Headline vs. Core Inflation: CPI and PCE Explained

Inflation is the rate at which a price index changes over time. “Headline” usually means an all-items index, including food and energy. In the United States, the monthly CPI and PCE releases each report an all-items measure. A headline reading describes the full index for its stated population and period; it does not tell every household how much its own bills changed.

“Core” is often used for an index that excludes food and energy prices. BLS publishes CPI for all items less food and energy, commonly called core CPI. BEA publishes the PCE price index excluding food and energy, commonly called core PCE. These are published analytical measures, not a claim that food or energy stopped counting as part of inflation. The BLS CPI explanation confirms that its official headline CPI continues to include both.

The reason for looking at both is straightforward: food and energy prices can move sharply from month to month. A storm, harvest change, supply interruption, or global energy shock can push headline inflation around. Core measures set those categories aside in a separate calculation so readers can inspect other price changes. That can help describe a trend, but it does not make a core reading more important for every decision.

Keep the scope visible. The definitions here refer to U.S. CPI and PCE series. Other countries may use “core” for different exclusions or measures. When a report says only “inflation,” look up the named index, its coverage, and its period before comparing the number.

Headline CPI tracks price changes for the consumer population covered by a particular CPI series. CPI-U, for example, covers all urban consumers. Its all-items basket includes food, energy, housing services, transportation, medical care, and other goods and services. BLS publishes many specialized series alongside the all-items index, including all items less food and energy.

BEA defines PCE as goods and services purchased by, or on behalf of, people living in the United States. Its consumer-spending overview summarizes that scope. The all-items PCE price index includes food and energy along with other covered consumption items. CPI and PCE do not have identical coverage, data sources, weights, or index formulas, so their headline rates can differ even over the same dates. For those structural differences, see the separate guide to CPI, PCE, and the GDP deflator.

“Headline” is not a synonym for “unadjusted.” A release can show a seasonally adjusted month-to-month change for an all-items index and a year-over-year change calculated from an unadjusted index. Read the series label and the release table: headline/core tells you which items are in the measure, while seasonal adjustment and the comparison period describe how its change is reported.

Core CPI is the CPI aggregate for all items less food and energy. BLS does not remove those prices from the official all-items CPI; it publishes both series. The core aggregate is useful when a reader wants to inspect price movements outside those two categories. Food and energy still matter to households and remain part of headline inflation.

Core PCE is the PCE price index excluding food and energy. BEA's definition is based on PCE categories: “food” includes items such as groceries and beverages purchased to consume away from the seller's premises, while food services and accommodations are not part of that excluded food category. Energy includes gasoline and other energy goods, plus electricity and gas utilities. See the BEA definition of core PCE for the category detail.

Because the starting indexes differ, core CPI and core PCE are not two labels for the same basket. Their component coverage, weights, source data, and calculation methods differ. A core CPI change cannot be substituted for a core PCE change without naming that switch. The same is true of headline CPI and headline PCE.

Think of “excluding food and energy” as a rule for constructing a particular index, not as a method for subtracting a fixed food-and-energy contribution from a headline number. The remaining components are aggregated under the index's own methodology. Published headline and core rates are separate series, not a simple subtraction exercise.

An all-items index answers the broad question of how covered prices changed, including categories that affect household budgets directly. A core measure can help with a different question: were price increases concentrated in a few volatile categories, or were changes also visible across the rest of the basket? Looking across the component table and several periods is more informative than treating one core print as a verdict.

The Federal Reserve says it tracks multiple “core” measures because short-term inflation readings can be erratic and policymakers try to judge whether developments are likely to persist. It also stresses that food and energy make up an important part of household budgets. Core is therefore a tool for interpreting the pattern, not a reason to disregard grocery, electricity, or gasoline costs. See the Fed's inflation FAQ.

Core measures summarize price-index data; they do not settle the separate model question of how inflation and unemployment may move together. The Phillips curve guide covers that relationship and its limits.

Core does not mechanically identify which changes will reverse. Food or fuel prices can affect transportation, shipping, production, and then prices of other items. Those later changes may appear in core categories. Conversely, a large one-time price change in a category that remains inside the core index can move core even if it does not persist. The measure separates selected categories; it does not label every price movement temporary or permanent.

The FOMC's longer-run 2 percent objective is defined using the annual change in the PCE price index, not core PCE. Policymakers watch core and other measures as evidence about the inflation process, while the objective remains tied to headline PCE. The distinction is documented in the Federal Reserve's explanation of how it evaluates inflation.

The phrase “underlying inflation” can refer to several analytical approaches. Exclusion measures such as core CPI and core PCE leave out specified categories, regardless of whether those categories had unusually large or small price changes that month. That gives the measure a stable definition over time, but a shock outside food or energy can still affect it.

Median and trimmed-mean measures take a different approach. The Cleveland Fed's median CPI ranks component price changes for a month and selects the change at the center of the expenditure-weighted distribution. Its 16 percent trimmed-mean CPI removes the tails of that distribution. The Dallas Fed's trimmed-mean PCE removes selected shares of expenditure weight from the lower and upper tails of the component price-change distribution. Which components fall in those tails can change from month to month, and food or energy may remain if their price changes are not in the tails. See the Cleveland Fed median CPI and the Dallas Fed trimmed-mean PCE series description.

No method wins under every condition. An exclusion measure can be influenced by a one-off increase in included items. A trimmed measure can be affected by which components and weights fall into its cutoffs. Use a small dashboard—headline, core, components, and perhaps a median or trimmed measure—when the goal is to understand breadth and persistence. Each additional measure is another lens, not an independent fact about what prices “really” are.

First record the exact index: CPI-U all items, CPI-U less food and energy, PCE all items, or PCE excluding food and energy. Then note the time comparison: month over month, year over year, or a shorter change converted to an annualized pace. Those rates answer different questions and should not be placed side by side without labels.

For a month-over-month change, compare this month's index with the previous month's. For year-over-year inflation, compare with the index twelve months earlier. A monthly rate expressed at an annualized pace compounds one month's change as though it repeated for a year. For example, an invented 0.4% monthly increase annualizes to about 4.9%: (1.004^12 − 1) × 100. That is an annualized pace derived from one month, not the inflation observed over the last year or a forecast.

Seasonal adjustment also matters. Seasonally adjusted data are often useful for reading recent monthly momentum; year-over-year comparisons are commonly reported using non-seasonally adjusted indexes. Always compare like with like and note the release vintage, especially when reading PCE estimates that can be revised. The related guide to disinflation and deflation explains why a slower positive rate is different from a falling price level.

Suppose a fictional all-items CPI index rises from 300.00 to 301.20 in one month. The change is (301.20 ÷ 300.00 − 1) × 100 = 0.40%. Suppose the separately published core CPI index rises from 310.00 to 310.62. Its change is (310.62 ÷ 310.00 − 1) × 100 = 0.20%. The headline series rose faster in this invented month, which could be consistent with food and energy adding pressure; the example does not identify the actual component contributions.

The different starting index levels are arbitrary reference scales. The calculation uses each series' own prior value, not the difference between 301.20 and 310.62. The inputs are hypothetical and are not current CPI data, forecasts, or a claim about typical outcomes.

Now suppose headline inflation is 0.2% year over year while core inflation is 0.3% year over year. This is possible: different prices moved at different times during the year, and the cumulative twelve-month change need not match this month's direction. Compare the component details and several month-to-month readings before describing the pattern. One release cannot establish whether inflation pressure has broadly strengthened or will persist.

No single core reading is the cost of living for every household. A family that spends more on food, fuel, or utilities can experience a different budget change from an index that excludes some of those categories. That does not make the index wrong; it means the index and household answer different questions.

Core also is not a causal diagnosis, a forecast, or a trading signal. It can help summarize part of the observed price data, but it does not by itself explain the source of inflation or predict what the next release will show. For an index-to-index comparison, keep CPI and PCE distinct; for purchasing-power questions, state the population and period; for monetary-policy claims, distinguish the Fed's headline PCE objective from the analytical measures it monitors.

Before sharing a number, write down: the named index, headline or core definition, population or spending scope, dates, month-over-month or year-over-year convention, seasonal-adjustment status, and source vintage. This small checklist prevents a difference in labels or time windows from being mistaken for a contradiction.

Primary references: BLS CPI and core CPI explanation; BEA core PCE definition and PCE coverage; Federal Reserve inflation FAQ and objective; Federal Reserve Bank of Cleveland Median CPI; Federal Reserve Bank of Dallas Trimmed Mean PCE series description.

This guide is limited to U.S. CPI and PCE measures. It defines common analytical terms and does not recommend a financial product or establish a rule for another country's statistical series.