CPI, PCE, GDP Deflator, HICP: The Four Official Inflation Measures Compared

The four official inflation measures — CPI, PCE, GDP deflator, HICP — diverge structurally by basket, weighting and scope. Reading any one without knowing the other three produces systematic misreadings.

CPI, PCE, GDP deflator and HICP measure overlapping but non-identical economic objects. The methodological gaps are structural, not noise — and they explain divergences of 100 basis points or more between official numbers describing the same economy.

A reader who knows only the headline CPI inherits the BLS’s basket choices, weight conventions and scope decisions without realizing it. The PCE deflator says something different about the same household. The GDP deflator says something different again. Knowing the four lets the reader pick the measure that fits the question.

Side-by-side: what each index actually covers

Consumer Price Index (CPI)

The Bureau of Labor Statistics publishes the U.S. CPI monthly, sampling roughly 80,000 prices across about 23,000 outlets and 200 categories. The CPI measures out-of-pocket spending by urban consumers — the headline series CPI-U covers about 93% of the U.S. population (BLS technical notes). It uses a Laspeyres-type framework with weights revised annually since 2023 (biennially before that). The basket is fixed in the short run, which means CPI captures little household substitution between goods.

Personal Consumption Expenditures (PCE) deflator

The Bureau of Economic Analysis derives the PCE deflator from National Accounts data. Coverage is broader: it includes spending by or on behalf of households (employer-provided health insurance, Medicare and Medicaid payments, employer-paid retirement contributions). PCE has used chain-weighting since 1996, updating quantity weights essentially every quarter. PCE is the Federal Reserve’s preferred gauge since 2000, in part because chain-weighting limits substitution bias.

GDP deflator

The GDP deflator, also published by BEA, prices everything produced domestically — household consumption, business investment, government spending and net exports. Imports are excluded by construction. The GDP deflator is the most comprehensive index but also the slowest (quarterly) and the least relevant for household cost-of-living analysis.

Harmonised Index of Consumer Prices (HICP)

Eurostat’s HICP, in force since 1997, harmonizes consumer-price measurement across euro-area members. It uses a Laspeyres framework with weights updated annually based on national accounts consumption data. Unlike CPI, HICP excludes owner-occupied housing entirely from the headline (a separate experimental owner-occupied housing index, OOHPI, has been published since 2016 but is not yet integrated). HICP covers all consumers — urban and rural — in each member state.

Why the indices diverge: three structural drivers

Substitution bias and the weighting method

Annual Laspeyres-style weighting (CPI, HICP) tends to overstate inflation versus chain-weighting (PCE) by 20 to 40 basis points per year, on average, in the U.S. data. The reason is mechanical: when relative prices shift, households substitute, and a chain-weighted index reflects that substitution while a Laspeyres index does not. Over a decade, the cumulative wedge can exceed 3 percentage points on the price level — a figure that matters for indexed contracts and for central-bank target reading.

Scope: what the basket includes

CPI weights are derived from the Consumer Expenditure Survey (out-of-pocket urban consumption); PCE weights come from National Accounts (broader, includes third-party payments). The biggest single difference is health care: PCE assigns roughly 17% weight to health, against about 8% in CPI, because PCE captures Medicare and employer-provided insurance. When healthcare prices behave differently from other components, the two indices diverge sharply. Shelter is another major source of divergence: the CPI assigns shelter roughly 33% weight (mostly via owners’ equivalent rent), against 16% in PCE, and HICP excludes owner-occupied housing entirely.

Geographic and behavioral coverage

HICP harmonization across 20 euro-area members imposes uniform definitions but cannot eliminate the underlying differences in consumption patterns. French food weight in HICP is about 16%, German weight closer to 12%; energy weights vary similarly. A common HICP shock therefore translates differently into national headline numbers — visible in the 2022 cycle when German HICP peaked at 11.6% (Destatis, October 2022) versus 7.3% in France (INSEE, February 2023).

How the divergences played out in 2021-2024

Headline gaps at peak

U.S. CPI peaked at 9.1% YoY in June 2022 (BLS); U.S. PCE peaked at 7.2% in the same month (BEA) — a 190 basis-point wedge at the cycle high. The disinflation that followed was faster on PCE than CPI: by September 2024, PCE was at 2.1% while CPI was at 2.4% (BLS, BEA). The gap reflected the heavier shelter weight in CPI; services and goods were also moving in opposite directions through this window.

The euro area’s harder shocks

Euro-area HICP peaked at 10.6% in October 2022 (Eurostat) — far above the U.S. CPI peak — driven by the energy-import composition and the absence of owner-occupied housing dampening. Disinflation was also faster: HICP fell below 3% by November 2023, while U.S. CPI took until June 2024 to reach the same level. The divergence between regions was not merely a difference in policy reactions; it was partly a difference in what the indices measured.

Use cases: matching index to question

Cost-of-living analysis

For analyzing household purchasing power, CPI (or HICP for euro-area households) is generally the better fit because it tracks out-of-pocket spending. Purchasing-power erosion calculations are most accurate when they reference the consumer index actually faced by the household.

Monetary-policy reading

To read the Federal Reserve, watch PCE and core PCE. The Fed’s 2% target is on PCE — a CPI-equivalent target would effectively be 2.3-2.5% given the typical wedge. The Fed’s 2% target implicitly accepts a higher CPI rate; reading the FOMC statement against CPI rather than PCE produces systematic misinterpretation. Core PCE filters food and energy noise.

Macroeconomic accounting

For converting nominal GDP into real GDP and analyzing growth contributions, the GDP deflator is the right tool — but it does not measure household cost of living and is therefore unsuited to that question.

Indexed contract design

The choice of reference index in an indexed contract is a multi-billion-dollar decision over long horizons. TIPS and OATi mechanics reflect specific index choices: TIPS reference U.S. CPI-U, OATi the euro-area HICP excluding tobacco. A 30-basis-point annual wedge between two candidate indices compounds to roughly 9% over 30 years.

Why central banks pick different targets

The Fed’s preference for PCE since 2000

Before 2000, the Federal Reserve focused on CPI. The shift to PCE was justified in a February 2000 Monetary Policy Report on three grounds: chain-weighting captures household substitution; the broader scope (including third-party medical payments) better reflects total consumption; the weight base is updated more frequently. The wedge between the two — averaging roughly 30 to 50 basis points with CPI running higher — means the Fed implicitly tolerates a higher CPI rate than its 2% target suggests. A reader who tracks CPI against the 2% target will systematically conclude the Fed is too loose.

The ECB’s HICP and the price-stability mandate

The ECB’s choice of HICP was a treaty-level decision: HICP is the only euro-area-wide price index, and the ECB’s mandate is euro-area price stability. The 1998 definition — “below but close to 2%” — was reframed in July 2021 as a symmetric 2% medium-term target after the strategy review. The choice between excluding or including owner-occupied housing in HICP has been a multi-year ECB project; current HICP excludes it, but a parallel index (HICP-H) including OOH has been published since 2024 and may eventually be incorporated into the headline.

The CPI-PCE wedge during disinflation cycles

The 2022-2024 disinflation displayed an unusual pattern: PCE fell faster than CPI, the gap between them widening from roughly 190 basis points at the June 2022 peak to nearly 100 basis points by mid-2023 before normalizing. The reason was almost entirely shelter: CPI’s high shelter weight (~33%) combined with sticky housing dynamics meant the index lagged the rest of the disinflation. PCE, with a 16% shelter weight, captured the broader cooling earlier. Reading “the inflation rate” without specifying which one would have produced opposite policy conclusions in early 2023 — one suggesting the Fed had won, the other suggesting more work to do.

Reading the gap, not just the level

The CPI-PCE wedge is informative on its own. When CPI runs sharply above PCE — as in mid-2022 — the gap is mostly about shelter (CPI weights it more). When PCE runs sharply above CPI, the gap is usually about healthcare (PCE weights it more). Tracking the difference between the two tells the analyst which component is moving the fastest. The same logic applies cross-Atlantic: HICP versus euro-area HICP-at-constant-tax-rates, or HICP versus the new HICP including owner-occupied housing (when published) reveals what the headline is and is not capturing. The complete inflation guide develops the regime framework that makes these comparisons productive; the inflation sub-pillar aggregates monthly readings across measures.

Last updated — 1 May 2026