On Friday I attended the Retail Price Index “improvement” consultation which was held in Westminster. I wish today to give readers my feeling about the meeting which was well attended (approximately 80 people) and led to a fascinating and wide-ranging debate. Let me first explain why we were there and the best way is to use the original announcement from the National Statistician.
The National Statistician will consult on a number of possible options for improving the Retail Prices Index (RPI).
As a result of work to understand the reasons for the differences between the RPI and the Consumer Prices Index (CPI) estimates of inflation, the National Statistician is to invite users’ views on the way the RPI is constructed. The differences between the RPI and CPI under consideration are those caused when different formulae are used to calculate average prices where there is no information about precise expenditure. This is known as the ‘formula effect gap’.
If you had left the meeting and gone out into the street and asked people what inflation was you would be likely to get a range of views above all the current inflation measures. It is therefore odd I pointed out that we are only reviewing the measure which tends to give the highest reading! After all on such analysis it looks the most accurate.
The Royal Statistical Society is not happy either
The President of the Royal Statistical Society wrote to express it dissatisfaction with what has been proposed. Firstly on the grounds that the consultation period should be longer.
The potential changes could have a direct and material effect on a substantial number of people and organisations,,,,,,, so we regret that the timescale is less than the full 12 weeks.
So we have for some reason and truncated time frame and we also have serious questions about the scope of the consultation. The emphasis here is mine.
The other concern is the apparent limits of the consultation. The information published so far states that only the RPI formulae are under consideration, not those in the CPI, and that only the choice of formulae is being considered. The Jevons formula extensively used in the CPI is not without its critics as well as its supporters; even within the constraints of HICP rules it is not the only option and these constraints will not apply to the new “CPIH”.
Apologies for the acronyms where HICP is CPI and CPIH is the new proposed measure including a type of housing price measure which I discussed on the 24th of September. If they had set out to make them look confusing they would struggle to do a better (worse) job!
The section I have highlighted is important as I am sure it seems odd to you that if you have a difference between two measures you only review one of them! There has been something of a public relations push from the Office for National Statistics and the National Statistician to attack the Carli index used as a 27% component of the RPI but it seems to have slipped their mind that there are also doubts about the Jeevons index which represents some 63% of the CPI. We should be aiming to do what the Royal Statistical Society suggest below.
the work on the formula effect continues so that ultimately we do arrive at the best possible conclusion for both indices.
Note the “both” not just the one which tends to give the higher inflation reading.
The meeting itself
I do not wish to get too bogged down in the statistics but there were some fascinating interventions of the subject of the RPI/CPI gap which has been of the order of 0.9%. A very powerful intervention came from a gentleman who quoted the Office for National Statistics own 2010 technical manual.The emphasis is mine.
In the case of AR (using an arithmetic mean), it can be shown that in certain circumstances its use, when combined with chain-linking of the within-year indices, introduces a small upward bias in the overall price index. This phenomenon is known as ‘price bounce’ or ‘formula bias’.
This was powerful because in the context of that document a small upward bias would be 0.1% and now suddenly we were supposed to believe two years later that it was 0.9%. So what about the other 0.8%? To which apart from shuffling in their seats a little uncomfortably there came no response from the representatives of the Office for National Statistics or the National Statistician. But the driving force behind the logic of this position would be that we are in a “position of ignorance” as to whether RPI over-measures inflation or CPI under measures it.
It was also argued that due to its greater sensitivity the much maligned Carli index used in the RPI does pick up price changes which the ones used in the CPI do not.
There was also an intriguing intervention from an official from the German Bundesbank who speaking in a private capacity argued that the use of a laspeyres index such as the CPI introduces biases at the elementary level.
Methinks the lady doth protest too much
The representatives of the ONS and the National Statistician kept repeating that they were independent of government so much so that when I quoted from Queen Gertrude in the play Hamlet I got applause and quite a few “hear,hears” from members of the audience who were no doubt thinking along similar lines.
What about the Consumer Price Index?
The official bodies continually pressed the prominence of the CPI with the ONS representative telling us this.
(it) rapidly established a position of prominence
He forgot to point out that it got it from being used as the official measure which is not quite the same thing. An effective riposte to this came from the audience who stated that we had been provided with the uses of the RPI but not the CPI which was odd from something of such “prominence”. They were not deterred in any way by this and informed us that CPI
was widely acknowledged to be state of the art
He did not tell us by whom and apart from himself there were no other supporters for this line.
What does “improve” mean?
A challenge was made on the basis that the use of the word “improve” implied a moral judgement and would the National Statistician be kind enough to explain what it was? We were told that no moral judgement was meant. To which an individual quoted the National Statistician’s own code of practice.
The Authority’s overall objective is to promote and safeguard the production and publication of official statistics that serve the public good.
Unfortunately our official bodies had plainly contradicted themselves as “the public good” clearly suggests some form of moral judgement.
What happened to the period 1996-2010 when it was estimated that clothing price inflation was under measured?
There was no response to this.
A likely legal challenge
It was pointed out that because of the large distributional effects that would be caused by this move it would be very likely that the UK government would be sued in the courts. This would be expensive and it is by no means certain it would win as the 2010 Equality Act could intervene. Over the weekend since the meeting this has been brought further into focus by Argentina losing in the law courts to Elliots Associates.
Groups that could sue would be index-linked government bond holders, those with inflation linked pensions,those with inflation linked savings, and those with inflation linked annuities. In short we are discussing those who have long-term contracts finding that the terms have been changed on them “mid-term” so to speak.
I hope that I have given you a good flavour of the discussion. If it feels one-sided then it was. There was one brief show of support for some of the moves from a representative of the Bank of Nova Scotia but that was it. Also just be clear all responsibility for this article is mine as it is based on my memories and the notes I made.
A separate issue was mentioned which was a change in collection of the values of private rents was going to be made. I made the point that this was admitting some of the criticisms made of the rental equivalence method of measuring house prices. How can something get approval when the underlying data is so unsure? In short, in my opinion the various changes either mooted or in the process of being made fall a long way short of the “public good” mentioned in the National Statistician’s Code of Practice. Sooner or later the use of Rental Equivalence will be completely discredited. Just at the point we most need reliable unbiased statistics we seem to be actually moving away from them.