Why housing shortages cause homelessness

Diona’s mom gave her one week’s notice. There were too many kids and too little space. If 17-year-old Diona didn’t find someone to stay with, she would be forced to check herself and her 3-month-old son into a homeless shelter. Because she was a minor, doing so would put her baby ‘into the system’. She was determined not to let that happen.

She found a family willing to take her in through a Christian group at her high school. There, she shared a bedroom with her own baby and a bathroom with three other children. There wasn’t enough privacy. She stayed three months with them, and then three months with another family. Each time, the tension of dissonant lifestyles steadily wore down her willingness to live by others’ rules and eat unfamiliar foods.

But the six months transformed her into a legal adult, able to take full legal responsibility for her son. And as a legal adult, Diona became legally homeless, and she was able to check herself into a shelter for single moms. Her story echoes the widely repeated observation that ‘people don’t become homeless when they run out of money, they become homeless when they run out of relationships’.

Being a homeless mom put her on the fast track for a housing voucher. She received one and found an apartment several months later. Since then, she’s moved on to a better apartment within walking distance of her retail job and her son’s preschool.¹

Diona charted an odyssey through her social and legal world that’s impressive for a teenager. But imagine for a moment a much simpler story: Diona stayed with her mother. That story is short and boring. It’s common and unappreciated. And if Diona’s family had happened to live in a cheaper part of the country and rented a bigger house, it might have been her story.

In the United States, the primary definition of homelessness includes those who sleep outdoors or in a tent, car, or recreational vehicle, or who are in a homeless shelter or transitional housing provided by a homeless services agency. This often differs from the colloquial use of the phrase, which connotes a vivid human portrait: a person who has lived on the street or in shelters for a long time, who spends his days begging or loafing, who likely suffers indignities, abuses, ill health, and toilet insecurity, and likely has mental illness, a drug addiction, or both.

In truth, many of the people who a passerby might call homeless aren’t homeless at all – they spend their nights in a home (perhaps an imperfect one) while spending their days in public. And many of the homeless are undetectable as such in daily life.

The American cities with the highest housing prices have the worst homelessness problems. YIMBY (Yes in My Backyard) advocates highlight this correlation to argue for policies that increase housing supply. But, when you think about it a bit, it’s not clear exactly how high rent contributes to homelessness. It’s not like $800 per month apartments are any more affordable to most homeless people than $1,000 per month apartments. And homelessness is frequently associated with mental health or drug abuse problems. This is why non-YIMBY progressives insist that only more generous vouchers or subsidies can help and non-YIMBY conservatives argue that only behavioral change can help by tackling alcoholism, drug abuse, and mental health problems.

The stories and data in this essay show the missing link between homelessness and housing costs: people without money who avoid becoming homeless do so mostly by staying with others, usually their own parents. This happens outside the formal housing market. But parents’ and others’ ability to offer space is limited by what they can afford in the market. Where housing costs are moderate, friends and family have bigger homes. When they are higher, friends and family don’t have space to share, and this is often what puts a vulnerable person onto the streets.

Why are housing costs linked to homelessness?

Housing advocates are convinced that expanding the supply and lowering the price of market-rate housing will decrease homelessness. But many who work with the homeless are not so sure. Their clients often have too little income to afford even the cheapest apartment.

In the recent book Homelessness Is a Housing Problem, Clayton Page Aldern and Gregg Colburn argue that although individual risk factors can help explain who becomes homeless, they cannot empirically explain the rate of homelessness in a city or region.² Although unemployment and drug abuse are risk factors for individual homelessness, they are negatively correlated with the rates of homelessness across cities. In fact, the cities with the highest rates of homelessness, such as San Francisco and New York, are those with the highest wages, least unemployment, lowest poverty rates – and highest housing costs. My chart below updates Aldern and Colburn’s work with post-pandemic data.

But Aldern and Colburn do not explain how, exactly, high housing costs translate into homelessness. The obvious explanation would be that low-income people can afford cheap housing in low-cost cities but are forced onto the street in high-cost ones. Surprisingly, this doesn’t stand up to scrutiny: there are more people in cheaper cities, on average, whose incomes are far too low to afford the local median rent – for example more San Antonians earn less than half of their city’s median rent than do San Diegans.³ And the number of single adults with low incomes is, in any case, far greater than the number of people counted homeless in any city.

The ‘homelessness is a housing problem’ argument is empirically strong but it is too zoomed out to tell us how, exactly, market prices are linked to people who might spend some time homeless. We needed a framework to think about how homelessness interacts with housing markets, one that makes sense of both the personal stories of homeless people and the overall systems that those stories take place within.

Housing costs and individual risk factors both matter – and the puzzle piece that links them is the role of family and friends. When housing is cheap, relatives and friends tend to have more space in their homes, enabling them to keep someone at risk of homelessness off the street and on their medications. When space is tight, the people forced out are those who are hardest to live with.

A theory of shelter

Sometimes Genesee needs a break from her boyfriend, and she goes to the other side of West Palm Beach to stay with her mom, Sabine. Once, however, she was beefing with both her boyfriend and her mom at the same time. In the moment, it was easier to check into a women’s shelter for two nights than to immediately resolve either conflict. During that brief period, she would have been counted as homeless. To think about the full spectrum of living situations, we need to move beyond the ‘housed versus unhoused’ binary and think about personal factors.

Genesee’s story highlights the role of factors other than money in keeping people housed. It wasn’t a sudden shortage of cash that pushed her into temporary homelessness, it was the emotional cost of arguments with her boyfriend and mother. These relational shelter transactions take place outside the formal market for housing, with its leases, mortgages, and listings. Such transactions are undertaken in currencies like amiability, chores, affection, and forgiveness – but they rely on upstream market transactions. Sabine’s ability to offer her daughter a backstop depends on her own ability to afford an apartment with a little extra space.

Sabine’s spare bedroom is an example of what social scientists call ‘decommodified housing’. It is not available to the highest bidder. Nor is it rented out fairly, without regard to race, religion, or sex, as market housing must be under US law. It’s for her loved ones.

Genesee’s shelter stay also shows how homelessness as used by statisticians departs from its richer colloquial meaning. In the US, two agencies count people that they call homeless (or, in the current euphemism, ‘experiencing homelessness’):

• The Department of Housing and Urban Development (HUD) orchestrates an annual Point-in-Time count each January, striving to enumerate all those who are sleeping that night in shelters or ‘unsheltered’, a designation which includes tents, shacks, and vehicles. The 2022 Point-in-Time count identified 582,500 people experiencing homelessness, and the 2023 count found 650,104.
• The Department of Education collects data on child homelessness, which is defined much more capaciously to include kids who are ‘doubled up’ by necessity – that is, living in a dwelling with more than one family – or staying in hotels. Schools gather this information about their students as best they can. During the 2021–2022 school year, administrators counted 1,205,292 students who were homeless – 12 times the number of children in the Point-in-Time count.

There is no point arguing about which definition of homeless is ‘correct’. Rather, as Alexander the Grate, a homeless man, contends, everyone has ‘some degree of shelter insecurity’, and everyone is working within their own financial, and psychological, budget.

We can imagine a ladder of possible sleeping situations for any given person – the best thing is at the top, the worst at the bottom. For different people, or at different times, the rungs might be in a different order. Most, but not all, of the lowest rungs would be labeled homeless by one definition or another, but not all. One person might prefer the privacy of a tent to the solidity of their mother’s basement; another might vacillate depending on the weather. Becoming housed isn’t always an improvement.

When the state of Vermont offered free motel rooms to any homeless person for the duration of the Covid-19 pandemic, the number of homeless people more than doubled. Most of the new homeless had been housed previously, but in situations less attractive than being homeless in a motel.⁴

Learning from the housed

To identify solutions to homelessness, scholars and policymakers have usually interviewed, surveyed, counted, and otherwise directly studied the homeless population.⁵ This can be helpful for directing casework and funding. But understanding the overall prevalence of homelessness requires flipping the telescope around to look through the big end. Doing so reveals that the vast majority of very poor people avoid homelessness. And they often do so by staying with friends and family.

Even among the homeless population, the highly visible hardcore or chronic homeless are a minority. And the vast majority of the homeless are connected to a formal source of income, either work or welfare. A team of economists including Bruce Meyer used federal administrative records to learn that about a fifth of the homeless had five-figure annual incomes while three quarters were enrolled in food assistance. When Meyer’s team statistically matched the homeless to a sample of the ‘single, housed poor’ with the same demographic characteristics, they found comparable levels of employment, job instability, welfare receipt, and disability insurance.

To understand how the single, housed poor manage to stay housed, I tweaked the Meyer team’s method. I identified single, housed poor people of working age in the American Community Survey (ACS),⁶ the most comprehensive public database of Americans’ incomes, family relationships, employment, and other information. I reweighted this sample so that it statistically matches the demographic and income data that Meyer’s team found for the homeless.⁷ Now we can distinguish who lives with whom and in what cities.

I’ve found that 39 percent of poor, single, housed adults are the sole adults in their households. A minority have children. Another 33 percent live with their parents. Others are scattered across noninstitutional group homes, roommate situations, or doubled up in multifamily households. Among the single, housed poor, three percent report that their household pays no rent.⁸

Turning to geography, the most glaring fact is that expensive places like Washington, DC, and San Jose have fewer single, housed poor people, as the scatterplots on the previous page show. Those who match the demographic profile of the homeless, but manage to stay housed, tend to do so in low- or moderate-cost metros. And the single, housed poor who do live in expensive metros are less likely to have their own place and more likely to live in group quarters.

Cities that reduce homelessness will likely do so by allowing people to keep themselves housed in imperfect situations. In more than half of cases, that involves sharing housing with someone, usually parents. The remainder of this essay will unpack the details of this.

Parents

As long as Sherman takes his medications, he’s a stable, pleasant guy with a solid job and a wry sense of humor. A brief experiment living on his own a few miles away left him deeply depressed. So he’s back with his parents in suburban Ohio, sleeping in the bedroom he’s had since age seven. Sherman becoming homeless is unthinkable to all three of them – but perhaps it shouldn’t be.

Unlike Diona’s mom, Sherman’s parents had extra space. That’s very common among homeowning, empty-nest parents, meaning that a great deal of vacant housing is in the hands of the two people who are most likely to love and forgive an adult child in dire circumstances. But in more expensive regions, fewer parents (or other loving figures) have that resource, either because they couldn’t afford to buy a house in the first place or because there are demands from multiple family members to share the legacy residence.

To measure the vacant space that might be available for nonmarket, relational transactions, I measure how many empty bedrooms there are in occupied homes in each city and county.⁹ Empty bedrooms are more common in poorer, low-rent cities than richer, high-rent ones. These data tell Sherman’s story at a macro scale: middle-aged people in suburban Ohio, for example, are more likely to have an extra bedroom they can offer to an adult child than are middle-aged people in suburban Los Angeles. Consistent with this story, the chart below shows that the homelessness rate is strongly negatively correlated with the prevalence of empty bedrooms.

Friends and family

As well as their parents, the single, housed poor are much likelier than the general adult population to live with other family members, roommates, or friends. But unlike their parents, these housemates are usually just as poor and in need of help affording shelter.

Estrella and the 13 or 14 of her extended family members and acquaintances who share a modest suburban apartment may be an extreme case. They are all immigrants from Venezuela, and most if not all of them lack permission to work in the US. Estrella spends some of her days standing outside a local grocery store asking for work. It’s a strategic location; she is rewarded more often with groceries than work. Somehow, the ever-shifting crew of residents scrape enough money together to keep the apartment.

In other cases, doubling up is a more hierarchical arrangement: a homeowner or leaseholder makes space available to an extended family member or friend in need. Anecdotally, this is usually a temporary arrangement, like Diona’s six-month sojourn in suburban spare bedrooms.

Under the Department of Education’s broad definition of homelessness, three quarters of homeless students are doubled up with another family in arrangements like this. Most education researchers treat doubling up as a problem. But it’s also a solution: these shared homes blur the line between market and nonmarket sectors, because they normally rest on a kinship or friendship basis, but with everyone expected to contribute.

Regardless of income, having extra space gives both sides a chance to disengage without leaving the home, to set their own standards, and maintain some privacy. Not only are those good things in themselves, they also make it easier for both sides to tolerate a temporary arrangement for longer, keeping vulnerable people housed.

This is not news to people who serve the homeless professionally. One former service provider noted that the people who came to his Detroit office were those who ‘ran out of couches to sleep on while they figured their shit out’. All I’m adding to this well-known pattern is to hypothesize that time runs out faster when the couch is in the living room rather than a spare bedroom.

Living alone

My friends came home once to find a large kitchen knife stuck in their front door. They believed the culprit was Stephen, a next-door neighbor who had recently been shouting obscenities at them. The police got involved. Everyone understands that Stephen has a serious mental illness. But he’s not going anywhere as long as he pays his taxes: he owns his home.

Although this essay has emphasized that most single, housed poor people live with other adults, two fifths live alone or with their own children. Like Stephen, more than a quarter of these sole adults own their homes, the majority without a mortgage.

But most – like Sabine and Diona after she received a housing voucher – rent apartments or houses with a mix of personal income and government assistance. The ACS does not report housing assistance, and other forms of assistance are seriously underreported. Nonetheless, three quarters of the single, housed poor report receiving at least one form of assistance.¹⁰ This differs little across living situations.

Only a minority of this group is in nonmarket housing. Even those with subsidies (vouchers or project-based) are market participants facing competition from many others in similar situations. Thus, it should come as no surprise that this living situation is the least common in high-cost metropolitan areas relative to the rest of the country.

Living rent free

Gwendolyn is a gentle, artistic college grad who suffers from social anxiety. She lives in a cozy camping trailer, behind an uninhabitable relic of a house that her parents own. If ACS interviewers happened to find her, they’d record her as paying ‘no cash rent’. But if she’d parked the trailer on public land, she’d theoretically be counted homeless.

Like Gwendolyn, a surprising number of Americans report paying no cash rent.¹¹ People in this situation might be doubled up or, more often, living in a house that belongs to a family or friend. They might chip in with chores or help maintain a house that would otherwise sit vacant. Like doubling up, living rent free is one way that poor people use their connections to avoid homelessness.

Living rent free is ten times as common among the single, housed poor sample described above as it is in the adult population at large. Of the single, housed poor sample living alone, five percent live rent-free like Gwendolyn. And, unsurprisingly, living rent free is more common in cities with low market rents. This is another connection between housing prices in the formal market and the incidence of homelessness.

Counterarguments

All the data shown are consistent with this essay’s hypothesis: most people at risk of homelessness manage to remain housed by staying with others. The higher rate of homelessness in high-cost areas is mostly explained by the inability of the family and friends of potentially homeless people to afford extra living space. But there are other reasonable explanations for why homelessness is higher in areas with high housing costs.

Mismeasurement

Counting the homeless is hard. Many of them have good reasons not to be found. When I volunteered for the Point-in-Time count, my team managed to interview one man who lived in an obscure area behind buildings only because we happened to spot him entering. In Detroit, a homeless services navigator observed ‘dozens of people’, mostly men, who lived in abandoned houses. In David Simon and Edward Burns’s nonfiction Baltimore account The Corner (which inspired the TV shows The Corner and The Wire), hardcore addicts drift from one half-abandoned row house to another. Such squatters are theoretically homeless, but administrators cannot safely and effectively count them.

Squatting can be widespread, of course, only in cities with a large number of abandoned buildings – which is to say, cities where buildings are cheap. In low-cost rural areas, homeless people may hide in natural areas far from roads.

The extreme, visible encroachment of homeless people onto public sidewalks in Los Angeles and San Francisco is partly – perhaps largely – on account of those cities’ very high land values. Private land and old buildings are simply too valuable to neglect, so homeless people are squeezed out of the quiet corners where they might prefer to camp. A side effect of visibility is enumerability.

How much of the homelessness gap between high- and low-cost areas does mismeasurement account for? We don’t know. But it can’t be the entire story. Data from the Department of Education, which covers doubled-up and sheltered families as well as the visibly homeless, and the experiences of homeless service providers confirm that California’s homelessness epidemic really is much bigger than in other places, not just much more visible.

Migration and program generosity

A popular narrative (in every city, as best I can tell) is that a lot of the homeless are not native to that city. If people are migrating to places with better weather or more generous safety nets, and those places happen to be more expensive, we would observe a spurious correlation between high housing costs and homelessness. What little evidence exists suggests migration is a secondary factor.

Surveys in San Francisco, Los Angeles, and Austin found that between 8 and 22 percent of homeless people in those cities first became homeless out of state. That’s not nothing, but it’s not enough to explain the large differences between cities.¹²

Advocates for the homeless cavil at the suggestion that homeless people are not locals, which carries the unsubtle suggestion that they should be shipped off. And many are, in fact, shipped off. The Guardian’s 2017 investigation of ‘Greyhound therapy’ (offering people a one-way bus ticket out of town) sketched the scope of the practice. But it also noted that the vast majority of homeless people are picked up by family members in lower-wage (and thus lower-cost) places when they get off the Greyhound bus. Whether it’s practiced callously or thoughtfully, city-initiated migration weakens the correlation between homelessness rates and housing cost.

More study of measurement and migration would be welcome. Given the existing evidence, however, neither appears likely to explain as much of the variation between high- and low-cost cities as the hypothesis that families and friends in low-cost cities have more space in their homes and can keep vulnerable loved ones housed.

Limitations

This essay has been about those who live close to the blurry boundary between being housed and unhoused. Such people are not the poorest, most desperate, or most in need of assistance. They are several rungs from the bottom of the proverbial housing ladder. Their options are not great, but they have options.

My friend and fellow housing advocate M Nolan Gray kept quiet for a couple years about a spell living in his car as a PhD student in Los Angeles. To talk about it as part of his pitch for zoning reform felt like claiming unmerited victimhood.

Most homelessness discourse is about the chronic, hardcore homeless. But the truth is that people in Nolan’s situation are those whose problems will be most thoroughly solved by the types of reforms that he and I support. Making housing more abundant and allowing cheaper options will lower prices, expanding the ability of family and friends to provide homes. If Los Angeles had nationally average housing costs, Nolan would have likely found a cheap room in a shared house, much as I did when I was a graduate student in low-rent Rochester, New York.

Housing policy in general is limited in how much it can offer the chronically homeless. Lack of shelter is a salient problem for them, and it worsens and exposes some of their other problems, but no homeless service professional fools herself into believing that a house key will cure alcoholism or schizophrenia.¹³ In a better-functioning housing market, it will only be slightly easier to address these challenges.

What this means for policy

There’s a rather sterile debate in homelessness policy: major funders and the left-leaning majority support ‘Housing First’, while right-leaning skeptics and some traditional service providers poke holes in the dominant view. To progressives Samantha Batko and Kathryn Reynolds, homelessness is a ‘math equation’ that just needs an unprecedented increase in spending. On the opposite side, Michael Shellenberger views homelessness as a behavioral problem that should be addressed with tough love. The framework put forward in this essay offers a way past this debate: in a moderate-cost city, housing reaches those at the margins of homelessness primarily through family members and friends willing to double up.

That implies that city leaders facing extreme housing cost and homelessness crises should not expect any program, even the most effective for participants, to make a major dent in homelessness rates. Dysfunction in the formal housing market is squeezing out those who would otherwise find housing in informal ways. That population is vastly larger than the formally homeless. Trying to solve homelessness without addressing the housing market is like trying to towel a floor dry when the roof has blown off.

The good news is that Housing First and other homelessness strategies work a lot better in a city with healthy institutions of housing supply. The top success story for Housing First is Houston – which is not coincidentally also the top success story for market-based housing affordability. Housing First requires housing, which the city and its nonprofit organizations have largely purchased at market rates. Replicating Houston’s success in Seattle or Los Angeles would be much more expensive for this reason alone.¹⁴ Houston’s reasonable-priced market rate housing also helps people who managed to stop being homeless to stay housed even if their fortunes decline.

To unlock the benefits of Houston-level house prices will require pricey cities to build many more homes. One way of doing so is to take a leaf out of Houston’s book (and New Zealand’s), and let low-density neighborhoods become townhomes. Another way is to allow larger apartment blocks in multifamily areas. It may require regulatory and procedural reforms such as removing requirements for expensive finishes and large yards and ensuring that new housing can be approved by city staff rather than city council votes.

As these policies are implemented, Diona, Genesee, Alexander, Sherman, Estrella, Stephen, Gwendolyn, Nolan, and countless others like them will be better housed. And traditional homelessness policies will become more meaningful too, refocused on those whose personal problems are beyond the capacity of their families and friends.