The Affordability Paradox: When the Ratio Falls but the Renter Falls Behind
Renter affordability is the subject here: how the cost of housing, household income, location, and payment trouble combine to shape what renters actually experience. The specific question is why an area’s rent-to-income ratio can look comparatively better while renters who live there still feel more squeezed. The practical judgment this supports is simple but demanding: before deciding that renters are better off, identify which renters, which measure, and which time period a statistic describes.
The question matters because “rent” is often treated as if it were one number with one meaning. A headline may report that rents are growing more slowly. A local statistic may show that rent takes a smaller share of income. A payment dataset may show that fewer active renters incurred a late fee. Each statement can be accurate within its own definition, yet the statements need not describe the same population or the same kind of improvement. They can even move in opposite directions without one of them being a mistake.
Affordability is a relationship, not a price tag
A rent level tells us what a dwelling costs. Rent growth tells us how that price changed. Neither tells us by itself whether a household can comfortably pay it. A $200 increase has a different meaning for a household with a high and stable income than for one whose income is already largely committed to necessities. That is why the central measure in the Federal Reserve study is often not rent alone, but rent compared with income.
Even “rent-to-income” has more than one possible subject. It can describe a household’s own rent and income, or it can describe an area average built from the people who happen to live there at a given time. The second measure can change because prices change, because incomes change, or because the mix of residents changes. If higher-income renters move into a lower-rent area, the area’s average income may rise even while landlords raise rents in response to new demand. The resulting ratio may look less troubling without showing that a renter who stayed gained any room in their budget.
A cost-burden measure asks a related but different question. In the Federal Reserve analysis, a renter is cost-burdened when more than 30 percent of income goes to rent. A continuous ratio can rise without crossing that threshold, or many households can cross the threshold while the median moves only modestly. The threshold is useful because it counts a consequential boundary; it is incomplete because it compresses different degrees of burden into a yes-or-no category.
Payment data add another lens. A late fee or a non-sufficient-funds fee records a payment event, not a full household budget. It can reveal cash-flow stress and whether a problem repeats, but it does not tell us everything about rent burden, other debts, housing quality, or what ultimately happened to the household. The Consumer Financial Protection Bureau’s data are valuable partly because they observe payment status at a higher frequency than many household surveys. They are not designed to be a representative national renter survey.
The levels of observation matter as much as the labels. An area statistic is assembled from a population that may gain or lose residents. A household ratio follows the relation between one renter’s resources and housing cost, but a repeated cross-section does not prove that the same renter supplied both years of the comparison. A payment record follows an account within the vendor’s reporting system, not necessarily the renter through every landlord, move, or change in income. These are not flaws that make the measures useless. They are the conditions under which each measure has to be interpreted.
This is why the primary subject is not simply whether rents rose. It is how a housing shock is distributed across places and households, and how that distribution is represented by different measurements. A statistic can be accurate and still answer a narrower question than the one a reader has in mind. “Did the area become more affordable?” may be a question about the average resident; “Can the renter who stayed keep paying?” is a question about an incumbent household. The two questions require different evidence.
The distinction can be stated as a map. Area rent and the area’s income composition produce one kind of signal. A household’s own rent and income produce another. A threshold counts a particular boundary in that distribution. A payment record captures a moment or sequence of cash-flow difficulty. These signals can inform one another, but none should silently stand in for the others.
The pandemic made the distinction visible
The pandemic-era housing shock supplied a particularly clear test because the geography of housing demand changed. The Federal Reserve study finds a pattern consistent with higher-income remote workers moving from high-rent, high-density places toward places with lower nominal rents. Their jobs could remain tied to a high-income labor market even after their homes moved elsewhere. That brought purchasing power into places that had previously been cheaper.
The result was not simply “affordable places stayed affordable.” New-lease rents rose more in places with lower pre-pandemic rents. But rent growth was not differentially higher in places that had lower pre-pandemic rent-to-income ratios. The distinction matters: movers could be searching for a lower monthly price while still bringing incomes that were high relative to the destination’s existing population. A place could be cheap in dollars and still become more expensive for the renters who already lived there.
The timing also matters. New-lease asking prices respond quickly to a change in demand. The rent paid by a household in the middle of a lease usually changes later, when the lease is renewed or the household moves. The Federal Reserve paper therefore compares timely RealPage new-lease data with longer-term American Community Survey measures of rent spending for multifamily renters. These are complementary clocks, not interchangeable readings of one monthly household experience.
The household results show why a cooling market headline cannot by itself settle the affordability question. From 2019 to 2022, the median rent-to-income ratio for the national multifamily renter sample rose from 30.0 percent to 32.2 percent. For lower-income renters, defined relative to the income distribution in their local PUMA, it rose from 63.0 percent to 79.7 percent. The share of all renters classified as cost-burdened rose from 49.9 percent to 53.5 percent; among lower-income renters, it rose from 85.9 percent to 88.3 percent.
Those figures do not mean that every renter experienced the same change. They do show why an analysis focused only on nominal rent growth is incomplete. A rent increase is absorbed differently depending on income, and lower-income households have less room to absorb a given shock before it displaces another essential expense.
The rent-spending comparisons reinforce that point without implying that every place followed the same path. Between 2019 and 2022, average multifamily renter spending rose 19.4 percent in the lowest pre-pandemic rent quartile, 22.2 percent in the second, 20.5 percent in the third, and 16.3 percent in the highest. When places were grouped by pre-pandemic rent-to-income ratios instead, the increases were much closer together, ranging from 18.7 to 19.4 percent. Lower-income renters saw somewhat larger increases in the rent-based groups, including a 24.8 percent rise in the second quartile. The pattern says that lower nominal rent did not shield renters from rising costs; it does not by itself say which households moved or how their incomes changed.
There is also a difference between a population result and a personal forecast. The Federal Reserve study focuses on multifamily buildings with five or more units and uses PUMAs, which are the smallest areas with timely rent and income information in this analysis. RealPage coverage is not nationally representative, even though the covered PUMAs contain most of the ACS multifamily renter population used in the comparison. The study omits the experimental 2020 one-year ACS product. Those choices do not erase the observed pattern, but they do limit how far it can be extended—to rural renters, single-family renters, or a particular household’s future bill.
What the numbers can—and cannot—tell us
The second source shows that broad payment conditions among observed active renters later improved. In the CFPB data, the fraction of renters with a late fee in the previous twelve months rose from 15.4 percent at the end of 2021 to 23 percent at the beginning of 2023, then fell to 14 percent by November 2024. The agency says this pattern could suggest improving financial conditions among active renters, and reports that the decline was not primarily driven by changes in the composition of its rental-payment data.
That is meaningful evidence of improvement within the stated denominator. It is not proof that every renter recovered, or that the pressure captured by the Federal Reserve’s 2019–2022 affordability measures disappeared. The CFPB rate uses a trailing twelve-month window and includes renters with recent transaction history in the vendor data. Its population, clock, and outcome differ from the Federal Reserve’s multifamily ACS sample and area comparisons.
The Federal Reserve’s most revealing result is similarly easy to overread. The lowest pre-pandemic rent quartile saw the smallest increase in its area median rent-to-income ratio—0.6 percentage points, from 30.3 to 30.8—while the second, third, and fourth quartiles saw increases of 2.5, 2.1, and 1.5 points. It would be tempting to call that evidence that the cheapest places became more affordable. The authors instead caution that population composition is likely doing substantial work. Higher-income arrivals can push the area ratio down while demand pushes local rents up. The data do not identify the outcome for the renters who stayed.
This is the core paradox, but it is not a contradiction in the data. An area ratio can improve because the people counted in the numerator and denominator have changed. A payment-incidence rate can fall because fewer observed active renters encounter a fee, while a subset of renters continues to repeat late payments or carry balances. A lower-income household can become more burdened even as the average income of its area rises. Each statement refers to a different layer of the rental system.
The strongest conclusion available at this stage is therefore conditional. Renter affordability may improve for an aggregate population on one measure and one time scale while deteriorating for lower-income or incumbent renters on another. To decide which interpretation applies, a reader must ask four questions before accepting the headline: Who is in the population? What clock does the measure use? What is the denominator? Could migration or selective observation have changed the composition?
The evidence here cannot follow the same renter simultaneously through a move, a lease renewal, an income change, a rent-to-income ratio, and a payment delinquency. That missing longitudinal view is not a minor technical footnote; it marks the boundary between an area-level signal and a claim about lived experience. The next step is to examine each measure on its own terms, beginning with the difference between a new lease, an existing renter’s bill, an income-adjusted ratio, a threshold, and a payment event.