Rent nobody pays outweighs food, gas and new cars combined

Editor’s note — 2026-08-02. This article replaces an earlier version published at this address on July 14, 2026. The original explained the Consumer Price Index without citing a single source or stating a single figure from it. This version is rebuilt on the Bureau of Labor Statistics’ own published weights and links to every document it relies on. The earlier version’s conclusions are not retracted; they were unsupported, which is the defect being corrected. Root cause: the commissioning instruction, not the reporting.

On July 14, 2026, the Bureau of Labor Statistics reported consumer prices fell 0.4 percent in June on a seasonally adjusted basis (the June release), the sharpest one-month drop since April 2020, when it fell 0.8 percent. Shelter barely moved. It rose 0.1 percent, which BLS calls “the smallest 1-month change reported for that index since January 2021,” even though shelter alone carries 35.6 percent of the index’s weight under the December 2025 weights.

That split is not a quirk of one month: owners’ equivalent rent of residences, a price nobody is ever billed for, carries more weight in the CPI-U than food, gasoline and new vehicles combined.

The line item nobody is billed for

BLS updates these weights every year. The December 2025 table, computed on a 2024-weights basis, carries the ones behind this summer’s headlines (Relative importance of components in the Consumer Price Indexes). Owners’ equivalent rent of residences is 26.2 percent of the CPI-U. Food is 13.7 percent, gasoline is 2.9 percent and new vehicles are 3.8 percent, and together they come to 20.4 percent, which is 5.8 percentage points less than owners’ equivalent rent alone. Anyone can redo the subtraction next year.

Narrow the definition to owners’ equivalent rent of primary residences only, dropping vacation homes from the count. The share falls to 25.2 percent. The category still leads food, gasoline and new vehicles combined, by a smaller margin.

What the survey question sets

Ask a homeowner what their house would rent for, and the Bureau of Labor Statistics listens for exactly one reason: the Consumer Expenditure Survey question behind owners’ equivalent rent sets how much the line counts in the basket. It does not set the price. BLS is direct about it. These responses “are not used in estimating price change for the shelter categories, only the weight.” Owner-occupied homes are never priced at all.

Every month, the housing survey behind the index samples rental units only, never the owner-occupied homes it is trying to describe (Measuring Price Change in the CPI: Rent and Rental Equivalence). From that rental sample, BLS models what each owner’s home would earn if it were rented out. That modeled figure, not a homeowner’s guess, is what moves the index every month.

The Handbook of Methods corroborates this independently. OER is built to track the rental value of the owner-occupied housing unit, not its purchase price. The document says the same about its data. Its figures come from actual rents, not the implicit rents owners believe they would charge. A survey of owners sets how much the line counts. A sample of renters sets how it moves.

Treating an owned home as a capital good

BLS has a reason for building the index this way, and the reason is not a dodge. The CPI program treats an owned home the way it treats a factory or a delivery truck: as a capital good. Not something a household consumes each month. In BLS’s own words, owned housing is “capital (or investment) goods distinct from the shelter service they provide, and therefore not as consumption goods.” A house purchase is treated as investment, not consumption. It sits outside a cost-of-living index the same way a stock purchase would.

Mortgage interest, property taxes and most maintenance are excluded for the same reason. What the index prices instead is the shelter service the house produces, priced the way a rental market prices any other shelter service. It is a coherent accounting choice. It is not the choice most people assume BLS made.

Not the index the Fed targets

CPI is not the only official inflation number Washington runs on. The Federal Reserve’s own longer-run target is different. It is 2 percent, defined on the price index for personal consumption expenditures, not on the CPI-U described here. The Social Security Administration runs on a third variant. Benefit increases are indexed to the Consumer Price Index for Urban Wage Earners and Clerical Workers (CPI-W). This year’s cost-of-living adjustment came in at 2.8 percent (cost-of-living adjustment).

Why the headline missed your month

Energy carries 6.4 percent of the index. In June it fell 5.7 percent. That dragged the seasonally adjusted headline down to a 0.4 percent monthly decline. Owners’ equivalent rent carries 26.2 percent of the index. In June it moved only 0.2 percent. A small category swung hard. The heavier category barely twitched. That is the same asymmetry from the top of this piece, and now the mechanism is attached. Weight decides how much a price’s monthly wiggle can move the whole number.

Weighting is one reason a household’s own month can feel nothing like the headline. Basket composition and geography are others. A renter across town and a driver filling up in June lived through different Junes than the nationwide index describes. The arithmetic above explains a share of that gap. It does not explain all of it.

How this piece was made. Off Script Wire uses AI to research and draft. This article’s reporting, analysis and prose were produced that way, then checked line by line against the primary documents it links to by a reviewer who did not write it. The byline is the person who directed the work, reviewed it and is answerable for it. Our standing statement on AI use and verification is on the editorial standards page.