By issuing gilts linked to the Retail Prices Index (RPI) the Government exposes itself to inflation risks on interest payments. In this box, we looked at how changes to the formula for calculating RPI would affect our forecast.

The RPI is a long-standing measure of inflation that is used to uprate a variety of payments and taxes, notably including those related to index-linked gilts (ILGs). In 2010, a change in the collection method for clothing prices resulted in a significant increase in the gap between the RPI and CPI measures of inflation, prompting a recognition that the RPI embodied an unsuitable formula for aggregating individual price quotes.a Over the past four years, the ONS estimates that the effect of the unsuitable formula has raised annual RPI inflation by an average of 0.7 percentage points. This box looks at the potential consequences for our fiscal forecast were the RPI to be changed in a way that removed this formula effect and thus narrowed the gap with CPI inflation.

The stock of index-linked gilts stood at £429 billion in 2018-19 (19.8 per cent of GDP). Both coupons and the value of the principal due for ultimate repayment rise with RPI. Both elements are accrued each year, though investors only receive the principal uplift on redemption. This means RPI inflation tends to be the most important near-term influence on our accrued debt interest forecast. What would a change to the formula for RPI mean for the holders of ILGs? The ready reckoners published alongside our most recent forecast imply that if RPI inflation were 0.7 percentage points a year lower – equivalent to the effect of removing the formula effect – debt interest spending in 2019-20 would be £3.1 billion (0.1 per cent of GDP) lower. The effect builds somewhat each year, reaching £4.4 billion (0.2 per cent of GDP) in 2023-24.

Assessing the fiscal implications of such a change is, however, further complicated in that three of the oldest ILGs outstanding contain a clause requiring the Treasury to offer to redeem them at (uplifted) par in the event of a change to the RPI that the Bank of England judges to be “fundamental… and materially detrimental” to bondholders. Given current market prices are far above uplifted par for these bonds, it seems unlikely that many bondholders would exercise this option. In sum, were such a change to go ahead, it would be likely to lower debt interest spending by progressively larger amounts over time. The large reduction in the overall return to investors over the lifetime of each ILG would generate a large drop in market prices for those bonds, with potentially significant implications for existing investors’ balance sheets. The change could also prompt wider market instability were it to be seen as a breach of trust, though as the Government would simply be curtailing an unintended windfall that seems unlikely.

A change to the RPI would have several other effects on our fiscal forecast. Many payments to the Government and other regulated prices are uprated using the RPI – including the annual revalorisation of various excise duties and increases in regulated rail fares – and the RPI is also used in the calculation of the interest accrued on most student loans. The business rates multiplier was switched from RPI to CPI inflation uprating in April 2018. In contrast, many payments from the Government to the public, other than those on ILGs, are uprated with CPI – including many welfare payments, personal tax credits, and, since May 2019, NS&I index-linked savings certificates. Income tax and NICs thresholds are also linked to CPI inflation.

Looking just at ILGs, excise duties and accrued interest on student loans (under the current accounting treatment) – the largest elements of the public finances still linked to the RPI – the net effect of a 0.7 percentage point a year drop in RPI inflation relative to CPI inflation would be to reduce borrowing by £2.3 billion in 2020-21, falling slightly to £1.9 billion in 2023-24.

In 2013, the UK Statistics Authority (UKSA) acknowledged the problems with the RPI by withdrawing its status as a National Statistic, though it continues to publish it. The RPI is also unusual in that it requires the Chancellor’s consent before any major changes can be implemented, rather than changes being the sole domain of statisticians.b The wider implications for ILG holders of addressing the shortcomings of RPI represent a disincentive to change. In light of the continued lack of action in addressing the problems with the RPI, the House of Lords Economic Affairs Committee recently concluded that UKSA should request “a fix to the clothing problem” and that the “Chancellor should approve this change regardless of the effects on index-linked gilt holders”.c That may hasten a methodological change in the RPI.

a The CPI largely uses the ‘Jevons’ and ‘Dutot’ methods to aggregate prices at the most basic level, while the RPI largely uses the ‘Carli’ and ‘Dutot’ methods. The Carli method tends to inject spurious inflation into the index and is therefore prohibited by Eurostat regulations except in ‘exceptional cases’. The ONS provides more detail in: ONS, Shortcomings of the Retail Prices Index as a measure of inflation, March 2018.
b See The Statistics and Registration Service Act 2007, 2007.
c House of Lords Economic Affairs Committee, 5th Report of Session 2017–19 HL Paper 246: Measuring Inflation, January 2019.