CPI vs PCE: how they differ and why the Fed targets PCE

The consumer price index (CPI) and the PCE price index both measure how fast consumer prices rise, and they use much of the same price data. They give different numbers because they cover different spending and weigh it differently. CPI, from the Bureau of Labor Statistics (BLS), tracks what urban households pay out of their own pockets. The PCE price index, from the Bureau of Economic Analysis (BEA), tracks spending by all households and by others on their behalf, such as employers and Medicare paying for health care. The Federal Reserve defines its 2% inflation goal in PCE terms.

That is CPI vs PCE in three sentences. This guide explains where the gap between them comes from, how big it usually is, why the Fed chose PCE, and when each one comes out next. Every comparison of the two numbers here uses the same month for both, because CPI for a month comes out about two weeks before PCE for that month, and setting the newest CPI next to an older PCE figure mixes two different months.

What is the difference between CPI and PCE?

Sources: BLS CPI release, Oct 14, 2026; BEA, Sep 30, 2026; BEA FAQ 555; BLS relative importance table; Cleveland Fed, Mar 2, 2020. "Core" means excluding food and energy. Readings as of Oct 26, 2026.

The two indexes are not independent. The Fed noted in 2000 that the PCE index "draws extensively on data from the consumer price index", and the Cleveland Fed says it "makes use of much of the data from the CPI". The differences come from how those prices are combined.

Why do CPI and PCE give different numbers?

BEA and BLS sort the gap into the same four groups: scope, weights, formula and "other effects" (BEA FAQ 555). BEA publishes a reconciliation table each month that rebuilds one index from the other in steps, so each effect can be measured.

Scope: whose spending counts

BEA puts it this way: "PCE measures 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."

Health care is the biggest example. In CPI, medical care counts only what consumers pay themselves. In PCE it also includes care paid for by employer health insurance and by Medicare and Medicaid, so medical care carries much more weight in PCE (BLS, 2011). PCE also counts things nobody pays for directly, such as the implicit cost of free bank services. All told, "about 25 percent of PCE spending is not captured by the CPI", by the Cleveland Fed's count. CPI-U also leaves out rural households; PCE covers everyone living in the US.

Weights: how much each item matters

Because PCE's total is larger, the items both indexes share get smaller weights in PCE. Housing is the clearest case. In the latest BLS table, for December 2025 with 2024 weights, shelter is 35.625% of CPI-U, and owners' equivalent rent alone is 26.204% (BLS). The Cleveland Fed estimated in 2020 that housing is "roughly 16 percent" of PCE. That is an older figure, and we have not found a newer one from BEA, but the order of magnitude is the point: a change in rents moves CPI about twice as much as PCE. Gasoline also weighs more in CPI.

The weights also come from different places. CPI's come mainly from the Consumer Expenditure Survey, which asks households what they buy; PCE's come from business surveys such as the Census Bureau's retail and services surveys (BLS, 2011).

Formula: fixed basket or changing basket

"The PCE price index is based on the Fisher-Ideal formula, while the CPI is based on a modified Laspeyres formula," says BEA. A Laspeyres index prices a fixed basket. A Fisher index, in BLS's words, "reflects consumer substitution among detailed items as relative prices change": when one item gets expensive and people buy more of a cheaper one, the index follows them. That tends to make a fixed-basket index run a little higher.

BLS also publishes a CPI built on a similar principle, the chained CPI (C-CPI-U, Törnqvist formula). BLS says it and PCE "tend to behave more similarly over time", though they still differ in scope and weights. In September 2026 the chained CPI rose 3.5% over 12 months against 3.7% for the regular CPI-U, both unadjusted; the chained figures for the past 10 to 12 months are still subject to revision (BLS).

Other effects

The rest: differences in seasonal adjustment, in price sources and in method. Airfares are an example. PCE bases them on passenger revenue and miles traveled, while CPI prices sampled routes (BLS, 2011).

How the four effects added up in two periods

BLS published a reconciliation for two periods: 2002 to the first half of 2007 (average annual rates) and the fourth quarter of 2010 (annualized rates). Each line shows how much an effect added to the gap between PCE and CPI, in percentage points; a negative number pushed PCE up relative to CPI.

Source: BLS, Focus on Prices and Spending, May 2011, summarizing BEA research. Effects are rounded, so the steps don't always add up exactly.

BLS explains both columns. In 2002–07 most of the weight effect came from shelter, whose prices outpaced overall inflation, and the scope effect came mostly from medical care, which weighs more in PCE and rose faster than other prices. In late 2010, gasoline drove the weight effect instead, and medical inflation was milder, so the scope effect shrank. The lesson is that weights and scope often pull in opposite directions, and which one wins depends on which prices are rising fastest at the time.

Which is usually higher, CPI or PCE?

Usually CPI. The Cleveland Fed found that "since December 1978, 12-month CPI inflation has run 0.30 percentage points (ppts) above 12-month PCEPI inflation". The gap has been anything but constant: about a full point in late 1981, and an average of 0.35 points the other way during most of 2009. Over the whole period, CPI inflation "has fallen short of PCEPI inflation 20 percent of the time" (Cleveland Fed, Mar 2, 2020). Those figures are for headline inflation.

August 2026 is an unusual month. It is the latest month both agencies have published, and the usual order doesn't hold. Change over 12 months:

Sources: BLS, Oct 14, 2026 (August figures as stated in the September release); BEA, Sep 30, 2026.

The headline rates are equal, and core PCE is running 0.6 points above core CPI. This guide doesn't assign a cause to that gap: the four effects above can push either way, and periods with PCE above CPI have happened before.

Why does the Fed use PCE instead of CPI?

The Fed switched in its Monetary Policy Report of Feb 17, 2000. Until then, a footnote explained, "the FOMC has framed its inflation forecasts in terms of the consumer price index." It gave three reasons for moving to the PCE index (Federal Reserve):

1. Formula. The PCE index "is constructed from a formula that reflects the changing composition of spending and thereby avoids some of the upward bias associated with the fixed-weight nature of the CPI."
2. Coverage. "The weights are based on a more comprehensive measure of expenditures."
3. Revisions. Past PCE data "can be revised to account for newly available information and for improvements in measurement techniques", so "the result is a more consistent series over time."

The Fed did not call PCE perfect. The same footnote says it is "not entirely free of measurement problems" and that the Committee would keep watching "a variety of aggregate price measures".

The goal itself was written down in January 2012, in the Statement on Longer-Run Goals and Monetary Policy Strategy, which the Committee most recently reaffirmed effective Jan 27, 2026. It names a rate of "2 percent, as measured by the annual change in the price index for personal consumption expenditures" (FOMC). That is why news reports call PCE the Fed's preferred inflation gauge: it is the measure the goal is defined in.

Because CPI usually runs higher, a 2% PCE goal has historically corresponded to CPI inflation of a little more than 2%. That is the use the Cleveland Fed had in mind for its 0.30-point estimate: translating forecasts made in CPI terms, such as those implied by inflation-linked bonds, into PCE terms.

Does the Fed target headline or core PCE?

Headline. The Statement's wording, "the annual change in the price index for personal consumption expenditures", refers to the full index, food and energy included.

Core PCE excludes food and energy, whose prices swing more from month to month. It is widely watched as a guide to where inflation is heading, and BEA publishes it in every release next to the headline figure. But the goal is defined in the full index, and the difference matters now: in August 2026, headline PCE was 3.4% and core 3.0%.

What are the latest CPI and PCE readings?

As of Mon, Oct 26, 2026:

• CPI, September 2026 (BLS, Oct 14): up 3.7% over 12 months, from 3.4% in August. Core CPI was up 2.6%, from 2.4%. Over the month, seasonally adjusted, CPI rose 0.6% and core 0.3%. Energy prices rose 5.2% in the month and 20.4% over the year, with gasoline up 9.1% in September. Shelter was up 3.2% over the year. CPI-W, used for Social Security, rose 3.9% (BLS).
• PCE price index, August 2026 (BEA, Sep 30): up 3.4% over 12 months and 0.3% over the month; core PCE up 3.0% and 0.2%. That release included BEA's annual update, which revised personal income and outlays back to January 2021 (BEA).
• Next: September PCE comes out Thu, Oct 29, at 8:30 am ET, together with the first estimate of third-quarter GDP. It will complete the September pair. That day's other releases and earnings reports are in this week's calendar.

When are CPI and PCE released?

Both come out at 8:30 am ET, before the stock market opens. That is 12:30 UTC until US clocks change on Nov 1, 2026, 13:30 UTC from then until Mar 14, 2027, 12:30 UTC again until Nov 7, 2027, and 13:30 UTC after that.

Each row pairs one month's data, so you can see how long the wait is between the two readings for the same month.

Sources: BLS CPI schedule, BEA 2026 release schedule and BEA 2027 release schedule, all checked Oct 26, 2026. All releases at 8:30 am ET. The agencies can change dates, so their schedules are the final word.

PCE usually follows CPI by about two weeks: 12 to 16 days in this table, with March 2027 the exception at 21. Four of the PCE mornings also bring BEA's first (advance) estimate of GDP: Jan 28, Apr 29, Jul 29 and Oct 28, 2027, covering the fourth quarter of 2026 and the first three quarters of 2027.

Where is CPI still used?

The Fed's choice doesn't make CPI the lesser number. CPI is written into law and contracts in ways PCE is not:

• Social Security. The cost-of-living adjustment "is defined as the percent increase between the third quarter average of the CPI-W for a given year and the previous peak third-quarter average of the CPI-W" (BLS). BLS says CPI affects the income of more than 108 million people through such laws, including over 67 million Social Security beneficiaries and over 41 million SNAP recipients.
• Taxes. CPI is used to adjust the federal income tax structure so that inflation alone doesn't push people into higher brackets (BLS).
• TIPS. TreasuryDirect says it adjusts the principal of Treasury Inflation-Protected Securities "using this version of the Consumer Price Index from the Bureau of Labor Statistics" (TreasuryDirect).
• Contracts. BLS notes that many collective bargaining agreements tie wage increases to CPI.

So both numbers matter to markets. CPI comes out first and sets most legal adjustments; PCE comes about two weeks later and is the measure the Fed's goal is written in. Other explainers are in Guides.