Spending Up, Outcomes Down: What Secretary Turner's Math Hides

And the bill the House just put on paper to do it.

By Common Ladder · June 3, 2026 · 14 min read

HUD released the 2025 Annual Homelessness Assessment Report this week: 745,652 people experiencing homelessness on a single night in January 2025, a 3 percent decrease from the record high of 771,480 in 2024.¹ The decline is real. Secretary Turner immediately characterized it as evidence that his administration's November 2025 pivot away from Housing First — toward transitional housing, work requirements, and treatment mandates — is producing results.

The pivot began ten months after the January 2025 count was taken. Turner is attributing a result to a cause that had not yet occurred.

That sequencing error would be a footnote if it were isolated. It is not. The same argument — that record federal spending produced record homelessness, therefore Housing First failed, therefore a new approach is warranted — appears in Turner's May 14 Congressional testimony, in the FY27 House THUD appropriations bill that followed seven days later ($10.659 billion below FY26 enacted, a 10.4 percent cut), and in the FY2026 CoC Notice of Funding Opportunity published June 1, which rebalances HUD's primary homelessness grant stream away from the permanent housing interventions the evidence has been supporting for twenty years. This piece presents the analytical case against the syllogism in four parts. The Senate has not yet acted on the House mark. The full-committee markup is June 4.

Seven days, two moves: what just happened

On May 14, testifying before the Senate Appropriations Subcommittee on Transportation, Housing, and Urban Development, Secretary Turner defended the administration's proposed cuts with a single talking point: "CoC spending surged by more than fifty percent under Biden, yet street homelessness rose twenty percent since 2020 — despite record funding, communities reported record homelessness." Senator Murray countered, on the record, that "all of the data, all of the research shows that economic factors are driving homelessness." Senator Gillibrand pressed separately on why the 2025 Annual Homelessness Assessment Report had not been released. It was released this week: 745,652 people experiencing homelessness on a single night in January 2025, a 3 percent decrease from 2024's record high. Turner immediately claimed the decline as vindication of his administration's approach. His policy pivot began in November 2025. The January 2025 count predates it by ten months.

On May 20, the House Committee on Appropriations released the FY2027 THUD bill summary. On May 21, the subcommittee marked the bill up: $92.224 billion in total discretionary allocation — $10.659 billion below FY26 enacted, a 10.4 percent cut. The framing in the bill summary, that the legislation "invests in housing and wrap-around services for the nation's most vulnerable… women, children, and veterans," uses the same self-sufficiency lexicon Turner employed at the May 14 hearing. Full committee markup is June 4. NLIHC and CBPP both report bipartisan resistance to the deepest cuts in the May 12 and 14 hearings, but the House majority's mark is now on paper.

The shape of the move is this: set the analytical case aside, and the sequence is testimony → appropriations bill → grant program restructuring, with each step citing the one before it. The analytical case does not hold. The syllogism — spending rose, outcomes worsened, therefore the spending was for the wrong things — collapses on four counts.

First: aggregate federal spending does not produce community-level outcomes

Whether more federal dollars produce fewer homeless people in a given Continuum of Care depends almost entirely on what those dollars purchase, in which housing market, allocated by what kind of governance body. The peer-reviewed evidence on this is reasonably settled.

A two-way fixed-effects analysis of 380 CoCs over nine years (Kim & Sullivan, Public Administration Review, 2023) finds that boards with higher proportions of non-government stakeholders produce measurable reductions in total and unsheltered homelessness. A 2025 study of 343 CoCs (Jenisa & Jang, Systems) finds that per-capita CoC funding intensity is positively associated with standard system performance measures — and that the effect attenuates sharply in high-rent markets. A HUD PD&R analysis (Nisar et al., 2019) finds that housing-market tightness explains, in some specifications, more variance in CoC outcomes than program characteristics.^2, 3

Translation: between 2020 and 2024, the federal government increased CoC funding, and community-level homelessness rose in many places, because asking rents in those places rose faster than any plausible federal intervention could have absorbed. The aggregate spending number Turner cites contains no information about whether money reached permanent housing in tight markets, whether vouchers were able to lease up, or whether new dollars were diluted across activities that produce fewer placements per dollar. Without that allocation analysis, the spending-up-outcomes-down correlation is a number. The 10.4 percent cut is a number too. It will produce its effects the same way: through what it removes, where, and from whom.

Second: what actually drove the post-2020 increase — and what the 2025 AHAR shows

The cleanest causal estimate of what drove the 2020–2022 surge comes from Leifheit and colleagues (JAMA Network Open, 2025): the 11 percent average state-level homelessness increase observed in that window would have approached 20 percent without eviction moratoria. The moratoria prevented roughly 9 percentage points of the increase.⁴ When they ended — alongside the wind-down of pandemic emergency rental assistance and the eventual discontinuation of Emergency Housing Vouchers — pent-up evictions entered the system. The 2024 PIT count recorded 770,000 people experiencing homelessness on a single night, an 18 percent increase from 2023.

That surge is what Turner was selecting on before this week. Now HUD has released the 2025 AHAR, and Turner has a new problem. The 2025 count — taken in January 2025 — shows 745,652 people experiencing homelessness, a 3 percent national decline from 2024's record. Turner immediately attributed the decline to his administration's policy pivot away from Housing First. The pivot began in November 2025. The January 2025 count predates it by ten months. He is attributing a result to a cause that had not yet occurred.

The veterans data inside the same report is more precise. Veterans homelessness fell 8 percent between 2023 and 2024.⁷ HUD's own reporting attributes that decline to HUD-VASH — the HUD-Veterans Affairs Supportive Housing program, which provides permanent supportive housing vouchers through partnerships with the VA. HUD-VASH is explicitly a Housing First program. The most sustained positive trend in Turner's own data was produced by the intervention model he is now dismantling. The 2025 AHAR does not vindicate the pivot. It refutes it, on its own numbers and its own timeline.

A serious diagnosis of the post-2020 surge would credit the moratoria, ask why their expiration was allowed to unwind into the largest single-year increase since the PIT count began, and identify which interventions — emergency rental assistance, vouchers, rapid rehousing — buffered the impact where they were funded at scale. Applied to the January 2025 data, the same analysis would ask what actually produced the 3 percent decline. Moratoria were long expired, so it was not that. The most credible candidate is HUD-VASH's continued Housing First operations for veterans. That is an evidence-aligned policy conclusion. It is not the conclusion the House mark and the June 1 NOFO will fund.

Third: what the federal money was actually buying

The single largest body of evidence on what works for chronically homeless adults is the evidence on Housing First. The five-city At Home/Chez Soi randomized trial in Canada (n=2,148) found that Housing First participants spent 73 percent of follow-up time in stable housing, compared with 32 percent for treatment-as-usual — a 41-percentage-point gap that held in long-term follow-up.⁵ The Santa Clara Project Welcome Home RCT found that 86 percent of intervention participants were ever housed during the four-year trial, compared with 36 percent of controls. PSH retention rates of 80–90 percent at twelve months are the field standard; treatment-first comparable programs are reliably in the 30–50 percent range.

Housing First does not, on average, directly improve mental health symptoms or reduce substance use in the housed population. The 2019 BMJ Open meta-analysis is clear: "large improvements in housing stability… no clear differences" on mental health or substance outcomes compared with treatment-as-usual. That is not a failure of Housing First. It is the design. Housing First produces the precondition — stable shelter — under which treatment becomes possible. The model never claimed to be a treatment intervention. Confusing housing stability with treatment outcomes is the conceptual error underlying every "self-sufficiency" argument made against it.

The cost case has been established since Culhane, Metraux, and Hadley's 2002 study of 4,679 PSH placements in New York: roughly 94 percent of annual housing costs were offset by reductions in shelter, hospital, emergency, and corrections use within two years. The Chicago hospital RCT (Sadowski et al., JAMA, 2009) found 29 percent fewer hospitalizations and 24 percent fewer emergency department visits among PSH recipients. The Community Guide's 2022 systematic review of twenty studies put median Housing First cost at $16,479 per person per year and median economic benefit at $18,247.⁶ Housing First was not the spending that did not work. It was the spending that worked best.

What the cut and the NOFO are about to do

The June 1 NOFO prioritizes transitional housing, foregrounds the language of "self-sufficiency," expands eligibility to faith-based providers and specialty courts, and steps back from the 2016 requirement that CoC-funded projects adopt Housing First principles. The 2016 requirement was not arbitrary: it was the regulatory codification of the evidence summarized above. Removing it does not return the system to neutrality. It subsidizes the treatment-first compliance regime that the evidence has been discrediting.

The House mark moves the same argument up the funding chain. A 10.4 percent reduction to HUD's discretionary budget does not fall evenly. Tenant-Based Rental Assistance, Project-Based Rental Assistance, and homeless assistance grants are the largest line items in the THUD discretionary stack. NLIHC's preliminary read of the bill summary is that the homeless assistance grant total is held closer to flat than other accounts, but the surrounding rental assistance accounts that keep CoC clients housed downstream of placement absorb most of the proposed cut. The end of that pipeline is the same person on the same lease. If the rental assistance does not arrive, the lease ends. The CoC counts that as a return to homelessness regardless of why the lease ended.

Layered on top, the HUD time-limits and work-requirements proposed rule — comment period closed May 1 — would allow public housing agencies and Project-Based Rental Assistance owners to impose two-year time limits and 40-hour work requirements on "work-eligible" adult households. NLIHC and CBPP estimate 3.7 million people, including 1.9 million children, in households that would be exposed. The pattern across the three simultaneous federal actions is consistent: shift the federal homeless assistance system away from the populations for whom the evidence is strongest, and impose compliance conditions on those whose housing it still funds.

What evidence-aligned investment actually produces

California has been running the counterfactual. The state has appropriated roughly five billion dollars across Homeless Housing, Assistance and Prevention rounds since 2018. Round 6 closed May 13 with $760 million distributed across all 42 eligible regions. The state credits the program with more than 100,000 transitions to permanent housing since 2023 and a 9 percent statewide drop in unsheltered homelessness — the first such decline in fifteen years. The National Alliance to End Homelessness's independent analysis surfaces a specific subfinding worth keeping: eleven California CoCs saw youth homelessness decline by at least 50 percent between 2019 and 2024.

Round 7, previously projected at $500 million, is absent from Governor Newsom's proposed FY27 California budget. State forecasters project a roughly 20 percent statewide homelessness increase if the investment cycle stops. The state-level mirror of the federal situation is now clearly visible: the funding that produced the 9 percent unsheltered decline — the cleanest available counterexample to Turner's spending-up-outcomes-down framing — is politically contingent in both jurisdictions where it has been working. (Note: California HHAP outcome figures are state-reported and PROVISIONAL; independent peer-reviewed evaluation is pending.)

What the field can do before June 4

• Senate Appropriations — Transportation, Housing, and Urban Development subcommittee

  The House mark is on paper. The Senate has not yet moved. Hold the line on the homeless assistance grant total, restore the rental assistance accounts the bill summary obscures, and require HUD to explain on the record — before conference — how its NOFO pivot is consistent with what the 2025 AHAR actually shows: that the strongest positive trend (veterans, −8%) was produced by HUD-VASH, a Housing First program. Bipartisan resistance was already on the record in the May 12 and 14 hearings. Make it the conference position.

• HUD — Office of Special Needs Assistance Programs

  If the June 1 NOFO publishes in the form announced May 4, CoCs across the country will reorganize their program mix in the direction least supported by the evidence. Either restore the 2016 Housing First requirement, or publish, alongside the NOFO, the specific evidence base on which the pivot to transitional housing is grounded. The field will treat the absence of that evidence base as exactly what it is.

• CoC executive directors

  The NOFO will create operational pressure to retreat from Housing First fidelity. Do not do it. The evidence on which your high-performing programs were built has not changed. The federal language around them has. Document your population-level outcomes against the system performance measures before the funding pressure forces the retreat. Your own data is your strongest asset in the next 18 months.

• Foundation program officers

  The funder guide companion to this piece sets the standard explicitly: tie grant eligibility to evidence-aligned program design, fund Housing First fidelity directly, and resource CoC capacity to run honest by-name lists and produce the data the federal system is about to stop demanding. The retreat from evidence at the federal level only matters if the field participates in it. Foundations decide whether it does.

• State housing finance agencies and Medicaid managed-care organizations

  If the federal system contracts, state-level decisions about Medicaid health-related social needs, housing trust funds, and managed-care housing partnerships become the marginal investment. Treat the next two years as the period in which those commitments either become structural or they don't. The California HHAP record is the case study — with the active warning that what gets built can be defunded by the next budget cycle.

• Journalists, policy shops, and field organizations

  The argument the next four weeks needs is a precise public account of why the spending-up-outcomes-down framing is a misreading of the evidence — not a defense of any prior administration's record. The data are clear. The studies are named. The mechanism is identifiable. And the 2025 AHAR — Turner's own data — provides the timeline proof. After June 4, the argument is being made about a different artifact.

The evidence is what the evidence is. The post-2020 surge tracked the end of eviction moratoria. Housing First produced housing stability at high rates and at costs offset by reductions in emergency and incarceration use. The 2025 AHAR shows a 3 percent national decline — from a count taken ten months before Turner's pivot began. The strongest sustained trend in that report was produced by HUD-VASH, a Housing First program. The Secretary is using his own data to take credit for results produced by the model he is dismantling.

Between now and June 4, the analytical record needs to be visible in the rooms where appropriations decisions are made. The Senate has the evidence it needs to hold the line. So do foundation program officers deciding whether to backfill what the federal government stops funding. So do CoC executive directors deciding how to respond to a NOFO asking them to walk away from what works. The math is in this piece. The window is June 4.