Five Ways to Measure Whether Renters Are Better Off
Renter affordability is measured here through several related but nonidentical signals: the price of a new lease, the rent paid by an existing tenant, rent compared with income, the share crossing a cost-burden threshold, and recorded payment trouble. The specific problem is that these measures use different populations and clocks. Learning to identify those boundaries lets a reader decide what an affordability number actually says before deciding whether renters are better off.
A useful statistic is not simply a number with a source attached. It is a number produced by a measurement choice. The choice determines whose experience enters the calculation, when that experience is observed, and what comparison is being made. Two measures can both be carefully calculated and still answer different questions. Much of the apparent contradiction in housing news comes from treating those questions as if they were the same.
A rent statistic is an observation with a boundary
The first distinction is between a rent level and an affordability measure. A rent level tells us the monthly price of a dwelling or the average amount paid by a group. Rent growth tells us how that price changed over time. These measures describe the housing market, but they do not incorporate the resources available to the households paying those prices. A place can have low rents and still be difficult for its residents if local incomes are lower still.
The second distinction is between a new-lease price and the rent paid by the average existing renter. A new-lease asking price is a current market signal: it describes what a prospective tenant faces when looking for a home now. Most existing tenants are not repricing their lease every month. A lease may hold the rent fixed for a period, commonly a year, and the household may then renew, move, or negotiate under different conditions. For that reason, measures that include existing renters usually lag new-lease measures by about a year.
The Federal Reserve study uses this difference deliberately. Its RealPage data provide timely new-lease prices for multifamily buildings, while the American Community Survey supplies a longer-term measure of rent spending for renter households. National new-lease growth fell to about negative 2 percent in 2020 and later reached roughly 15 percent year over year in spring 2022. That surge identified a rapid change in the market for new tenants. It did not mean that every renter’s bill rose by 15 percent at the same moment. The average existing renter’s exposure arrived through the lease cycle.
The third distinction is between rent and rent-to-income. A ratio asks how large the housing payment is relative to the resources available to pay it. The ratio can be calculated for a household or summarized across a population. Those are not interchangeable. A household ratio links one renter’s payment to that renter’s income. An area ratio describes the relationship among the people and prices included in that area’s average. It may change when rents change, when incomes change, or when different households enter and leave.
This is why a rent-to-income ratio can be more informative than rent alone but still require interpretation. Suppose an area attracts households with higher incomes. The income side of the local ratio may rise faster than the rent side, making the average ratio look stable or improved. That could reflect a genuine improvement for new arrivals, a compositional change that says little about people who stayed, or both. The ratio is not wrong; it is answering a population-level question rather than a direct question about each incumbent household.
The fourth distinction is between a continuous ratio and a threshold measure. In the Federal Reserve analysis, cost burden means spending more than 30 percent of income on rent. A median rent-to-income ratio can move from 30 to 32 percent while the number of households crossing 30 percent changes substantially, depending on how the distribution is shaped around the threshold. Conversely, the cost-burden rate can rise even when the median moves only slightly if many households are near the boundary. The ratio preserves more information about degree; the threshold makes a particular level of pressure easy to count.
Neither measure is a complete account of a household’s finances. The study uses gross rent, which includes utilities, and excludes renters with zero income or no cash rent from the cost-burden calculation. Lower-income renters are defined relative to the income distribution in their PUMA, not by one nationwide dollar cutoff. Those definitions are not technical decoration. They tell us which households are being compared and prevent a local-relative category from being mistaken for a universal measure of poverty.
The fifth distinction is between affordability and payment status. A late fee records that a landlord charged for a late payment; a non-sufficient-funds fee records that there were not adequate funds when a payment was processed. These observations can reveal cash-flow stress, especially when they recur. But they do not directly measure the rent-to-income ratio, the quality of the housing, other debts, or whether a household ultimately moved or was evicted. Payment status is a different outcome that can complement affordability measures without replacing them.
The five measures therefore form a set of windows rather than a ladder from bad to good. A new-lease price is not an early version of a household ratio. A cost-burden rate is not a more accurate median. A late-fee rate is not a direct count of all households unable to afford rent. Each window becomes useful when the reader keeps its frame visible.
The denominator decides how much a number can carry
The fastest way to misread an affordability claim is to skip the denominator. “Renters” can mean all renter households, multifamily renters in the areas covered by a particular data source, active renters with recent transaction history, or renters who already experienced a particular fee. The denominator narrows the population. A percentage calculated within that population cannot automatically be generalized to everyone outside it.
The Federal Reserve evidence illustrates the issue at the geographic level. The analysis focuses on multifamily buildings with five or more units and uses Public Use Microdata Areas, or PUMAs, because they are the smallest areas with timely information on both rents and renter incomes for the study. The RealPage data are not nationally representative, although the covered PUMAs contain more than 95 percent of the ACS multifamily renter population used in the analysis. The results are therefore informative about a large multifamily, largely metropolitan population, not a direct census of every rural, single-family, or small-building renter.
The CFPB evidence uses a different denominator. Its rental-payment data are drawn from a vendor’s database and are limited, for the trend calculations, to renters with transaction history in the previous twelve months. The late-fee percentage is also a trailing twelve-month measure: it asks whether a renter incurred a fee at any point in that window, not whether the renter was late on one specified day. That design can detect recurring exposure and produce a high-frequency view of changing conditions, but it should not be compared with a point-in-time survey rate as if they shared a denominator.
Conditional statistics narrow the denominator further. Among renters with at least one late fee, the relevant question is how often additional late fees occurred. Among renters with a positive outstanding balance, the relevant question is how large the balance was for that group. Those statistics are useful precisely because they examine severity or recurrence among affected renters. They are not the average fee or balance carried by all renters. If smaller balances disappear from the observed group, the median balance among those who remain can rise even while the overall share with a balance falls.
The time window matters just as much. New-lease prices can respond to a demand shock before existing tenants see a renewal. A 2019–2022 ACS comparison spans a multi-year change and is not a monthly account of a particular household. A rolling twelve-month late-fee measure records a history of events rather than a single current condition. Comparing these measures without aligning their clocks can make a delayed effect look like a disagreement or make two different periods look simultaneous.
Composition is the fourth check. An area average can change because the underlying prices changed, because incomes changed, or because the people counted changed. A household-level repeated observation would be better suited to answering whether the same renter became more or less able to pay. The available Federal Reserve comparison does not provide that panel. The CFPB payment data follow observed account statuses, but a renter can leave the reporting system for reasons the data cannot identify. Neither source should be asked to answer a same-household question it was not designed to answer.
The Federal Reserve tables show how much the grouping rule matters. When places are sorted by pre-pandemic rent, average multifamily rent spending rose 19.4 percent in the lowest quartile, 22.2 percent in the second, 20.5 percent in the third, and 16.3 percent in the highest between 2019 and 2022. That ordering invites a story about cheaper places receiving more rent pressure. When the same broad comparison is organized by pre-pandemic rent-to-income instead, the increases are much closer, from 18.7 to 19.4 percent. The second table is not correcting the first; it is answering a different question about what “affordable” means.
The comparator can therefore change the apparent geography without changing the underlying rent bills. A place may belong to the low-rent group because its nominal prices are low, yet not belong to the low-rent-to-income group if its residents also have low incomes. Conversely, a high-rent place can have a relatively strong income base. Sorting by the ratio recognizes that affordability is relational, but it still leaves the composition problem: the area’s income base may be changing because households move. A careful reader should ask not only which groups are compared, but why that grouping is relevant to the claim being made.
This also explains why a single national median can be insufficient. The Federal Reserve reports that lower-income renters’ average rent spending rose by 24.8 percent in the second rent quartile, compared with 17.6 percent in the fourth. A national average could contain both experiences without showing the difference. The continuous ratio and cost-burden rate add information about that distribution, while payment transitions can show whether the pressure is a temporary event or a repeating problem. No one measure is redundant when it captures a different part of the mechanism.
The evidence could become stronger with a panel that followed the same renter across location, income, lease status, and payment history. Such a panel could distinguish a rent increase faced by an incumbent from a higher-income household newly counted in the area. It could also test whether a late fee was followed by recovery, repeat delinquency, a move, or another outcome. Until that evidence exists, precision about what each denominator includes is not excessive caution; it is the method for keeping an accurate observation from becoming an inaccurate generalization.
These four checks—population, clock, denominator, and composition—turn a vague claim into a precise one. Consider the statement that rental conditions improved after the pandemic peak. The CFPB data support a decline in the fraction of active observed renters with a late fee in the previous twelve months, from 23 percent at the beginning of 2023 to 14 percent in November 2024. That is a meaningful improvement within that payment-data population and measure. It does not, by itself, show that the median lower-income renter’s rent-to-income ratio returned to its earlier level, that an incumbent in a rapidly changing area became safer, or that every kind of rental stress declined.
The same discipline applies to an apparently negative result. The Federal Reserve’s national multifamily renter median ratio rose from 30.0 to 32.2 percent between 2019 and 2022, while the lower-income group rose from 63.0 to 79.7 percent. Those are strong signals of worsening income-adjusted pressure in the measured population. They do not establish that every household saw the same increase, nor do they describe the later payment trend in the CFPB data. The measures speak to different layers of the problem.
A defensible reading therefore keeps the claim no broader than the observation. A new-lease number can tell us about current market pricing. An existing-renter measure can tell us how average spending changed over a longer interval. A ratio can tell us how cost relates to income. A threshold can show how many households crossed a defined boundary. A payment record can show whether observed cash-flow trouble is recurring. To move from any one of these to a conclusion about renter welfare, the reader must add the missing context rather than silently assume it.
That is the practical measurement model for the rest of the discussion. Once each statistic has a clear boundary, the next question is no longer simply which places had the lowest rent. It is what process caused lower nominal-rent places to receive stronger rent pressure, and why that geographic pattern need not track rent growth relative to local incomes.