The contractor layer

The story has changed

For three years the Multnomah County homeless-services story was an underspending story. Metro’s 2020 Supportive Housing Services measure was raising more than the joint city–county office could obligate; balances accumulated at the county; Metro periodically reminded the county, in public, that the money was sitting; the county replied that ramping a new system takes time; the press alternated between the two framings.

That story is over. In FY2024 Multnomah County spent $143.5M against $140.4M of Metro SHS revenue collected. In FY2025 it spent more than it collected, drawing down accumulated balance. In FY2026 the county used $3.8M of unbudgeted SHS carryover to partially fill a budget gap created by the loss of pandemic-era state and federal dollars.

The dollars are now out the door. The question is what they bought.

What replaced underspending

The structure that replaced underspending is contracted delivery at scale. The county’s Homeless Services Department is, in budget terms, mostly a pass-through. The department holds the contracts; a small number of nonprofit providers do the actual work — running the shelters, doing outreach, holding the leases on supportive housing units, providing the behavioral-health services on which most placements depend.

This structure is not a Multnomah County invention and not a scandal. It is how every large U.S. county delivers homeless services. The provider network exists because direct county delivery would be slower, more expensive, and politically harder to staff up. What is specific to Multnomah County in 2026 is the combination of three things:

1. The dollar volume is now large — Metro SHS plus county general fund plus the residue of pandemic-era federal dollars.
2. The Joint Office of Homeless Services as a city–county shared entity has been wound down; the City of Portland filed in late 2024 to end the joint arrangement, and the county now holds the relationship alone.
3. Per-program outcomes reporting at the provider level remains uneven, so it is not possible from public records to compute cost per placement or cost per twelve-month retention for most of the contractor table.

Underspending was the easy story because the dollar amount was the entire fact. A contractor-layer story is harder because the central fact is what is not publicly legible.

What is and is not on the public record

The county publishes annual SHS reports with system-level outcome figures: people newly placed into housing, people retained at twelve months, shelter bed-nights provided. These figures are real and the county has improved the cadence and detail of their publication over the last two years.

What is not published, in any form a member of the public can recompute from, is the per-provider version of those figures. The contractor table on this page lists eight providers. Two — Central City Concern and Path Home — publish program-level annual reports detailed enough that a reader can compute cost per outcome from public documents. The other six do not. Their bed-night totals appear in county aggregate reports; their placement rates, retention rates, and exit destinations do not.

This is the sense in which the dollar disappears at the contractor boundary. The county is publishing the system-level numbers, and most of the providers are publishing the input-level numbers (how many beds, how many shifts, how many staff), but the per-program cost-per-outcome arithmetic that would let a member of the public say this provider is delivering and this one is not is not available from public records for most of the system.

The defense and the rebuttal

The defense for this state of affairs is real and offered seriously. It runs roughly as follows.

The population being served is, by definition, the population that has fallen out of every other safety-net relationship — including the ones that produce cleaner outcome data. Many placements involve clients with severe behavioral-health complications, recent incarceration, or both; the same provider running the same program in the same building has very different success rates with very different referral cohorts; provider-level cost-per-outcome reporting, naively published, would create incentives for providers to refuse the harder-to-serve clients.

This is not wrong. Cost-per-outcome reporting that does not adjust for client acuity does in fact create those incentives, and the providers most willing to take the hardest clients would in fact appear to perform worst on a naive metric. This pattern is well-documented in behavioral health, in supportive housing, and in workforce reentry; it is one of the main reasons U.S. county homeless-services contracts default to input-based deliverables (beds, hours, units of service) rather than outcome-based ones.

The rebuttal is that “naively published outcome data would mislead” is an argument for publishing risk-adjusted outcome data, not an argument for not publishing per-provider outcome data at all. Several large U.S. counties — including King County in Washington and Maricopa County in Arizona — now publish acuity-adjusted provider outcomes alongside contract value. The methodology exists, has been peer-reviewed, and has been operationalized at jurisdictions of comparable scale. Multnomah County does not currently publish this.

The structural reading is not that the providers are hiding outcomes. It is that the public record is configured to make the system illegible at exactly the level — the contract — where the most consequential decisions are made.

The reorganization didn’t fix the legibility problem

The dissolution of the Joint Office of Homeless Services as a shared city–county entity, and the consolidation of operational responsibility inside the county’s Homeless Services Department, was sold publicly as a fix for accountability. A single chain of command, a single point of public answerability, fewer interagency disputes about whose budget covered what.

In a narrow sense it did fix one thing: the county is now unambiguously the entity to which the public can address the question of how SHS dollars are being spent, where in the prior structure two governments and a joint office could each plausibly route the question elsewhere. That is a real improvement and worth keeping.

In the broader sense — the contractor-layer legibility — the reorganization changed nothing. The legibility problem was never inside the joint-office governance structure. It was, and is, in the contracts themselves: in what the county requires providers to publicly report as a condition of the obligation, in what the county itself publishes about what is delivered against each contract, and in whether either of those publication regimes lets a member of the public recompute the per-program arithmetic.

A reorganization that does not change the contract terms cannot change what the public can see at the contract level. None of the published FY2024–25 SHS contracts that we have read add new public outcome-reporting requirements compared with the equivalent FY2022–23 contracts.

Absorption labor

There is a second public-record gap, and it is more important than the first.

The most substantial part of Multnomah County’s homeless-services delivery system, in terms of person-hours, is not on any contract. It is the unfunded slack absorbed by the public and quasi-public spaces that the funded system underperforms onto. This is the work the public library does when it functions, daytimes, as a de-facto warming center. It is the work the coffee-shop staff do when they let an unhoused regular charge a phone, or buy them coffee, or call a non-emergency line on their behalf when something is wrong. It is the work the laundromat owner does when they let someone sit out of the rain. It is the work the church does in the hours when its day program is not formally funded.

This labor is not optional. It is the substrate that keeps the funded system from generating worse public outcomes than it does. If you removed it — if the libraries enforced loitering rules, the coffee shops kicked everyone out, the laundromats locked the doors — the funded contractors would not absorb the slack. The placement rates would fall, the retention rates would fall, the visible-on-the-street population would rise.

A complete read of homeless-services delivery in Multnomah County is therefore a read of two layers, not one. The legible layer is the contractor table on this page. The illegible layer is the absorption labor performed, mostly without compensation, by the rest of the public and quasi-public infrastructure of the county. The next investigation in this series will document the second layer with the same level of specificity that this one applies to the first.

What we are saying, plainly

The Multnomah County Supportive Housing Services system in 2026 is no longer underspending. It is spending most of what comes in, and is now drawing down accumulated balance against budget gaps elsewhere in the county portfolio.

What that money buys, at the level of the individual contract and the individual provider, is not in the public record at the level required to evaluate it.

The reorganization that consolidated the Joint Office into the county’s Homeless Services Department was a real improvement on the question of who is answerable to the public. It was not a change to the question of what the public can see.

The contracts can be rewritten. Acuity-adjusted per-provider outcomes can be published. Other counties publish them. The fact that Multnomah County does not is a choice, not a constraint.