data / Field guide
How is inflation measured?
See how agencies turn sampled prices and spending weights into a price index, then calculate inflation for a clearly defined period.
Inflation is measured by tracking the prices of a defined group of goods and services, combining those changes with spending weights, and calculating the percentage change in the resulting price index. The result depends on the index's population, scope, formula, and time period.
A price index combines many observations
For the Consumer Price Index, the Bureau of Labor Statistics samples urban areas, outlets, specific goods and services, and rental housing. BLS collects prices during the month and uses Consumer Expenditure Survey data to estimate how much weight categories receive. Categories with larger shares of covered spending generally have more influence on the combined index.
This is not a simple average of every price tag. The sample represents a defined population, and the item being priced is specified so changes can be compared over time. CPI-U covers all urban consumers as BLS defines them, while CPI-W covers a narrower population. Other indexes, including the PCE price index, use different scope, source data, weights, and formulas.
- Price sample: observations for selected items, outlets, places, and housing units.
- Expenditure weights: estimates of the relative importance of covered categories.
- Index formula: rules for combining detailed price movements into higher-level measures.
Sources: [bls-cpi-faq] [bls-cpi-overview]
Weights make the result economically meaningful
Consider a deliberately simplified two-category basket: 60 percent housing and 40 percent groceries. If the housing measure rises 5 percent and groceries fall 2 percent, a fixed-weight calculation gives 0.60 times 1.05 plus 0.40 times 0.98, which equals 1.022. The basket rose 2.2 percent, not the unweighted average of 1.5 percent.
Real official indexes are much more detailed and do not reduce to this classroom formula. BLS uses interrelated samples and formulas at different levels, updates spending information, and publishes thousands of series. The example demonstrates only why weights matter. A category can post the largest percentage change without making the largest contribution if its spending weight is small.
- Illustrative weighted factor: (0.60 × 1.05) + (0.40 × 0.98) = 1.022.
- Illustrative basket change: (1.022 - 1) × 100 = 2.2 percent.
- Contribution depends on both a category's movement and its relative weight.
Sources: [bls-cpi-faq]
Changing products require quality treatment
A product can disappear or be replaced by a model with different features. Comparing the sticker prices directly could mistake a quality difference for pure price change. BLS may select a replacement and, where appropriate, estimate the value of changed characteristics. Its quality-adjustment page explains that a hedonic adjustment removes the estimated part of a price difference attributed to quality.
Quality adjustment is not a blanket claim that improvements cancel inflation. BLS lists different methods for different item categories, and many categories have no hedonic adjustment. Estimation introduces uncertainty, but ignoring replacement and quality change would also distort a constant-quality comparison. Method notes matter most when a claim depends on a specific product category.
- A higher price for a substantially improved item is not necessarily all price inflation.
- A quality estimate is a statistical treatment, not an observed discount at checkout.
- Check the agency's category-specific method before making claims about an adjustment.
Sources: [bls-quality] [bls-cpi-faq]
The published rate comes from two index levels
Once an index exists, the basic percent-change calculation is straightforward: subtract the earlier index level from the later level, divide by the earlier level, and multiply by 100. The hard part is choosing comparable levels from the same series. A twelve-month CPI change uses the same month in consecutive years, not January and December of one year.
Monthly changes are often seasonally adjusted to reduce recurring calendar patterns, while twelve-month CPI changes are commonly reported without seasonal adjustment. Displayed index levels may be rounded more than the internal values used for publication, so a hand calculation can differ slightly. Some series can also be revised. Record the exact series, endpoints, adjustment, release vintage, and units before presenting a rate.
- Formula: ((later index - earlier index) / earlier index) × 100.
- Never mix an ending level from one index with a starting level from another.
- A national population average does not measure every household's personal experience.
Sources: [bls-cpi-faq] [bls-cpi-overview]
Primary sources
- Consumer Price Index Frequently Asked QuestionsU.S. Bureau of Labor Statistics · accessed 2026-10-19
- Consumer Price Indexes OverviewU.S. Bureau of Labor Statistics · accessed 2026-10-19
- Quality Adjustment in the CPIU.S. Bureau of Labor Statistics · accessed 2026-10-19
Prepared with AI assistance and reviewed against the linked primary sources. This is public financial education, not personal financial or investment advice.
FISCAL VITALS
Understand the chart behind a headlineMeasureSource and dateContextWhat the number can tell youIllustrative preview, not a native app screenshotBring the context with you
Understand the chart behind a headline
Planned chart notes will identify an index's population, scope, weights, period, and adjustment before showing its rate.
Fiscal Vitals is being built to keep the measure, source, date, and explanation together in one calm view.
Coming soon for iPhone and Android. It is not yet available on the App Store or Google Play, and no release date has been announced.
See the illustrative product preview