Home truths

America’s September inflation report brought another increase in the cost of shelter just as workers’ purchasing power weakened. The Bureau of Labor Statistics reported on October 14th that consumer prices rose 3.7% from a year earlier and 0.6% over the month, after seasonal adjustment. Petrol drove much of the immediate shock. Housing presents a slower-moving problem: the price of accommodation, the cost of buying a home and the income available to pay for either can follow different paths.

The shelter index rose 0.4% in September. Owners’ equivalent rent, its measure of the service provided by owner-occupied homes, increased 0.3%. The broader shelter category also includes accommodation away from home, such as hotels. Reading its monthly movement as the change in the cost of buying or renting a permanent home would already be a mistake.

A service, not a purchase

The BLS treats a dwelling as both an asset and a source of accommodation. CPI measures the latter. Buying a house is an investment transaction; living in it means consuming a continuing service. For tenants, rent offers an observable price. For homeowners, statisticians estimate what comparable accommodation would cost to rent, whether or not the owner still has a mortgage.

Actual rents collected from rental properties drive changes in owners’ equivalent rent. Homeowners’ own estimates of their properties’ rental value serve a different purpose: they help establish expenditure weights. These determine how much importance owner-occupied shelter receives in the consumer basket. The monthly price change is not calculated by asking owners how much they think their house has appreciated.

Mortgage interest and house-purchase costs fall outside this consumption measure. They remain central to a buyer’s decision. A prospective borrower must finance a particular property at the terms available to them, while an existing owner may retain an earlier fixed-rate loan. Similar homes can therefore provide similar housing services while imposing different cash payments. Rental equivalence is designed to compare the services, rather than reconstruct each household’s finances.

For a prospective buyer, the omitted information is consequential. A shelter index cannot reveal the deposit required, the mortgage payment they could obtain or the homes available within their budget. Adding these figures to an affordability assessment would answer a different question from CPI; substituting a house-price index for rental equivalence would not repair that distinction.

Renewals remember

For tenants, the mismatch arises partly from timing. A new lease measures a transaction at the market’s edge. A rent index covering occupied homes also includes households whose contracts have yet to reset. A moderation in newly agreed rents can consequently coexist with increases for people renewing older, cheaper leases.

Cleveland Fed researchers examined this mechanism in October 2024. Their model linked the gap between new and continuing tenants’ rents to how often people moved and how much landlords passed market changes into existing contracts. During the pandemic, the authors found, rents for continuing tenants had risen more slowly than those for newcomers. Subsequent increases could narrow that gap even after new-tenant inflation had slowed.

This is a distinction between rent levels and their rate of increase. Slower growth in a newly offered rent does not necessarily make it cheaper than an existing tenant’s contract. That helps explain why a cooling market can take time to appear in an all-tenant measure. It does not establish where the gap stands in September 2026: the Cleveland analysis describes an earlier episode.

The BLS’s rotating housing sample also means its monthly reading is not a census of new deals. Comparisons require consistent coverage of tenancy, location and housing type. A listing for an empty flat and the rent paid by a long-standing occupant may both be accurate observations without measuring the same transaction.

The income side

September’s earnings release supplies a timely reason to look beyond rents alone. Real average hourly earnings for employees on private non-farm payrolls fell 0.5% from August, seasonally adjusted, according to the BLS. Prices rose faster than average hourly pay. This is an aggregate measure of earnings per hour, not an estimate of every renter’s income or disposable household resources.

To examine the latter, Cleveland Fed economist Lara Loewenstein studied income left after rent in research published this March. Her observation period was 2019–2023. The bottom third of renters by income saw their real income after rent fall by more than 7% a year, over a quarter in total; the top third saw essentially no change. The result shows why a common inflation rate need not translate into a common ability to absorb it.

The study compares groups in successive surveys rather than tracking identical households. It cannot establish the present hardship of a particular tenant. Combined with the current earnings release, however, it identifies what a housing-affordability assessment needs to measure: rents, the resources available to pay them and the money left for other necessities. September’s shelter index answers the first question only partially. Neither its statistical consistency nor a later slowdown in inflation would, by itself, make the rent affordable.