Table of contents
In our Transforming short-term turnover statistics: October 2017 article, we published a proposed programme of transformation to deliver improvements to the UK’s economic statistics. For retail sales, the focus of early research and development was:
the potential to redesign existing methods and systems
to replace some direct data collection with use of administrative data
to expand the coverage of retail sales
to move to regularised reporting periods
sampling through the Statistical Business Register
Section 2 discusses each of these areas in more detail.Back to table of contents
2. Research and development
Data collection systems and methods
Starting with our Retail sales, Great Britain: February 2022 bulletin, we transferred production of the statistics to a new processing platform, which will provide a more efficient and sustainable way of working. As part of this process, we also reviewed and improved our methods. This included improvements to the imputation of the mail order industry in internet sales and the adoption of an outlier approach called winsorisation across all variables. These improvements are explained further in the February 2022 release.
Use of administrative data
In the context of administrative data, the proposal was to explore the potential use of HM Revenue and Customs (HMRC) Value Added Tax (VAT) data in retail sales. This built on work to integrate aggregate VAT data into the National Accounts.
The monthly retail sales bulletin is published in a timely manner, around 20 days after the reference period, and is an important early economic indicator.
As explained in Transforming short-term turnover statistics: October 2017, a substantial amount of VAT data are unavailable when we process Month 1 (t+1) and Month 2 (t+2) retail estimates and it is not until Month 3 that a high proportion of VAT returns are available. We have therefore been working with the Economic Statistics Centre of Excellence to research methods that estimate for the most recent periods to explore the potential use of VAT data in initial retail sales estimates.
The Nowcasting in the presence of large measurement errors and revisions publication by Labone and Weale (March 2022) tested various approaches. However, it was found that the VAT-based nowcasts of early estimates tended to be revised downwards as more data became available, with revisions substantially larger than those seen from survey data-derived first estimates.
This work suggests that there is currently limited value in using VAT data in early retail sales estimates, given the bias introduced and the impact on revisions. Also, internet sales, which have increased substantially since the coronavirus (COVID-19) pandemic and provide an important measure of consumer behaviour, are not available from VAT data. A move away from survey data would reduce the accuracy of internet sales statistics.
VAT data could potentially be used to supplement survey returns when they are at a more mature stage. The benefit of using VAT would be an improvement to the robustness of estimates (reducing sample error) and the potential for further disaggregated data, such as sub-national estimates. However, at present this would only be possible several months after the reference period was first published. We will continue to consider the cost versus benefit of this approach and will provide an update to users by the end of 2022.
It was our intention to extend the geographic coverage of retail sales to include Northern Ireland enterprises, creating a UK index. The Northern Ireland Statistics and Research Agency (NISRA) have since developed a retail sales index for Northern Ireland, first published on 16 March 2022. We will work with NISRA to understand how we might incorporate the NISRA estimates into our release to allow UK reporting. In the interim, a link to the Northern Ireland estimates is available within our monthly retail release.
There were also plans to move from the production of a retail sales index to a retail trade index, combining the retail sales, wholesale and retail trade of motor vehicles sectors. This change in output coverage was to fulfil the legislative requirement of the Eurostat Framework Regulation Integrating Business Statistics (FRIBS). Following the UK’s decision to leave the EU, there is no longer a legislative requirement to make that change. We will consider this as part of longer-term developments, as users tell us there are currently more important priorities.
Retail Sales Index (RSI) sampling through the Statistical Business Register
The Statistical Business Register is still being developed. We will continue to feed into discussions on the development of the new Statistical Business Register, considering feedback received from users to inform requirements.
Calendar reporting period
Retail sales data are collected for standard trading periods that follow a four-week, four-week, five-week cycle with estimates published on an average weekly basis. We will engage with users on the need for monthly reporting periods and on the ability of responders to provide this. We will communicate next steps by the end of 2022.Back to table of contents
3. Future developments
Over the next few months, we will engage with users to gather feedback on the details provided in this article. We will ensure that we remain aware of user needs as well as securing a platform for providing progress updates.
We will engage with important government users through our existing stakeholder groups, and wider users’ feedback will be gathered through consultations linked to our monthly publications.
For any further detail or to provide user feedback please email: email@example.comBack to table of contents
Contact details for this Article
Telephone: +44 1633 455602