Every month the United States produces four widely watched inflation numbers – headline CPI, core CPI, headline PCE, and core PCE – and every month cable news picks whichever one makes the best chyron. They are not the same, they are not interchangeable, and the Federal Reserve only officially targets one of them.
TL;DR: CPI and PCE both measure U.S. price inflation, but they use different formulas, different baskets, and different sources of price data. The Bureau of Labor Statistics builds CPI from an urban household’s out-of-pocket spending; the Bureau of Economic Analysis builds PCE from a broader “personal sector” that includes what employers, Medicare, and Medicaid pay on your behalf. “Core” versions of each strip out food and energy. The Federal Open Market Committee’s 2 percent inflation target is defined on the headline PCE price index, but the Core PCE reading gets the most trading-desk attention because it strips out the noisiest components. In June 2026, all four readings were still running above 2 percent.
Why there are four inflation numbers, not one
Two statistical agencies measure the same phenomenon – the average change in the prices of things Americans consume – using very different methods, because they were built for different jobs. The Bureau of Labor Statistics’ Consumer Price Index was designed to adjust wages and benefits for the cost of living. The Bureau of Economic Analysis’ Personal Consumption Expenditures price index was designed to deflate the consumption component of GDP so real growth could be measured.
Because they were engineered for different purposes, they answer slightly different questions. That is why the Fed watches PCE – it fits the same national-accounts framework the Fed uses to think about output and productivity – while wage negotiations and Social Security cost-of-living adjustments still lean on CPI.
The “core” versions of each simply strip out the two most volatile categories – food and energy prices – to give a cleaner read on the underlying trend. A hurricane that spikes gasoline for a month, or a bird-flu outbreak that doubles the price of eggs, will move headline inflation without really telling you anything about the trajectory of prices in the rest of the economy. Core inflation is the answer to “ignore the weather, what is the trend?”
CPI: the household checkout receipt
BLS publishes CPI-U (the “U” is for urban consumers) monthly, about two weeks after the reference month ends. Underneath the headline number is an enormous survey operation: BLS field staff collect roughly 80,000 prices every month from stores, service providers, and rental units in 75 urban areas. Weights come from the Consumer Expenditure Survey, which asks households what they actually spent money on. Starting with the January 2023 CPI report, BLS switched to updating those weights every year (previously every two years) so the basket keeps up with changing consumer behavior.
The math CPI uses is a modified Laspeyres formula – it holds the basket roughly fixed and asks what it would cost today versus a base period. The BEA explicitly identifies this as one of the key differences with PCE in its comparison guidance: “The PCE price index is based on the Fisher-Ideal formula, while the CPI is based on a modified Laspeyres formula.”
Core CPI: strip the volatile stuff
Core CPI removes food and energy from the basket and re-weights the rest. It is not more or less “real” than headline – your electric bill is very real – but it is far less noisy. In practice, most Wall Street commentary on the CPI print focuses on the core month-over-month change and on shelter, because shelter alone is roughly a third of the CPI basket and it moves slowly.
PCE: the whole personal-sector basket
BEA publishes the PCE price index once a month as part of the Personal Income and Outlays release, roughly four weeks after the reference month. The data come partly from CPI itself (BEA reuses BLS price collection for many categories) and partly from producer price indexes and administrative sources. What is fundamentally different is the scope: PCE covers, in BEA’s own words, “spending by and on behalf of the personal sector, which includes both households and nonprofit institutions serving households; the CPI measures out-of-pocket spending by households.”
That “on behalf of” clause is the reason healthcare weighs so much more heavily in PCE than in CPI. When your employer pays your health insurance premium, or when Medicare pays your hospital bill, that shows up in PCE but not in CPI. In CPI, only the copay you actually swipe your card for counts.
PCE also uses a Fisher-Ideal chain-weighted formula. The basket is re-weighted every quarter to reflect what consumers are actually buying now. When beef gets expensive and shoppers switch to chicken, PCE captures that substitution much faster than CPI does. That is a meaningful methodological advantage when relative prices are moving a lot – which is exactly what happens during energy shocks and supply crunches.
Core PCE: what the Fed actually watches
The FOMC’s official 2 percent target is on the annual change in the headline PCE price index. In its Statement on Longer-Run Goals and Monetary Policy Strategy – most recently reaffirmed on January 27, 2026 – the Committee writes that “inflation at the rate of 2 percent, as measured by the annual change in the price index for personal consumption expenditures, is most consistent over the longer run” with its mandate. Core PCE gets more airtime on trading desks because it is less noisy, but the target itself is on the total.
Four differences at a glance
| Feature | CPI (BLS) | PCE (BEA) |
|---|---|---|
| Publisher | Bureau of Labor Statistics | Bureau of Economic Analysis |
| Formula | Modified Laspeyres | Fisher-Ideal, chain-weighted |
| Weight updates | Annual (since 2023) | Quarterly |
| Scope | Out-of-pocket urban household spending | All personal-sector consumption (incl. employer- and government-paid) |
| Housing weight | Larger (shelter ~1/3 of basket) | Smaller (~1/6 of basket) |
| Healthcare weight | Smaller (under 10%) | Larger (~1/5 of basket, incl. employer plans) |
| Release timing | ~2 weeks after month-end | ~4 weeks after month-end |
| Fed’s 2% target? | No | Yes (headline PCE) |
What June 2026’s readings actually show
The most recent prints for the reference month of June 2026 make the divergence concrete. Both were released within about two weeks of each other, and both are still telling the Fed the same story: inflation has come off its highs but is not yet at target.
A few things stand out. First, both PCE readings are running above both CPI readings this month. Over long stretches CPI historically tends to print a touch higher than PCE – the wedge is often attributed to CPI’s larger housing weight and its slower substitution response – but in any single month, either can lead. Second, the gap between headline and core is much wider in CPI (0.9 points) than in PCE (0.4 points). That tells you food and energy did more work in the CPI print than in the PCE print – a direct consequence of the two indexes weighting those categories differently.
Why core is less noisy than headline
The whole reason “core” exists is that food and energy prices swing on things monetary policy cannot control – weather, oil politics, seasonal harvests. To see that, compare month-over-month percentage changes for headline versus core CPI over the first half of 2026. Core is dramatically smoother.
The March 2026 headline print jumped 0.87 percent in a single month. Core rose only 0.20 percent. In June, headline fell 0.42 percent while core was basically flat. Every one of those big headline moves was food or energy doing the swinging – which is exactly the story the “core” filter is designed to see through.
Why the Fed prefers PCE
The FOMC formally shifted its preferred inflation gauge from CPI to PCE in the early 2000s and codified the 2 percent target on PCE in 2012. The reasoning, laid out repeatedly by Fed research staff and by the BEA in FAQ 555, comes down to three things:
- Broader coverage. PCE captures spending that CPI misses – most importantly, the employer- and government-paid share of medical care. In an economy where about half of health spending is not out-of-pocket, that matters.
- Substitution. Chain-weighting lets PCE reflect that consumers substitute between goods when relative prices change. CPI’s Laspeyres formula holds the basket more rigid, which biases it slightly upward during periods of large relative-price moves.
- National-accounts consistency. PCE plugs into the same GDP framework the Fed uses to think about real output, so real GDP and real consumption use a consistent deflator.
The tradeoff is release timing – PCE lands about two weeks after CPI – and public familiarity. Most Americans have never heard of PCE, so the Fed has to translate its target into “the inflation number you have heard of” in public communication.
Common mistakes when reading these numbers
- Confusing headline and core in headlines. A journalist may report “inflation fell to 3.5 percent” – that is headline CPI. The Fed is watching Core PCE, which was 3.3 percent for the same month. Both statements are true; only one is directly comparable to the target.
- Treating CPI and PCE as substitutes. They are not – especially not month to month. Compare like with like: if you are talking about the Fed’s target, use PCE; if you are talking about the cost of your grocery bill, CPI is closer.
- Ignoring seasonal adjustment. Month-over-month prints are almost always quoted seasonally adjusted (SA). Year-over-year changes are typically quoted from the not-seasonally-adjusted (NSA) index because the seasonal pattern washes out over 12 months. Mixing them up is a common mistake.
- Reading too much into any single print. Both series are revised – CPI seasonal factors get restated, PCE gets revised as source data updates. The FOMC looks at three- and six-month annualized trends more than the single-month wiggle.
What to learn next
Once the CPI/PCE distinction clicks, the natural next stops are: the Trimmed Mean PCE published by the Dallas Fed (an alternative to core that trims the biggest outliers each month rather than always dropping food and energy); the Sticky-Price CPI from the Atlanta Fed (which separates goods whose prices change quickly from those that change slowly); the GDP deflator (which covers the whole economy, not just consumption); and the Employment Cost Index and Average Hourly Earnings, both of which the Fed reads alongside inflation as measures of the wage side of the price story. Understanding how each of those relates back to the Fed’s PCE target is where most macro traders and buyside economists spend a lot of their time.
Sources
- BEA FAQ 555 – How does the PCE price index differ from the CPI?
- BEA – Personal Consumption Expenditures Price Index (June 2026 headline PCE +3.7% YoY)
- BEA – Personal Income and Outlays, June 2026 (release date July 30, 2026)
- BLS – Consumer Price Index home page
- BLS – CPI-U All Items, seasonally adjusted (CUSR0000SA0)
- BLS – CPI-U Less Food and Energy, seasonally adjusted (CUSR0000SA0L1E)
- FOMC – Statement on Longer-Run Goals and Monetary Policy Strategy (adopted Jan 24, 2012; reaffirmed Jan 27, 2026)
- TradingEconomics – United States CPI history (June 2026 headline CPI 3.5%, core 2.6%; historical series)
Disclosure: This article is for informational purposes only and is not investment advice.