In the third quarter of 2006, one inflation gauge rose 3.0 percent at an annual rate. Another rose 3.7 percent. Same economy, same quarter, a gap of 0.7 percentage point. That old example from the Bureau of Economic Analysis is a good place to start. It is a methodology illustration, not news. It shows why two credible inflation numbers can sit side by side and still disagree.

A crumpled grocery receipt and a folded rent renewal notice side by side on a kitchen table, two different records of what prices did to one household.
A grocery receipt and a rent renewal notice laid next to each other, two different records of what prices did to one household.

Who publishes each index and what it tracks

The Consumer Price Index, or CPI, comes from the Bureau of Labor Statistics. It tracks prices paid by urban consumers. Think of the checkout line, the rent notice, the utility statement. It centers on out-of-pocket spending by households.

The Personal Consumption Expenditures price index, or PCE, comes from the Bureau of Economic Analysis. Its net is wider. It covers spending by and on behalf of the personal sector, which includes households and the nonprofit institutions that serve them. Some of that spending never touches a household wallet directly.

The Federal Reserve's inflation page tracks PCE, and the Cleveland Fed publishes median measures for both indexes. BLS publishes CPI. Both are federal statistical sources, and BLS and BEA have each documented how the two differ. A headline number always comes from a specific agency with a specific scope. Find that first.

Why the two numbers can differ

BEA sorts the differences into four buckets: formula, weight, scope, and other effects. They can push in opposite directions, so they sometimes offset each other. That is why a small gap can hide large differences underneath.

The formula effect is about math. PCE uses a Fisher-Ideal formula. CPI uses a modified Laspeyres formula. In plain terms, the two handle shoppers switching products differently. As a simple illustration of the idea, if beef gets pricey and households buy more chicken, the two formulas can respond to that shift in different ways.

The weight effect is about importance. Each index gives each category a share of the total, so a category can count for more in one index than in the other. The scope effect covers items counted in one index and not the other, and BEA adjusts for those when comparing. The remaining bucket, other effects, holds seasonal adjustment differences, price differences, and leftover residuals.

A worked example and why gaps can be large

Back to 2006. BEA's breakdown of that 0.7 point gap is the useful part. The weight effect added 0.84 point. Other effects added 0.29. The formula effect added 0.16. A net scope effect of negative 0.50 point pulled the gap back down. Imputed rent for owner-occupied housing, an estimate of what homeowners would pay to rent their own homes, supplied 0.56 point of the weight effect by itself.

So housing treatment was a big lever in that quarter. A rent notice counts differently depending on whose ruler you use.

A one-quarter gap this size does not mean one index is broken. Each is the output of its own method.

More recent commentary says the gap can move. PIMCO reports that the year-over-year PCE-minus-CPI gap moved from a historically negative 30 to 40 basis points to positive 60, which it describes as one of the largest reversals since 1985. A basis point is one hundredth of a percentage point. PIMCO also reports a core PCE three-month annualized pace of 2.4% in November 2025 and 4.4% in March 2026. Those are one firm's readings of the data. Treat its views on the outlook as opinion, and check the official series yourself.

How to read an inflation headline

Ask four questions, in order.

  1. Which index? A CPI figure and a PCE figure are not interchangeable, even if both are called inflation.
  2. Headline or core? Headline counts everything. Core leaves out food and energy, the volatile items. That is a choice of scope, and a good story says so.
  3. What time frame? A monthly change, a quarterly change stated at an annual rate, and a year-over-year change can look very different from the same data.
  4. Who is forecasting? Outlooks from investment firms reflect their own views. Return to the agency release.

The time frame deserves extra care. As plain arithmetic, a 0.3 percent monthly rise sounds tiny, yet repeated for twelve months it compounds to roughly 3.7 percent. Neither figure is wrong. They answer different questions. The Cleveland Fed also publishes an infographic comparing the two indexes, plus median CPI and median PCE indicators that give another angle on the broad trend.

What these sources do not tell consumers

The official sources explain how each index is built and how the Fed and statistical agencies use them. They do not say which one a household should use for budgeting. They do not map either index onto one family's real costs, which depend on where you live, whether you rent, and what you buy.

They also do not give the latest month's values, so check the current agency release for any specific figure. Nor do they cover credit, savings rates, or Social Security adjustments.

What you can take from them is a way to read the numbers. You can see why two credible figures differ, what a headline is measuring, and where the methodology is published. This is general education, not personal financial advice. For your own situation, check the official sources or talk with a qualified professional.

Primary sources: the BEA explanation of the PCE and CPI differences, the BLS research paper comparing the two, and the Cleveland Fed's explainer on consumer price data.