Pathway 01 · Learn

What causes homelessness

Ask a random American why people become homeless and you'll hear a familiar answer: addiction, mental illness, bad choices. The evidence says something different. Personal struggles determine who falls through; the housing market determines whether anyone falls at all.

The reframing

If addiction caused homelessness, the cities with the highest rates of addiction would have the highest rates of homelessness. They don't. West Virginia and Arkansas have some of the highest rates of opioid misuse in the country and some of the lowest rates of unsheltered homelessness. San Francisco and Seattle have lower rates of opioid misuse and dramatically higher rates of unsheltered homelessness. The same is true for mental illness: the prevalence is roughly even across regions; the homelessness is not.

What does correlate with regional homelessness rates? The price of housing. The tighter the housing market, the more homelessness — almost regardless of what the population looks like underneath it. This is the central finding of Gregg Colburn and Clayton Page Aldern's 2022 book Homelessness is a Housing Problem, which uses HUD point-in-time count data and metro-level economic data to test every popular theory of homelessness. Housing cost wins, repeatedly.

This is not a claim that personal factors don't matter. They do — they determine which specific people experience homelessness when housing markets tighten. But the size of the homeless population in a region is set by the housing market, not the sum of personal vulnerabilities.

The housing market is the dominant variable

The clearest way to see this: compare US metros. Honolulu, San Francisco, Los Angeles, New York, Boston, Seattle, and Washington DC have the highest rates of homelessness. They share one thing — the most expensive housing in the country. Detroit, Cleveland, and Birmingham have higher rates of poverty, mental illness, and substance use than any of those cities and far lower rates of homelessness. The difference is supply and price.

The mechanism is straightforward. When rents rise faster than wages, the bottom of the rental market disappears. A person who could afford a $700 apartment in 2015 may have nowhere to land at all in 2025. Add a job loss, a medical bill, or a relationship breakdown to that situation and the result is homelessness — not because the person changed, but because the housing market did.

The National Low Income Housing Coalition's annual Out of Reach report calculates the "housing wage" — what a worker would need to earn to afford a modest rental without spending more than 30% of income. In 2024 that figure was roughly $32/hour for a two-bedroom nationally. The federal minimum wage is $7.25. In the most expensive metros the gap is far larger. A person working full-time at the going wage for retail, food service, or in-home care often cannot afford any unsubsidized apartment in the city where they work.

This is the structural condition. Inside it, the personal triggers below decide who falls.

Proximate triggers: what pushes someone over the edge

Most people who become homeless do so because of a discrete event — not a slow decline. The most common triggers, in roughly the order they show up in the data:

• Eviction. Roughly 3.6 million eviction cases are filed in the US in a typical year (Eviction Lab, Princeton). A meaningful fraction of evictions end in homelessness within months. Matthew Desmond's Evicted documents this in depth.
• Job loss or income shock. The majority of US households can't cover a $1,000 emergency from savings; for renters at the low end of the market a single missed paycheck can start the cascade.
• Medical crisis. Health-related bankruptcies still account for a large share of US bankruptcies. A serious illness can wipe out income (you can't work) and add cost at the same time.
• Relationship breakdown or domestic violence. A leading cause of homelessness for women and children specifically. See the demographic detail on pathway 2.
• Aging out of foster care. About 20,000 young adults age out of US foster care every year; roughly one in four experiences homelessness within four years.
• Release from incarceration. People leaving prison or jail face barriers in housing applications (background checks) and are at high risk of homelessness in the first months of release.
• Discharge from inpatient psychiatric or hospital care without housing arranged.
• Discharge from the military without re-integration support — though VA programs have substantially reduced this.

None of these are uniquely American problems. What is uniquely American — and varies dramatically across US cities — is whether someone who experiences one of these triggers can find another apartment within their means before they end up on a friend's couch, in a car, in a shelter, or on a sidewalk.

How structural and individual causes interact

The cleanest way to think about it: structural conditions set the size of the cliff. Individual events determine who walks toward it. In a housing market with adequate supply at every income level, a person who loses a job finds a cheaper apartment, doubles up with family, or applies for assistance that bridges the gap. In a market with no apartments at any price the person at the bottom can afford, the same job loss produces homelessness.

This explains why the same person, with the same struggles, may live housed in one city and unhoused in another. It also explains why investments in services without investments in housing don't move the needle: you can treat someone's mental illness, but if there is no apartment for them to be treated in, they remain on the street.

What the data on mental illness actually shows

Roughly 21% of adults experiencing homelessness in the US have a serious mental illness (SAMHSA/HUD data). That is meaningfully higher than the general adult prevalence of around 5–6%. So mental illness is overrepresented — but it is not the dominant cause. The other 79% of adults experiencing homelessness do not have a serious mental illness. And many people with serious mental illness are not homeless. The variable that swings homelessness rates is the gap between the median rent and the income of the most vulnerable, not the prevalence of psychiatric diagnoses in the population.

There is also a directional question that the casual framing gets wrong. Homelessness is itself a major cause of mental health deterioration. Living on the street, in shelters, or in chronic instability produces or worsens anxiety, depression, PTSD, and psychosis. A snapshot of mental illness among the homeless population is partly a snapshot of what homelessness does to people, not only what brought them there.

What the data on addiction actually shows

Substance use is similarly elevated but similarly secondary. Estimates vary, but roughly 30–40% of adults experiencing homelessness have a substance use disorder — again, higher than the housed adult population, but not a majority and not a stable explanation across cities.

The causal arrows run in both directions. Substance use can precede and contribute to housing loss. Housing loss reliably increases substance use: it is one of the most effective predictors of relapse for people in recovery, and unsheltered life pushes many people toward substances they would not have used otherwise. Programs that require sobriety before housing have far worse outcomes than programs that provide housing first and treatment after — covered in detail in pathway 3.

Counter-arguments and where they fail

The most common counter-arguments, and what the evidence says:

"It's because progressive cities tolerate it."

This claim runs into a basic problem: politically conservative cities with strict policing also have homelessness, and the variable that predicts homelessness rates is housing cost, not local political ideology. Phoenix, Houston, Charlotte, and Las Vegas all have substantial homeless populations. What separates the highest-rate metros from the rest is housing cost.

"Generous benefits attract homeless people."

The "magnet" theory. The evidence does not support it. HUD and academic studies of where homeless individuals come from consistently find that most are local — they became homeless in or near the place they were already living. Interstate migration of homeless individuals is small relative to local production of homelessness.

"It's a choice."

For a small number of people who reject shelter rules or services, there is an element of choice in not entering shelter on a given night. There is not a meaningful "choice" to be homeless in the first place. Surveys of people experiencing homelessness consistently find that the overwhelming majority want stable housing and would accept it if offered without conditions they couldn't meet.

"If we just built more shelters, it would end."

Shelters are emergency tools — necessary but not solutions. Without an exit into housing, shelters cycle people through the same crisis indefinitely. The places that have reduced homelessness substantially (Houston, Helsinki, Bergen County NJ — see pathway 3) have done so by pairing shelter capacity with rapid placement into permanent housing.