You want to know the difference between RPI and CPI – from an expert?

Sometimes twitter just says wow! This is one of those times!


Response to Department for Work and Pensions consultation on Security and Sustainability in Defined Benefit Pension Schemes


1. The Royal Statistical Society (RSS) is a learned society and professional body, with more than 8,000 members in the UK and across the world. We are one of the world’s leading organisations to promote the importance of statistics and data, and have done so since we were founded in 1834. As a charity, we advocate the key role of statistics and data in society, and we work to ensure that policy formulation and decision making are informed by evidence for the public good.

2. The Society wishes to comment on just one question, question 4h:

“Should the Government consider a statutory over-ride to allow schemes to move to a different index, provided that protection against inflation is maintained? Should this also be for revaluation as well as indexation?”

While the question does not specify which index or indices are to be over-ridden or used as a replacement it is obvious from current circumstances and the text in the consultation document that the intention is to allow schemes that are locked into RPI to move away from it.

3. We are not commenting on whether there should or should not be a statutory override since this decision requires expertise outside our remit. But we feel it appropriate to comment on the statistical qualities of the three main indices (RPI, CPI and CPIH) and their suitability, or lack of it, for uprating or revaluation in pension schemes. None of the three currently meets the need fully in our opinion. While the limitations with the RPI are well known (and sometimes exaggerated), those with the CPI (and by extension CPIH) are often overlooked.

4. An index used for this purpose should not just be technically adequate but should also be capable of enjoying a degree of public confidence. While public confidence is important for all statistics it is crucial for consumer price indices since through indexation they directly affect people’s income and expenditure and this is particularly important in this sensitive context.


5. The RPI is the only index of the three which was designed with uprating (compensation for inflation) purposes in mind as one of its key aims. Features of its design are appropriate for this need. This includes its exclusion of the top 4% of the population by income since this mitigates the problem found generally in consumer price indices where high spending households have greater weight in the index. Its treatment of insurance is also suitable. It also measures owner occupied housing in a way which relates better to actual household costs than CPI and CPIH. Partly for the above reasons and partly as it is long-established it appears to enjoy more public confidence than CPI or CPIH.

6. However the RPI’s use of the Carli index for some categories of expenditure at the first stage of aggregation has rendered it liable to overestimate inflation in certain circumstances. Changes to the methods of collecting clothing prices made in 2010 unfortunately gave rise to conditions in which Carli noticeably overestimates.

7. Technically it should be possible to make adjustments which would reverse the problem caused by the 2010 changes. However clauses in the prospectuses of index-linked gilts (ILGs) issued prior to July 2002 (one of which will be extant until 2030) state that if any change is made to the coverage or the basic calculation of the index which, in the opinion of the Bank of England, constitutes a fundamental change which would be materially detrimental to the interests of the stock-holders, holders can demand immediate redemption of the gilts. This would potentially have a significant adverse impact on public finances so the Statistics and Registration Service Act, 2007, states that such changes can only be made with the consent of the Chancellor of the Exchequer. Adjustments to reverse the 2010 problem seem likely to fall foul of these provisions and therefore cannot be made in practice.


8. The CPI, which is also the EU Harmonised Index for Consumer Prices (HICP) for the UK, was not compiled as a compensation index but for macroeconomic purposes. This was made very clear in the initial EU regulation setting out the framework for the HICP:

“Whereas there is a need for the Community and particularly its fiscal and monetary authorities to have regular and timely consumer price indexes for the purpose of providing comparisons of inflation in the macro-economic and international context as distinct from indexes for national and micro-economic purposes…” [1]

Accordingly it is fully weighted by overall consumer spending so that higher spending households have greater weight. Insurance is weighted only according to the difference between premiums and claims.

9. An acknowledged and serious disadvantage is that it excludes owner occupier housing costs. It uses the Jevons (geometric mean) formula widely at the first stage of aggregation. While more robust than the Carli it can underestimate in certain circumstances, albeit its underestimation tends to be much less severe than Carli’s overestimation. The use of an index using Jevons was originally justified in 2010 as it was said to proxy the way consumers switch towards items which rise less sharply in price but this argument has since been discredited. The use of an argument which was subsequently discredited and the potential for underestimation means that Jevons is regarded with suspicion in some quarters.

10. For the reasons in the preceding paragraphs, the CPI is unlikely to meet the public confidence test. In addition since, as the UK’s HICP, it will remain subject to EU rules, this could also be a politically sensitive issue. Finally, its use would be perceived as unfair given that pension funds will continue to receive RPI-related returns on ILGs.


11. CPIH is essentially the same index as CPI with the addition of owner occupied housing measured according to the rental equivalence theory and (now) council tax. The inclusion of council tax is a positive point for uprating purposes. However the use of rental equivalence (using rents of comparable properties) as a means of measuring owner occupier costs, while arguably justifiable in economic theory, does not always seem credible, particularly to the general public. The behaviour of CPIH so far, showing an inflation rate not very different to that of CPI, does not help. Like CPI it uses Jevons.

12 Thus CPIH could therefore also fail the public confidence test and would also not help the unfairness perception if pension funds were still enjoying RPI related returns on ILGs.

A longer term solution: the Household Costs Index (HCI)

13. In the longer term the solution should be the Household Costs Index currently being developed by the ONS. This will have the same aim as the RPI but should avoid its problems and be designed for the 21st century. It will also be available for different household groups so will offer the possibility of having different indices for pensioners and those of working age if desired. However, it will take some time for this index to be developed, for its feasibility and credibility to be accepted and for teething problems to be ironed out.


14. We urge that if a legislative override for pension schemes locked into RPI is proposed arrangements for the moment should be temporary allowing a switch to the HCI (or to anything else appropriate) once that becomes feasible. In the meantime the government might wish to consider a formula such as “RPI minus x”. The RSS would be happy to take part in any discussions about what “x” should be.


[1] Council Regulation (EC) no 2494/95 of 23 October 1995 concerning harmonized indices of consumer prices.

Submitted by RSS’s Policy and Research Manager, 12 May 20

About henry tapper

Founder of the Pension PlayPen,, partner of Stella, father of Olly . I am the Pension Plowman
This entry was posted in actuaries, pensions and tagged , , , , , , , , , . Bookmark the permalink.

7 Responses to You want to know the difference between RPI and CPI – from an expert?

  1. Phil Castle says:

    Very interesting/informative, especially the implication with RPI on certain Gilts and teh general knock on problems.

  2. David Quinn says:

    I have read a lot of comment over the last few years, including all the arguments on the RPI CPI User Group website, and the consensus would not accord with the statement that Jevons underestimates less than RPI overestimates.

    There are many indeed who would claim that Carli is fundamentally more accurate than Jevons.

    It is in fact very hard to justify Jevons at all without recourse to the economic theory i. e. Substitution, which this article says has been unproven. To reduce a complex argument to crude terms the RPICPI User Group consensus came out as RPI over reading by about 0.2% and CPI under reading by about 0.8% giving a formula effect of 1%.

    When Paul Johnson was presenting to the RPICPI Uservices Group on his study into the best measure he was told than many statisticians there thought Carli/RPI better than Jevons/CPI he said “Not the ones I talk to”.

    Since he was probably talking mainly to ONS this was a very glib dismissal and I have been wary of his pronouncements ever since.

    Finally, John Astin, who developed HCIP has said it was never meant to be an uprating index. So the conclusion that an HCI is required but will take some time is dead right and that in the meantime an RPI -x% should be put in place is a sensible compromise.

    In truth though no government will do it since CPI is a useful way of robbing most pensioners and those on benefits.

  3. Derek Benstead says:

    I disagree with paragraphs 6 and 9. My understanding of the changes to the collection of clothing data is that it resulted in a reduction in CPI, not an increase in RPI. I was going to write a paragraph disputing paras 6 and 9, but having got to the bottom of the page I see I don’t need to, David Quinn got there first. I support David’s comments.

  4. In some of my modelling of the effects of the change in the real value of state benefits, following the move from using RPI to CPI, I’ve had problems with benefits for homeowners. As I’m using future expected values from the OBR, where can I find HCI and CPIH figures for future estimates?

  5. Mike Otsuka says:

    Here’s a letter of mine that the THES published last year on the duplicity of UK University employers in their use of RPI versus CPI:

    Is that RPI or CPI?
    November 3, 2016

    In defence of its offer of a 1.1 per cent inflationary pay increase, our employer rejects the retail price index as a statistically unsound measure of inflation and presses for benchmarks against other measures of inflation that have shown more modest increases. But in its defence of an increase in tuition fees from £9,000 to £9,250, our employer claims that this is justified on grounds that it “is limited strictly to an inflationary rise of 2.8 per cent in 2017-18”, thereby “maintaining the income that universities receive in real terms”. Astonishingly, however, here our employer endorses an RPI-based measure of inflation, even though this measure (RPIX=RPI minus mortgage costs) employs the very methodology that it rejects as statistically unsound in pay negotiations.

    If an RPI-based hike in tuition fees is needed to maintain income in real terms, then so is an RPI-based increase in pay. The figure of 2.8 per cent was based on an Office for Budgetary Responsibility forecast for RPIX in 2017-18. RPI and RPIX are running at 2.0 per cent and 2.2 per cent, respectively, whereas the consumer price index and CPIH (=CPI plus housing costs) are running at 1.0 per cent and 1.2 per cent, respectively. Our employer is therefore offering us a rise in pay along the lines of CPI. If it is unwilling to increase this offer to RPI, on the grounds that RPI is a flawed measure of inflation, it will be a hypocrite if it does not decrease the 2017-18 increase in tuition fees to the OBR’s forecast of a 1.7 per cent rise in CPI for that period.

    Our employer can’t have it both ways: not RPI for me and CPI for thee.

    Michael Otsuka
    Professor of philosophy
    London School of Economics

    • David Quinn says:

      The government of course does the same thing – e.g. rail fares go up by RPI but pensions etch by CPI. It is a deliberate and barefaced con. I have been reading that John Pullinger, the National Statistician, has said the CPI (and CPIH) is an economic measure of inflation, and therefore by implication not suitable for uprating. THE ONS has not kept up with this yet and the government will resist it.

Leave a Reply