UK Statistical System Code of Practice Office for Statistics Regulation
UK Statistics AuthorityPublicationsAPCP – Technical Panel – Meeting of 8 July 2022

APCP – Technical Panel – Meeting of 8 July 2022

Published
1 September 2022
Last updated on
12 September 2022

Members in attendance

  • Mr Grant Fitzner (ONS)
  • Dr Martin Weale
  • Mr Rupert de Vincent-Humphreys
  • Prof Paul Smith
  • Dr Jens Mehrhoff
  • Prof. Ian Crawford
  • Mr Peter Levell

Secretariat

  • Mr Chris Payne (ONS)
  • Mr Huw Pierce (ONS)

ONS Contributors

  • Mr Ben Hillman (ONS)
  • Ms Joanna Corless (ONS)
  • Ms Helen Sands (ONS)
  • Ms Emma Halshaw (ONS)
  • Ms Sofia Poni (ONS)

Observers

  • Mr Matt Corder (ONS)
  • Ms Abi Casey (ONS)

Apologies

  • Ms Corinne Becker Vermeulen
  • Prof. Guy Nason
  • Prof. Bert Balk
  • Mr Mike Hardie (ONS)
  • Mr Liam Greenhough (ONS)

1. Introduction and apologies

  1. Mr Fitzner opened the meeting and passed on apologies from members unable to attend.
  2. Mr Payne confirmed the position of any outstanding actions.

2. Terms of Reference for the APCP-T

  1. Mr Fitzner invited comment from the panel on the relationship between APCP-T and NSCASE. A panel member belonging to both bodies reported that the initial view of NSCASE would be that it would only examine issues with consumer prices after they were first
    considered by the Technical or Stakeholder advisory panels. However, as an independent body NSCASE may also refer issues raised with it to the panels.
  2. The panel discussed expanding the remit of APCP-T to cover other price domains – for example output prices and deflators. The panel were broadly in favour, and Mr Fitzner proposed that the current terms of reference should be updated to reflect this.
  3. Panel members asked if discussions from APCP-S could be relayed back to APCP-T more fully. Mr Fitzner proposed sharing the agenda from APCP-S meetings with Technical Panel members in advance, alongside any papers subject to their confidentiality status. If any
    topics from an APCP-S meeting were of particular interest a readout could be provided. Minutes could also be provided in draft form.
  4. A panel member queried the requirement for comments in the minutes to be attributed to named individuals. Mr Fitzner had no objection in principle but preferred to align with practice of other advisory committees, stressing that the purpose of the minutes was to
    capture the extent of consensus on options and recommendations.
  5. It was agreed that the three-year review process for appointments to the panel would be formalised and letters sent to members.
  6. Mr Fitzner thanked the panel for their suggestions. They will be incorporated into a new draft that will be circulated to panel members for their feedback.

ACTION: APCP-S Secretariat to share agenda, papers, and minutes of future meetings with APCPT members
ACTION: APCP-T Secretariat to update Terms of Reference and circulate to panel members.
ACTION: APCP-T Secretariat to formalise the three-year review process for panel membership

3. Multilateral Index Methods: Introducing the GEKS-Törnqvist

  1. Mr Hillman described the user engagement materials under development relating to the GEKS-Törnqvist method. These have been created at the suggestion of both the Technical and Stakeholder panels to raise understanding of the method among the wider user
    community.
  2. Several panel members commented on the level of assumed knowledge for the reader and made suggestions for additions to the glossary. An explainer document published by the Australian Bureau of Statistics was recommended for reference. Specific suggestions for the body of the article were:
    1. explaining the GEKS method at the start of the article
    2. including a section on how multilateral indices can sometimes produce counterintuitive results.
    3. including a wider range of motivations for using multilateral indices besides handling missing items.
  3. Mr Fitzner proposed creating two resources, a short summary for users less familiar with traditional index number methods and a longer article for a more technical audience.
  4. Mr Hillman outlined some issues created by using a 25-month mean splice, chiefly the requirement for back-data covering the 25 months prior to introducing a new item. The panel were invited to evaluate options for addressing this: delaying introduction until 25
    months had elapsed, using a 13-month window for year 2 or using an expanding window. A panel member asked if any empirical research had been carried out for these options. Experience from other countries would suggest it is unwise to rush a new product group into production, however an individual product presents less risk.
  5. A panel member noted the requirement for new products only to be introduced at scheduled basket updates, meaning that some delay would be inevitable. Another argued that any of the options under consideration would perform better than the current arrangement for new products under manual collection.
  6. A further point was made, that a consistent approach to weighting retailers would mean that switching from manual to scanner data would not impact the overall index

4. Research Indices Using Scanner Data: Progress Report

  1. Ms Corless gave a progress update on research indices calculated using scanner data, and outlined some considerations raised through this work: calculating the average price of a product with reference to different store types and relaunch linking.
  2. The panel first considered the topics of average price calculation and store type stratification. The panel raised queries about the approaches to weighting and aggregating strata, the spread of growth rates and extent of substitution across different store types,
    and the distribution of product shares across store types over time. The Eurostat Guidance on Multilateral Methods in the HICP was referenced.
  3. Panel members shared views on the influence of the chosen index number method. One member observed that a key property of multilateral index methods was mitigating the effect of product churn, and therefore churn should have less influence over the decision
    on stratification approach. Another suggested that fixing weights at too high a level in the hierarchy would detract from the value of a superlative index method.
  4. There was no consensus on the best approach to outlet stratification, a theoretical preference for averaging at the lowest level (by outlet) was contrasted with the practical advantages of averaging by store type. There was support for ONS carrying out further
    analysis at the outlet level to determine the best approach.
  5. Comparing the results for the retailers presented, one member queried if the grouping of stores for one retailer may be incorrect. They suggested using a clustering approach to differentiate between store types where the retailer did not provide this information.
  6. The panel then discussed product relaunches. A member shared experiences of working with a third-party data provider that publishes a database of self-identified product relaunches.
  7. Responding to a query about the quality and scale of relaunch linking within the data, Ms Halshaw and Mr Hillman confirmed the approach and outcomes of ONS’ investigation. They reiterated that all relaunch linking is confirmed manually in the proposed process.
  8. A panel member expressed interest in results presented in levels rather than index values to investigate discrepancies between regions, store types and outlets.
  9. Ms Sands explained how relaunches can be identified using barcodes even if the unique product identifier is carried forward to the new product by the retailer. Ms Corless described how a comparison between retailer approaches could be performed by using the barcodes of common products.
  10. The panel and Ms Corless discussed the prevalence and treatment of multibuy discounts, referring to previous ONS work on the topic.
  11. The panel discussed the implications of aggregating regional indices to give a national figure and whether this creates inconsistences with other approaches to disaggregation (e.g. by income decile), suggesting that the choice would depend on which levels of aggregation were of most interest.

5. AOB and date of next meeting

  1. Mr Fitzner advised that the GEKS-Törnqvist explainer material will be published following further redrafting. The scanner data research paper was produced for the panel’s information only, however aspects of the analysis within may be published separately in
    future.
  2. The panel discussed ONS’ outreach work in respect of public understanding of price statistics, several solutions for online presentations were mentioned. Ms Poni described some ongoing preliminary work in ONS exploring opportunities for public facing and cross
    government knowledge sharing.
  3. The next meeting will be held on Friday 7th October 2022.

6. Actions

  1. APCP-S Secretariat to share agenda, papers, and minutes of future meetings with APCP-T members.
  2. APCP-T Secretariat to update Terms of Reference and circulate to panel members
  3. APCP-T Secretariat to formalise the three-year review process for panel membership
  4. Panel members to contact the Secretariat if they are interested in contributing to ONS outreach work.