Dear Ed,

Thank you for your recent letter in response to our comments on the OSR’s reviews of the Winter CIS and other issues. Although I accept that there has been some progress in demonstrating OSR’s independence from the ONS, I believe there is further to go before Merle and others are likely to accept the OSR as a truly independent regulator. For example, you have asserted that it is important for the OSR to provide encouragement to producers, which seems to me to confuse training with regulation. Possibly, I have failed to convey BSC’s concerns clearly and I will write further on this subject in due course.

Presently, BSC are more concerned with the issues associated with the measurement of inflation, particularly because the Assembly of the 22nd January 2025 does not include any reference to this important subject.

You may recall that we wrote to you on the 14th December 2023, following the discussions on the HCI’s on 12th December 2023, hosted by the Royal Statistical Society in collaboration with the Resolution Foundation. That meeting had highlighted a recent correction for inaccurate rental data, re-igniting the controversy created by the use of a fabricated ‘rental equivalence’ measure to reflect the increased cost of housing as experienced by owner occupiers. Our letter had therefore suggested that accreditation should be removed from CPIH until such time as a more appropriate measure should be adopted. Subsequent events are such that we now consider that accredited status should be removed from the CPI as well as the CPIH and that both should be redefined as statistics under development. I provide below our reasons for this suggestion, which may be considered unwise for the CPI because of its use as the target for the Bank of England’s responsibilities to control inflation. However, we would argue that the continuing use of the RPI to uprate some prices, such as rail fares, provides a sufficient precedence for the Bank to continue to use the CPI, whether accredited or not, bearing in mind that accreditation is the single most important function of your office in support of our National Statistics. Primarily, our objection to these macro-economic measures is due to the use of lagged elements, such as GEKS-T, to measure contemporary inflation rates, a practice we consider contrary to the expectations of the public.

For your information it was the Consultation on the Reform to Retail Prices Index (RPI) Methodology in 2020 that first drew BSC’s attention to the variety of different price indices, a factor that had contributed to our formation as a Community Interest Company in February of 2021. Subsequently, we have reflected the importance of inflation by devoting more attention to it than to any other topic; organising 3 seminars; Inflation Seminar 2022, 2023 Inflation Seminar and 2024 Inflation Seminar – Are we using the right measures? These meetings have provided valuable insights from over 20 speakers drawn from a wide variety of professions including journalists, businessmen, economists and statisticians. We have also benefitted from two public opinion studies provided pro bono by Opinium Research Ltd, examining the public’s experience of the effects of inflation and the public’s awareness of the various indices used to measure the phenomenon. The results of these studies and the videos of the seminars are all available on the BSC website, as are other relevant documents.

In summary there is evidence of considerable confusion as to the purpose of the various measures and of the methods employed in their creation. In particular we identify the following primary concerns:

1. There is a lack of public understanding as to what measures of inflation are used to control / influence various aspects of people’s lives.
2. Even amongst persons with experience of these issues, there remains considerable disagreement as to the detailed methods to be applied and which measure(s) are most appropriate to use for which purpose.
3. Anyway, the primary measures presently in use and proposed for use as accredited statistics are flawed in that they are not based solely on recent evidence of changes in pricing but also include lagged data from previous periods.

It has, however, been the OSR’s recent review of PIPR that has most clearly evidenced that accreditation should now be removed from CPI and CPIH. The OSR’s report of that review explains that the ONS had developed PIPR to overcome some limitations of its previous private rental sector statistics as identified in the UK Statistics Authority (UKSA)’s Systemic Review on Housing and Planning Statistics in 2017. As a result, PIPR has now replaced ONS’s Index of Private Housing Rental Prices (IPHRP) and Private Rental Market Statistics (PRMS). Commenting that “An important use of the new PIPR estimates is for estimating the owner occupiers’ housing costs (OOH) element of the Consumer Prices Index including OOH (CPIH)”. We consider this latter statement itself provides sufficient evidence to merit removal of the status of accreditation from CPIH, although the position with CPI may be less clear cut. However, BSC believe that the inclusion of data from previous periods to estimate current rental values for the housing cost element of the CPI also invalidates that measure.

It may also be worth noting that both the Stakeholder and Technical Committees had attended a jointly held teleconference on 27th October 2023, where a presentation of an Impact analysis and readiness assessment of private rental prices was given by Ms. Natalie Jones of the ONS. The minutes of that teleconference were not published by UKSA until 28th June 2024, by which time the OSR had already started the review of PIPR. However Ms. Jones’ presentation, as provided to last October’s meeting, has not been published, a fact that had hampered contributions to the OSR’s review. This point is noted in the OSR’s report of their review of the PIPR which has recommended that the details provided to the ACPC committees should now be published, as should other details previously withheld because they were claimed to contain time sensitive information. The OSR’s review also makes it clear that there is significant further work required before the PIPR can be considered for accreditation.

This recommendation does, of course, leave UKSA without any accredited index for measuring inflation, a position that we consider undesirable and our letter to the OSR of December 2023 had asked if there were plans to consider the HCI’s for accreditation because of the extensive stakeholder support for that possibility and because the HCI’s are a clear example of the better statistics we seek. We certainly believe they should be produced each month for comparison with the other inflation indices so that users can become more familiar with them and to use them as they may consider appropriate.

Meanwhile the HCI’s are not believed to be ready for accreditation and we therefore suggest that it is time to re-consider the status of the RPI. From our own review of the evidence of the past few years we have been unable to trace the reasons for selecting CPIH as a replacement for the RPI, particularly in the light of some of the controversies mentioned above. We understand, however that there had been a concern with the RPI resulting from its sensitivity to reduced mortgage rates following the economic crash of 2008 / 2009 and the results of amending the level for some items of clothing in January of 2010.

Amongst other things, work in 2010 had highlighted that the use of the Carli formula gave a significantly higher Inflation rate for the RPI as compared to the use of Jevons for the CPI a result which was considered to be unacceptable. However, the decision to use Jevons rather than Carli fails the important test of maximising the public’s understanding of how changes in prices are measured and BSC believe that statistics must be understood by the public if they are to serve the public good. We also believe that measurement of the inflation rate should be readily understood by the public, because it should reflect their own lived experience. Clearly, that is through direct experience of changes in prices as expressed by the Carli formula and not through some derivative of those changes as with Jevons.

Aside from the controversy relating to the formula, it is very difficult to identify any serious commentary that demonstrates the RPI as being ‘flawed’. Indeed at the meeting of the Technical APCP held on 8th November 2019 a panel member disagreed with the description of a “methodological flaw” in RPI, arguing that the mismatch between the data collection and index construction methodologies, did not represent a fundamental flaw with the methodology.

Further, there has never been a full analysis of the 887 responses to the 2020 consultation, which are available for downloading as ZIP Files of responses, with a short report on the consultation also available in the same link. This evidences that, in the main, response to the consultation was not supportive of the proposal to redefine the RPI to use the methods of the CPIH; a move that the Chief Executive of the Royal Statistical Society had characterised as “The … plan to replace RPI with CPIH is a clear case of using the wrong tool for the job”. Moreover the APCP Stakeholder Panel meeting of 31st January 2020 was also not fully supportive with minute 2.1.iv stating: “Some Panel members opposed the UKSA’s proposal to bring the data sources and methods of CPIH into RPI, especially since the HLEAC report seemed to offer a compromise. Other Panel members supported the move.” Whilst one member expressing the argument that “CPIH is an ‘economic measure’ where RPI is directly meaningful for households and consumers. It is also important to have a measure that properly understands inflation as experienced by households.” By contrast the APCP Technical Committee was largely supportive of the proposals, although the minutes of the panel meeting held on 8th November 2019 has John Astin’s opposition recorded as “Mr John Astin stated that he disagreed with the plans to bring CPIH methods into RPI and did not have any intention of participating in this process.”

Reading the minutes of these meetings and other background documents it is difficult to identify any specific role the Office for Statistics Regulation has had in this process. As far as we can tell there has been no review of any of these measures since 2017, moreover it is difficult to understand what influence the advisory committees have had. Whilst we recognise that those committees are not responsible for the decisions made concerning the usage of statistics, whether accredited or not, one must assume that they are there to assist UKSA in the development and evaluation of public policy as required by the 2007 Act. The decision to change the definition of the RPI from measuring inflation as experienced by the average household to a macro-economic (whole economy) measure has not been effectively communicated to the public and we consider that the Authority is therefore in breach of its responsibilities under section 7.2 of the Statistics and Registration Service Act of 2007, clauses 1 and 2 of which state:

1. In the exercise of its functions under sections 8 to 21 the Board is to have the objective of promoting and safeguarding the production and publication of official statistics that serve the public good.
2. In subsection (1) the reference to serving the public good includes in particular—
   1. informing the public about social and economic matters, and
   2. assisting in the development and evaluation of public policy

The decision is therefore in breach of the Trust pillar of the code of Conduct and should be called into question by the OSR accordingly.

It is also arguable that the decision of the coalition government in 2010 to replace the RPI with the CPI to upgrade benefits has made a major contribution to many of the subsequent social difficulties experienced, including the increased use of foodbanks in recent years. As our present chancellor said at the time of the coalition government’s decision:

“It is disingenuous as well to argue that CPI is a better measure of inflation than RPI for those on benefits. Those in that group spend more on food and fuel, so the average inflation is higher, not lower, than either RPI or CPI.” Rachel Reeves, December 2010

Whilst the influence of that decision on so many subsequent matters can only be surmised, it is a fact that most other countries, in the EU and elsewhere, do not employ a macroeconomic measure as their measure of inflation as it affects the average household. Why should we?

We therefore consider it perfectly legitimate to reconsider the RPI as an accredited statistic, particularly considering the range of data sources now available – a factor that should obviate any repeat of difficulties generated by inadequate source data.

Hopefully, the OSR will now exercise its independence from the production process by undertaking a thorough review of the whole issue of inflation measurement for the public good. There can be no more important issue for regulatory consideration, not only because of the recent experiences of inflation but also because of the failure to bring the issue under control; with the likely consequence of further policy errors resulting from flawed methods.

Yours sincerely,

Tony Dent

On behalf of Better Statistics CIC

CC to Kate Barker, Chair APCP Stakeholder Committee: Kate.Barker@BTinternet.com

& Grant Fitzner, Chair APCP Technical Committee: Grant.Fitzner@ons.gov.uk

Related links: Ed Humpherson to Tony Dent: Consumer Prices Accreditation Status – Office for Statistics Regulation