium (BRC) which produce the Retail Sales Monitor in collaboration with KPMG; and also from the Confederation of British Industry which produce retail trade statistics as part of their Distributive Trade Survey. More information on the differences between these two indicators and the RSI can be found in the article Retail Sales Index (RSI) and Other Surveys.
Other indicators include the British Retail Consortium’s (BRC), United Kingdom Retail Sales Monitor and the Confederation of British Industry’s (CBI) Distributive Trade’s Survey. The data produced by both are regularly used along with the RSI as key indicators of retail sales growth. However, there are differences between the indicators in relation to coverage and compilation that can lead to discrepancies between published figures.
13.1 BRC United Kingdom Retail Sales Monitor (RSM)
The RSM is produced using data supplied by retailers from across the UK's retail industry. Sales values from a range of 12 product categories are provided weekly to KPMG who then aggregate them into annual growth rates for each month on a total and like-for-like basis. The like-for-like figure strips out the effects of expansions and store openings and closures and is presented as a measure of retail industry performance. The total figure reflects retail industry growth. The headline growth rates are published in the second week after end of the month reported on.
From the Retail Sales Statistical Bulletin, the most appropriate measure to compare with the RSM total sales series is the RSI value of retail sales excluding automotive fuel for large stores, non-seasonally adjusted.
13.2 CBI Distributive Trade’s Survey (DTS)
The DTS is a qualitative survey which looks to capture short and medium-term trends in the UK Retail Sector. On average 68 retailers, who account for approximately a third of total employment in the retail sector, respond to the survey. Respondents are required to answer a series of questions regarding sales volumes using a scale of up, down or same as last year.
15.4. Coherence - internal
Not available.
The cost of producing these retail variable is £531,000.
The compliance cost or burden is £230,000
17.1. Data revision - policy
Retail Sales non-seasonally adjusted data is revised as needed. This typically occurs due to late data returns, updated respondent information, replacing adjustments with actual data, and reclassifications of respondents to the appropriate category either within or out of retail. Changes are not made to any non-seasonally adjusted data prior to 2001. Revisions to non-seasonally adjusted data will directly impact on the seasonally adjusted estimates.
Revisions and sampling variations are a consequence of the trade-off between timeliness and accuracy. All estimates are subject to statistical error which refers to the uncertainty inherent in any process or calculation that uses sampling, estimation or modelling. Estimates for the most recent month are provisional and subject to revision because of:
• late responses to the Monthly Business Survey - Retail Sales Index;
• revisions to seasonal adjustment factors which are re-estimated every month and reviewed annually;
• changes from the annual seasonal adjustment review; and
• annual updating of the business register that forms the basis for the sample for the RSI (usually occurring in January) and
• other methodological improvements.
Policy regarding revisions to Retail sales are available in the monthly release on the website.
Provision of information about revision and advance notice of major changes in methodology
Revisions to all published seasonally adjusted estimates are not constrained or restricted as they are for other short term indicators and the national accounts. Revisions are allowed to occur naturally each month along the full length of each time series as a consequence of the direct seasonal adjustment of the component time series. Aggregate seasonally adjusted estimates will then be derived. Any revisions from the directly seasonally adjusted time series will then occur as part of the higher level aggregation. This approach ensures that the RSI uses all available information and provides the most up-to date estimates for recent periods. A report on Revisions to Retail Sales Index and the RSI revisions policy are published on the ONS website.
18.1. Source data
Type of source
Statistical survey
Frame on which the source is based
Interdepartmental business register (IDBR).
Sample or census
Sample
Criteria for stratification
Stratified sample by activity into different employment strata.
Threshold values and percentages
Approximately 5,000 units are sampled each month. There is full enumeration for units with at least 100 persons employed and those between 0 to 99 persons employed and over GBP 60 million turnover a year. A random sample is taken of units employing 0-4, 5-9, and 10-99 employees with less than GBP 60 million turnover a year.
Frequency of updating the sample
New sample every year in January and is maintained on a monthly basis via gains and losses.
18.2. Frequency of data collection
Periodicity of data collection
Data collection takes place monthly.
18.3. Data collection
Questionnaires used in the survey
A standard questionnaire is used each month to collect retail turnover.
Planned changes in national questionnaires
Not applicable
Data collection media
Electronic data collection for all businesses.
Planned changes in data collection methodology
Not applicable
18.4. Data validation
Intra-dataset checks
Returned data are passed through a series of validation checks which includes an automatic selective editing procedure. The selective editing approach means that the editing process should be more efficient and effective since it will only edit potential errors that have a significant impact on final outputs. Selective editing is an internationally recognised method that uses a data based approach to assess the influence of business estimates on the aggregate outputs. Under selective editing key variables on the questionnaire are defined and scores derived for these. The scores compare the value provided on the questionnaire with expected values, where the expected values are generally estimated using past data or other available information related to the variable, for example, from administrative sources. The scores from ‘key’ variables are then combined to derive a score for every questionnaire. This derived, single score for the business’s return is then tested against a methodologically set threshold. If the score is higher than the defined threshold then the questionnaire will fail and be flagged for manual editing. Thresholds have been derived and set to ensure minimum bias is introduced from not editing values that may have been edited under the previous edit all returns system. Thus selective editing does result in an adverse impact on output quality.
In order to ensure that all errors are captured, before questionnaires pass through selective editing they are subjected to automatic editing and then a number of user defined checks. For example, these will include checks to detect changes to reporting periods and implausible returns such as turnover being less than zero.
This is queried by the editing & validation (E&V) team and can also be queried by the results & prosessing (RAP) team. The respondent may be contacted to verify whether the movement is genuine. This is recorded and retained on the database for future reference when analysing the index.
In turn, data are analysed at a macro level to determine the contributing factors to the movement in the value index and its published categories. Together, this micro and macro approach to editing identifies outliers and anomalies within the returned data.
18.5. Data compilation
Estimates for non-response
These estimates are made based on responses from units of a similar size using ratio imputation as the method.
Estimates for grossing-up to population levels
With the exception of the largest units which are fully enumerated, employment bands 0-4, 5-9, and 10-99 employees with less than GBP 60 million turnover a year are grossed up to the population.
Type of index
Annual chain-linking is used. Estimates will be linked together using a three month average as opposed to a single month, in order to reduce volatility.
Method of weighting
For each year, relative proportions (in other words weights for sales values of each industry within all retailing) are calculated using actual data from the Retail Sales Inquiry. Commodity weights from the annual business survey (ABS) are used as a benchmark to be used in conjunction with monthly Industry weights from consumer price indices (CPI) to produce a constant price index.
Planned changes in production methods
None
18.6. Adjustment
WDA, SA and other calculations
Seasonally adjusted estimates are derived by estimating and removing calendar effects (for example Easter moving between March and April) and seasonal effects (for example increased spending in December as a result of Christmas) from the non-seasonally adjusted (NSA) estimates. Seasonal adjustment is performed each month, and reviewed each year, using the standard, widely used software, X-13-ARIMA. Before adjusting for seasonality, prior adjustments are made for calendar effects (where statistically significant) such as: returns that do not comply with the standard trading period, bank holidays, Easter, and the day of the week Christmas occurs.
While the number of trading days in a set period remains the same, the standard reporting periods of 4-4-5 weeks introduce a ‘phase-shift’ effect, associated with the fact that the standard periods do not match calendar months and move slightly each year relative to the calendar month. In a typical year, the standard periods total 52 weeks or 364 days. As a result, the standard periods fall back one day each year (or two days for a leap year).
Every five or six years, the reporting year is brought back into line with the calendar year by adding an extra week, normally to January.
The seasonally adjusted estimates also have corrections for bank holidays; Easter effects which are caused when Easter falls late in March or early in April and the day of the week Christmas occurs. Prior connections are applied as necessary. Corrections are estimated and applied where there is a statistically significant effect.
The calculation of the RSI has an adjustment to compensate for calendar effects which arise from the difference in the reporting periods.
Not available.
Name of indicator/source
The retail sales index is compiled from the results of the monthly business survey- retail sales index. The purpose of the indicator is to measure the short term movements of retail sales.
27 November 2018
List and definition of variables
On a monthly basis the questionnaire collects information on retail turnover including VAT. It is also broken down by broad product group for 52 enterprises.
Planned changes in information collected
None
Accounting conventions
Turnover data is recorded for a fixed 4 or 5 week period.
Reporting unit
Enterprise
Observation unit(s)
Enterprise
The survey measures value and volume of retail sales in Great Britain.
Geographical area covered
Great Britain- excluding Northern Ireland.
Reference year of this report
2016 Monthly.
A report on RSI standard errors and response rates titled Improving quality information for the Retail Sales Index was published in January 2015. The report shows that the median year-on-year standard error is estimated at 0.9% and the month-on-month standard error is estimated at 0.5%.
The RSI is first published using a 61% response rate, this equates with approximately 88% of sampled turnover. The sample of 5,000 from a population of nearly 200,000 may seem small but together these 5,000 retailers cover approximately 95% of all retailing turnover captured on the IDBR.
Indices (2016=100). Growth rates in %
Estimates for non-response
These estimates are made based on responses from units of a similar size using ratio imputation as the method.
Estimates for grossing-up to population levels
With the exception of the largest units which are fully enumerated, employment bands 0-4, 5-9, and 10-99 employees with less than GBP 60 million turnover a year are grossed up to the population.
Type of index
Annual chain-linking is used. Estimates will be linked together using a three month average as opposed to a single month, in order to reduce volatility.
Method of weighting
For each year, relative proportions (in other words weights for sales values of each industry within all retailing) are calculated using actual data from the Retail Sales Inquiry. Commodity weights from the annual business survey (ABS) are used as a benchmark to be used in conjunction with monthly Industry weights from consumer price indices (CPI) to produce a constant price index.
Planned changes in production methods
None
Type of source
Statistical survey
Frame on which the source is based
Interdepartmental business register (IDBR).
Sample or census
Sample
Criteria for stratification
Stratified sample by activity into different employment strata.
Threshold values and percentages
Approximately 5,000 units are sampled each month. There is full enumeration for units with at least 100 persons employed and those between 0 to 99 persons employed and over GBP 60 million turnover a year. A random sample is taken of units employing 0-4, 5-9, and 10-99 employees with less than GBP 60 million turnover a year.
Frequency of updating the sample
New sample every year in January and is maintained on a monthly basis via gains and losses.
Monthly
Timeliness
The retail sales index is disseminated around 19 days after the end of the reference period.
Timetable of data collection
Questionnaires are sent out three days before the end of the reference period. Closedown for the provisional month is ten working days after the end of the reference period. Late forms are included in subsequent result runs.
ONS follow the A (ideal) method for the compilation of these statistics to as described by the Short Term Statistics Regulation - no other European country follows this method.
Comparable time series are available going back to 1988 for the following headline aggregates and their sub-sector series:
• All retailing excluding automotive fuel;
• Predominantly food stores;
• Predominantly non-food stores; and
• Non-store retailing.
For the following aggregates, comparable time series are available going back to 1996:
