Every month the ONS publishes several inflation figures — CPI, RPI, CPIH — and they are always different. Most people assume they measure the same thing. They don't. The differences are methodological and the choices made — which costs to include, which formula to use, which items to put in the basket — have real consequences for wages, benefits, rail fares, index-linked bonds, and your standard of living. This page explains what each measure is, why they differ, whether the basket has been changed to favour lower readings, and what inflation actually looks like when you include the biggest cost most people face: rising taxation.
The Four Main UK Inflation Measures Explained
CPI
Consumer Prices Index. The Bank of England's 2% target measure. Covers a basket of ~760 goods and services. Does NOT include mortgage interest payments, council tax, or owner-occupier housing costs. Uses geometric mean formula which typically produces lower results than arithmetic mean. Covers a wider population including students in halls and care home residents.
Used for: Bank of England target. State pension and benefits uprating (from 2011). Government bond inflation adjustments (CPI-linked gilts).
RPI
Retail Prices Index. The older measure, first published 1947, official from 1956. DOES include mortgage interest payments, house depreciation, and council tax. Uses arithmetic mean formula — which mechanically produces higher results. Covers only private households (excludes students, care homes). Typically runs 0.8-1.2 percentage points above CPI.
Used for: Index-linked gilt coupon payments. Rail fare increases. Commercial rent reviews. Many wage agreements. Student loan interest. Road tax. Phone and broadband contract increases.
Status: "Not a national statistic" since 2013 — meaning ONS does not endorse it as meeting international standards — but still officially published monthly.
CPIH
CPI Including Owner Occupiers' Housing Costs. The ONS's preferred measure since 2017. Extends CPI to include an estimate of what it would cost a homeowner to rent their own property ("rental equivalence"). Also includes council tax. Accounts for approximately 17% of the basket. More representative for homeowners than CPI.
Used for: ONS's headline preferred measure. From 2030, CPIH methods will replace RPI's calculation — RPI will effectively become CPIH under a different name.
RPIX
RPI Excluding Mortgage Interest Payments. Was the Bank of England's target measure 1992-2003 (target: 2.5%). Mortgage interest was excluded to prevent a "vicious circle" — where the Bank raised rates to fight inflation, which caused mortgage costs to rise, which pushed up RPI, which required even higher rates. Replaced by CPI target in 2003.
Used for: Still published monthly. Historical comparison. Some contractual references.
The Basket of Goods — What Goes In and What Comes Out
Every year the ONS updates the "basket" of goods and services used to calculate inflation — removing items that fewer people buy, adding new ones that reflect modern spending. This is a legitimate and necessary process: a 1990 basket that included coal, film for cameras and telegrams would produce meaningless results today.
But basket changes can also — accidentally or deliberately — affect the measured inflation rate. If you remove items whose prices are rising fast and replace them with items whose prices are stable or falling, you produce a lower inflation reading. The question is: has this happened?
Has the Basket Been Skewed to Show Lower Inflation?
The ONS maintains that basket changes are made to keep the measure representative — not to manipulate the result. The independent UK Statistics Authority oversees the process. There is no credible evidence of deliberate manipulation.
However, there are legitimate criticisms that are not conspiracy theories:
The "substitution bias" problem: When prices of a particular item rise sharply, consumers buy less of it and switch to cheaper alternatives. CPI captures this substitution (via the geometric mean formula) — but some economists argue this understates real inflation, because people are being pushed into cheaper goods they wouldn't otherwise choose. Their standard of living has fallen, but the index doesn't fully capture it.
Housing costs excluded from CPI: CPI does not include mortgage interest payments or the actual cost of buying a home. For most UK households, housing is their single largest cost. When interest rates rose from 0.1% to 5.25% in 2022-23, the CPI barely moved on this component — but millions of mortgage holders saw their monthly payments increase by hundreds of pounds. The ONS's own preferred measure, CPIH, attempts to correct this via "rental equivalence" — but critics argue this is also an imperfect proxy for actual homeowner costs.
The "quality adjustment" problem: When a product is improved, statisticians sometimes reduce its effective price in the index to reflect the improvement. A new car with more safety features might be recorded as cheaper in real terms even if the actual price rose. This is methodologically sound but tends to produce lower measured inflation in technology and manufactured goods.
The Biggest Inflation Nobody Measures — Taxation
Every discussion of inflation focuses on the cost of goods and services. Almost no discussion includes what is, for most working people, their single largest cost: taxation. Income tax, National Insurance, council tax, VAT, and stamp duty collectively represent more of most households' outgoings than food, energy and transport combined.
When the government raises taxes — through rate increases, threshold freezes, or stealth means — the real cost of your income rises even if CPI stays at 2%. The ONS inflation measures capture none of this. They are designed to measure the cost of a basket of goods, not the cost of funding your life including your tax burden.
The most significant tax increase of the past decade has been the freeze on income tax thresholds. The personal allowance has been frozen at £12,570 since 2022 and will remain there until at least 2028. The higher rate threshold has been frozen at £50,270. With wages rising by approximately 5% annually, millions of people have been pulled into higher tax bands without any change to the headline tax rate — "fiscal drag" at work.
Real Inflation Including Tax Rises — 2016 to 2026
The table below shows cumulative CPI inflation versus the cumulative real cost increase for workers at different income levels — taking into account not just rising prices but also rising effective tax rates from fiscal drag and explicit tax changes. Figures are approximate and illustrative, based on HMRC tax data and ONS earnings data.
| Year | CPI % | RPI % | Tax burden rise £40k earner | Real cost rise £40k earner | Tax burden rise £80k earner | Real cost rise £80k earner | Tax burden rise £120k earner | Real cost rise £120k earner |
|---|---|---|---|---|---|---|---|---|
| 2017 | 2.7% | 3.6% | +0.2% | 2.9% | +0.3% | 3.0% | +0.3% | 3.0% |
| 2018 | 2.5% | 3.3% | +0.2% | 2.7% | +0.2% | 2.7% | +0.2% | 2.7% |
| 2019 | 1.8% | 2.5% | +0.1% | 1.9% | +0.1% | 1.9% | +0.1% | 1.9% |
| 2020 | 0.9% | 1.5% | 0% | 0.9% | 0% | 0.9% | 0% | 0.9% |
| 2021 | 2.6% | 4.8% | +0.3% | 2.9% | +0.4% | 3.0% | +0.5% | 3.1% |
| 2022 | 9.1% | 12.6% | +1.2% (NI rise + freeze begins) | 10.3% | +1.8% | 10.9% | +2.2% | 11.3% |
| 2023 | 7.3% | 10.7% | +1.4% (threshold freeze) | 8.7% | +1.9% | 9.2% | +2.3% | 9.6% |
| 2024 | 2.6% | 4.0% | +0.8% (freeze continues) | 3.4% | +1.1% | 3.7% | +1.4% | 4.0% |
| 2025 | 3.4% | 4.5% | +0.7% | 4.1% | +1.0% | 4.4% | +1.3% | 4.7% |
| 2026 (est) | 2.6% | 3.4% | +0.5% | 3.1% | +0.8% | 3.4% | +1.0% | 3.6% |
| CUMULATIVE 2016-2026 | ~35% | ~52% | ~+5.4% | ~40% | ~+7.6% | ~43% | ~+9.3% | ~45% |
Figures are approximate and illustrative. Tax burden rise calculated from HMRC effective tax rate changes including threshold freezes, NI changes and explicit rate changes. These are estimates based on ONS/HMRC data — individual circumstances vary significantly. Not financial advice.
What If the Basket Had Stayed the Same? — The Frozen Basket Exercise
This is a fascinating and technically difficult question. If the ONS had used exactly the same basket of goods in 2026 as it used in 2016 — without any substitutions, removals or quality adjustments — would the measured inflation rate be higher or lower?
The ONS does not publish this counterfactual, and constructing it precisely requires access to individual item-level price data going back 10 years. However, we can draw some conclusions from what we know about which items were removed and added:
Items removed that were rising rapidly in price: Several items were removed in the 2020-2023 period when their prices were rising sharply — hot rotisserie chicken (2024, removed when poultry prices were high), oven-ready joints (2025, removed when meat prices were elevated). Removing items at their price peak reduces the basket's measured inflation in subsequent years.
Items added that were falling in price: Electronic goods, streaming services, VR headsets, and other technology items were added in years when technology prices were broadly falling. Adding deflationary items to the basket reduces overall measured inflation.
Academic estimates: Research by the Resolution Foundation and the Bank of England suggests that methodological differences between UK CPI and a "fixed basket" measure could add approximately 0.3-0.7 percentage points per year to measured inflation — amounting to roughly 3-7 percentage points cumulatively over a decade. This is not proof of manipulation — these are the normal effects of substitution bias and quality adjustment — but it does mean the "real world" cost of living has risen faster than CPI suggests for most people.
Who Gets the Best and Worst Deal From the Different Measures?
The choice between CPI and RPI is never neutral — it creates winners and losers depending on who benefits from higher or lower indexation.
Government wins from CPI being lower than RPI: The state pension is uprated by the triple lock (CPI, earnings or 2.5%). Benefits are uprated by CPI. Using CPI instead of RPI saves the government billions annually versus what would be paid under RPI uprating. This is why the switch from RPI to CPI for benefits uprating in 2011 was controversial — it represented a real-terms cut in benefits over time.
Rail passengers lose from RPI being used for fares: Rail fares are increased by RPI — which is consistently above CPI. If CPI were used instead, rail fares would have risen roughly 10-15 percentage points less over the past decade. Passengers pay the RPI premium; the government saves the CPI difference on benefits.
Index-linked gilt holders win from RPI: The government's index-linked bonds pay RPI. Higher RPI means higher coupon payments — costing the taxpayer more, but rewarding bondholders (primarily pension funds and wealthy investors). This is partly why the government is moving RPI toward CPIH from 2030 — it will reduce its debt servicing costs.
Mortgage holders get neither: CPI doesn't include mortgage costs and RPI includes them in a way that doesn't fully reflect modern variable/fixed mortgage structures. CPIH uses rental equivalence — which is the closest to capturing housing costs but is still an imperfect proxy for what actual mortgage holders pay.