CPI vs. PCE: Why the Fed Watches a Different Inflation Number

America runs on two inflation numbers. The one in the news is the Consumer Price Index, published mid-month by the Bureau of Labor Statistics. It sets Social Security COLAs, adjusts TIPS and I Bonds, escalates leases, and powers this site’s calculator. The one that moves interest rates is the Personal Consumption Expenditures price index (PCE), published about two weeks later by the Bureau of Economic Analysis, and it is the index the Federal Reserve’s 2% target refers to. The two usually tell the same story at different volumes: PCE inflation has run a few tenths of a percentage point below CPI inflation in most years, and in dramatic years the gap widens: in June 2022, CPI peaked at 9.1% while PCE topped out near 7%.

The gap isn’t an error. The indexes answer subtly different questions, and the differences come down to four things: whose spending counts, how it’s weighted, what formula combines it, and what happens after publication.

Scope: out-of-pocket vs. on-your-behalf

The CPI measures what urban households pay out of pocket. PCE measures all goods and services consumed by households, no matter who pays. The difference sounds academic until you look at healthcare: when your employer’s insurance plan or Medicare pays a hospital, that spending is invisible to the CPI but fully counted in PCE.

The consequence is a very different-looking basket. Healthcare is roughly twice as large a share of PCE as the medical care share of the CPI, while housing, which makes up about a third of the CPI as explained in how inflation is calculated, carries roughly half that weight in PCE. So in years when rent surges, CPI runs hotter; when medical costs surge, PCE feels it more. PCE’s population is broader too: the CPI covers urban consumers (about 93% of people), while PCE covers everyone, including rural households and nonprofit institutions serving them.

Weights: asking households vs. counting sales

The CPI’s weights come from the Consumer Expenditure Surveys: households recording what they buy. PCE’s weights come from the business side of the economy: the sales and revenue data underlying GDP. Businesses’ books and households’ diaries never quite agree (people are famously bad at remembering what they spend on alcohol, for one well-documented example), and those discrepancies feed directly into the weight differences between the indexes.

Formula: fixed basket vs. moving basket

The CPI is a Laspeyres-type index: between weight updates, it prices an essentially fixed basket, assuming you keep buying roughly the same things as prices shift. PCE uses a Fisher-ideal chained formula that updates expenditure shares continuously. When beef gets expensive and shoppers shift to chicken, PCE follows them almost immediately.

Substitution is real behavior, so the chained formula tracks the cost of maintaining satisfaction more closely, and it mechanically produces a lower measured rate. This “formula effect” is a substantial, persistent chunk of the CPI-PCE gap, and it’s the same logic behind the Chained CPI that now indexes federal tax brackets.

Revisions: carved in stone vs. always provisional

A published CPI figure (not seasonally adjusted) is never revised, which is essential for an index that settles contracts, since nobody wants last year’s rent escalator retroactively changed. PCE, as part of the GDP accounts, is revised repeatedly: monthly updates, annual revisions, and periodic comprehensive overhauls. For a contract that’s a bug; for understanding the economy it’s a feature: PCE can incorporate better data later, while the CPI must live forever with its first estimate.

Why the Fed chose PCE

The Federal Open Market Committee made it official in its January 2012 statement of longer-run goals: the 2% target is measured by the annual change in the PCE price index. The reasoning follows directly from the differences above: PCE covers all consumption for the whole population rather than out-of-pocket urban spending, handles substitution realistically, and can be revised toward accuracy. In practice the Fed’s internal conversation centers on core PCE (excluding food and energy) and on research measures like the Dallas Fed’s trimmed-mean PCE, which strip out noise to reveal the trend.

One practical implication for reading the news: because PCE typically runs a few tenths cooler, “2% PCE” corresponds to roughly 2¼–2½% on the CPI. When headlines say inflation is above target, check which index they mean, because the CPI can sit at 2.4% while the Fed’s measure is effectively at its goal.

Why the CPI still runs your life

None of this demotes the CPI. The dollar amounts that actually hit your account are almost all CPI-indexed: Social Security benefits (via CPI-W), TIPS and I Bond returns (via CPI-U), tax brackets (via Chained CPI), union contracts, commercial leases, alimony escalators. And the CPI’s headline 12-month rate, the one on our current inflation rate page updated with every monthly release, remains the number the public means by “inflation.”

There’s also a longevity argument: consistent CPI data reaches back to 1913, while PCE begins in 1959. If you want to know what $100 during the Great Inflation is worth today, or anything else across a century of prices, CPI is the only official series that can answer, which is why this calculator is built on it. The exact series and formulas are documented in the methodology.

How wide is the gap in practice?

Since 2000 the CPI has outrun PCE by roughly 0.3 percentage points a year on average. That is small in any single month, but compounding. Run the same twenty-five years through both indexes and the CPI shows meaningfully more total inflation; over a full generation of prices the difference amounts to several percentage points of purchasing power. The direction is remarkably consistent: years in which PCE runs above CPI are rare, because the formula effect (substitution) pushes one way and the weight differences usually cooperate.

The gap is at its most informative when it stretches. Housing-driven inflation, with 2022 and 2023 as the textbook case, widens it, because shelter counts for roughly twice as much in the CPI. Healthcare-driven inflation narrows or flips it. During the 2009 deflation year, both indexes went negative, but CPI fell further for the same reason it rises further: the categories households pay for directly, gasoline above all, were the ones collapsing.

Which one should you use?

A practical rule: match the index to the cash flow.

• Estimating what your salary, savings, or a past price is worth across years? Use CPI: it’s what this calculator uses, what your contracts likely reference, and the only official series that reaches back to 1913.
• Judging whether the Fed is likely to raise or cut rates? Watch core PCE, because that’s the number the FOMC is graded on, while the CPI can overstate how far policy is from target.
• Writing an escalation clause? CPI, always: it’s never revised, so both parties can settle on a number that will still exist unchanged in ten years.

The one-table summary

Same economy, two lenses. When they diverge sharply, as during 2022’s rent-driven surge or the 2009 deflation scare, the gap itself is information: it tells you which prices, paid by whom, are doing the moving.