CPI (Consumer Price Index) measures the cost of a basket of goods — but the basket is updated periodically, the weighting of items within it changes over time, and "what's in the basket" is a political and methodological decision that significantly affects what the resulting inflation figure actually represents
The previous articles on this site covered inflation adjustment basics, financial goal inflation adjustment, central bank targeting, inflation winners and losers, and personal inflation vs headline CPI. This article addresses how CPI is constructed — the basket methodology, the index number problem, and why the same economy can produce different measured inflation rates depending on which specific CPI variant is used.
What a "basket of goods" actually is
The Consumer Price Index tracks price changes for a specific collection of goods and services — called the "basket" — that represents typical consumer spending. Price collectors (in the UK, the ONS; in the US, the Bureau of Labor Statistics) visit thousands of shops and websites monthly to record current prices for specific products.
The basket's composition: the UK CPI basket includes approximately 700+ goods and services, spanning: food and non-alcoholic beverages, alcohol and tobacco, clothing and footwear, housing, household goods, health, transport, communication, recreation and culture, education, restaurants and hotels, and miscellaneous goods and services.
The basket is weighted by how much consumers typically spend on each category — transport is weighted more heavily than, say, medical equipment, because the average household spends more on transport. If petrol prices rise sharply, and transport has a high weight, CPI rises more than if a less-important category rose by the same percentage.
How the basket is updated — and why it matters
Basket weights and contents are updated periodically (annually in most countries) based on survey data of actual household spending patterns. If households begin spending significantly more on streaming services and less on physical media, the basket is revised to reflect this.
The update lag creates measurement issues: if consumer spending patterns shift quickly (as during COVID-19, when spending on services like restaurants collapsed while online shopping surged), the basket weights may not reflect current reality for a year or more. During 2020-2021, CPI measured "a basket" that still heavily weighted restaurants and travel — categories that few people were using. This potentially underestimated inflation in the categories people were spending on.
Deliberately excluding items: UK CPI excludes mortgage interest payments (a significant housing cost). The CPIH variant includes owner-occupiers' housing costs (OOH). During periods of rising interest rates — which both control inflation and increase mortgage costs — CPI may significantly understate the actual cost increase for mortgage holders.
The Laspeyres, Paasche, and chain-linking problem
CPI measurement has a known statistical problem: the index number problem.
Laspeyres index (traditional CPI): uses a fixed base-period basket. If apples cost £1/kg in the base year and the basket includes 10 kg, the apple component is always measured against 10 kg × current price. Problem: consumers substitute — if apples become expensive, people buy more pears. The Laspeyres index misses this substitution, overstating inflation (because it assumes you still buy as many expensive apples as before).
Paasche index: uses current-period quantities. Captures substitution but understates inflation (because it assumes you always bought the substituted cheaper goods, crediting you with savings you hadn't yet made).
Chain-linking: modern CPI methods "chain" the index annually — each year uses the previous year's basket as the base, then the chain links are multiplied together. This gradually incorporates substitution behavior without assuming either a fixed old basket or a current one.
The practical consequence: chained CPI (used in the US since 2002 for some government programs) consistently measures about 0.25-0.3 percentage points lower per year than unchained CPI, because it accounts for substitution. Over 10 years, this difference accumulates to ~2.5 percentage points — meaningful for programs indexed to CPI (Social Security, inflation-linked bonds).
CPI vs RPI vs CPIH vs PCE: the UK and US alphabet soup
UK inflation measures:
CPI: the main headline measure; excludes mortgage interest payments; used for Bank of England's 2% target and most government indexing.
CPIH: CPI plus owner-occupiers' housing costs (estimated via "rental equivalence" — what the homeowner would pay to rent their home). More comprehensive housing measure but based on an imputed cost rather than a directly-measured payment.
RPI (Retail Price Index): the older UK measure; includes mortgage interest payments; uses a different averaging formula (arithmetic mean rather than geometric mean) that tends to produce slightly higher inflation readings. Now considered a "national statistic" rather than an "official statistic" due to methodological concerns, but still used for some legacy applications (student loan interest, some National Savings products, index-linked gilts issued before 2030).
US inflation measures:
CPI-U: Consumer Price Index for All Urban Consumers — the main US headline measure.
PCE (Personal Consumption Expenditure deflator): the Federal Reserve's preferred measure; uses a broader spending universe (including spending on behalf of consumers by employers and government, like employer-paid health insurance); chain-weighted; typically 0.25-0.5% below CPI-U.
Why your "personal inflation" differs from headline CPI
The basket represents average spending. Your spending pattern almost certainly differs from the average:
- Homeowners with large mortgages face higher housing cost inflation during rate-rise periods than CPI shows
- Non-drivers face no petrol price exposure but may face higher transit cost exposure
- Households with children spend more on childcare and education
- Renters face rent inflation that CPI measures; homeowners (with fixed-rate mortgages) face minimal direct housing cost change during rate rises
The ONS provides a "personal inflation calculator" allowing individuals to adjust the basket weights to their own spending patterns — producing a personal CPI that may differ meaningfully from the headline figure.
How to use the Inflation Calculator on sadiqbd.com
- For financial planning: use the calculator to understand how the headline CPI compounds over time — if planning savings or investments, applying inflation to future values makes the target more realistic
- For salary negotiations: compare wage increases against the relevant CPI variant for your spending pattern — a 5% pay rise against 7% CPI (weighted for renters in a city with rapidly rising rents) may feel different from the headline figure
- For comparing periods: the calculator can adjust historical prices to current terms — allowing meaningful comparison of costs across decades, though recognizing that what's in the basket has changed over those decades (no mobile phone component in 1980 CPI)
Frequently Asked Questions
If the UK stopped using RPI, why does it still appear in contracts and products? Legacy commitments predate the statistical decision. When the ONS downgraded RPI in 2013 and again when reform discussions continued through the 2020s, existing contracts that specified RPI (student loans, existing index-linked gilts, some defined benefit pension schemes) couldn't simply be changed unilaterally — they were legal contracts. The UK government has announced that gilts issued after 2030 will use CPIH rather than RPI, gradually reducing RPI's footprint. But for the many existing instruments that specified RPI, those contracts continue to use it until they mature. The difference between RPI and CPI can be 0.5-1 percentage point per year — over a 30-year index-linked gilt, this compounds significantly.
Is the Inflation Calculator free? Yes — completely free, no sign-up required.
Try the Inflation Calculator free at sadiqbd.com — calculate the real purchasing power of any sum across any time period.
