Two Pillars: Ending Homelessness in Toronto by Housing People and Stopping the Inflow

Housing First for people already homeless, and prevention to stop new homelessness, are the two prioritizations the evidence supports; every other policy recommendation is support structure for one or both. This paper states the system's own current scale and human cost in verified figures, makes the evidence and cost case for each pillar, re-sorts a full 50-recommendation catalogue against them, and closes with a costed ask by audience.

Section 1 — The State of the System, in Verified Figures

A system already large, and still growing

Toronto's homelessness response is not a small or emerging problem the city is still building capacity to meet — it is already one of the largest municipal social-service systems in the country, and it is still growing. The City's Toronto Shelter and Support Services (TSSS) division ran a gross 2025 operating budget of $897.957 million, of which roughly 73% was funded by senior governments CSRL2-001; TSSS's own 2025 Budget Notes lay out a fully internally-consistent multi-year table reconciling gross expenditure, revenue, and net cost from 2022 through 2025 CSRL2-391. That budget serves a population whose official count hit a record 15,418 people in the October 2024 Street Needs Assessment, before falling to approximately 12,180 by October 2025 — a 21% one-year decline [CL-002, CL-101]. That decline should not be read, on its own, as a system successfully resolving chronic homelessness: the same SNA data attributes most of the movement to a fall in refugee-claimant shelter use, not to a fall in chronic homelessness specifically, and the underlying pressure on the system is not falling at anything like the same rate.

That underlying pressure shows up most clearly in the shelter system's own count of who it is actually serving. 78% of shelter users meet the system's own chronic-homelessness threshold — 180 or more cumulative days in shelter — a definition that is itself stricter than the federal three-year CMHC/HIFIS chronic-homelessness standard would produce, meaning this is not an inflated figure by comparison to how other jurisdictions count it CSRL2-003. A system whose users are overwhelmingly chronic, not transitional, is a system for which "more shelter beds" functions as a description of the status quo rather than a plan to change it.

The waitlist for the housing that would actually resolve this is large and growing in parallel. Toronto's rent-geared-to-income (RGI) social-housing waitlist stood at 92,965 active applications in Q3 2024, rising to 105,115 by the end of 2025 [CL-016, CL-161, CL-311, CL-326]. These are applications and households, not a headcount of individual people — a distinction the City's own Housing Data Hub draws explicitly; citing this figure as "105,115 people" would overstate the population affected and should be avoided.

The mortality toll, and why the revised figures matter

The human cost of this system operating at its current scale and structure is not abstract. 223 people died while experiencing homelessness in Toronto in 2021 — Toronto Public Health's current, standing figure, of whom 132 were confirmed shelter residents at the time of death [CL-015, CL-781]. This figure has a documented history worth stating plainly rather than leaving implicit: TPH's own reporting originally put the 2021 total at 216, before revising it upward to 223 in a later data release. 223, not 216, is correct and current.

The more recent years are, if anything, a starker revision. New data-sharing between Toronto Public Health and the Office of the Chief Coroner substantially raised the official 2022 and 2023 mortality counts: 2022 revised from an earlier 189 to 331; 2023 revised from an earlier 150 to 300 [CL-296, CL-792]. These are not competing estimates to weigh against each other — they are the same reporting system's own corrected, current baseline, arrived at by integrating coroner death-investigation data the earlier counts did not have access to. The honest reading is not that mortality is declining from a 2021 peak; it is that better data has revealed the true toll was, and by these figures still is, considerably higher than the originally-reported numbers suggested.

What this section does not resolve

This section states the scale and human cost of the current system; it does not yet make the case for an alternative, or attempt to net the system's cost against what an alternative would cost — that comparison is Section 2 (for the currently-homeless population) and Section 5 (for the system as a whole). Nor does it carry any hard-block dependency: the source documents behind these figures are solidly archived (Statistics Canada / City of Toronto Street Needs Assessment materials, filed under RES-0254), and none of these specific figures require the Indigenous-consultation or legal-appeal caveats that apply elsewhere in this paper (Sections 4 and 6 respectively).

Section 2 — Pillar One: Housing First

The core argument

If Section 1 establishes that the current system is large, structurally overwhelmed, and increasingly deadly for the people inside it, this section makes the case that there is a better-evidenced and less expensive alternative for exactly that population: Housing First. The international trial evidence and Toronto's own cost data converge on the same conclusion from two independent directions — a randomized controlled trial literature that isolates the model's effect, and a Toronto-specific cost comparison that shows what the current alternative already costs.

The trial evidence

Canada's own At Home/Chez Soi trial — the largest Housing First randomized controlled trial in the world at the time it ran — found a net cost, after offsetting savings elsewhere in the system, of $6,311 per person per year for high-need Housing First participants receiving Assertive Community Treatment, against a gross intervention cost of $20,367 offset by $14,056 in reduced use of other services (a 69% offset) CSRL2-346. This is not a claim that Housing First is free, or that it produces savings from day one in every budget line — it is a documented, trial-measured net cost figure for the specific high-need population the trial studied.

The same trial's Winnipeg site — whose study cohort was 71% Indigenous — found a housing-stability effect for Housing First over treatment-as-usual generally: 45% of Housing First participants were housed all of the time in the study's last six months, versus 29% of treatment-as-usual participants CSRL2-042. This is the trial's own Housing-First-vs-control finding, not a stratified Indigenous-vs-non-Indigenous outcome comparison — the cohort's composition is real and worth naming, but the 45%/29% gap itself measures the intervention's effect, not an ethnicity difference, and should not be read as the latter. This is outcome data from a clinical trial, not governance or self-determination content — it does not draw on the same source lineage this paper's Section 4 treatment of REC S-11 gates below, and no caveat is owed on it here; the distinction matters and is kept deliberately explicit rather than blurred.

The Toronto cost case

Trial evidence from elsewhere is corroborating; Toronto's own numbers are the headline case, and the two independent Toronto-specific costings point the same direction.

The system currently costs $58,972 per person per year for a homeless person with mental illness in Toronto, per Latimer et al.'s 2017 five-city study — this is Toronto's own city-specific figure, not a national or cross-city average [CL-028, CL-343]. $53,144 is a separate, frequently-confused figure from the same study — Vancouver's city-specific cost, not Toronto's; the two are not interchangeable, and $58,972 is confirmed as Toronto's own number via the study's full city-by-city breakdown (Vancouver $53,144; Winnipeg $45,565; Toronto $58,972; Montreal $56,406; Moncton $29,610). This is the headline figure this paper uses for Toronto's Housing First cost case, kept as a concrete dollar figure rather than a ratio-only framing because it is independently verified and harder to dismiss than a ratio alone.

A second, independent Toronto-specific corroboration comes from the City's own Auditor General: a 2022 audit found emergency shelter (including hotel/COVID programming) costs more than 3 times as much as supportive housing, and 7 to 10 times as much as subsidized housing or rental-subsidy programs CSRL2-228. This multiplier is not a restatement of the $58,972 figure — it is a second, independently-arrived-at statement of the same underlying conclusion, from the City's own audit function rather than an academic study, and this paper uses it as corroboration alongside the dollar figure, not as a replacement for it.

What building the alternative actually costs, with an honest caveat

The cost case above compares an existing system to an alternative; it says less about what it costs to build the supply of housing Housing First requires. Two Toronto capital figures are relevant here, and both carry a caveat that should travel with them rather than be dropped for a cleaner story. Toronto's Modular Housing Initiative delivered permanent supportive housing units, with private kitchens and bathrooms, at an average cost of $309,000 per unit — but this is not a clean benchmark: the City's own June 2023 Auditor General audit found this program ran 63% ($33 million) over its adjusted budget, citing insufficient pre-construction planning and improperly-reviewed change orders [CL-226, CL-341]. (The $309,000 figure itself carries a second, separate caveat: it traces to a secondary synthesis source, not yet independently confirmed against a primary document — the ledger tags it VERIFY, not VERIFIED. The AG's own overrun finding is independently solid regardless.) Separately, the Homelessness Services Capital Infrastructure Strategy (HSCIS) — the City's larger, still-in-progress capital program — carries a planning-stage target of $421,875 per bed (partitioned sleeping space, not a private unit), a figure independently confirmed against the City's own 2024 Budget Notes math ($675M ÷ 1,600 beds), but as of the 2026 budget notes only 7 of the program's 20 planned sites had been acquired [CL-225, CL-344]. Neither figure should be cited as a settled, completed-program cost; both are cited here with their caveats intact rather than dropped for a tidier presentation.

A gap this section does not close: Houston and Denver

Two frequently-cited North American Housing First results — Houston's claim of more than 36,000 people housed since 2012, and Denver's supportive-housing social impact bond reporting 79% housed and 86% housing retention — do not yet have a checkable primary source in hand and are not cited here as fact. This is stated as a plain, disclosed gap, not smoothed over: neither figure appears above. A checkable primary source (Houston's own Coalition for the Homeless reporting, or the original Denver social impact bond evaluation) would strengthen the North American comparator case; until then, Toronto's own cost case above, plus the Canadian trial evidence, carries this section's argument without needing them.

What this section does not resolve

This section does not resolve an archival gap: Latimer 2017 (the source of the $58,972 figure) is verified but its underlying source documents have not yet been formally archived, and the Auditor General's Modular Housing audit has the same gap — both are checked and confirmed figures, just not yet on file. This section also does not carry an Indigenous-consultation caveat on the Winnipeg outcome data cited above: that data is a trial outcome finding, not governance or self-determination content, and does not draw on the same source lineage as Section 4's REC S-11 discussion below.

Section 3 — Pillar Two: Prevention

The core argument

Housing First is the answer for people already inside the shelter system. Prevention is the answer for everyone else — and it is dramatically cheaper. Toronto and its comparator jurisdictions already run small, real programs that prove the mechanism works at the level of an individual case; the policy question is scale and reach, not invention.

The foundational cost comparison

Stephen Gaetz's 2012 synthesis of Canadian cost data remains the clearest statement of why prevention is the higher-leverage dollar. On average: a shelter bed costs $1,932/month, a provincial jail bed $4,333/month, a hospital bed $10,900/month — against $701/month for a City of Toronto rent supplement or $199.92/month for social housing CSRL2-614. The same synthesis found that providing support to someone who would otherwise become homeless saves the system an estimated $9,390/year, and roughly $350,000 over that person's lifetime CSRL2-614. These are 2012 figures, complementary to — not a replacement for — the more recent Latimer-derived Housing First cost figures used in Section 2; the two evidence bases point the same direction from different angles and different eras.

Real programs, real numbers

Toronto is not choosing between "prevention in theory" and "Housing First in practice" — it already runs prevention programs at real, if limited, scale:

• Toronto Rent Bank: $10.8M annual budget for 2026 CSRL2-024; 2,350 grants issued in 2024 CSRL2-168. (Flag: the 2,350 figure is ledger-verified but its corroboration to a single clean primary source is incomplete — a CBC report on the 2026 budget increase implies a consistent ~2,400 baseline but does not pin the figure directly. Cite it, but don't treat it as airtight the way the $10.8M budget figure is.)
• Eviction Prevention in the Community (EPIC): 809 tenancies sustained in 2024, 1,050 projected for 2025 CSRL2-169. EPIC is case-management-based, a distinct mechanism from the Rent Bank's direct financial grants — the two are complementary, not overlapping, levers.
• BC Rent Bank (a real, same-country comparator, not a Toronto figure): every $1 invested returned $5 in savings; $27.5M saved in 2023–24; 600+ people diverted from becoming homeless CSRL2-448. This corroborates the underlying prevention economics from outside Toronto's own reporting — useful precisely because it isn't the same city grading its own program.

Where the inflow is coming from, and where it isn't yet being stopped

Institutional discharge is a documented, quantified pipeline into homelessness that Toronto's existing prevention programs do not yet reach. Ontario's jail-release-into-homelessness rate nearly doubled from 8.8% (2016) to 17.3% (2021/22); at Toronto South Detention Centre specifically, the 2021/22 rate was 23.4% CSRL2-216. (Section 4 carries the recommendations this evidence supports — mandatory discharge housing plans and benefit continuity — since they belong to POLICY-001's own recommendation set, not to new drafting here.)

The scale of the funding gap, nationally

The Parliamentary Budget Officer's own May 2024 costing found that an additional $3.5B/year in federal spending — roughly seven times the current Reaching Home budget — would be needed just to halve, not end, chronic homelessness nationally [CL-082, CL-126, CL-425]. This figure is national, not Toronto-specific, and it costs halving the problem, not solving it; it should not be cited as if it were a Toronto number or a total-solution number.

The highest-leverage unrealized fix

Toronto's prevention infrastructure has a specific, identified, and currently unrealized upgrade: Rent Bank and EPIC outreach remain referral-based, not automatically triggered. The Landlord and Tenant Board does not share N4 (eviction notice) filing data with the City in real time, so there is no automatic connection between an eviction notice being filed and Rent Bank outreach beginning CSRL2-170. This is not a request for a new program — the Rent Bank and EPIC already exist and are already ledger-verified as effective at the individual-case level. It is a request to connect an already-working intervention to the earliest possible trigger point, which is a data-sharing and jurisdictional-coordination problem, not a program- design problem.

Summary framing for Section 4

Prevention, like Housing First, is not short of evidence — it is short of scale and of the administrative connections (like the N4 trigger) that would let already-proven programs reach people before they enter the shelter system rather than after. Section 4 sorts POLICY-001's full 50-recommendation set against both pillars, including the specific prevention-tier recommendations this evidence supports (Rent Bank scale-up, no-discharge-to-homelessness protocols, corrections and hospital discharge planning, and the LTB's own prevention mandate).

Section 4 — The 50 Recommendations Re-Mapped to the Pillars

Why this section is real, previously-unstarted work

POLICY-001 sorts its 50 recommendations by who must act: Municipal (M-01–M-16), Provincial (P-01–P-07), Federal (F-01–F-05), Systemic & Structural (S-01–S-12), Public Space (PS-01–PS-05), and Daytime Engagement (DT-01–DT-05). That sort answers "whose desk does this land on." It does not answer this paper's own question — "which pillar does this serve." No such mapping exists in the source recommendation catalogue. What follows is that mapping, built fresh against POLICY-001's own text, not inherited from any prior draft.

Method. A recommendation is sorted to Pillar One (Housing People) if its mechanism is moving someone who is currently homeless into permanent or supportive housing — Housing First delivery, housing-stock capital, exit-matching, or outcome incentives tied to housing placement. It is sorted to Pillar Two (Stopping the Inflow) if its mechanism is keeping someone who is not yet homeless from becoming homeless — income adequacy, eviction prevention, or institutional-discharge planning. Everything else — accountability and measurement infrastructure, workforce and data systems that cut across settings, legal/policy reforms that remove a structural barrier without themselves housing or preventing anyone, and governance or fiscal plumbing both pillars depend on — is sorted to systemic enabler. A handful of calls are genuinely close; those are flagged inline with the reasoning, not silently forced into a clean bucket for tidiness.

Citation note: only recommendations already cross-checked against the verified claim ledger carry a CL- anchor below. The rest are cited to POLICY-001's own recommendation ID directly (e.g. "M-09") — a full ledger cross-check of all 50 recommendations' individual supporting claims is a larger, separate verification task, and no new CL- ID has been invented here for anything that lacked one.

A citation correction worth stating plainly: REC S-11's ledger anchors are CL-366 (TICAB governance authority), CL-367 (ALFDC's 20%/25% set-aside), and CL-369 (the $13M PH25.4 allocation) — all genuinely about S-11 and verified. Three other claim IDs sometimes associated with this recommendation in earlier source material — a derived COHB-recipient estimate, the federal Value of Statistical Life standard, and the Lamport Stadium encampment-clearance lawsuit — have nothing to do with S-11 and are not cited here as anchors for it.

Pillar One — Housing People (11 recommendations)

Housing First delivery, housing-stock capital, and the mechanisms that move currently-homeless people into that housing.

Classification note — COHB (P-03/F-04): a first read might sort COHB to prevention, since a housing benefit sounds preventive. POLICY-001's own text is specific that COHB's documented function in Toronto is a shelter-exit tool — its 2024 suspension is linked to a 26.7% drop in housing exits, not a rise in new entries. Sorted on what the evidence says it does, not on what a benefit-program label suggests it should do.

Pillar Two — Stopping the Inflow (10 recommendations)

Income adequacy, eviction prevention, and institutional-discharge planning — keeping people who are not yet homeless from becoming homeless.

Classification note — F-05: asylum claimants are, in POLICY-001's own framing, already inside the shelter system in large numbers (53% of Toronto shelter users as of Oct 2024). Faster work-permit issuance doesn't move them into permanent housing (Pillar One) — it moves them toward self-sufficiency and off continued shelter reliance, which is the same underlying mechanism as prevention: reducing new/continued dependence on the shelter system. Sorted to Pillar Two on that basis, not because it targets people who have never touched the system.

Systemic enabler — supports both pillars, or the structural conditions either depends on (29 recommendations)

Accountability and measurement infrastructure; workforce and data systems that cut across shelters, hospitals, courts, and welfare offices; legal and policy reforms that remove a structural barrier without themselves housing or preventing anyone; and fiscal or governance plumbing both pillars need to function.

Classification notes on the less obvious systemic-enabler calls:

• M-10, M-13, M-14 (winter bed-matching, Managed Alcohol Program capacity, gender- differentiated shelter capacity) are shelter-system operations and equity fixes, not Housing First. Pillar One in this paper's own Section 2 framing is specifically the Housing-First alternative to shelter — these three recommendations improve the shelter system itself, which is a genuinely distinct function from either pillar's core mechanism. Sorted here rather than force-fit into Pillar One.
• P-06 bundles a decriminalization/repeal ask with a proposed fund redirect to Housing First. The fund-redirect component leans Pillar One; the core ask (repealing a criminalization statute) is a structural/legal-barrier removal in the same family as M-16. Sorted as systemic enabler on the primary ask, with the Housing First component noted rather than hidden.
• P-07 / S-08 (Bill 23 / development charges) fund new housing supply broadly — both permanent supportive housing (Pillar One) and deeply affordable rental supply that prevents entry (Pillar Two) draw on the same development-charge-funded pipeline. Sorted as an enabling condition both pillars depend on, not assigned to either. CL-412 (the Toronto Region Board of Trade's December 2025 finding that GTA development charges are themselves up 176% since 2011, with municipalities facing $250–290B in decade infrastructure need) complicates a simple "restore DC revenue" framing and should travel with this recommendation wherever cited.
• S-06 (Peer Navigators) is deployed explicitly across shelters, hospitals, courts, welfare offices, and street outreach — a genuinely cross-cutting workforce spanning both housing-adjacent and prevention-adjacent settings, unlike S-07's Housing-First-specific co-op placements. That breadth of deployment is why S-06 and S-07 are sorted differently despite both being workforce recommendations.
• S-10 (community mental health reinvestment) supports housing retention for Pillar One and reduces crisis-driven entries for Pillar Two simultaneously — a genuine dual-support case, not a residual "didn't fit elsewhere" placement.
• Tier 5 (PS-01–05) and Tier 6 (DT-01–05) make an inter-divisional cost-shifting problem visible (PS) and address a daytime support gap for people currently experiencing homelessness (DT). Neither tier's mechanism is housing placement or inflow prevention as such; both are sorted here as the honest answer rather than stretched to fit a pillar they don't actually serve.

The caveat on S-11 — carried explicitly, not dropped

REC S-11 (dedicated Indigenous-led housing capital and operating funding) is real and well-evidenced — the Toronto Indigenous Community Advisory Board (TICAB) already holds final approval authority over Indigenous-specific policy and assessment tools within Coordinated Access CSRL2-366; the Aboriginal Labour Force Development Circle (ALFDC) administers a Council-mandated 20% funding set-aside and a 25% housing-match target CSRL2-367; and a 2025 Council decision (PH25.4) authorized a concrete $13M allocation to Indigenous-led housing projects, co-led with Miziwe Biik Development Corporation and ALFDC CSRL2-369. On the evidence alone, this recommendation belongs in this paper.

But the underlying research documents this governance architecture partly through Indigenous organizations' own public statements — for instance, TICAB's own stated priority for "Indigenous-by-Indigenous" housing, quoted directly in POLICY-001's own S-11 text — and no Indigenous organization whose public statements this material draws on has yet reviewed how those statements are being used here. This is a relationship precondition, not a research gap, and this recommendation is not presented as clear for public use until that review happens.

This caveat must travel with S-11 every time it is cited — inline, not as a single footnote a reader could miss, matching the same discipline this paper applies to the 2026 Ontario Superior Court ruling discussed in Section 6 (cited there as a fact-of-the-world with its own under-appeal status stated every time it appears, never as settled doctrine). Any excerpt, summary, or derivative of this paper that includes S-11 must carry this caveat inline; one that cannot carry it should not include S-11 at all.

Section 5 — What It Costs vs. What the Status Quo Costs

The core argument

Sections 2 and 3 have already made the per-person and per-case argument: Housing First and prevention both cost less than what the current system already spends per person it fails to house or fails to divert. This section makes the same argument at a system level, and adds the piece Sections 2–3 deliberately left aside: what building the alternative actually costs to construct, not just to run, and how that construction cost compares in kind — not just in size — to a status quo that is not a cheap baseline but an already-expensive, and still-growing, recurring operating cost.

The per-person comparison, restated as a cost decision

Toronto's current system-wide cost for a homeless person with mental illness is $58,972 per person per year [CL-028, CL-343]; Housing First's trial-measured net cost for the same high-need population is $6,311 per person per year CSRL2-346 — both figures already established in Section 2 and repeated here only to make the point explicit: the "cost" of Pillar One is not an added expense against a cheap status quo. It is a lower number than what the status quo already costs, for the same people, today.

A methodological point worth making explicit rather than skated past: it is tempting to derive a Toronto-wide per-person cost by dividing the $897.957 million TSSS budget CSRL2-001 by some population figure, but the answer changes drastically depending which one is chosen. Dividing by the 21,929 unique individuals who used the shelter system at some point in 2023 CSRL2-004 suggests roughly $40,950 per person; dividing by the 9,010-person average nightly occupancy for the same year CSRL2-379 suggests roughly $99,660 per person — two arithmetically valid calculations, from two equally real and independently verified figures, that differ by more than double. This arithmetic is illustrative only, not a new ledger claim — it is presented here specifically to show why it is unstable, not as a citable system-wide average. This is exactly why this paper relies on Latimer's $58,972 figure: it is a targeted, peer-reviewed, published cost study of a specific high-need subpopulation, not a self-computed system-wide average that shifts by a factor of two depending on an arbitrary choice of denominator.

The methodological guardrail: no grand total

This paper deliberately does not sum all 50 of POLICY-001's recommendations into a single grand "total cost of fixing homelessness" figure. Two documented examples show why that discipline matters:

• P-01 and P-02 alone (raising Ontario Works to $1,500/month and ODSP to $2,000/month, indexed) — an income-adequacy measure, not a homelessness-specific intervention — combine to an estimated ~$5.25 billion per year CSRL2-267. This is a derived, internally-synthesized calculation, not an independently-verified figure — the ledger itself tags it INFERENCE/ VERIFY, not VERIFIED, and flags it as order-of-magnitude only, not accounting for family composition variation. It is presented here, honestly hedged, for exactly one reason: it already exceeds Toronto's entire TSSS shelter budget by roughly 5.8 times — a number this large cannot be summed into a "cost of the two pillars" total without misrepresenting an anti-poverty investment as a homelessness-system cost. P-01/P-02 belong in this paper's ask (Section 6) on their merits; they do not belong in a homelessness-system total.
• Toronto's own 2010-vs-2025 budget comparison — $60.692 million gross in 2010 against $897.957 million gross in 2025, a 14.8x nominal increase — is methodologically invalid without a major scope caveat CSRL2-545: the 2010 figure covers only POS shelter contracts, Personal Needs Allowances, Habitat Services, and motel contracts, while the 2025 figure covers the full current TSSS mandate, including refugee-response systems, street outreach, encampment response, and capital infrastructure programs that did not exist at comparable scale in 2010. Citing the 14.8x figure as pure cost inflation, without that scope caveat, would be a real methodological error, and this section does not make it.

The national and provincial scale of underfunding

Two separate, differently-scoped national findings both point at a large funding gap, and they are kept separate here deliberately rather than combined into one ratio. Stephen Gaetz's 2014 State of Homelessness in Canada report found the national cost of homelessness ran approximately $7 billion per year, against roughly $119 million per year in federal response funding at the time CSRL2-637. A full decade later, the Parliamentary Budget Officer's May 2024 costing found that an additional $3.5 billion per year — roughly seven times the current $561 million/year Reaching Home average — would be needed just to halve, not end, chronic homelessness nationally [CL-082, CL-126, CL-425]. These are not two data points on the same trend line: one is a 2014 total-cost-vs-total-response ratio, the other is a 2024 estimate of additional spending needed to achieve a partial (50%) reduction against a different funding baseline. Both point the same direction — the gap is large and has not closed — but averaging or otherwise combining them into a single number would overstate precision neither report offers.

Provincially, HelpSeeker's modelling for AMO found that ending chronic homelessness in Ontario would require an estimated $11 billion over 10 years, against the current Homelessness Prevention Program's roughly $700 million/year trajectory ($7 billion over the same 10 years) — a $4 billion gap CSRL2-144.

What building the alternative costs, restated as front-loaded capital

Section 2 already established, with its caveats intact, the two Toronto capital figures for building Housing First's housing supply: Modular Housing at $309,000 per unit (AG-flagged as a 63%/$33M cost overrun, not a clean benchmark) and HSCIS's planning-stage target of $421,875 per bed (only 7 of 20 planned sites acquired as of the 2026 budget notes) [CL-226, CL-341, CL-225, CL-344]. Restated here in cost-decision terms: both figures describe capital spending — money that builds a fixed unit or bed and, subject to maintenance, is largely done once spent. The $897.957 million TSSS operating budget CSRL2-001, by contrast, is recurring operating spending that must be re-appropriated every single year, has grown for years running, and has no structural mechanism that causes it to shrink on its own. A dollar of capital spending and a dollar of perpetual operating spending are not the same kind of cost, even before either side's caveats are applied — this is the "front-loaded capital against ongoing status-quo operating cost" framing this section's core argument names, and it holds even under each figure's own honest caveat (the AG-flagged overrun, the still-in-progress site acquisition, the imprecision of any single per-person system average discussed above).

What this section does not resolve

This section does not attempt a single unified "total cost of the two pillars" figure — that is precisely the move this section's own guardrail above declines to make, and drafting around that discipline rather than into it would undercut the paper's own credibility. It also does not resolve the archival gap already flagged in Section 2 (Latimer 2017 and the AG Modular Housing audit are verified but not yet formally archived). None of this section's figures require the Indigenous-consultation or under-appeal caveats that apply respectively to Section 4 (S-11) and Section 6 (the Waterloo encampment ruling discussed there).

Section 6 — The Ask, by Audience

Why this section is a synthesis, not new fact-gathering

Section 4 already sorted all 50 of POLICY-001's recommendations by which pillar they serve. POLICY-001 itself, independently, already sorts the same 50 recommendations by who must act: Municipal (M-01–M-16), Provincial (P-01–P-07), Federal (F-01–F-05), and three further tiers — Systemic & Structural (S-01–S-12), Public Space (PS-01–PS-05), and Daytime Engagement (DT-01–DT-05) — that POLICY-001 does not assign to a single level of government. This section re-addresses that existing WHO-must-act sort as four audiences — council, province, federal, civil society: the M-, P-, and F-tiers map directly onto council, province, and federal respectively; the S-, PS-, and DT-tiers — which POLICY-001 itself already treats as a distinct catch-all rather than assigning to one office's in-tray — are grouped here as the civil-society ask, since most of that tier's 22 recommendations depend on sustained public, sector, or cross-institutional pressure rather than a single legislative body's own initiative. No new facts are introduced in this section beyond Sections 1–5; every recommendation below already appears, with its anchor, in Section 4.

The council ask (POLICY-001's M-tier, 16 recommendations)

Toronto City Council holds direct authority over roughly a third of POLICY-001's full recommendation set, spanning both pillars and the systemic-enabler bucket: Pillar One items include fixing TCHC/social-housing vacancy management (M-01, ~2,200 more people housed from existing stock alone), converting POS shelter contracts to reward housing exits (M-02), building real centralized Coordinated Access and a comprehensive By-Name List (M-03, M-04), and redirecting 25% of HSCIS capital to modular permanent supportive housing (M-08); Pillar Two items include scaling the Rent Bank with an automatic N4 eviction-notice trigger (M-05) and a no-discharge-to-homelessness protocol for City hospitals (M-06); systemic items include open-book POS contract accounting (M-07), an annual public Functional Zero progress report (M-09), the City's own already-accepted Auditor General recommendations on winter shelter bed-matching (M-10), a wage floor extension for existing shelter operators (M-11), the Ombudsman's 14 refugee-claimant shelter-access recommendations (M-12), Managed Alcohol Program capacity-planning (M-13), gender-capacity parity (M-14), a Safe Streets Act-related amnesty (M-15), and a police directive against homelessness-status stops (M-16, carrying the same VERIFY, not VERIFIED, caveat on its underlying sample size that Section 4 already flags for CL-785).

This ask is not hypothetical — Council has already had, and declined, a directly comparable one. On December 18, 2024, Council received the City Ombudsman's report on refugee-claimant shelter access without discussion, without directing staff to implement any of its 14 recommendations, and voted 14–9 against a motion to reconsider — the first time since the Office of the Ombudsman's creation that a City Manager declined to endorse its recommendations at all [CL-451, CL-606]. This is stated here as documented Council behaviour on the specific M-12 ask, not as a claim about Council's motives or intentions on the other 15 M-tier items — but it is the honest, on-the-record baseline against which this ask should be read.

A second, live Council-level decision point intersects this ask directly, and it is where this paper is most careful to state a fact-of-the-world without making a legal argument. In November 2025, Council amended its own Interdivisional Protocol for Encampments (Motion MM34.4) to cap the number of shelter offers before removal at three and set a 24-hour removal timeline for encampments near a school, daycare, or playground — a policy the City's own Housing Rights Advisory Committee formally objected to in writing, calling it a "deliberate incursion on established human rights principles" [CL-453, CL-713]. Since that vote, a separate and external legal development has become directly relevant to the same underlying question: in May 2026, the Ontario Superior Court of Justice, in a case concerning a Waterloo Region encampment, found that clearing an encampment without adequate alternatives violates Charter rights, and — for the first time in Canadian jurisprudence — that homelessness itself constitutes an analogous ground under section 15 equality-rights analysis (2026 ONSC 2971, Gibson J., May 21 2026) [CL-494, CL-495]. This ruling is under appeal by the Government of Ontario and the Region of Waterloo as of June 2026, was announced jointly with an explicit rationale tied to a transit-construction project, and has no appeal hearing date yet set — it is not final, settled law, and this paper does not treat it as such CSRL2-496. It is also politically live in a way directly relevant to Council's own posture: the sitting Premier publicly called the ruling "the most ridiculous...[he] had ever seen" and said he wished he could get the judge's address to "send encampments to his backyard," prompting counsel in the case to write the Attorney General asking him to publicly reaffirm judicial independence CSRL2-497. Cited here as a documented fact-of-the-world — an under-appeal, non-final ruling relevant to an ongoing Council-level policy debate, and the public reaction it has provoked — not as a legal argument this paper is equipped to make: that question is for an actual lawyer, not resolvable by further drafting. This ruling remains under appeal, discussed further below.

This ruling's ledger anchors are CL-494 (exact citation and procedural history), CL-495 (the analogous-ground finding), CL-496 (the remedy and under-appeal status), and CL-497 (the Premier's public reaction). Two other claim IDs occasionally associated with this case in earlier source material — Ontario Sunshine List compensation data for a shelter operator's executive and a City general manager — are real, verified ledger rows with nothing to do with this ruling and are not cited here as anchors for it.

The provincial ask (POLICY-001's P-tier, 7 recommendations)

The provincial ask centres on income adequacy and institutional responsibility: raising Ontario Works to $1,500/month and ODSP to $2,000/month, both indexed (P-01, P-02 — an anti-poverty investment on the order of $5.25 billion/year per the honestly-hedged CL-267 figure discussed in Section 5, not a homelessness-system cost); restoring and protecting the Canada-Ontario Housing Benefit without encampment-clearance conditionality (P-03); implementing the Province's own 2021 Auditor General homelessness-strategy recommendations on targets and timelines (P-04); legislating a no-discharge-to-homelessness duty across provincial institutions — hospitals, corrections, youth care (P-05); repealing or reforming the Safer Municipalities Act and redirecting its $75.5 million to Housing First (P-06); and reversing Bill 23's damage to development-charge revenue for affordable and non-profit housing (P-07). On P-07 specifically, Toronto's own November 2022 staff estimate found the City stood to lose approximately $230 million per year in development charges, community benefit charges, and parkland levies combined if Bill 23 passed, including $120 million per year specifically in affordable-housing funding tied to the Housing Now program CSRL2-389. This ask should not be read as a simple call to restore full development- charge authority, however: the Toronto Region Board of Trade's own December 2025 finding that GTA development charges are already up 176% since 2011, against $250–290 billion in decade infrastructure need, complicates that framing and should travel with P-07 wherever it is cited CSRL2-412 — the Board's own position is reform and partial cost-shifting to senior governments, not simply restoring municipalities' pre-Bill-23 development-charge authority, and this paper does not force a false alignment between the two positions.

The federal ask (POLICY-001's F-tier, 5 recommendations)

The federal ask is the smallest tier by count but among the largest by dollar figure: making the Interim Housing Assistance Program (IHAP) permanent at full cost for asylum-claimant shelter reimbursement (F-01); committing the federal government's own $674.5 million HSCIS capital ask through the National Housing Strategy bilateral, converting to roughly 1,600 supportive beds at the same planning-stage caveat already carried in Section 2 and Section 4 (F-02); scaling Toronto's Reaching Home allocation from its current roughly 6.5% federal contribution toward a share proportionate to actual need (F-03); making the Canada-Ontario Housing Benefit permanent with protected municipal access — the federal companion to P-03's provincial ask (F-04); and closing the gap between the federal work-permit portal's stated 6–8 day issuance time and the 3–4 month to 1-year-plus reality settlement agencies report for asylum claimants (F-05). A federal-level ask is also where this paper's own methodological caution about mixing eras matters most, since two genuinely different framings are both legitimate and both easy to conflate: Gaetz's 2016 State of Homelessness in Canada report frames a $43.788 billion, 10-year federal investment ask explicitly as costing "$50 per Canadian per year, or less than $1 a week" CSRL2-619 — a ledgered, legitimate rhetorical framing device, distinct from and not a restatement of the PBO's 2024 $3.5 billion/year halving-not-ending figure discussed in Section 5. Both are cited here on their own terms, not combined into a single number.

The civil-society ask (POLICY-001's S-, PS-, and DT-tiers, 22 recommendations)

The largest tier by count does not belong to a single office's in-tray, which is exactly why it is grouped here as the ask directed at civil society broadly — service providers, advocacy coalitions, professional bodies, and the public whose sustained attention makes cross- institutional change possible: mandatory discharge-housing plans and benefit continuity from corrections, hospitals, and child welfare (S-01, S-02, S-03); extending outcome-based shelter contracting to stop cherry-picking easy-to-house clients (S-04); cross-system data integration across TCHC, hospitals, ODSP/OW, and corrections (S-05); 200+ funded Peer Navigator positions deployed across shelters, hospitals, courts, welfare offices, and street outreach (S-06); a Housing-First-specific student co-op pipeline (S-07); repealing or substantially amending Bill 23 as P-07's legislative companion, carrying the same CL-412 caveat (S-08); LTB adjudicator efficiency restoration plus a Wales-style statutory duty to prevent homelessness at the point of eviction threat (S-09); community mental-health reinvestment to the level promised at deinstitutionalization (S-10); fidelity-funded Individual Placement and Support (S-12); an inter-divisional "true cost of homelessness response" reporting and cost-sharing mechanism across TSSS, TTC, and TPL (PS-01–PS-05); and a network of dedicated daytime engagement hubs integrated with library and recreation services (DT-01–DT-05).

One item in this tier carries this paper's other hard-block caveat, and it must be repeated here rather than assumed carried from Section 4. S-11 — dedicated Indigenous-led housing capital and operating funding beyond the current 20% floor — is real, well-evidenced, and belongs in this ask on the merits: the Toronto Indigenous Community Advisory Board already holds final approval authority over Indigenous-specific policy within Coordinated Access CSRL2-366, the Aboriginal Labour Force Development Circle administers a Council-mandated 20%/25% funding set-aside CSRL2-367, and a 2025 Council decision authorized a concrete $13 million allocation co-led with Miziwe Biik Development Corporation and ALFDC CSRL2-369. But the underlying research draws on Indigenous organizations' own public statements, and no organization whose statements this material draws on has yet reviewed how those statements are being used — a relationship precondition, not a research gap, and this recommendation is not presented as clear for public use until that review happens. Any excerpt, summary, or derivative of this civil-society ask that includes S-11 must carry this caveat inline; one that cannot carry it should not include S-11 at all.

What this section does not resolve

This section is drafted without the "poverty-industry" framing found in some earlier source material — that is a deliberate editorial choice, not an oversight, and reflects this paper's own standing default toward neutral, verifiable framing over rhetorical framing. It does not attempt to independently verify every recommendation's supporting claim beyond what Section 4 and this section's own corrections above already checked — a full cross-check of all 50 recommendations' individual supporting claims remains a larger, separate verification task. This paper is a draft pending human review before publication.