Homeless shelters are simply more expensive.
Mother Jones reports on how Utah "has decreased the number of homeless by 72 percent—largely by finding and building apartments where they can live, permanently, with no strings attached."
It's a terrific long story by Jim McAuley, and my favorite part is where it notes the paradox behind how the state typically hands out housing aid. Generally, they require someone to get clean before getting housing assistance. But this is precisely the opposite of how things actually work in real life, as New York University psychologist Sam Tsemberis points out. Housing creates stability; so why not hand it out first? Then you might discover, as Utah has, the attendant health-and-welfare costs of coping with people's instability shrinks to the point where it's smaller than the cost of the housing itself.
"Going from homelessness into a home changes a person's psychological identity from outcast to member of the community," Tsemberis says. The old model "was well intentioned but misinformed. It is a long stairway that required sobriety and required stability in order to get into housing. So many people could never achieve that while on the street. You actually need housing to achieve sobriety and stability, not the other way around.
Like all social-welfare experiments, though, these programs are complicated, so — as McAuley reports — it's not super clear how you'd replicate this approach in the state where serious homelessness clusters, like California.
This piece puts me in mind of "Million-Dollar Murray", a 2006 New Yorker piece by Malcolm Gladwell that came to a similar conclusion.