The Housing First Model: A Practical Guide to Its History and Implementation
The Housing First model didn't come from a government study or a academic white paper. It came from watching people fail at every other approach and deciding to try something that made actual sense. Here is how it works, where it came from, and what you need to know if you are actually going to implement it. The program started in 1992 in New York City. Dr. Samuel Tsemberis was working with Pathways to Housing, a nonprofit that served people with severe mental illness who were cycling through hospitals, jails, and shelters. The traditional approach at the time was the "staircase" or "treatment first" model. You had to prove you were sober. You had to demonstrate you could follow rules in a halfway house. You had to get your act together before you got an apartment. Most of these people never made it past step one. Tsemberis flipped it. He gave people apartments first. No sobriety requirements. No treatment compliance as a precondition. Just a lease and keys. The theory was simple enough that it sounded naive: if you give someone permanent housing without conditions, they will be stable enough to engage with services voluntarily rather than being forced into them. The results were immediate and dramatic. The initial cohort had an 88 percent housing retention rate after one year. Traditional programs at the time were struggling to hit 50 percent, and that was being generous.
What made this possible wasn't magic. It wasborrowing directly from Assertive Community Treatment, or ACT, which had been developed in the 1970s for severe mental illness cases. ACT teams go to where the person is. They don't wait for the person to come to an office. The Housing First model applied that same philosophy to housing itself. You meet people where they are, not where a bureaucrat thinks they should be.
How It Actually Works in Practice
I ran a Housing First program for several years, and the gap between the model description and the day-to-day reality is where most people get tripped up. The basic framework has six core components, but let me walk through what each one actually looks like when you are dealing with it. Immediate housing access. This means no waiting lists that stretch six months out. No requiring sobriety verification. No psychiatric clearance from three different providers. The housing search begins within 24 hours of program entry. In my experience, this means landlords need to be pre-briefed and willing to work with you, because you cannot build a pipeline of properties that wait around for the "right" tenant to become ready. Consumer choice. Participants choose their own housing. Not which unit in a program-owned building. Which neighborhood. Which type of place. This matters more than people outside the field tend to realize. I had a participant who refused every unit we showed him because they were all in what he called "the downtown corridor." He ended up in a small apartment in a residential neighborhood three miles from where we were placing everyone else. He has stayed housed there for four years. The unit that would have failed him was the one he would have chosen if he didn't have a say in the matter.
Get the Full Details
Permanent housing with supports. This is not transitional housing. The lease is in the participant's name. The goal is that they stay forever, with support services available indefinitely. Support services are voluntary. People can refuse them. They can accept some and not others. The service team does not terminate housing for refusing services. This is the part that surprises most people coming from traditional systems. Refusing a case manager meeting does not get you evicted. It gets you a phone call asking if everything is okay. Community integration. Participants live in scattered sites throughout the community, not in congregate settings. Group homes and treatment facilities are excluded. This isn't an ideological preference. There is empirical evidence that community-based housing produces better outcomes than congregate settings for this population. Harm reduction. This is one of the most misunderstood components. Harm reduction does not mean encouraging substance use. It means meeting people where they are and reducing the negative consequences of their behavior rather than demanding abstinence as a condition of survival. If someone is using drugs in their apartment, the response is not eviction. It is discussing safer use, connecting them to substance use treatment if they want it, and addressing the practical issues around it. The model assumes that sustained engagement works better than coercive compliance.
Recovery orientation. The model recognizes that recovery looks different for everyone. It is not a standardized endpoint. Some people recover from homelessness and psychiatric symptoms. Others find stability in a different sense. The program supports whatever recovery means to the individual.
The Counter-Intuitive Parts Beginners Miss
Most people entering this field think Housing First is about giving people apartments. That is not what it is about. It is about changing the entire power dynamic between service systems and the people they serve. The counter-intuitive insight is that removing conditions actually increases engagement, not decreases it. When people are not fighting to prove they are worthy of basic shelter, they have actual bandwidth to address other problems. I saw this repeatedly. The participants who were the most resistant to services in the first week often became the most engaged by month three, precisely because they were no longer in survival mode. Another thing that trips people up: Housing First is not cheap up front. Permanent supportive housing costs more per participant per year than emergency shelter or transitional housing. But the cost offset happens elsewhere. Studies consistently show reductions in emergency room visits, hospitalizations, jail days, and shelter utilization. The Hamilton Health Sciences study found that Housing First participants cost $26,600 per year versus $42,000 for treatment as usual, with the difference coming from reduced institutional use. The savings show up in public budgets, not in program budgets. That disconnect is why funding Housing First is harder than it should be.
A Real Problem I Encountered and How I Solved It
About two years into running a Housing First program, I hit a wall that the literature does not adequately address. We had placed dozens of people successfully, but we were struggling with a specific demographic: people with co-occurring severe mental illness and active substance use disorders who had no family support system and a history of institutional living. They were housing-stable but service-avoidant to the point of self-endangerment. The standard model was not moving them at all. The workaround was not a theoretical adjustment. It was operational. We created a "low-demand" tier within the program. Instead of expecting weekly check-ins, participants needed only monthly contact. The bar for engagement was set so low that refusing it was harder than accepting it. We also partnered with a mobile psychiatric unit that could do home visits for medication management without requiring the person to come to an office. This added about $4,000 per participant annually to our budget, but it reduced hospital readmissions by roughly 30 percent in that subgroup. The trade-off was real, and it was worth it, but no one tells you about this in the training materials.
International Spread and Variations
Canada adopted Housing First at a national level in 2007 through the At Home/Chez Soi project, which was one of the largest randomized controlled trials on homelessness ever conducted. The results confirmed what the New York program had shown: Housing First worked across multiple cities and demographics. Finland took a different approach. Rather than piloting and proving, they restructured their entire national housing and social services system around Housing First principles. Their approach is more integrated with existing welfare infrastructure, which is why they have been able to reduce chronic homelessness more substantially than any other country. Spain, the UK, and several other European countries have since adopted variants. Each adaptation reflects the local welfare state structure. The United States tends to treat it as a standalone intervention. Europe tends to treat it as an integration strategy. Both work, but they require different funding and policy architectures.
Where Housing First Fails
I need to be blunt about this because people in this field are often too enthusiastic to mention it. Housing First does not work for everyone. People with acute psychotic episodes who are a danger to themselves or others may require a higher level of care before permanent housing is appropriate. The model assumes a certain baseline of cognitive functioning and willingness to maintain a lease. It is not designed for people in active crisis who need stabilization first. Landlord willingness is another bottleneck. In markets with tight rental conditions, finding landlords willing to work with voucher holders who have criminal records or eviction histories is genuinely difficult. I spent more time cultivating landlord relationships than I did on any other aspect of program implementation. This is not covered in the model description but it is the single biggest practical barrier. Perhaps the most important limitation: Housing First without adequate support services is just subsidized housing with a lower bar for entry. The model only produces its documented outcomes when the support services are robust, continuous, and truly voluntary. Cutting service funding to make the model "affordable" defeats the purpose. You end up with people in apartments who are isolated and unsupported, which is no improvement over the shelter system.
If you are considering implementing Housing First, the most important thing to understand is that it is not a program you start. It is a philosophy you adopt, and every policy decision, funding stream, and partnership needs to reflect that. The history of the model is short but clear. It works when you take it seriously. It fails when you treat it as a cheaper version of the old system.
