Lord Forsyth of Drumlean's speech on Measuring Inflation

The following is Lord Forsyth's speech in full:

"It may sound like a dry, technical report. But it is a revealing tale of how complacency on the part of the statistical authorities, and opportunism on the part of governments, has led to a situation where the use of inflation statistics is leaving many people worse off.

The Committee’s attention was drawn to problems with the calculation of the Retail Prices Index through a series of articles by Chris Giles in the Financial Times, a long-time campaigner on these issues.

We asked the Governor of the Bank of England about these problems during his annual evidence session with the Committee last year. He told us the RPI had “known errors”, should not be further embedded in Government contracts and that if there was anything the Committee could do to advance this process, we would be providing “a real service.”

I hope that we have risen to the Governor’s challenge.

We began our inquiry in June 2018 and reported in January this year. Before I discuss our findings I would like to thank the Committee staff who produced the report, Luke Hussey, Ben McNamee and Lucy Molloy.

Differences between RPI and CPI

Chapter 1 of the report describes the history of consumer price indices. There are two main measures of consumer price inflation in use in the UK: the Retail Prices Index, which was introduced in the 1950s, and the Consumer Prices Index, introduced in the 1990s following the Maastricht Treaty.

Both indices are based on the changes in price of a fixed basket of goods and services. But there are a number of differences between what the indices cover and how the price changes are calculated.

One of the main differences is how the two indices calculate the average price change for unweighted items in the basket, that is those items where the proportion of household spending on those items is not available for the level at which prices are collected.

I’ll give an example to illustrate. The surveys used to calculate the indices will reveal the proportion of household spending on potatoes as a class of goods. But they won’t reveal the spending split between different varieties of potato, say King Edward or Maris Piper.

This means the price changes of different varieties of potato—the price of King Edwards may rise by more than Maris Pipers say—cannot be weighted according to household spending. A method is therefore required to calculate the average price change when expenditure weights are not available.

The RPI does this for some items through the Carli formula, which uses the arithmetic mean. The CPI however largely relies on the Jevons formula, which uses the geometric mean.
Statisticians have been debating which is the preferred formula since the 19th century. We set out the arguments in Chapter 2 but decided, probably wisely, not to enter that debate.

This difference in how to calculate price changes for unweighted items leads to a difference in the inflation rate which each index produces. The gap between the CPI and RPI recorded inflation rate, which is attributable to this methodological difference, is referred to as the ‘formula effect’.

The clothing price change

In December 2009, the ONS estimated the formula effect was responsible for RPI recording the rate of inflation as being 0.54 percentage points above the rate recorded by CPI. But by December 2010, the difference had increased to 0.86 percentage points, a 0.32 increase.

What caused this substantial change?

In 2010 the ONS changed the way it collected prices for clothing. It increased the sample size of clothing it recorded prices for, relaxed the rules on what types of clothing were considered comparable and began collecting prices during the January sales.

This was believed to be a routine, methodological change. But it had a strange effect on the recorded price change of clothing in the RPI.

From 1987 to 2009, the average annual price change in women’s clothing as measured by the RPI was a 2.5 per cent decrease. But from 2010 to 2017, the average annual price change was an 11.1 per cent increase.

Something had clearly gone amiss. The ONS held its hands up and admitted it had made an error. Witnesses were agreed generally that the interaction of the Carli formula with the new method of price collection was to blame for the over-estimation of price rises.

Although the error also affected CPI, it affected RPI more. The ONS said the change was responsible for 0.30 percentage points of the 0.32 increase in the formula effect.

Effect of the error

This may sound very technical so far. But that error created real-life winners and losers.

Who won? Holders of index-linked gilts. These gilts are linked to RPI and the resulting 0.3 percentage point increase in the index led to an undeserved windfall—Chris Giles has estimated that the value of interest payments received by index-linked gilt holders has increased by around £1 billion each year.

And who lost? Commuters. Rail fare increases are linked to RPI. And students. The interest on student loans is linked to RPI.

The increased difference between inflation as recorded by RPI and CPI also encouraged governments to engage in the practice known as ‘index-shopping’. Benefits, tax thresholds and public sector and state pensions were all switched from being uprated by the higher RPI to the lower CPI in 2011.

Statutory duties of the statistical authorities

This brings us to the most surprising part of the story. The UK Statistics Authority, of which the ONS is its executive arm, has refused to correct the error, despite admitting it made a mistake.

Why not? As the correction of the error would likely reduce the rate of RPI inflation, this would adversely affect holders of index-linked gilts. The 2007 Act requires the Authority to obtain the consent of the Chancellor of the Exchequer to such a change.

Sir David Norgrove, the Authority’s chairman, told us the 2007 Act meant there was no point requesting the change as the Chancellor would just say no. Last week he wrote to me to say was unable to reply to our report after all this time as discussions with the Government continue.

The National Statistician, John Pullinger, who retired last week, suggested to the Committee that section 7 of the 2007 Act required him to take into account the interests of those who would be affected negatively by any change, such as index-linked gilt holders.

This is not the Committee’s interpretation of section 7.

Section 7(1) gives the Authority the objective of “promoting and safeguarding the production and publication of official statistics that serve the public good.”

Section 7(3) says the Authority should promote and safeguard “the quality of official statistics”, “good practice in relation to official statistics” and “the comprehensiveness of official statistics”.

Section 7(4) says that references to the quality of official statistics includes “their impartiality, accuracy and relevance” and “coherence with other statistics.”

The Committee’s reading of this is that the Authority is at risk of failing in its statutory duties by its refusal to attempt to correct the clothing error in RPI, an error which it openly admits.

It is not for the Authority to pre-empt the decision of the Chancellor, as its chairman suggested.

The Chief Secretary to the Treasury told us it was difficult for the Chancellor to say yes or no to a proposal he hadn’t received. And the Chancellor told us he was happy to hear from the Authority.

And the Committee was unconvinced by the National Statistician’s suggestion that he should take into account the interests of index-linked gilt holders when deciding whether to make a change. It is not clear from section 7 that this is a relevant consideration to be taken into account.

We believe the Authority is required by its statutory duties to attempt to fix the issue with clothing prices.

RPI as a legacy measure

The decision not to correct the error is part of a wider neglect of RPI by the statistical authorities. Following a review of inflation indices, the Authority removed National Statistics status from RPI in 2013 and now treats RPI as a ‘legacy measure’.

Following the recommendations of a 2015 review by Paul Johnson, it has resolved to make no further methodological improvements to RPI.

But this is a surprising stance given RPI remains in widespread use.

Paul Johnson told us he had changed his mind since his 2015 review. He said his recommendation on making no further improvements to RPI was predicated on RPI being phased out. He said that given this hadn’t happened, the Committee should ask the Authority to correct the RPI.

We therefore called for the UK Statistics Authority to resume its programme of periodic methodological improvements to RPI.

Towards a single measure of inflation

The Governor of the Bank of England, when he asked us to look into this, suggested that with three official measures of inflation—RPI, CPI and the CPIH, “it would be good to consolidate the focus onto one.”

We agree and believe that in the future there should be one measure of general inflation that is used by the Government for all purposes.

But to achieve this work is required on how best to capture owner-occupier housing costs in inflation indices. Witnesses criticised the approach of the RPI, which uses mortgage interest costs, and the approach of CPIH, which uses rental equivalence. CPI does not account for owner occupier housing costs (save for minor repairs).

We said the UK Statistics Authority, together with its stakeholder and technical advisory panels, and a consultation of a wide range of interested parties, should agree on a best method of capturing owner occupier housing costs.

Once a method has agreed, the Authority, again after consultation, should decide which index to recommend as the Government’s single general measure of inflation. We would like to see this adopted within five years.

Sir David Norgrove told us that RPI is “not a good measure of inflation” and “does not have the potential to become one”. We disagree and believe an improved RPI would be a viable candidate for the single general measure.

Index-shopping

A single general measure of inflation would prevent governments from index shopping. Table 2 on page 39 of the report shows how when the Government in making payments to the public, CPI is the index used to uprate payments. But when the public is making payments, RPI is used to uprate payments.

The latest example of this was when National Savings and Investments index-linked saving certificates were switched from RPI to CPI in May. It appeared to be a switch motivated by its favourability towards the Government, rather than a principled approach to uprating.

The present Government has however taken some steps to address the imbalance. Business rates were changed to be uprated by CPI rather RPI. And discussion have taken place around uprating rail fares by CPI instead of RPI.

A single general measure would remove the temptation to index shop. As the single general measure of inflation will take time to be implemented, the Government needs to take interim action to stop this unfair practice. It should switch to CPI for uprating purposes in all areas where it is not bound by contract to use RPI.

The exception to this recommendation is the interest rate on student loans. As recommended in our Treating Students Fairly report, this should be reduced to the ten year gilt rate so it reflects the Government’s cost of borrowing (it was disappointing that the Augar Review did not recommend a switch from RPI on student loans).

Index-linked gilts

This interim switch to CPI should also apply to new issuances of index-linked gilts. We heard evidence there was sufficient demand for CPI-linked gilts, and Ben Broadbent from the Bank of England dismissed concerns from the Debt Management Office that the existence of CPI and RPI-linked gilts would lead to market fragmentation.

We heard concerns about the effect a change to the calculation of RPI would have on existing index-linked gilts, the last of which is due to mature in 2068, private sector bonds and pension schemes.

A sudden change such as redefining RPI as CPI or CPIH, as some witnesses discussed, would be inappropriate. But once the single general measure of inflation is in place, the Government and the UK Statistics Authority should decide whether RPI, if it is not the chosen measure, should continue to be published in its existing form or whether a programme of adjustments should be made to RPI so that it converges on the single general measure.

To avoid disruption, any programme would take place gradually, over a sufficiently long time, to a plan that was clearly communicated at the outset.

Government response

That concludes our recommendations. The Spring Statement said that the Government would respond to the report by the end of April. The Chancellor wrote to me on 30 April to say that the issues raised in the report are “complex and wide-ranging. The breadth, complexity, and importance of the issues mean that your report requires further consideration.” He said the Government would respond to the Committee’s report “as soon as it is practicable to do so.”

This report cannot remain unanswered. It raises serious questions about decision-making by the statistical authorities. The Government and UK Statistics Authority need to address the challenges our report has highlighted.

I beg to move."