Many savers will be breathing a sigh of relief at the recent news the Office for National Statistics has decided to make no changes to the methodology for calculating the retail prices index (apart from a slight change to the way it calculates rental expenditure).
There were fears, following a consultation at the end of last year, that the official statistician, Jil Matheson, would decide to devalue the RPI to bring it closer in line with the consumer price index inflation measure.
The consultation started on the premise that the RPI must over-record inflation because it is normally higher than the CPI. The consultation presented various options which would have resulted in a lower figure for the RPI in future, but the ONS decided it would recommend making no changes to the statistical calculation.
This is most welcome, since many investors have their incomes tied to rises in the official measure of retail prices – any move to force the figure lower would have resulted in lower incomes for millions of people over time.
There must have been significant political pressure on the official statistician to agree the RPI should be lowered. Reducing the RPI would save the exchequer money, because the yields on index-linked gilts and national savings inflation-linked bonds are tied to the RPI, not the CPI.
In addition, many private sector employer pensions are tied to RPI inflation, so those pensioners would have received reduced incomes in future.
Furthermore, the government changed the measure of inflation used for uprating public sector and state pensions from the RPI to the CPI in April 2011, angering many pensioners. Therefore, bringing the official RPI measure of inflation into closer alignment with the CPI would have assuaged some of the concerns about lower pensions. As such, the ONS has publicly demonstrated its independence by deciding to stick with the current methodology for the RPI.
To have radically changed the traditional inflation measure, on which many so people’s incomes depend, could have jeopardised the inflation protection inherent in their income arrangements. It could also have caused problems in the index-linked gilts market, because investors could, in theory, have challenged the decision as a material change to their terms and demanded repayment.
There will, of course, be some downsides to the decision.
For example, some energy, transport and other price rises are linked to RPI, so they may be higher than they would be otherwise. However, it is very important the public has confidence in the official statistics and the inflation figures.
Of course, the reality is there are no truly ‘exact’ statistical measures of inflation – each measure is based on estimates and assumptions which are open to challenge.
That was one of the big problems inherent in the ONS consultation, because it started from the assumption the CPI inflation measure was ‘correct’. Using that logic, the fact the RPI was higher meant it was ‘wrong’.
However, the accuracy of the CPI calculation methods were not actually considered. This is a major reason why my own response called for no change to be made until further consideration had taken place.
In fact, it is not clear if the calculation methodology for the consumer price index is absolutely correct. Indeed, the responses to the consultation from many experts explained why using a geometric mean such as that in the CPI may under-record inflation in certain circumstances and why the RPI might sometimes be a better measure. In reality, of course, both could be flawed, but even then suggests reducing RPI to CPI levels would not be appropriate. In any case, the consultation responses overwhelmingly rejected the proposed changes to the RPI.
Two-thirds of the organisations who responded and nearly 87 per cent of the individuals supported the “no change” option, while only eight per cent of organisations and five per cent of individuals supported any of the other options which were proposed for adjusting the RPI calculation. Of course, the actual statistical methods are pretty complex, but the inflation number is so important that it should not be changed without strong reasons.
The technicalities revolve around whether a geometric mean (which calculates the average change in a basket of ‘n’ goods by taking the ‘nth’ root) or an arithmetic mean (which calculates the average as the sum of all the price changes divided by the number of goods in the basket) will most accurately reflect the actual change in prices for those goods.
There is no right answer to this, although it is true the international convention is to use geometric means.
The important thing to remember, however, is this measure is vital to vulnerable people’s incomes. The reason the RPI is used to uprate pensions and inflation-linked income products is to protect those incomes against future inflation and maintain their purchasing power. This is not just an esoteric argument about which statistical measure is most appropriate. If the inflation uprating is too low, people’s purchasing power will fall over time and they will not be able to afford the goods they need to buy.
Therefore, it is important to consider whether the CPI may under-record inflation, as well as whether RPI may over-record it. The consultation did not, however, do this.
It is certainly the case that geometric means such as the ‘Jervons’ measure used by RPI can under-estimate inflation in some circumstances, for example when the price rises of the goods in any basket are correlated, or if people are unable to substitute the goods they buy to find the cheapest alternatives, or if the lowest priced items in a particular basket are the ones which rise most in price.
Therefore, it could be that the CPI index under-records the inflation faced by consumers some of the time and that the RPI is a more accurate measure for many of those whose incomes we are trying to protect.
Often it is the poorest members of society who are unable to substitute as they are less able to find the lowest price goods each time they shop.
It is important to note that the national statistician has said she will continue looking into the accuracy of inflation statistics overall. That is most welcome.
She has also introduced a new measure of the RPI which does use the geometric mean – it will be called RPIJ (the ‘J’ refers to the Jevons geometric mean methodology which is used) and will be tracked over time alongside the traditional RPI, so we can gather more information about the relative advantages, disadvantages and trends in each measure.
Since these indexes are so vital for protecting many people’s future incomes against inflation, they need to reflect the actual price rises that consumers will face over time as well as possible.
If investors bought inflation-linked products in order to protect their future income in real terms and, therefore, ensure their purchasing power is protected, so the inflation measure used to calculate their income needs to reflect as accurately as possible the actual price rises consumers will face.
I am delighted the ONS will carry out more work to assess the accuracy of both the RPI and the CPI before making any radical changes that could affect the lives of millions.
Dr Ros Altmann is a former government policy adviser and pensions expert
Key oints
The Office for National Statistics has decided to make no changes to the methodology of calculating the retail prices index.
Consultation responses overwhelmingly rejected the proposed changes to the RPI.
The National Statistician has said she will be continuing the work to look into the accuracy of inflation statistics overall.