CPI Says Rent Rose Less Than Your Lease Did. Both Are Right.

The CPI shelter index prices the entire occupied housing stock, and by design it smooths new-lease spikes across millions of units that have not been repriced this month or even this year.

What the CPI counts as shelter

In February 2025 shelter carried a relative importance of 35.417 percent inside the all-items Consumer Price Index for All Urban Consumers, according to the BLS’s own supplemental tables. Rent of primary residence—the check a tenant writes—weighed in at just 7.473 percent. The rest, and the reason shelter dominates the index, is owners’ equivalent rent of primary residence. OER runs roughly 25 percent of the full CPI-U, making it the single most influential item the agency publishes. Both components come from the same Housing Survey. The BLS states that the survey produces two indexes: rent of primary residence and owners’ equivalent rent of primary residence. OER is imputed from observed rents, not from house prices, mortgage statements, or a homeowner’s guess about what the place might fetch.

Six panels and the delay that follows

The subindexes move slowly because the BLS collects the underlying data in a deliberate rhythm. The rental sample is split into six panels. Each panel is priced twice a year on a staggered schedule, so each month the bureau visits one panel, records the rent, and calculates a six-month price ratio—the change since that same unit was last priced. The monthly shelter indexes are then built from weighted averages of those unit-level ratios. A brand-new lease does not leap into the number; it enters the averaging gradually, unit by unit, as each panel rotates through its next pricing cycle.

The BLS states this is not a bug. Rent payments are less volatile than prices for food or fuel, and a large sample collected less frequently is more efficient than surveying the same properties every month. One-sixth of the rented housing unit sample is replaced every year, and a newly sampled unit is priced on the panel every six months after initiation. That keeps the sample fresh without eliminating the drag. A lease that resets in March will not fully imprint on the published rent and OER series for many months, and only after thousands of older, unchanged leases have been averaged in with it.

Why OER eats so much of the index

The sheer weight of OER over tenant rent surprises renters who think of shelter as the monthly check they write. The BLS explains it plainly: the average implied rent for homeowners is larger than the average rent that renters pay, and there are more owner households than renter households. Combine a bigger average housing cost with a larger stock of units, and you get a weighting that overwhelms the apartment-dweller’s line item. So when a city’s new-lease market jumps double digits, that surge has to feed through a measure where a quarter of the entire index is the slow-moving implied rent of people who are not moving at all.

The new-tenant yardstick is a separate animal

The BLS does know about the gap. Its New Tenant Rent Index, a research series, isolates price changes for units with a new lease signed during the reference period. The agency’s own documentation says the official CPI rent component measures the change in all rents—new leases, renewals, and rents locked mid-term. The new-tenant series strips away the inertia of sitting tenants. When asking rents scream higher, that index can spike while the all-rent CPI crawls. The BLS points out, without hedging, that the new-tenant measure is distinct from the official series specifically because the official series captures all rents, not just new leases. That is not a defect; the CPI aims to track what the occupied universe is paying, not what a prospective mover would face if they had to sign today.

What the index won’t tell you

A few misreadings never die. OER is not a homeowner survey. The BLS confirms it is measured from rents, not from home prices. A FAQ from the bureau states that the higher relative importance of OER is only a function of larger implied rent and more owners. The shelter weight itself does not describe your household’s budget share; it is an aggregate expenditure weight derived from a broad sample. Someone whose rent eats forty percent of take-home pay may look at a thirty-five percent shelter weight and conclude the statistic is broken. It is built for the urban average, not for one lease, one neighborhood, or one paycheck.

Every layer of the measurement design pulls the official shelter numbers away from the sharp edge of a fresh renewal. The six-panel rhythm, the six-month repricing window, the heavy OER weight, and the continuous averaging over millions of unchanged units all do the same thing: they dampen. A market shock that feels like a blowtorch to a tenant moving this week looks more like a distant glow in the CPI because the index is measuring the temperature of the entire lake, not the hot spot at the intake valve. That is why both the lease in your hand and the official statistic can be simultaneously right.