CPI vs PCE vs PPI: Which Inflation Gauge Matters for What

The three gauges at a glance

Sources for the table: BEA's FAQ on the CPI and PCE formulas and scope, the BLS PPI overview, and the Cleveland Fed's CPI-versus-PCE page.

CPI: what households pay out of pocket

The St. Louis Fed describes the CPI as measuring "the average change over time in the prices paid by urban consumers for a market basket of consumer goods and services." BEA's FAQ adds that the CPI measures out-of-pocket spending by households, so it leaves out spending made on their behalf. The Cleveland Fed notes that the CPI has been published since 1919\.

Shelter dominates the basket. In the BLS release for August 2026, the relative-importance column (dated July 2026\) shows shelter at 35.3% of the index and energy at 7.3%. Because the weight is so large, shelter inflation shapes the CPI more than any other category. In August 2026 shelter rose 3.0% over twelve months.

PCE: the Fed's measure

The PCE price index differs from the CPI in three ways, according to BEA and the Cleveland Fed.

Why the formula matters: a simple example. All numbers are hypothetical. Two goods, A and B, each cost $10 and a household buys 50 of each. B's price rises 50% to $15, and the household switches, buying 60 of A and 30 of B.

(Our calculation.) A Laspeyres-type index ignores the switch. A Fisher-type index partly reflects it. The example is exaggerated on purpose. The Cleveland Fed's 0.39-point average gap includes all the differences above, not the formula alone.

Which one does the Fed target? The FOMC's statement of 25 January 2012 says: "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 the Federal Reserve's statutory mandate." The Cleveland Fed notes that PCE has been tracked since 1959 and has been the Fed's preferred measure since 2000\.

• A fixed-basket (Laspeyres-type) index prices the old basket: it rises 25.0%.
• The same calculation on the new basket (Paasche) rises 16.7%.
• The Fisher index is the geometric mean of the two: 20.8%.

1. Scope. BEA says PCE "measures spending by and on behalf of the personal sector," which covers households and the nonprofits that serve them. The Cleveland Fed says it also covers rural households and includes spending on consumers' behalf, such as employer-provided health insurance, Medicare and Medicaid.
2. Weights. The St. Louis Fed estimates that medical spending paid by Medicare Part A, Medicaid, veterans benefits and employers was almost 70% of medical spending in 2022\. The St. Louis Fed puts health care at about 20% of total PCE (2022) but only about 8% of the CPI basket (it gives no reference year for the 8%). The CPI's smaller health share is offset by larger weights elsewhere, most notably shelter. The Cleveland Fed says shelter and health care are the two largest differences.
3. Formula. BEA says PCE uses a Fisher-Ideal formula and the CPI a modified Laspeyres formula. The Cleveland Fed adds that PCE weights are updated monthly while CPI weights are updated annually, and says this faster updating helps PCE capture substitution, which partly explains why PCE inflation tends to run lower. Since 2000, it says, annual CPI inflation has averaged 0.39 percentage points more than PCE inflation.

PPI: prices at the factory gate

The BLS describes the PPI as measuring "the average change over time in the selling prices received by domestic producers." It adds that "PPIs measure price change from the perspective of the seller," whereas the CPI measures it from the buyer's side. Companies report prices in a monthly survey.

Two features matter when you read it.

• Final demand vs intermediate demand. The headline PPI is final demand: products sold for personal consumption, business capital investment, government purchase or export. Intermediate demand covers goods, services and construction sold to businesses as inputs to production. BLS says the intermediate-demand stage indexes can be used to study how price changes pass between stages of production and final demand.
• Scope. PPI includes capital goods, exports and government purchases, so it is not a measure of what consumers pay. BLS also notes that indexes can be recalculated for up to four months after first publication.

August 2026: the three gauges side by side

Sources: BLS CPI release of 11 September 2026, BEA Personal Income and Outlays release of 30 September 2026, and BLS PPI release of 10 September 2026\. CPI and PPI figures are twelve-month changes as published (unadjusted); monthly changes are seasonally adjusted. The PPI "core" is the BLS series that excludes foods, energy and trade services.

Three observations.

Headlines matched, cores did not. CPI and PCE headline inflation were both 3.4%. But core PCE (3.0%) was 0.6 points above core CPI (2.4%), the reverse of the usual pattern of CPI running above PCE. The sources we used do not explain this gap and we do not guess at causes. It is a good reminder to compare the same measure across months before drawing conclusions.

Energy was a large part of the headline. The CPI energy index rose 16.3% over twelve months in August 2026, compared with 2.7% for food and 3.0% for shelter. That is why the headline numbers sit well above core CPI.

PPI ran well above CPI. Final demand PPI was 2.0 points above headline CPI and 2.3 points above core CPI (our calculation). That does not mean consumer prices will catch up. The PPI covers sales to businesses, government and exporters as well as consumers, and BLS presents the intermediate-demand indexes as a tool for studying price transmission, not as a forecast of the CPI.

A methodology change you should know about

The September 2026 PCE release was part of BEA's annual update of the national accounts, with revisions starting in January 2021\. BEA's June 2026 preview describes three changes to how prices are measured:

BEA's preview does not quantify the effect on inflation. RBC Economics estimated that the changes would lower core PCE inflation by about 0.18 percentage point, and cautioned that a lower-than-expected reading should not be read as a sign of disinflation. That is a bank's estimate, not an official figure. The practical point: when two PCE readings are compared across the release date, they may not use the same method, so check the revision notes before comparing old and new figures.

• For portfolio management and investment advice, BEA stops deflating spending with the PPI and uses a quantity extrapolator based on the BLS Current Employment Statistics (CES).
• For legal services, a BEA composite index replaces the CPI series from 2024, because of erratic, uncorroborated unpublished CPI values.
• For computer software and accessories, a BEA composite of CPI and PPI series replaces the CPI category.

Which gauge for which question

Practical rules for reading an inflation release

This article is general information, not financial advice. The formula example uses hypothetical prices.

1. Name the index. "Inflation was 3.4%" can mean CPI or PCE, and the two can diverge.
2. Separate the headline from the core, and look at the monthly change as well as the twelve-month change. A twelve-month figure can stay high for months after the monthly pace has slowed.
3. Check the revision notes. PPI and PCE figures can be revised after first publication, and PCE revisions can change methods as well as data.
4. Do not treat PPI as a forecast of CPI. Use it as an input to understanding costs.
5. Compare like with like. Core CPI against core PCE gives you the gap that matters for the Fed.