Yes. There is a real literature on this, and the short version is: **generous housing and homeless services can help the people who receive them and still fail—or even worsen—the city’s overall problem.** The “magnet” story is only one piece. The more common backfire is a **stock-and-flow problem**: people keep falling into homelessness faster than expensive programs can house them, while Housing First-style units with no treatment requirements can become long-term, high-cost holding pens rather than exits.
The evidence is mixed by *who* is moving and *what* is being offered. It is not a myth, and it is not a simple “they all hitchhiked here for free apartments” cartoon either.
## What the studies actually show
**1. Families respond to generosity. Single adults, much less so.**
Economist Igor Popov’s work using federal Continuum of Care grant formulas (an old, quasi-random funding rule) is the cleanest U.S. evidence. Extra homeless-assistance money:
- Houses people who would otherwise be unsheltered.
- Does **not** pull a large new population of *single adults* into the local homeless count.
- **Does** pull *homeless families*. Roughly two-thirds of the extra family caseload comes from families who became homeless in another community and then moved. An extra $100,000 in local grant funding was associated with on the order of 70–85 more people in homeless families using that community’s system.
That is a welfare-magnet result, but it is concentrated in families, who are more mobile and more likely to shop for shelter and services.
**2. Broader social-service expansions can raise the count.**
A paper on ACA Medicaid expansion found homelessness rates rose about 12% in expansion states versus non-expansion states, driven by unsheltered and chronically homeless people. The authors treat that as evidence that people with few resources will move toward places that deliver more health and social services.
**3. Permanent supportive housing (PSH) reduces the local count only modestly.**
Panel studies of PSH beds nationwide find one extra PSH bed associated with something like 0.04–0.12 fewer people in the homeless count—far less than “one bed, one person off the street.” Researchers list migration toward expanding PSH as one reason the local effect is muted.
**4. Most people are still local.**
The large UCSF California survey found about 90% of people experiencing homelessness lost their last housing *in California*, and 75% were in the same county. Interstate “they all came from Ohio” migration is overstated. Short-distance and intra-metro movement is not.
**5. Housing costs dominate the *level* of homelessness.**
Pew and a long line of research show high rents and tight housing supply are the strongest predictors of *how many* people are homeless in a metro. Services layered on top of a broken housing market do not cancel that. They can change *where* people sleep and *who* shows up in the local system.
So the honest mechanism is usually some mix of:
- Inflow from unaffordable rents, evictions, addiction, and mental illness faster than outflow.
- Families (and some singles) relocating toward better shelter, Medicaid, and cash-like services.
- Housing First units that stabilize housing status without requiring treatment—so the same people stay expensive to serve, overdose in city-funded buildings, or cycle.
- Visible street homelessness that persists even as the city “houses” thousands, because the pipeline never empties.
That is the backfire: taxpayers fund a large industry, individual clients often get a roof, and the city’s street problem and bill keep growing.
## Three cities where high taxpayer housing/services spend did not close the gap
These are not the only examples. They are the clearest West Coast cases of big dedicated taxes, Housing First / low-barrier housing, and counts that did not fall in proportion to the money.
### 1. Portland metro (Multnomah + Metro’s Supportive Housing Services tax)
This is the cleanest “we opened the door and it overflowed” case.
In 2017–18 Multnomah County tried to shelter **every** homeless family. Less than half the families who checked into shelter said their last address was in Portland or Multnomah County. Shelters overflowed, the county put nearly 100 families in motels, the budget blew up, and the open-door policy ended. The Seattle Times called it one of the few well-documented magnet episodes.
Then voters passed Metro’s 2020 Supportive Housing Services tax (1% on high earners and large businesses) on the promise of ending the crisis in ten years. It has raised on the order of **$1.4–1.6 billion**. The region has added thousands of shelter beds and placed thousands of households into housing. Inflow still outran outflow: for a stretch, about 14 people became homeless for every 10 placed. Point-in-time counts in the tri-county area jumped sharply (one comparison is a 61% increase 2023–2025); later regional estimates put the homeless population around 12,000 on a count night and higher in administrative data. Voters are now deeply skeptical the tax “ended” anything.
What backfired: an explicit “shelter everyone” rule pulled families from outside the county; a huge dedicated tax improved services and still could not outrun new homelessness plus regional in-migration. The tax did not fail to house people. It failed to shrink the problem the public was sold.
### 2. Seattle / King County
King County’s Regional Homelessness Authority and the city have spent on the order of **$1 billion+** through the authority over several years, plus large city Human Services budgets (Seattle’s homelessness-related spend has been in the $190 million range in recent years, on top of county and Health Through Housing sales-tax programs). Shelter and housing inventory went up. The point-in-time count still rose: roughly 13,400 (2022) → 16,900 (2024) → **18,365 (2026)**, with about **64% unsheltered**. That is thousands more people after years of elevated spending. A forensic audit of the authority also found overspending and weak controls—an accountability failure layered on a results failure.
City materials have also noted that a large share of unsheltered people surveyed in the Seattle metro said their last stable residence was **not Seattle**—i.e., a regional magnet into the city that runs the biggest shelter and service system.
What backfired: Housing First plus high tech-driven rents produced a large unsheltered population that services could process but not shrink. Money bought more beds and more nonprofit capacity; the street count and the bill both went up. Seattle is the textbook “booming city, booming rents, booming homeless budget” case.
### 3. San Francisco
San Francisco has spent at a level few cities can match. The Department of Homelessness and Supportive Housing budget has been in the **$700–830 million** range in recent years, heavily funded by Prop C / “Our City, Our Home” business taxes plus general fund. From 2007 through the mid-2020s the homeless count rose on the order of 40%+ even as spending exploded. Critics (and a civil grand jury) have argued the system put vulnerable people at risk, exits to stable housing lagged, and some city-funded housing became sites of heavy drug use—UCSF/DPH work has been cited showing a large share of overdose deaths among people with a fixed address in city-supported housing.
The 2026 point-in-time count finally showed movement the other way: total homelessness down about 4% (8,323 → 7,973) and **unsheltered down 22%** to the lowest since 2011, with a higher sheltered share. Family homelessness, however, rose. That recent dip came after a shift toward more shelter, outreach, and treatment-oriented “breaking the cycle” tactics—not after another decade of the same unconstrained Housing First mix. Even so, the city still runs a ~$800 million system for under 8,000 people on a given night and more than 20,000 system users over a year. Per-person cost is enormous relative to outcomes over the previous decade.
What backfired for years: unmatched per-capita spending plus low-barrier housing and services coexisted with a growing, highly visible street population and weak accountability. The magnet piece is smaller than the folklore (most people last housed in California), but SF was still a destination inside the state for services, weather, and a high-tolerance street environment. The recent improvement is itself evidence that *how* the money is used matters as much as how much is spent.
---
**Los Angeles** is the larger cousin: Measure H, then Measure A, plus HHH and city budgets that grew from roughly $140 million toward **$1 billion** a year at the city level, with county spending on top. The county housed well over 100,000 people over time and still sits near **73,000** on the latest count after a stall. Same pattern: huge placements, huge inflow, huge bill.
## How to read this without the culture-war version
Programs that give a specific person a unit usually help *that person* stay housed in the short run. Several LA studies of rapid rehousing and placements find lower return to the homeless system for the treated individual. That can be true at the same time the *city’s* count and costs rise.
The backfire is at the system level:
| What cities hoped | What often happened |
|---|---|
| More PSH / RRH / shelter → fewer people on the street | Inflow from rents + addiction outruns outflow; count stays high or rises |
| Low-barrier housing is humane and cheaper than jail/ER | Units fill; treatment is optional; overdoses and chaos concentrate in and around the buildings |
| Generous services are a local public good | Families (and some singles) relocate toward the generous CoC; neighboring places free-ride |
| “Housing is the solution” | Housing without enough supply, work/treatment expectations, or enforcement becomes a parallel welfare system that never empties |
If the policy goal is “house the people already here and shrink the street problem,” the research points toward: tight targeting, treatment requirements for the chronically addicted/mentally ill, enough market housing so low-income people are not pushed out, and not advertising an open-ended entitlement that neighboring counties can dump into. If the goal is “maximize beds with no conditions,” the West Coast decade is the experiment—and the counts and tax bills are the result.
Henry McClure
785.383.9994
