CPI vs RPI: why the UK has two inflation numbers

The three numbers and what they measure

The UK does not publish one inflation rate. It publishes several, and the differences between them are deliberate rather than accidental. The Consumer Prices Index rose by 3.1 percent in the 12 months to August 2026, up from 2.9 percent in July 2026. On a monthly basis CPI rose by 0.5 percent in August 2026, compared with a rise of 0.3 percent in August 2025. The Consumer Prices Index including owner occupiers' housing costs, known as CPIH, rose by 3.2 percent in the 12 months to February 2026. The Retail Prices Index, reported by the House of Commons Library from ONS series CZBH, stood at 3.4 percent in August 2026 and 3.6 percent in February 2026. Core CPI, which excludes energy, food, alcohol and tobacco, rose by 2.6 percent in the 12 months to August 2026, while core CPIH rose by 2.9 percent, unchanged from July. Within CPIH, the goods annual rate rose from 2.2 percent to 2.7 percent and the services annual rate was unchanged at 3.6 percent.

Each measure exists because it answers a different question. CPI tracks the changing cost of a basket of goods and services for a representative household and is the measure the government has set the Bank of England a 2 percent target against. CPIH adds owner occupiers' housing costs, which brings in the cost of living in a home you own rather than rent, and it is used in contexts where council tax and housing costs matter to the comparison. RPI covers a broader basket that includes mortgage interest payments and council tax, and it is used to uprate most excise duties and the principal value and coupon of index-linked gilts. The ONS notes that RPI is not a National Statistic. That status matters, because it signals that the method behind the number does not meet current statistical standards, yet the number still governs contracts, duties and debt.

Why RPI runs higher than CPI

The largest single component of the wedge between RPI and CPI is the formula effect, and it is the part most consumer guides skip. At the lowest level of aggregation, where individual price quotes are combined into an average, RPI uses the Carli formula, an arithmetic mean of price relatives. CPI uses the Jevons formula, a geometric mean of the same price relatives. For an identical set of price changes, the Carli produces a systematically higher result than the Jevons, and the gap widens as the dispersion of those price changes widens. A simple illustration makes the mechanism clear. Take two items of equal weight, one of which doubles in price and one of which halves. The arithmetic mean of the two price relatives is 1.25, so the Carli records a 25 percent increase. The geometric mean is the square root of 2 multiplied by 0.5, which is 1, so the Jevons records no change at all. The same prices, two different answers, and the difference comes entirely from the averaging method rather than from anything happening in shops.

Two further differences widen the gap. The RPI basket is broader, including mortgage interest payments and council tax, which CPI excludes, so a period of rising mortgage rates feeds directly into RPI and not into CPI. The reference populations also differ: RPI covers a population that excludes the highest earners and many pensioner households, while CPI covers a broader representative sample. The OBR has previously estimated the long-run wedge between the two at around 0.9 percentage points. The practical consequence is that RPI is typically the higher number, and that matters because RPI governs much of what you pay while CPI governs much of what you receive. A rail fare, an excise duty or a student loan balance linked to RPI will generally rise faster than a benefit or a state pension linked to CPI.

Where each measure actually applies to you

The wedge is not an abstraction. It determines whether a given household is on the receiving end of the lower number or the higher one, and many households sit on both sides at once. Someone receiving a working-age benefit uprated by CPI and paying a rail fare uprated by RPI experiences the gap as a squeeze in both directions. Someone with an index-linked gilt holding experiences it as a return. The table below sets out the split so that the position of a given household can be identified directly.

The pattern is consistent: CPI tends to be the index applied to payments received from the state, and RPI tends to be the index applied to charges paid by households and to the returns paid to holders of index-linked debt. That asymmetry is the reason the formula effect is not a technical curiosity. A household whose income is uprated by CPI while a significant share of its outgoings is uprated by RPI is exposed to the wedge every year, and the exposure compounds. Over a decade, a wedge of around 0.9 percentage points is not a rounding error in a household budget. It is the difference between a real increase and a real cut.

What the headline number hides

A single headline rate is an average across a basket that no individual household actually buys. The August 2026 category rates show how widely the components diverged. Transport inflation rose to 4.6 percent from 3.6 percent in July, driven by motor fuels, where inflation rose to 23.0 percent from 15.5 percent. Petrol rose 9.1 pence a litre to 161.3 pence and diesel rose 14.2 pence to 181.8 pence. Housing and household services inflation was 4.9 percent, communication 5.3 percent, recreation and culture 1.6 percent, and food inflation was unchanged at 1.3 percent. Against a headline CPI rate of 3.1 percent, those category rates range from 1.3 percent to 23.0 percent.

A household's own inflation rate depends on what it buys, and the same headline figure can mean very different things for a driver and for someone who does not own a car. A household that fills a tank regularly felt the 23.0 percent motor fuels rate directly, while a household with no car felt none of it and may have experienced an overall rate closer to the food and recreation figures. The headline number is a useful summary of the average, but it is not a description of any particular household. For context on how far these rates can move, UK inflation rose almost continuously from under 1 percent in early 2021 to 11.1 percent in October 2022, a 41-year high, before easing over the following two years. The composition of that episode, and of the August 2026 figures, differed sharply between categories even when the headline moved in one direction.

Why economists argue about which index should guide policy

A Bank of England staff working paper published on 25 September 2026 argues that the aggregation used to measure the cost of living is not necessarily the one that should guide monetary policy. Staff Working Paper No. 1,208, 'Inflation composition and monetary stabilization', by Michele Andreolli, Natalie Rickard, Paolo Surico and Chiara Vergeat, argues that a consumer price index measures purchasing power, while a stabilization index should weight sectoral prices by what they reveal about states in which the policy rate has high marginal welfare value. The two diverge when the sectors generating price movements differ from the sectors through which interest rates move demand, income and marginal costs.

The paper uses euro-area data and finds that discretionary consumption and employment adjust most after a contractionary monetary policy shock, while necessity prices respond more, concluding that expenditure weights need not be stabilization weights. Two cautions apply. The empirical work uses euro-area data rather than UK data, so the findings are not a statement about UK inflation. And Bank of England staff working papers set out research in progress by staff and are not Bank policy. The argument is a contribution to an ongoing methodological debate, not a change in the remit or the target. The government has set the Bank of England a 2 percent CPI inflation target, and that target remains the framework within which policy is set.

What changes in 2030

From 2030 at the earliest, CPIH methods and data sources will be introduced into the RPI, and the supplementary and lower-level indices of the RPI will be discontinued, following the UK Statistics Authority's response to the joint consultation on reforming RPI methodology. The OBR states that after February 2030 its RPI forecast grows in line with its CPIH forecast. In effect, the formula effect that has kept RPI above CPI for decades is scheduled to disappear from the headline series, because the method producing it will be replaced by the method used in CPIH.

Holders of index-linked gilts objected, and the reason is straightforward. Index-linked gilts pay a return linked to RPI, and the wedge between RPI and CPIH has been part of that return. Aligning RPI with CPIH removes the wedge they had been paid, reducing the value of the link. The OBR notes that RPI is used to uprate most excise duties and the principal value and coupon of index-linked gilts, so the change reaches both government revenue and government debt. For anyone on an RPI-linked contract, pension or student loan, the practical step is to check now rather than in 2030. Read the wording of the contract to establish whether it references RPI specifically, whether it references RPI as defined at a particular date, and whether it contains a fallback clause if the index is reformed or discontinued. Contracts that name RPI without a fallback are the ones where the 2030 change is most likely to be felt, and the time to identify them is before the reform takes effect rather than after.

Source: ONS: consumer price inflation bulletin.

Related coverage on Kael Tripton: UK inflation rises to 3.1 percent as fuel and airfares climb, What Is Inflation?, Bank of England Decision on 17 September: Hold at 3.75% Expected, Rise to 4% Possible, What It Means for Your Mortgage and Savings, 5 Million UK Homeowners Face Higher Mortgage Costs: What the Bank of England Report Means Across the UK, Student Loan Interest Rates for 2026-27: Plan 2 Capped at 6%.

Frequently asked questions

Why does the UK publish more than one inflation figure?

Each measure answers a different question. CPI tracks the changing cost of a basket of goods and services for a representative household and is the measure the government has set the Bank of England a 2 percent target against. CPIH adds owner occupiers' housing costs, which brings housing into the comparison. RPI covers a broader basket that includes mortgage interest payments and council tax, and it is used to uprate most excise duties and the principal value and coupon of index-linked gilts. The ONS notes that RPI is not a National Statistic.

Why is RPI higher than CPI?

The largest single component is the formula effect. At the lowest level of aggregation RPI uses the Carli formula, an arithmetic mean of price relatives, while CPI uses the Jevons formula, a geometric mean. For the same set of price changes the Carli produces a systematically higher result, and the gap widens as price dispersion widens. Two further differences widen it: the RPI basket includes mortgage interest payments and council tax, and the reference populations differ, with RPI excluding the highest earners and many pensioner households. The OBR has previously estimated the long-run wedge at around 0.9 percentage points.

Which inflation measure applies to benefits and which applies to rail fares?

CPI is used for benefits uprating and is the measure against which the Bank of England's 2 percent target is set. RPI is used to uprate most excise duties, the principal value and coupon of index-linked gilts, student loans and rail fares. The pattern is that CPI tends to be applied to payments received from the state, while RPI tends to be applied to charges paid by households and to returns paid to holders of index-linked debt. A household can sit on both sides of that split at once.

Does the headline inflation rate reflect what a household actually experiences?

No. The headline rate is an average across a basket that no individual household actually buys. In August 2026, against a headline CPI rate of 3.1 percent, motor fuels inflation was 23.0 percent, communication 5.3 percent, housing and household services 4.9 percent, transport 4.6 percent, recreation and culture 1.6 percent and food 1.3 percent. A household's own inflation rate depends on what it buys, so the same headline figure can mean very different things for a driver and for someone who does not own a car.

What happens to RPI in 2030?

From 2030 at the earliest, CPIH methods and data sources will be introduced into the RPI, and the supplementary and lower-level indices of the RPI will be discontinued, following the UK Statistics Authority's response to the joint consultation on reforming RPI methodology. The OBR states that after February 2030 its RPI forecast grows in line with its CPIH forecast. Holders of index-linked gilts objected because aligning RPI with CPIH removes the wedge they had been paid. Anyone on an RPI-linked contract, pension or student loan should check the contract wording now, including whether it contains a fallback clause.