Pathway 03 · Learn

What actually ends homelessness

There is more good evidence on what ends homelessness than on most social-policy questions. Housing First works. Permanent supportive housing works for the hardest cases. Rapid rehousing works for short crises. Prevention is the cheapest tool. The gap between what we know and what we do is mostly a question of funding scale and political will.

We know what works

This is the pathway most likely to surprise people. The general public conversation about homelessness tends to assume the problem is so complex and so personal that no one really knows what to do. Researchers and providers know what to do. They have known for two decades. The interventions have been tested in randomized trials, scaled in cities, and replicated in countries. They produce results that would be considered overwhelming evidence in any other field.

The core insight: housing is the intervention. Not housing as a reward for doing other things right — housing as the starting point that makes everything else possible. People who are housed sleep, eat, take medication, hold jobs, and engage with services at rates that people who are unhoused cannot. The single most cost-effective thing a community can do for a person experiencing homelessness is end the homelessness — give them an apartment, with the supports they need, and start from there.

Housing First, the core principle

Housing First is not a specific program. It is a principle: place people in permanent housing as quickly as possible, with no preconditions about sobriety, mental health treatment, or "housing readiness." Then offer supportive services — voluntarily — once the person is stable in housing.

The principle was developed in New York in the 1990s by Sam Tsemberis and tested through the Pathways to Housing program. The original randomized controlled trial compared chronically homeless adults with mental illness assigned to Housing First versus those assigned to traditional "treatment-first" programs that required sobriety and treatment compliance before housing. The Housing First group had 88% housing retention at two years; the treatment-first group, around 47%. The Housing First group also showed comparable mental health outcomes despite not being required to engage with treatment.

Housing First has since been replicated in dozens of US cities and several countries. It has consistently outperformed treatment-first models on housing retention, often at lower per-person cost. It is now the official approach of HUD, the VA, USICH, and most major homelessness research bodies.

Permanent Supportive Housing (for chronic homelessness)

Permanent Supportive Housing (PSH) is the Housing First principle applied at scale to the chronically homeless population — long-term homelessness combined with a disabling condition (serious mental illness, substance use disorder, chronic physical illness). PSH provides a permanent apartment plus voluntary on-site or visiting support — case management, mental health care, addiction recovery, primary care.

The evidence base on PSH is unusually strong for a social intervention. Studies consistently show:

• 75–90% housing retention at two years, across populations widely considered "untreatable" in the prior treatment-first paradigm.
• Sharp reductions in emergency room visits, hospitalizations, and jail nights — typically 40–60% reductions in public-system use for the same individuals after housing.
• Net cost savings or cost neutrality in most studies, because the cost of housing plus supports is lower than the cost of cycling the same people through ERs, jails, and shelters.

PSH is the intervention that ends chronic homelessness. The reason chronic homelessness has not been ended at the national level is not that PSH doesn't work; it is that PSH is funded at roughly a tenth of the scale required to meet the need. Pathway 4 covers the funding gap.

Rapid Rehousing (for short crises)

Most homelessness is not chronic — see pathway 2. For the 80% of people experiencing homelessness in a year who have a single short episode, the appropriate intervention is Rapid Rehousing (RRH): time-limited rental assistance (typically 3–24 months) paired with housing search support and light case management, designed to move a household out of homelessness quickly and into a unit they can sustain on their own once the assistance ends.

RRH works well for the population it's designed for: people whose homelessness is caused by a discrete shock (job loss, eviction, breakup) rather than a long-term disabling condition. Outcomes are strong when RRH is well-targeted and weaker when it's applied to people who actually needed PSH. The current US system over-applies RRH to chronic cases because RRH is cheaper per household — a mismatch that produces worse outcomes for both populations.

Prevention and diversion (the cheapest tool)

The cheapest intervention for homelessness is preventing it before it starts. Prevention programs identify households at imminent risk — typically through eviction filings, utility shut-offs, or 211 calls — and intervene with short-term financial assistance, mediation with landlords, or relocation help. Diversion programs work with people who have just become homeless and help them quickly resolve the situation (often by mediating with the family member they were doubled up with, or providing one-time assistance to access a new unit) without ever entering the shelter system.

The COVID-era Emergency Rental Assistance Program (ERAP) was the largest US natural experiment in homelessness prevention. The evidence is consistent: when at-risk households received timely rental assistance, evictions and downstream homelessness fell substantially. Programs that combined cash assistance with legal representation (right-to-counsel in eviction court) reduced evictions further.

Coordinated Entry (the system glue)

None of the above works at scale without a system to match people to the right intervention. Coordinated Entry is the HUD-required process by which every Continuum of Care assesses people experiencing homelessness and prioritizes them for available resources — PSH for the chronic, RRH for the transitional, diversion for those who can be resolved upstream.

The communities that have made real progress have invested in their coordinated entry as a system, not a form. Houston's success (below) is largely a story of getting every provider in the region onto a single by-name list, prioritizing the most vulnerable, and matching them to whatever resource opens next — instead of letting each agency operate its own intake.

Where it has worked: Houston, Helsinki, Bergen County

Houston

Houston has reduced homelessness by more than 60% since 2011, the largest sustained reduction of any major US city. The approach: an unusually well-coordinated CoC, a shared by-name list, a Housing First philosophy, and the political stamina to keep at it across multiple mayoral administrations and federal funding cycles. Houston still has homelessness — the housing market is tightening and progress has stalled in recent years — but the trajectory remains the strongest in the US among large metros.

Helsinki and Finland

Finland is the only OECD country to have substantially reduced long-term homelessness at the national level. The Finnish Y-Foundation and the national Housing First policy converted shelters and transitional housing into permanent housing units and gave them to formerly homeless people. Chronic street homelessness has been nearly eliminated. The model is studied frequently and replicates poorly in the US for two reasons: Finland has a national rental market that didn't require building housing from scratch, and Finland has a national funding mechanism that doesn't depend on annual appropriations.

Bergen County, New Jersey

Bergen County became the first US community to achieve and sustain "functional zero" for chronic homelessness and for veteran homelessness, certified by Community Solutions' Built for Zero initiative. Functional zero doesn't mean nobody is ever homeless — it means the system can house any new person who becomes chronically homeless faster than new people enter that status. The county did it with a population just under 1 million and a tight regional housing market by aggressive coordination and persistent local political support.

The cost case

One of the more counterintuitive findings: ending someone's homelessness is usually cheaper than letting it continue. A chronically homeless adult cycling through emergency rooms, psychiatric stays, jail nights, and shelter beds typically costs the public system $30,000–60,000 per year (the exact figure varies by region and methodology). A unit of PSH for the same person costs roughly $15,000–22,000 per year, all in.

This is why the cost-effectiveness case has been made for two decades and still doesn't move policy at scale: the costs are spread across different agencies (Medicaid, corrections, hospitals, the shelter system) and the savings would accrue to those same agencies, but the up-front investment has to come out of the housing budget. Without budgeting structures that capture the cross-agency savings, the housing-side cost looks like new spending rather than a reallocation.

What doesn't work, and why we keep trying it

Encampment sweeps

Police clearing an encampment does not produce a single new unit of housing. It moves people from one location to another, often discarding their possessions (ID, medication, sleeping gear), and frequently disconnects them from outreach workers and case managers who knew where they were. Studies of repeated sweeps find that they make subsequent engagement with services harder, not easier. They also cost a great deal of money — police time, sanitation, encampment cleanup — for no durable reduction in homelessness.

Criminalization

Laws that ban sleeping in public, sitting on sidewalks, or panhandling do not reduce homelessness. They produce arrests, fines, and warrants that make it harder for the cited person to get an ID, pass a background check, or apply for housing — actively increasing the duration of their homelessness. The 2024 Supreme Court Grants Pass decision allows cities to enforce these laws regardless of shelter availability, which has accelerated the trend without changing the evidence on what it produces.

Treatment-first models

The pre-Housing First model required people to demonstrate sobriety, mental health treatment compliance, or "housing readiness" before getting an apartment. This approach has been tested for decades and consistently produces worse outcomes than Housing First on every measure that matters — housing retention, mental health, substance use, and cost. It persists in some communities for ideological reasons or because it matches a moral intuition about deserving and undeserving recipients of aid.

Mass shelter without a housing pipeline

Adding shelter beds without a corresponding investment in permanent housing can ease the visible symptoms of homelessness temporarily but does not reduce it overall. People cycle through shelter beds because there is nowhere to exit to. The cities that have actually reduced homelessness invested in shelter and a robust permanent housing pipeline.

Why we keep trying things that don't work

Several reasons. First, the interventions that don't work are often visibly satisfying (a swept encampment "looks" like the problem is being addressed). Second, they don't require building anything — and building housing is slow, expensive, and politically contested. Third, the moral framing of homelessness as personal failure leads naturally to interventions that target the person rather than the housing market. Fourth, the agencies that fund "what works" don't capture the savings, so the cost-effectiveness case is structurally difficult to act on.