Homelessness Rises More Quickly Where Rent Exceeds a Third of Income

  • Communities where people spend more than 32 percent of their income on rent can expect a more rapid increase in homelessness.
  • Income growth has not kept pace with rents, leading to an affordability crunch with cascading effects that, for people on the bottom economic rung, increases the risk of homelessness.
  • The areas that are most vulnerable to rising rents, unaffordability and poverty hold 15 percent of the U.S. population – and 47 percent of people experiencing homelessness.

Editor’s note: On Dec. 11, 2018, Zillow hosted a roundtable discussion a roundtable discussion in Washington, D.C., based on this new research. Watch a replay at https://www.zillowgroup.com/thought-leadership/events/.

Communities where people spend more than 32 percent of their income on rent can expect a more rapid increase in homelessness, according to new Zillow-sponsored research on the size and root causes of the nation’s homelessness challenge. The research also estimates that the scale of homelessness nationwide has been undercounted by roughly 115,000 people, or 20 percent.

The U.S. Department of Housing and Urban Development (HUD) estimates that 546,566 people experienced homelessness in 2017, based on counts collected at local levels and reported nationally.[i] But prior research shows those counts to be imprecise and, in all likelihood, far too low. A new analysis by Zillow Research Fellow Chris Glynn of the University of New Hampshire, Thomas Byrne of Boston University and Dennis P. Culhane of the University of Pennsylvania estimates that far more people – 660,996 – likely experienced homelessness in 2017.

Rising rents have long been associated with climbing rates of homelessness. This research demonstrates that the homeless population climbs faster when rent affordability – the share of income people spend on rent – crosses certain thresholds. In many areas beyond those thresholds, even modest rent increases can push thousands more Americans into homelessness.

The U.S. median rent has risen 11 percent over the past five years,[ii] requiring an American renter earning the national median income to spend 28.2 percent of their earnings on the typical U.S. rental. That measure of affordability is up from 25.8 percent historically – and far above the 17.7 percent that median-income households buying a typical home today spend on their monthly mortgage payment.

The new research found two rent affordability thresholds that directly affect homelessness. The first threshold is 22 percent: Any uptick in a community’s rent affordability beyond 22 percent translates into more people experiencing homelessness. The second threshold is 32 percent: Any increase in rent affordability beyond 32 percent leads to a faster-rising rate of homelessness – which could mean a homelessness crisis, unless there are mitigating factors within a community.

30 percent rule not just an adage

It has long been a real estate rule of thumb that a person’s housing costs should not exceed roughly 30 percent of their income, and this research finds empirical evidence to support that adage at the community level. When the share of average income spent on rent in a community begins to meaningfully exceed that line, the risk of housing insecurity and/or homelessness rapidly increases. Establishing this link to community-level rent affordability in the first place provides an important nuance to conventional wisdom about the root causes of homelessness. The 32 percent threshold provides a crucial benchmark for policymakers to gauge exactly where their communities stand – and to adjust programs and resource allocations if they are approaching the threshold.

Across the country, the rent burden already exceeds the 32 percent threshold in 100 of the 386 markets included in this analysis, led by Monroe County in Florida, where the median market rate rent consumes 62.9 percent of the area’s median household income.

These thresholds also help explain why the story of incomes, rents and homelessness does not read the same everywhere. Prior research has operated largely under the implicit assumption that pulling the same levers with the same strength and in the same direction will have an identical effect on homelessness regardless of the community in question. This latest research suggests communities would be wise to take a more nuanced approach in how they contend with unique, local structural factors in seeking to reduce homelessness.

In pricey coastal markets[iii] including New York, Boston, Los Angeles, San Francisco and Seattle, rising rents have created a no-win situation for many financially strapped renters. Although incomes in those markets tend to be higher than the national median, income growth has not kept pace with rents, leading to an affordability crunch with cascading effects: Some high-income renters who typically rent more expensive apartments turn to lower-priced rentals, pushing middle-income renters into even less expensive housing. The lowest earners are forced to work multiple jobs, find multiple roommates and otherwise struggle to make ends meet.

Renters on the bottom rung are at risk of falling completely off the housing ladder if their rents rise even a small amount.

In those pricey coastal markets, renters earning the area’s median income already spend more than 32 percent of their income for rentals priced at the median market rate, crossing that critical second threshold and entering territory in which they should expect local homeless numbers to more rapidly balloon. In Los Angeles, for example, if affordability worsens by 2 percentage points – if renters are required to spend 51 percent of their income on a typical apartment, up from 49 percent at the time of this analysis – the number of homeless is likely to rise by an additional 4,227 people, or 6 percent above estimated 2017 levels.

These affordability thresholds – along with local poverty rates and the level of rent itself – act as signals for understanding how so many parts of the country can face a homeless crisis, even as the number of people in homelessness nationwide is falling. HUD’s annual counts have fallen consistently, and the new research estimates that the number of people experiencing homelessness has fallen by more than 90,000 since 2011.[iv]

Collectively, these signals help identify clusters of communities that share a similar risk of rising homelessness when affordability worsens, rents rise and/or poverty grows.

The cluster where people are most at risk of homelessness due to some combination of these factors includes New York, Boston, Los Angeles and Seattle, which all have crossed the 32 percent affordability threshold – as well as Las Vegas, St. Louis and Anchorage, which have not.

In this cluster, rent is 29 percent higher on average than the rest of the country, and the average homeless rate is much higher than in any other cluster. Almost half (49.7 percent) of renters in this cluster spend more than 30 percent of their income on housing. The CoCs in this cluster – one of six clusters identified by this research – are home to 15.1 percent of the total U.S. population, but a staggering 47.3 percent of the nation’s homeless population.

The clusters also help explain how some CoCs can have shockingly high rent (un)affordability but different homelessness effects.

For example, Washington, D.C., is in the cluster where worsening affordability is most likely to affect homeless numbers. Typical renters there spend 38 percent of their income on the area’s median rent. This research shows that if affordability worsens to 40 percent, 39 more people are predicted to experience homelessness.

The story is different in Chicago, where median-earning renters spend 36 percent of their income on the area’s median rent. Chicago is part of a different cluster, where the risk of increased homelessness in reaction to affordability changes is more muted. In fact, if rent affordability rose to 38 percent in Chicago, 54 fewer people are predicted to face homelessness. (We will explain factors that might drive homelessness down shortly.)

A third cluster of communities that includes the state of Rhode Island, central Minnesota and Provo, Utah, has the lowest average homeless rate, the best average affordability and the lowest poverty rate – and rents in this cluster are 8.9 percent lower, on average, than the country. While 37 percent of the U.S. population lives in the communities that make up this cluster, it has just 14 percent of the homeless population.

Spilling over

A kind of spillover effect may be one explanation for homeless numbers remaining steady or even falling, despite rising rents and worsening affordability. When prices go up in one community, people sometimes move to an area next door for a larger supply of lower-priced homes. After those rentals are filled, homeless rates in that community rise as well.

An example is Riverside, Calif., where rents have climbed more than 20 percent over the past five years – yet HUD’s homeless count has fallen. Next door to Riverside is the Los Angeles CoC, where the homeless count has risen 73 percent in a similar five-year span – some of which can be attributed to factors beyond Los Angeles’ poverty rate and rental market. Spillover from places like Riverside is a possibility.

Because homelessness often appears to involve a jumble of factors that are difficult to tease apart, it can be useful to quantify some of the main factors. Rent affordability is one, as is the poverty rate. The model for this research also produced a baseline estimate for the homeless population in each market – the population that the model indicates might be homeless regardless of housing costs and the poverty rate. It’s an informed, but hypothetical, statistical starting point in relation to the market’s peers before considering other factors. To that baseline, the model adds the effect of the poverty rate and rent affordability, as well as unknown and unobserved “latent” factors that could include everything from local policy efforts to social attitudes toward homelessness to the weather in a given locale. Each force can act as a headwind to push homelessness higher or a tailwind that pulls it down.

In some communities, such as Baltimore, the poverty rate plays a substantial role in the homeless count. In others, like Los Angeles, rent affordability is the larger player. Declining rent affordability affects homelessness in Los Angeles with eight times the force it does in Baltimore.

Houston is a fascinating counterpoint to both Los Angeles and Baltimore. Homelessness is a lot lower in Houston than the expected baseline generated by the model – and than other markets in the same cluster, with similar costs of housing and rates of poverty. In Houston, poverty pushes up homeless rates from the baseline, while rent affordability has little impact. In fact, homelessness has fallen in Houston despite rents rising. There’s such a strong mix of latent factors driving down homelessness – things unique to the community of Houston beyond housing affordability and poverty – that it can be useful to understand Houston’s dynamics at the ground level, in person.

Houston has largely beaten back what used to be a homeless crisis – making it a model for other communities. But it’s not the only one. The research identifies Houston as an outlier, not just for having fewer homeless people than it used to or fewer than its rental costs would indicate, but across the board. The outliers within each cluster, like Houston, bear further exploration.

Like Houston, Tallahassee, Fla., and Lincoln, Neb., have rising rents that nevertheless have not exacerbated their homelessness count. But unlike Houston, latent factors in Tallahassee and Lincoln work in the opposite direction, contributing to higher levels of homelessness than would be expected just by examining their rental costs and poverty rates. Again, the model has identified areas where homeless numbers do not react as expected to rent affordability and poverty rates – and they warrant ground-level research for lessons that might help other communities.

No universal template

Policy solutions to housing unaffordability require solutions as diverse as the markets themselves. The most effective solutions will involve a suite of policies and programs that tackle the multifaceted nature of this problem across a diverse range of housing markets. Places with the worst affordability already know that their current interventions and tools have been dulled by continually hammering against rising rental costs. Now they can see similar areas where that may not be the case.

Some policies will necessarily create more housing in general, while some will need to be more pointedly focused on creating affordable housing in particular. Other policies will need to increase accessibility to the affordable housing that already exists, by taking a hard look at how vouchers are working and what can be done to improve tenant-based assistance. Still other solutions will have little to do with housing at all, and will involve hard work on improving transit from where we live to where we work and other factors. The point is there is no one-size-fits-all approach, because there is no universal template for how homelessness evolves and responds in a given community.


[i] Oct. 2013-Oct. 2018

[ii] The Department of Housing and Urban Development (HUD)’s official homeless count for all Continuums of Care in 2017 was 553,742. This research examined 386 Continuums of Care, where 98.7 percent of the counted homeless population lives.

[iii] For this analysis, “markets” are HUD-defined geographies known as Continuums of Care (CoCs). Some CoC’s encompass a single city or county, others an entire state. HUD determines funding and resource allocation at a CoC level, and every CoC nationwide conducts its own point-in-time count of homelessness, typically on a single night in late January. These counts help inform HUD and national and local policymakers about the level of their homeless population and how it has changed over time.

[iv] HUD’s official homeless count for the 386 CoCs studied here dropped by almost 69,000 people from 2011 to 2017.

Link to original article

Related Posts