How to Read CPI: U.S. Inflation, Measure by Measure

An inflation release can show a slower annual rate and a faster monthly increase. Households can face rising prices while the headline says inflation has fallen. Neither is necessarily a contradiction: the comparison period and basket determine what the number means.

Start with the Bureau of Labor Statistics’ CPI release, then read the table headings before the commentary. This guide explains the recurring method. It does not label an old monthly observation as today’s inflation rate.

Identify the index before using it

CPI-U covers all urban consumers; CPI-W covers the narrower population of urban wage earners and clerical workers. The chained C-CPI-U uses the urban-consumer population but also accounts for substitution across spending categories and is revised as expenditure information becomes available. The BLS CPI FAQ explains these differences.

The frequently quoted national headline is the all-items CPI-U for the US city average. It is an aggregate measure, not a record of one household’s costs. A renter, a homeowner and a commuter can have different exposure to the same price changes.

The index level and the inflation rate are different quantities. To calculate a percentage change, divide the later index by the earlier index, subtract one, and multiply by 100. Use the same series and adjustment convention at both ends. A slower positive rate means prices are rising more slowly; it does not mean the price level has returned to its previous value.

Read the monthly and annual columns separately

For short-term momentum, use the published seasonally adjusted monthly change. Seasonal adjustment removes estimated recurring patterns, making consecutive months more comparable. The widely reported twelve-month CPI change is not seasonally adjusted. Keep those conventions in the label when charting or quoting a result. The BLS seasonal-adjustment FAQ describes the process.

An annual comparison changes as a new month enters and the month a year earlier drops out. This base effect can move the annual rate even when the newest monthly increase is unremarkable. Inspect both ends of the comparison before calling it an acceleration or a reversal.

Seasonally adjusted history can also be revised when seasonal factors are recalculated. Save the release vintage if you are explaining what readers knew at a past date. Do not compare today’s revised monthly series with an unrevised historical headline without checking the difference.

Headline, core and the meaning of the basket

Headline CPI includes all items. Core CPI excludes food and energy, whose movements can obscure developments elsewhere. Core still includes shelter. It is an analytical subset, not a description of all the expenses households must pay. These definitions are set out in the BLS CPI FAQ.

“Supercore” needs a definition whenever it appears. Analysts can use different baskets under that label. A series excluding shelter does not automatically exclude every energy service, and a CPI-based services measure is not interchangeable with its PCE counterpart. Cite the actual series or calculation before comparing two claims about supercore.

Weights explain why a small move can matter

A component’s price change and its contribution to overall inflation are different. A large increase in a small category may contribute less than a modest increase in a large category. Check the release’s component weights and contribution discussion together.

The BLS publishes relative-importance and weight information. The reference period matters: spending weights are updated, while relative importance also moves as component prices change. A percentage lifted from an old table should not be presented as a permanent feature of the basket.

Nor is the CPI simply a frozen shopping list. The BLS calculation methodology combines different formulas at different levels, including geometric means for most basic indexes. A precise contribution calculation should follow the published methodology, rather than multiply rounded headline numbers and expect an exact reconciliation.

Shelter measures a housing service

Rent of primary residence tracks tenant rents. Owners’ equivalent rent (OER) estimates the rental value of the housing service consumed by homeowners. It does not insert the purchase price of a home or the homeowner’s mortgage payment into CPI. The BLS rent and rental-equivalence factsheet explains the measurement.

The change in OER is derived from rents observed in the rental sample. It is not calculated by asking homeowners each month what their property could rent for. Homeowner survey responses help establish expenditure weights, a separate part of the process.

Existing leases and the sample’s collection schedule mean CPI shelter can move differently from advertised new-lease rents. That difference can be informative, but it does not establish a fixed lag or make either measure universally preferable. Define whether the question concerns rents paid across the housing stock or the price faced by a new tenant.

CPI and the Fed’s inflation target

The Fed’s longer-run 2% inflation objective is defined using the annual change in the PCE price index, as its official explanation states. Core PCE can help analyse underlying conditions, but the objective itself refers to the overall PCE index.

CPI and PCE differ in coverage, weights and formulas. PCE includes certain spending on households’ behalf, including healthcare spending; CPI focuses on households’ out-of-pocket spending within its scope. The BEA’s comparison of the two indexes explains the main differences. Continue with our PCE guide before translating a CPI print into a claim about the Fed’s target.

Make the reading reproducible

For any release, retain the observation month, publication date, index population, basket and adjustment status. Explain the components that account for the movement, using weights from the relevant period. Check revisions and collection notices before interpreting an unusual observation.

CPI is a statistical estimate. The BLS publishes variance estimates, which depend on the series and period. Do not attach a generic confidence interval to every monthly print, or turn a small difference from a forecast into certainty about the economy. A useful analysis explains what moved, how much the evidence resolves, and what another release could change.

Sources and revision

This guide links directly to BLS definitions, calculation methods, seasonal adjustment, shelter methodology and variance tables, alongside BEA and Federal Reserve explanations. The October 8, 2026 revision replaces undated basket weights and a stale monthly case study with a reusable release-reading method. It also clarifies OER measurement, the PCE target and the limits of a generic “supercore” label.