I spoke at an event about the Retail Prices Index (RPI) last week and made three points – that there is a misunderstanding about the formula effect, ONS is too influenced by economists’ ‘group think’ and weaknesses in governance. These can all be resolved easily, returning RPI to full use, if ONS and UKSA wants to. It was widely agreed that “the mess” had to be sorted out, and as the RPI cannot be killed off some modest changes to it are required.
The meeting was held at the RSS in London, on 13 June, to discuss the future of the RPI. (I wrote a preview of it last month.) The meeting was prompted in part by the publication of the “shortcomings” paper by ONS in March. (“Shortcomings of the Retail Prices Index as a measure of inflation“, 8 March). I, and a good number of experts, saw it it as a deeply flawed one-sided paper, unworthy of an independent statistics office.
The paper said: “In 2013, the RPI lost its status as a National Statistic. Our position on the RPI is clear: we do not think it is a good measure of inflation and discourage its use. There are other, better measures available ……. that the problems with RPI are many ……. RPI does not have the potential to become a good measure of inflation“.
The National Statistician opened the meeting by setting out his views which broadly reflected the “shortcomings” paper. There was then a chance for about a dozen respondents to speak and each had just five minutes. I made three points.
1 – Formulae
The prevailing view that Jevons, the geometric mean embedded in the CPI, is perfect, is simply not true. Two examples show that Jevons (too, along with Carli, the arithmetic mean) can produce odd figures:
- The CPI’s measurement of clothing pre 2010 was terrible. Over two decades to 2011, UK clothing prices apparently more than halved, when the CPI as a whole was up by more than a half. The CPI clothing component was more “wrong” relative to aggregate CPI pre-2010 than the RPI clothing is now, post-2010 – and for much longer, and it was a larger weight. So Jevons caused more damage to the measurement of UK’s clothing than Carli has.
- Rates of clothing inflation in many EU countries look implausible. Over the last decade Polish and Bulgarian aggregate CPIs showed almost identical growth yet one has clothing falling by 37% and the other rising by 17%. Lithuania clothing fell by 12%, Estonia rose by 41%. The Irish numbers are simply extraordinary. These are divergent, nonsense trends, despite using Jevons.
Jevons is not blamed for these trends, so why is Carli the reason for RPI’s problems now? It is the price collection of clothing that is the problem.
2 – Bias and group think in the ONS
ONS seems unable or unwilling to appreciate that:
- Inflation is a latent variable, and
- There are different uses and users of inflation indices.
The true rate of inflation is unknown yet ONS states stridently that the RPI is wrong and CPI is good. The evidence is, in contrast, that many experience inflation closer to RPI than CPI.
The ONS dismisses RPI in the way that a group of economists producing economic statistics for economists would. Yet, a responsible national statistical institute should realise that no one price index fulfils any one of the many possible purposes perfectly. No one index is perfect for all purposes. Household or pensioner experience of inflation is conceptually and numerically different to macroeconomic inflation. The differences in construction between RPI and CPI are not, as ONS says, problems to resolve, they have evolved over time, are there by design and justified.
3 – Politics and independence
ONS statisticians are statutorily independent but are perhaps more influenced by their political masters than they could be:
- Apart from the secretive “tetra-partite group” (quarterly meetings of the ONS, Treasury, BoE and OBR), attendees from the ONS, Bank and Treasury often outnumber others at advisory panel meetings. In contrast, the RPIAC in the 90s had only four representatives out of 21 from those bodies.
- The evidence given to the House of Lords Economic Affairs Committee (recording of the session is here, on 19 June) suggested that statisticians are fearful of changing the RPI in case it leads to fiscal costs from the redemption of index-linked gilts.
Statisticians should do the statistics and leave the politics to others. ONS needs to produce the RPI and produce fit for purpose official statistics – and not to worry about the gilt market. This is due process, as set out in legislation and practice.
Looking to the future, I think the ONS must:
- prioritise research into clothing prices collection, and, where necessary, change how it’s done. Start with the now infamous strappy tops which are just 1/30th of one percent of the index.
- publish a consumer prices development plan for all three measures, produced in the open.
This would be modest evolution: small, incremental iterations run alongside many other changes. This means that:
- the public and financial markets will know what’s coming.
- the Bank will not call any such change fundamental or materially detrimental
- the Chancellor won’t have to veto the change.
- even if he does, there are only three stocks that could be redeemed and, because the market is high, no one will take him up on the offer.
Any adverse reaction in financial markets would be very modest.
I would set UKSA the challenge of producing three families of inflation measures – CPI, RPI and the new HCI – fit for purpose in under five years.