Housing First: Evidence Base and Adoption Case

Give people housing before demanding they fix other problems first — what the strongest Canadian trial actually found.

What this page draws on: carried-forward (from that page's own internal records carried-forward documents — this library's prior synthesis document (v2.3), this page’s carried-forward master briefing (homelessness housing first), this library's prior executive synthesis (Executive Summary), all carried forward from earlier research per this page’s this page’s own scope note) and NEW (this run's live discovery, each with an inline source quote: exact quote + source + date accessed 2026-07-13). Date: 2026-07-13 · Coverage: not evaluated this review (breadth check against v1/master briefing is a separate follow-up). Cui Bono: 0 beneficiary entities identified directly in this page (pointer to homelessness-political-economy.md's 2 ESTABLISHED rows) — added 2026-07-14, a later review; see "Cui Bono" section below.

Voice note: Toronto already paid for the answers — this work reads it all, integrates it, and shows the work, including the uncertainties.

Indigenous context

Indigenous context: what Indigenous nations, organizations, and knowledge-holders have publicly said about this issue — the Indigenous Context Library (one of this library's own project records, added 2026-08-17).

---

Scope

The neutral scope question this document answers: what does the evidence base say about the Housing First model — does it work, for whom, at what cost, and what does adopting it at scale in Toronto actually require? This backgrounder covers: the Housing First model itself and what distinguishes it from the "staircase" model; the decisive Canadian trial evidence (At Home/Chez Soi); the cost-effectiveness case; national- and city-scale precedents (Finland, Houston, Medicine Hat) including their honest reversals; Toronto's own adoption record and structural constraints; and the FOR/AGAINST evidence on scaling it up. It does not cover, and hands off by name: encampment-clearance legal/enforcement policy specifically (homelessness-encampments), the deep-poverty/OW-ODSP income-adequacy case in full (a companion briefing per this page’s own master briefing), housing-supply/CLT production mechanics (housing-supply-affordability), or harm reduction/toxic-drug-crisis policy. No claim citations are required in this section — it frames the document rather than asserting facts about the world.

Current state

What Housing First is, and what it replaced

Housing First provides permanent housing immediately and unconditionally — no sobriety, treatment compliance, or "housing readiness" precondition — paired with flexible, voluntary support services, inverting the older "staircase" model that made housing a reward for progress through shelter and transitional stages first [the inherited master briefing's §Background]. Chronic homelessness (long-term or repeated, often with disability, mental illness, or addiction) is a minority of people experiencing homelessness but a majority of system cost, and is the model's core evidentiary target; transitional/episodic homelessness is more common and responds more to income and housing-supply interventions than to supportive housing specifically [the inherited master briefing's §Background]. "Functional zero" is a system-capacity standard — chronic homelessness rare and brief enough that the system houses anyone who becomes homeless within a short, defined window, with inflow matched by outflow — not a claim that no one is ever homeless [the inherited master briefing's §Background].

The decisive Canadian trial: At Home/Chez Soi

At Home/Chez Soi is the largest randomized controlled trial ever conducted to evaluate Housing First, run by the Mental Health Commission of Canada across five cities from 2009–2013, with a Toronto follow-up through 2017 [this library's homelessness master synthesis]. Toronto's own arm enrolled 575 participants (MAP Centre, Unity Health Toronto) [this library's homelessness master synthesis]. Across the trial's roughly 2,285 total participants (split into high-needs and moderate-needs arms), the approximately 950 high-needs participants — people with severe mental illness and chronic homelessness — saw 73% housed stably under Housing First versus 31% under treatment-as-usual, more than double [the inherited master briefing's §1]. Housing First was found superior to treatment-as-usual (Stergiopoulos et al., JAMA 2015) and cost-effective versus treatment-as-usual (Latimer et al., JAMA Network Open 2019), with benefits sustained through 2017 follow-up (Lachaud et al., Journal of Urban Health 2021) [this library's homelessness master synthesis]. The master briefing itself flags the headline 73%/31% figures with a [confirm exact figures against the primary MHCC report] note — a live open item this backgrounder inherits rather than silently resolves.

Cost-effectiveness

At Home/Chez Soi's net cost after offsets is approximately $6,311 per person per year — a Tier 1 figure used consistently across this page’s source documents — after roughly 69% of the intervention's gross cost was recovered through reduced shelter, hospital, and justice-system use [this library's homelessness master synthesis]. A National Low Income Housing Coalition systematic review separately found a median 1.8:1 societal benefit-to-cost ratio across all U.S. Housing First studies reviewed, and 1.3:1 for good-quality studies specifically [this library's homelessness master synthesis]. Toronto's own Auditor General found in 2022 that emergency shelters cost 3x more than supportive housing and 7–10x more than subsidized housing or rental subsidies [this library's homelessness master synthesis]. Chronic homelessness is already expensive to the public system — paid for in emergency-room visits, ambulance calls, policing, courts, and shelter beds — so Housing First's cost case is substantially a reallocation argument, not a request for entirely new money, though the "wrong-pocket" problem (capital costs land on one budget, savings accrue to different budgets and different orders of government) is a genuine financing obstacle [the inherited master briefing's §Costs & financing]. The master briefing names a concrete mechanism to break this deadlock rather than only naming the problem: a pooled-funding model in which the City, the Province, and health regions contribute to a single shared pot for housing-based health and support services, so the order of government whose budget realizes the savings (health, justice) directly co-funds the housing investment that generates those savings — "the mechanism that actually breaks the wrong-pocket deadlock rather than just naming it" [From this library’s earlier research from this page’s carried-forward master briefing (homelessness housing first), "Costs & financing" section — the master briefing's own proposed mechanism, not attributed there to any external source beyond the briefing's own analysis].

National-scale precedent: Finland, with an honest account of its reversal

Finland converted its national shelter system into permanent supportive housing starting in 2008, becoming the fourth-largest landlord in the country via the Y-Foundation, investing over EUR 270 million, and cutting shelter use 76% between 2008 and 2017 — a real, substantial, durable achievement not undermined by what follows [this library's homelessness master synthesis].

Correction/reconciliation (a later verification pass, 2026-07-13, live-verified): the EUR 21,000 and EUR 15,000 savings figures are not a genuine unreconciled conflict between two sources measuring the same thing — they come from two different studies at two different points in Finland's Housing First rollout. The EUR 21,000/person/year figure comes from a small pre-2008 pilot phase (fifteen homeless people observed for five months before moving into supported housing, then five months after) testing the model's financial efficiency ahead of the 2008 national rollout. The EUR 15,000/person/year figure (with a wider EUR 15,000–52,000 range) comes from a later, broader 2011 cost-effectiveness evaluation by Tampere University of Technology, covering the scaled national program rather than the original 15-person pilot (Connecticut General Assembly Office of Legislative Research, "Finland's Housing First Policy," 2023-R-0109, accessed 2026-07-13, citing the underlying Finnish sources). The two figures are therefore both plausibly correct for what each measures — a small early pilot versus a large mature program — rather than interchangeable estimates of the same quantity, and should not be cited as if either supersedes the other. Helsinki itself owns 70% of city land and mandates 25% affordable housing in new rental developments, a structural advantage Toronto lacks [this library's homelessness master synthesis]. High-fidelity Housing First staffing is itself a documented, corrected figure in this page’s own sources: Assertive Community Treatment (ACT) teams operate at approximately 1:10 staff-to-client for high-acuity clients, while Intensive Case Management (ICM) is specifically defined as fewer than 20 clients per worker (roughly 1:20) — a July 2026 correction replacing a previously-cited "1:3" ratio that did not match any established fidelity standard found across multiple peer-reviewed sources [From this library’s earlier research from this library's prior synthesis document (v2.3) §8, final bullet — figures corrected there against multiple peer-reviewed sources, no single external citation given beyond that correction note]. This is a directly relevant benchmark for any assessment of Toronto's own staffing and support-capacity requirements in scaling Housing First. But after 11 consecutive years of decline, Finnish homelessness rose for two consecutive years: 3,806 people in 2024 (+11% vs. 2023), then 4,579 in 2025 (+20% vs. 2024), per Varke, Finland's national housing finance agency [this library's homelessness master synthesis]. Long-term homelessness — the population Housing First targets most directly — rose 29% in 2025 alone [this library's homelessness master synthesis]. Y-Säätiö's CEO attributed the reversal to rising rents, cuts to housing-advice service budgets, and cuts to social security and income support [this library's homelessness master synthesis] — the same lesson this page’s own sources draw from Medicine Hat's reversal below: a well-designed Housing First system is not self-sustaining without continuous investment.

Correction (a later verification pass, 2026-07-13, live-verified directly against both primary sources, replacing the prior single-secondary-source citation): Varke (the Centre for State-Subsidised Housing Construction of Finland), Finland's own statistical authority, confirms directly: 4,579 homeless people living alone in 2025 (+20% year over year, "grew for the second year in a row" after over a decade of decline), 1,306 long-term homeless (+29%), 1,023 homeless women (+24%), 966 homeless people with a foreign background (+22%), 761 homeless people under 25 (+37%), and 68 homeless families (Varke, "Homeless people 2025," https://www.varke.fi/en/statistics/homelessness/homeless-people-2025, accessed 2026-07-13). Y-Säätiö's own press release, published under its own byline, independently confirms the same 4,579/1,306/20%/29% figures and states in its own headline and body text that "the rise in homelessness is the largest in recorded history" — this is Y-Säätiö's own characterization, not a secondary aggregator's paraphrase (Y-Säätiö, "Record Rise in Homelessness in Finland in 2025," published 2026-03-03, https://ysaatio.fi/en/news/the-rise-in-homelessness-is-the-largest-in-recorded-history/, accessed 2026-07-13). Independently re-verified 2026-07-16: both varke.fi and ysaatio.fi URLs directly re-fetched independently; all figures (4,579 total, +20%; 1,306 long-term, +29%; 1,023 women, +24%; 966 foreign-background, +22%; 761 under-25, +37%; 68 families) and both quotes match the live pages exactly, word-for-word.

Varke quote: "In 2025 in Finland there were 4,579 homeless people living alone, which is 773 more than in the year before... Homelessness among persons under the age of 25, women and persons with foreign background has also increased from the previous year. Also long-term homelessness rose for the first time in five years." Y-Säätiö quote: "Homelessness in Finland increased more in 2025 than ever before. The rise was record-breaking both in percentage terms and in the number of people affected, according to a report published by the Centre for State-Subsidised Housing Construction Varke."

⚠️ still being checked (unresolved, 2026-07-13): the prior draft of this document cited a claim, via a single secondary-source blog aggregator (LegalClarity), that "immigrants accounted for 26% of homeless individuals and 67% of homeless families in the Helsinki capital region" as of 2022. This specific figure could not be confirmed against a primary source in this review — Varke's own 2025 statistics report a related but different figure (966 homeless people "with foreign background" nationally in 2025, +22% year over year, out of 4,579 total — roughly 21% nationally, not 26% Helsinki-specific, and Varke does not appear to publish the family-specific 67% figure at all in the pages checked this review). Rather than repeat an unconfirmed figure or delete it silently, this document now flags it as: the specific 26%/67% (2022, Helsinki-specific) figures remain unconfirmed and should not be cited until traced to a primary Helsinki-city or Finnish-government document; the confirmed, primary-sourced, national-level figure — 966 people with a foreign background among Finland's 4,579 homeless in 2025, roughly 21% — should be used in its place if a foreign-background/homelessness figure is needed.

The same secondary source also documents a structural, non-population-driven part of the reversal: after Finland's PAAVO national programs ended, "national coordination weakened," support-service funding became municipality-discretionary rather than centrally earmarked, and quality became uneven across cities — a caution directly relevant to any Toronto design that assumes provincial/federal support funding, once secured, stays secured. [⚠️ Still being checked: single secondary source (LegalClarity), citing European Journal of Homelessness; not independently re-checked against the primary EJH article this run.]

City-scale precedent: Houston

Houston cut its homeless population 63% since 2011 (from roughly 8,500–8,950 to 3,325 by 2025) through a Housing First and Coordinated Entry approach — The Way Home, coordinating 100+ agencies — achieved without additional municipal spending beyond existing federal allocations [this library's homelessness master synthesis]. More than 36,000 people have been housed since 2012, with 90% retaining housing 2+ years; family homelessness fell 82% and chronic homelessness 69%; veteran homelessness was effectively ended by 2015 [this library's homelessness master synthesis]. Houston's rate (16 per 10,000) is well below comparator U.S. cities (LA 42, NY 47, Seattle 57, SF 82) [this library's homelessness master synthesis]. A recent modest increase is attributed to the expiration of federal pandemic-era relief — the same "gains require sustained investment" lesson as Finland and Medicine Hat [this library's homelessness master synthesis].

City-scale precedent: Medicine Hat — functional zero, with the mandatory caveat

Medicine Hat, Alberta (population ~63,300) became the first city in Canada to achieve functional zero chronic homelessness on June 2, 2021, confirmed by CAEH, Built for Zero Canada, and Community Solutions [this library's homelessness master synthesis]. Since 2009 the city has housed 1,323 unique individuals (358 chronically homeless, 328 children) through Housing First principles adopted that year, a real-time By-Name List, weekly case conferencing on every individual, and a self-defined "Housing Guarantee" since 2015 that no one would sleep rough or in shelter more than 10 days [this library's homelessness master synthesis]. Medicine Hat lost its functional-zero status by late 2021/early 2022 — as of December 2022, the city no longer qualified, driven by rising non-chronic homelessness from the affordable-housing shortage: people who had previously been housed could not sustain housing as rents rose, became newly homeless, and some became chronically homeless [this library's homelessness master synthesis]. This is explicitly framed in this page’s own sources as a failure of housing supply and income support, not of the Housing First model itself [this library's homelessness master synthesis]. Toronto is itself a Built for Zero Canada member community, and the BFZ Canada movement counts 33+ Canadian member communities, with London, Ontario having already achieved functional zero for veteran homelessness specifically [this library's homelessness master synthesis].

⚠️ Discovery note (not a load-bearing claim, flagged for a future pass): a 2026-07-13 search surfaced that the Government of Alberta ended Medicine Hat Community Housing Society's role as the region's Community-Based Organization effective March 31, 2025, and that a 2025 point-in-time count report exists (dated April 2026). No exact quote or primary URL was captured this review; this is flagged [not yet mined — genuinely uncovered] rather than asserted, and should be a target for the next discovery pass on this page.

Toronto's own adoption record

Toronto officially adopted Housing First as its philosophy, but in practice ran 3,808 housing exits in 11 months of 2024 against a nightly census of 12,000+, with 78% of shelter users meeting TSSS's own (180+ day) chronic-homelessness threshold — language adopted, not funded as an operating model [this library's homelessness master synthesis]. Toronto's 2024 Street Needs Assessment recorded a peak of 15,418 people on October 23, 2024; this page’s most recent inherited source reports the October 2025 point-in-time count showed a 21% decrease to approximately 12,180 [this library's prior executive synthesis], a decline this same source attributes almost entirely to a 47% drop in refugee claimants in the shelter system (6,600 at August 2024 peak to 3,500 by August 2025) rather than to improvement in chronic, long-term Toronto homelessness, which "has not improved" on the same account [this library's prior executive synthesis]. Outdoor/unsheltered homelessness is reported to be spreading even as the total count falls, with the proportion sleeping outside doubling in Etobicoke-York and North York between 2024 and 2025 [this library's prior executive synthesis].

⚠️ NEW (2026-07-13 discovery): the City's own November 2025 Winter Services Plan release gives a granular, dated snapshot of current shelter/housing-exit scale that updates and specifies the master briefing's more general figures.

Quote: "This winter, the City will open approximately 1,275 additional shelter and housing spaces... Approximately 370 permanent supportive and subsidized units[;] 244 Warming Centre spaces at five sites... Close to 490 new shelter and 24-hour winter respite site spaces[;] Approximately 175 additional surge capacity spaces... The City is currently providing accommodations to almost 9,000 people each night. The number of encampments has decreased from more than 500 at this time last year to about 330 this month. More than 700 people have been referred into the shelter system from encampments and 3,800 people have moved from the shelter system into permanent housing in 2025 so far. However, this momentum is at risk without stable funding agreements between all three orders of government." Source: City of Toronto, "City of Toronto releases details of 2025/2026 Winter Services Plan for people experiencing homelessness," news release dated 2025-11-07, https://www.toronto.ca/news/city-of-toronto-releases-details-of-2025-2026-winter-services-plan-for-people-experiencing-homelessness/ · accessed 2026-07-13. Independently re-verified 2026-07-16: direct live re-fetch of this exact URL; the full quoted block matches the live page word-for-word, including the "almost 9,000 people each night," "more than 500... to about 330," "700 people... 3,800 people," and 1,275/370/244/490/175 figures.

⚠️ NEW (2026-07-13 discovery): the City's Homelessness Services Capital Infrastructure Strategy (HSCIS) — the vehicle the master briefing criticizes on a per-space cost-comparison basis against modular housing — gives its own stated savings case and timeline.

Quote: "Toronto City Council approved a decade-long infrastructure strategy that will see up to 20 new shelters built by 2033... Each shelter will be smaller by design and serve approximately 80 people... The model will allow the City to transition out of expensive shelter hotels, leading to an anticipated savings of up to $127 per bed per night, representing $74 million in operating cost savings over 20 years." Source: City of Toronto, "City of Toronto unveils first six shelter locations under new 10-year strategy and expands winter services for people experiencing homelessness," news release dated 2024-12-11, https://www.toronto.ca/news/city-of-toronto-unveils-first-six-shelter-locations-under-new-10-year-strategy-and-expands-winter-services-for-people-experiencing-homelessness/ · accessed 2026-07-13. Read alongside this library's homelessness master synthesis's own finding that HSCIS spends 36% more per space than the City's Modular Housing Initiative for an inferior (partitioned rather than private) product — this is the same program described from two angles, not a conflicting figure.

Toronto: the case for and against

Section merged 2026-08-11 from a companion Toronto-specific brief (Lane L2a Toronto brief-merge pass).

FOR:

• The world's largest Housing First randomized controlled trial ran partly in Toronto: the At Home/Chez Soi trial's Toronto arm enrolled 575 participants (MAP Centre, Unity Health Toronto), with the full trial's high-needs arm (~950 participants) showing 73% housed stably versus 31% under treatment-as-usual [this library's homelessness master synthesis; backgrounder "The decisive Canadian trial"]. [confirm exact figures against the primary MHCC report — flagged in this page’s own master briefing, not independently re-verified this review.]
• Toronto's own Auditor General found emergency shelters cost 3x more than supportive housing and 7–10x more than subsidized housing or rental subsidies — a Toronto-specific, not merely trial-derived, cost argument for Housing First [this library's homelessness master synthesis].
• Toronto is a Built for Zero Canada member community with an existing coordinated-access tool (STARS Common Assessment Tool) already referenced in BFZ Canada's national resources — meaning the infrastructure Medicine Hat used to reach functional zero is not being built from scratch here [this library's homelessness master synthesis].
• Toronto's own 2020 Auditor General audit found that improving TCHC vacancy management by 50% would house approximately 2,200 more people and recover ~$7 million/year in currently wasted vacancy losses, with no new construction or program required [this library's homelessness master synthesis, REC 0] — see that page's recommendation cards a recommendation card for the full card.

AGAINST:

• Toronto's supportive- and affordable-housing supply is, on this page’s own framing, the binding constraint: "you cannot do Housing First without housing," and Toronto has a severe shortage of both [the inherited master briefing's §The strongest case AGAINST; backgrounder "Key tensions / tradeoffs"].
• The City's own Homelessness Services Capital Infrastructure Strategy (HSCIS) — its primary current capital response — costs $421,875 per partitioned sleeping space versus $309,000 per private unit with kitchen and bathroom under the City's own Modular Housing Initiative, a 36% premium for an inferior product, even though the City separately frames HSCIS as saving up to $127/bed-night versus shelter-hotels [this library's homelessness master synthesis; City of Toronto, 2024-12-11 release]. Both figures are cited; this brief does not resolve which comparison should govern.
• The Canada-Ontario Housing Benefit (COHB), which is provincially controlled, was effectively suspended April–October 2024 during a federal-provincial funding dispute, correlating with a 26.7% drop in housing exits from shelter that year (5,927 to 4,344) [this library's homelessness master synthesis] — a documented instance of a non-municipal funding lever directly constraining Toronto's own Housing First exit rate.
• Two independent precedents this page cites as proof Housing First scales — Finland nationally and Medicine Hat, Alberta at city scale — both later reversed for structurally similar reasons (rents/housing costs outpacing sustained investment), a caution against treating scale-up as a one-time fix rather than a continuously-funded commitment [backgrounder "National-scale precedent: Finland," "City-scale precedent: Medicine Hat"].

Symmetry note: both sides draw on multiple independent sources; this is not assessed as a lopsided evidence base on either side.

Toronto-specific figures: No committed L3 jurisdiction-specific fiscal data rows (FIR, CMHC, or Point-in-Time cost tables keyed to Toronto's SGC code) exist yet for this issue slug in this library's Toronto data layer; the figures below are drawn from the L4 backgrounder's inherited claims, which are themselves Toronto-specific primary-source figures (City budget documents, Auditor General reports) rather than L3-normalized data rows.

If a needed figure exists only at the provincial or national level (e.g., Ontario's 81,515-person 2024 known-homelessness count, or the At Home/Chez Soi trial's national design), it is stated above as such rather than presented as Toronto-specific.

Toronto-relevant precedents:

• Houston, Texas: cut homelessness 63% since 2011 (roughly 8,500–8,950 to 3,325 by 2025) via Housing First and Coordinated Entry, achieved without additional municipal spending beyond existing federal allocations, with 90% of the 36,000+ people housed since 2012 retaining housing 2+ years [this library's homelessness master synthesis].
• Medicine Hat, Alberta: achieved Canada's first functional-zero chronic homelessness on June 2, 2021, then lost that status by late 2021/early 2022 due to rising non-chronic homelessness from the local affordable-housing shortage — explicitly framed by this page’s sources as a housing-supply failure, not a Housing First model failure [this library's homelessness master synthesis].
• Finland (national): cut shelter use 76% between 2008 and 2017, then saw long-term homelessness rise 29% in 2025 alone (to 1,306 individuals), confirmed directly against Varke's and Y-Säätiö's own primary statistics (corrected 2026-07-13 from a prior single-secondary-source citation), after cuts to housing-advice and social-security budgets [this library's homelessness master synthesis; Varke, "Homeless people 2025"; Y-Säätiö, "Record Rise in Homelessness in Finland in 2025"; both accessed 2026-07-13].
• London, Ontario: has already achieved functional zero for veteran homelessness specifically, within the same Built for Zero Canada network Toronto belongs to [this library's homelessness master synthesis].

Municipal ask (upward): Per this library's issue index row C2★ (homelessness-encampments, the closest-mapped issue index row to this page), the Owner is: municipal Service Manager (CMSM/DSSAB) delivery, provincial funding (Homelessness Prevention Program), and federal funding (Reaching Home). Because non-municipal levels control significant levers here — the COHB (provincial) and Reaching Home/federal housing-investment transfers — this section records what has already been formally identified as a demand in this page’s sources, not a new call to action: this page’s own master briefing names a specific set of provincial accountability demands (REC 16: increase OW to $1,200/month indexed to inflation; increase ODSP to close the $11,503 poverty gap; reinstate rent control for post-2018 units; enact a provincial no-discharge-to-homelessness mandate; legislate a provincial chronic-homelessness target with accountability) [this library's homelessness master synthesis, REC 16]. this library's municipal-asks table has not yet been checked against this specific jurisdiction/issue combination this review — flagged as an open item rather than asserted as empty or populated.

Toronto bottom line: Toronto has the evidence — it ran part of the RCT that proved Housing First works, and its own Auditor General has independently confirmed the cost case — but has adopted Housing First as stated philosophy without funding it as the system's operating default, and the City's own most recent data (November 2025) states plainly that recent gains are "at risk without stable funding agreements between all three orders of government" [City of Toronto, 2025-11-07 release]. This is the single most defensible synthesis sentence this brief can state without exceeding a claim_type:recommendation framing it does not have the structure to carry here.

Toronto-specific uncertainties:

• The At Home/Chez Soi headline 73%/31% figures carry an unresolved [confirm] flag in this page’s own master briefing.
• No L3-normalized, SGC-keyed fiscal data rows exist yet for this issue in this library's Toronto data layer — all cost figures above are backgrounder-inherited primary-source figures, not committed L3 CSV rows.
• Whether Toronto's October 2025 PiT decrease is durable or will reverse if refugee-system pressure changes again is not addressed in either inherited source — flagged genuinely uncovered in the backgrounder.
• The Medicine Hat 2025 governance transition (Community-Based Organization role change) was surfaced in this run's discovery but not captured with an exact quote or primary source — flagged, not asserted.
• this library's municipal-asks table has not been checked against this specific jurisdiction/issue combination this review.

Key tensions / tradeoffs

A proven model against a binding housing-supply constraint. This page’s own sources describe At Home/Chez Soi and Finland/Houston/Medicine Hat as decisive evidence that Housing First works, with essentially no serious empirical dispute over efficacy [the inherited master briefing's §The strongest case FOR]. Set against that is the same sources' own honest account that "you cannot do Housing First without housing" — Toronto's severe supportive- and affordable-housing shortage means the model is, on this page’s own framing, "a proven engine with no fuel" [the inherited master briefing's §The strongest case AGAINST]. Both claims are drawn from the same source documents; this is a documented internal tension in the evidence about implementation feasibility, not a dispute about whether the underlying model works.

Toronto's declining headline count against an unimproved chronic core. The October 2025 PiT count shows a 21% system-wide decrease [this library's prior executive synthesis], which could read as vindication of current policy. But the same source attributes that decrease almost entirely to falling refugee-claimant numbers and states the chronic, long-term homelessness problem "has not improved," with outdoor homelessness spreading in some districts even as the total falls [this library's prior executive synthesis]. A reader relying only on the topline number would draw the opposite conclusion from a reader who reads the attribution — both figures are from the same source and both are cited here rather than only the more favorable one.

Finland and Medicine Hat as both proof-of-concept and cautionary tale. Both are cited in this page’s sources as the strongest evidence Housing First scales (a whole country; a Canadian city under Canadian housing-market conditions) [this library's homelessness master synthesis] and as the strongest evidence that scaled Housing First is not self-sustaining without continuous investment in housing supply and income support, since both reversed after initial success for structurally similar reasons (rents rising faster than support funding) [this library's homelessness master synthesis]. This page’s sources present both readings themselves rather than one being added by this backgrounder.

HSCIS's own savings claim against the master briefing's cost-efficiency critique of the same program. The City frames HSCIS as delivering $127/bed-night and $74M/20-year operating savings versus current shelter-hotel costs [City of Toronto, 2024-12-11 release, cited above]. The master briefing's own analysis finds HSCIS costs $421,875 per partitioned sleeping space versus $309,000 per private unit with kitchen/bathroom under the City's own Modular Housing Initiative — a 36% premium for an inferior product [this library's homelessness master synthesis]. Both figures can be true simultaneously (HSCIS may still save money versus hotels while remaining less efficient than the modular alternative); this document states both rather than resolving which comparison should govern the policy conclusion.

What the evidence does and doesn't support

Well-supported (independent sources/methods converging):

• Housing First produces materially better housing-stability outcomes than treatment-as-usual for chronically homeless, high-needs populations: the RCT evidence (At Home/Chez Soi), a national case (Finland's 2008–2017 76% shelter-use reduction), and two independent city cases (Houston, Medicine Hat) all converge on this, using different methods (RCT, national administrative data, city program evaluation) in different jurisdictions [this library's homelessness master synthesis].
• Housing First is cost-effective relative to emergency shelter and treatment-as-usual: Toronto's own Auditor General (3x/7–10x shelter-vs-housing cost ratio), the At Home/Chez Soi net-cost figure (~$6,311/person/year after 69% offset), and an independent U.S. systematic review (1.3–1.8:1 benefit-cost ratio) all converge, from three independent source types (municipal audit, Canadian RCT, U.S. multi-study review) [this library's homelessness master synthesis].
• Gains from Housing First are not self-sustaining without continued investment in housing supply and income support: this is independently demonstrated by Finland's 2024–2025 reversal and Medicine Hat's 2021–2022 reversal, two unrelated jurisdictions on two different continents reaching the same structural conclusion for similar underlying reasons [this library's homelessness master synthesis].

Thin or contested:

• The At Home/Chez Soi headline 73%/31% housing-stability figures are flagged by this page’s own master briefing as [confirm exact figures against the primary MHCC report] — inherited here as an open item, not independently re-verified this review.
• Reconciled (2026-07-13): Finland's EUR 21,000/person/year and EUR 15,000/person/year savings figures are not actually in conflict — they measure different things at different points in the program's history (a small pre-2008 pilot vs. a 2011 evaluation of the mature national program). See the correction inline above; this is no longer an open reconciliation item.
• Unresolved, flagged rather than repeated (2026-07-13): the prior draft's immigration-composition figure for Helsinki homelessness (26% of individuals, 67% of families, 2022) could not be confirmed against Varke's own primary statistics in this review and should not be cited until traced to a primary source; the confirmed replacement figure (966 people with a foreign background among Finland's 4,579 homeless nationally in 2025, +22% year over year) is now cited above in its place.
• The Medicine Hat MHCHS governance-transition item (Community-Based Organization role ending March 31, 2025) is flagged not yet mined above; no exact quote or primary source was captured this review.
• Denver's Social Impact Bond case study and Allegheny County's predictive-analytics case study both carry explicit ⚠️ ALL FIGURES REQUIRE PRIMARY SOURCE VERIFICATION flags in this page’s own inherited source and are accordingly not treated as well-supported in this document, though they are referenced in the recommendation cards below as illustrative model types rather than as Toronto-ready evidence.

International context

1. Treaties/frameworks touched. This page engages the right to adequate housing under Article 11(1) of the International Covenant on Economic, Social and Cultural Rights (ICESCR), domesticated via the National Housing Strategy Act (S.C. 2019, c. 29) — see homelessness-federal-funding-architecture.md's International context section for the Act's own statutory text, cited there directly against the Department of Justice's consolidated text and not re-fetched here. This page’s own point of engagement is the UN Committee on Economic, Social and Cultural Rights' General Comment No. 4 (1991), which frames "availability of services, materials, facilities and infrastructure" and "habitability" among the seven components of adequate housing under Article 11(1) [cited per homelessness-international-comparators.md's own citation of the same General Comment, accessed 2026-07-14] — directly relevant to this page’s own central distinction between the "staircase" model (housing as a conditional reward) and Housing First (housing provided immediately and unconditionally), since the General Comment's own framing of adequate housing as a component of dignity rather than a conditional reward is consistent with, though this document does not claim identical to, Housing First's own design premise.

2. Best global comparators. This page’s own "Current state" section already carries this corpus's most deeply evidenced international comparator material directly: Finland's national Housing First model (Y-Foundation/Y-Säätiö, 76% shelter-use reduction 2008–2017, with an honestly-stated 2024–2025 reversal) and Houston's city-scale precedent (63% reduction since 2011, with its own 2024–2025 reversal and enforcement-era complication) are both covered in this document's own dedicated sections above, cross-referenced consistently with homelessness-international-comparators.md (which owns comparator depth for the whole Domain H corpus and carries the fuller Rockford/Illinois and UK Rough Sleeping Initiative material this page does not repeat) and homelessness-federal-funding-architecture.md (which independently confirms Y-Säätiö's own current institutional description directly against its own website). Medicine Hat is this page’s own third, domestic comparator, already covered in depth above — cited here only to note that this document does not double it as an "international" comparator, consistent with homelessness-international-comparators.md's own Canadian/international split.

3. What Toronto/Ontario can steal shamelessly. Stated descriptively, not as a recommendation, and already substantively present in this document's own "Key tensions / tradeoffs" section: the specific, nameable structural lesson from both Finland and Medicine Hat is that Housing First's housing-stability mechanism is not itself in empirical dispute (this document's own "Well-supported" findings above), but both jurisdictions' reversals trace to the same specific, nameable design gap — support-service and income-support funding was not maintained at the same durability as the initial housing-placement investment. Finland's own structural detail (via homelessness-federal-funding-architecture.md's citation of Y-Säätiö's current site) is the specific transferable mechanism: a purpose-built non-profit housing developer (Y-Säätiö) paired with a standing national practitioner network (the Housing First Development Network) whose function is spreading operational practice, not just capital, across the sector — a distinct design element from simply funding units, and one this page’s own Toronto adoption-record section shows Toronto currently lacks an equivalent to (Toronto's HSCIS strategy funds capital; no standing cross-provider practice-network function analogous to Y-Säätiö's is documented in this page’s own sources).

Cui Bono — who profits from this problem persisting

Draft note: the sourced findings below are published pending independent legal review, which is currently under solicitation. Every row is a pointer to a named, already-published source finding — never this document's own allegation. This note is removed when legal review completes.

Per the Accountability Observatory's charter (Prime Rule: pointer, never author) and this library's standard page structure's binding requirement — 0 beneficiary entities identified directly in this page, and this document points instead to homelessness-political-economy.md's own Cui Bono table, which is the corpus's binding source for Toronto-specific beneficiary findings. This page’s own scope is the clinical/programmatic evidence case for Housing First — does the model work, for whom, at what cost — not an entity-level "who profits from the status quo persisting" analysis; this document's own Scope section already hands off the political-economy question to a companion document rather than absorbing it here. homelessness-political-economy.md's own Cui Bono table identifies two ESTABLISHED entities directly relevant to the cost comparison this page’s own "Cost-effectiveness" section draws on — the two anonymized hotel operators ("Hotel C"/"Hotel D") the 2022 Auditor General audit found overcharged the City for vacant emergency-shelter rooms (no registrable entity — the Auditor General's own report anonymizes the operator(s)), and Garda Canada Security Corporation/GardaWorld's $11.9 million encampment-support contract (per ACL-0009) — both of which sit on the "emergency shelter" side of this page’s own $49,640/year-shelter-vs-$6,311/year-Housing-First cost comparison, meaning both entities have a documented financial relationship to the status quo this page’s own evidence base shows is more expensive than the alternative it evaluates. This document does not re-list those rows as its own table, per this template's pointer-never-author discipline — a reader looking for the entity-level "who profits" answer should go to homelessness-political-economy.md directly.

Open questions / data gaps

• Genuinely uncovered: what specific supportive-housing unit count, at what capital and operating cost, would actually close Toronto's Housing First supply gap — the master briefing itself lists this as an open question rather than an answered one [the inherited master briefing's §Key uncertainties].
• Genuinely uncovered: whether Toronto's October 2025 PiT decrease (21%, attributed to falling refugee-claimant numbers) will itself reverse if refugee-system pressure rises again, or whether it represents a durable structural change — not addressed in either inherited source.
• Not yet mined: the Medicine Hat 2025 governance transition and its effect (if any) on functional-zero status or By-Name List operation — flagged above, no primary source captured this review.
• Not yet mined: a direct primary-source check of the Varke/Y-Säätiö 2025 Finland report itself, rather than the secondary summary relied on this review.

Claim-index appendix

Grouped by section used. Citations follow the format [Source §section] for carried-forward claims (this page has no formally registered claims yet — its carried-forward documents predate claims register wiring, per this page’s own scope note) and a full source quote for NEW claims.

• the inherited master briefing's §Background · carried-forward · Housing First definition, chronic vs. transitional, functional zero
• this library's homelessness master synthesis · carried-forward · At Home/Chez Soi trial design, Toronto arm, peer-reviewed citations
• the inherited master briefing's §1 · carried-forward (flagged confirm) · 73%/31% high-needs housing-stability figures
• this library's homelessness master synthesis · carried-forward · net cost $6,311/person/year; NLIHC 1.3–1.8:1 benefit-cost ratio
• this library's homelessness master synthesis · carried-forward · AG 2022 shelter-vs-housing cost ratio; Toronto adoption-as-language finding
• the inherited master briefing's §Costs & financing · carried-forward · wrong-pocket problem framing; pooled-funding model as the mechanism that breaks it (restored 2026-07-16, a later verification pass)
• this library's homelessness master synthesis · carried-forward (EUR 21k/15k figures now reconciled — see correction inline, 2026-07-13) · Finland national model and 2024–2025 reversal; ACT/ICM fidelity staffing ratios, §8 final bullet (restored 2026-07-16, a later verification pass)
• NEW (2026-07-13, a later verification pass, primary sources) · Varke, "Homeless people 2025"; Y-Säätiö, "Record Rise in Homelessness in Finland in 2025" · Finland 2025 figures (4,579 total, 1,306 long-term, +20%/+29%) and "largest increase in recorded history" framing, both now confirmed directly against primary sources rather than a secondary blog aggregator
• REMOVED (2026-07-13): the prior "26% of homeless individuals / 67% of homeless families" Helsinki immigrant-composition claim (LegalClarity, single secondary source) — could not be confirmed against Varke's own primary statistics; replaced with the confirmed primary figure (966 people with a foreign background nationally in 2025, +22% YoY), flagged in "Thin or contested" above
• this library's homelessness master synthesis · carried-forward · Houston case study
• this library's homelessness master synthesis · carried-forward · Medicine Hat functional zero achievement, mechanisms, loss of status, Toronto BFZ membership
• NEW (2026-07-13, flagged not yet mined) · Medicine Hat MHCHS governance transition, March 2025
• this library's prior executive synthesis · carried-forward · Toronto October 2025 PiT count, refugee-claimant attribution, outdoor-homelessness spread
• NEW (2026-07-13) · City of Toronto, 2025-11-07 news release · Winter Services Plan 2025/2026 figures
• NEW (2026-07-13) · City of Toronto, 2024-12-11 news release · HSCIS timeline and stated savings case
• this library's homelessness master synthesis · carried-forward · HSCIS vs. Modular Housing Initiative per-space cost comparison
• this library's homelessness master synthesis · carried-forward (“still being checked” in its source) · Denver SIB, Allegheny County predictive analytics — referenced as illustrative model types only
• International context / Cui Bono (added 2026-07-14): ICESCR Art. 11(1); General Comment No. 4 (1991) · cited by pointer to homelessness-international-comparators.md's own citation, accessed 2026-07-14 · Y-Säätiö/Housing First Development Network structural comparator · cited by pointer to homelessness-federal-funding-architecture.md's own direct-fetch citation · Cui Bono pointed to homelessness-political-economy.md's table (ENT-0009/ACL-0009; the "Hotel C"/"Hotel D" row carries no registrable entity, since the Auditor General's own report anonymizes the operator(s)), not re-derived here