Homelessness Population Estimate Methodology — Playbook
The official homeless count leaves people out by design — what each counting method measures, and what it misses.
What Toronto can actually do to make its own homelessness counts legible and complete — each move with its costs, its beneficiaries, and its receipts.
v2.0 · 2026-08-11
---
The honest bottom line
Every homelessness number you've seen for Toronto is real. It's also almost certainly not the number you think it is. The City runs at least four separate counting systems, each measuring a genuinely different thing. Toronto's Street Needs Assessment — the headline number, an estimated 15,418 in October 2024, down to an estimated 12,196 by October 2025 — is a single-night census that, by the City's own stated methodology, "did not include people experiencing hidden homelessness, such as people temporarily staying with friends or family." The 21% drop between 2024 and 2025 is mostly a falling refugee-claimant caseload, not a uniform improvement — outdoor homelessness's share actually doubled in two boroughs over the same period. Ask how many people used the shelter system last year and you get 24,200 (the full-year distinct-persons count); ask how many are actively cycling through right now and you get either 10,569 or 9,251 depending on which of this project's own documents you read — a genuine, unresolved internal gap, not a picked winner; ask how many beds are full tonight and you get roughly 7,800–9,800 depending on the night and season. The City's own Shelter System Flow Data page states the tension directly: nightly occupancy in 2023 was 9,010, and close to 22,000 different people used the system that year — three different populations measured over three different time windows, not three ways of saying the same thing. Buried in a government transition binder, reported by name and republished by eight regional outlets with no contradicting follow-up, is an "unofficial estimate" that put total Ontario homelessness at roughly 234,000 people — nearly nine times what the province's own Auditor General's math implies from the last hard municipal count. A government spokesperson called it "inaccurate" after it became public and never offered a corrected number. Two different ways of scaling that figure down to Toronto both land in the same rough place: 30,000 to 45,000 people, two to three times the official SNA count — not confirmed, but not invented either; it came from the government's own internal paperwork.
Commission a real Toronto-specific hidden-homelessness study. (a recommendation card) Use the Chicago Coalition for the Homeless's census-microdata "doubled-up homelessness" model — peer-reviewed, already proven in at least one major North American city, and never run for Toronto — instead of continuing to extrapolate from a disputed provincial figure.
Publish one plain-language page explaining which number means what. (a recommendation card) A single explainer covering SNA, Shelter Flow, nightly occupancy, and the By-Name List, cross-linked from every dashboard that currently reports one of these numbers without saying which one it is.
Neither requires inventing new data collection from scratch. One asks for a study using a method that already exists. The other asks the City to explain, in one place, what it's already publishing in four. What this doesn't settle: which number to use in any given argument — that's a downstream, card-level question — and whether the By-Name List's estimated 18–20% undercount and the Shelter Flow data's own approximately 18% figure are the same statistic counted twice or two separate numbers that happen to agree, which remains flagged and unresolved.
---
a recommendation card — A Toronto-Specific Hidden-Homelessness Census Study, Using an Existing Replicable Method
Card id: a recommendation card · Issue: homelessness-population-estimate-methodology · Backgrounder: our research file for that page · Trust: carried-forward (problem statement and comparator method)
Problem
Toronto's best current estimate of hidden/total homelessness (roughly 35,000–45,000, against an official SNA count of an estimated 15,418 or a more current estimated 12,196) is an extrapolation built from Ontario's own disputed, unofficial approximately 234,000 provincial estimate, not a direct Toronto measurement [backgrounder, "Ontario's disputed internal estimate..."]. The page’s own carried-forward sources names the specific gap this card addresses: no study has directly measured Toronto's hidden-homelessness rate against the provincial average, and a peer-reviewed, replicable methodology already exists to do so — the Chicago Coalition for the Homeless's census-microdata-based "doubled-up homelessness" model (Housing Policy Debate, 2022) [backgrounder, "International context"; carried-forward, an earlier synthesis document].
Action
Commission a Toronto-specific hidden-homelessness estimate using the Chicago Coalition for the Homeless's doubled-up-homelessness census methodology (or a directly comparable, published, peer-reviewed method), applied to Statistics Canada census microdata for Toronto, to produce a direct, defensible estimate in place of the current extrapolation range.
Jurisdiction split
- City does: commission or partner with an academic institution (e.g., a Toronto-area university's housing/urban-studies faculty, several of which are already active in this space per the page’s substrate) to conduct the study; provide Shelter System Flow Data and By-Name List data as inputs where methodologically appropriate and privacy-compliant.
- City demands of Province: none required for the study itself — this is fundable and executable at the municipal/academic level. A demand bucket applies only if the City seeks provincial data-sharing cooperation (e.g., OW/ODSP "no fixed address" administrative data, already partially used in the page’s provincial-level analysis) to strengthen the model; named as a possible, not required, ask.
- City demands of Feds: access to Statistics Canada census microdata at the geographic and variable granularity the Chicago model requires — this may require a federal data-access agreement (e.g., through a Research Data Centre) rather than open-data access; named generically since no specific current agreement was confirmed this review.
Cost
Order-of-magnitude: low-to-mid six figures CAD, anchored to the general cost profile of a single-city academic census-microdata study of this type (a defined research contract with one or two academic partners over 12–18 months) — no specific dollar figure for the Chicago study itself, or for a Toronto-specific replication, was located this review; this range is anchored to the general scale of comparable single-city administrative-linkage research studies already cited elsewhere in this page’s substrate (e.g., the ICES/Homeless Hub healthcare-cost linkage study), not to a specific published budget for either that study or this proposed one. Confidence in this cost range is accordingly low.
Funding path
No confirmed existing funding line. Plausible paths, named but not confirmed as available: a Reaching Home this library's internal records stream (federal), a City research partnership budget line (as already used for the ICES/Homeless Hub study), or a university-led grant (e.g., SSHRC) with the City as a data-access and dissemination partner. This card does not assert any of these is already committed.
Who benefits, and how
Researchers, policymakers, and advocacy organizations, via a defensible, Toronto-specific figure replacing an extrapolation anchored to a disputed provincial estimate — directly closing the gap the page’s own carried-forward documents names as the central unresolved question in this page’s evidence base [backgrounder, "What the evidence does and doesn't support"]. People experiencing hidden homelessness themselves, indirectly, via better-targeted prevention-spending arguments (per an earlier synthesis document’s own prevention-pool framing, inherited but not re-modeled in this card) once their scale is directly measured rather than extrapolated.
Who bears the cost, and how
Whichever funder ultimately supports the study (City research budget, federal research stream, or university grant funding) — opportunity cost against other research priorities in each case; no new tax or levy is proposed by this card.
Who benefits from the status quo
No beneficiary identified in the backing backgrounder's Cui Bono table — the backgrounder's own Cui Bono section states plainly that this page, being a measurement-methodology question rather than a service-delivery or procurement question, surfaced no ESTABLISHED or REPORTED beneficiary of undercounting specifically. This card accordingly carries no "who benefits from the status quo" line beyond this note, per the template's own instruction not to manufacture one.
ROI (four dimensions) — schema v2
(a) Financial ROI
Not directly estimable as a fiscal saving — this is a measurement/research investment, not a service-delivery change, and it does not itself avoid or generate a cost. An indirect case exists: a defensible Toronto-specific figure could improve the targeting of prevention spending (per an earlier synthesis document’s inherited prevention-pool illustration, itself explicitly labeled "not a verified figure" in its own source document), but this card does not re-model that illustration or claim a fiscal-savings figure from it; the City's existing practice of commissioning administrative-linkage studies (e.g., the ICES/Homeless Hub healthcare-cost study already in this page’s substrate) anchors the cost range above only, not a fiscal-return figure. Confidence: low — this is a research-investment action with no direct fiscal-return case to size.
(b) Economic ROI
Not yet estimable — no third-party economic-impact study of hidden-homelessness measurement research (as distinct from homelessness services themselves) was located this review; flagged as a genuine gap rather than inferred from an unrelated economic-impact figure. Confidence: low — stated as not yet estimable rather than guessed.
(c) Social ROI
A direct, Toronto-specific hidden-homelessness figure would resolve a documented, named gap in the evidence this page’s own substrate already identifies as its most significant open question [backgrounder, "Open questions / data gaps"] — better information for prevention policy, advocacy, and public understanding of the true scale of the crisis. This is a directional, evidence-quality claim, not a quantified wellbeing outcome; no study found this review quantifies the downstream social-outcome value of improved homelessness measurement specifically. Confidence: low-medium — the value of closing a named measurement gap is well-argued in the substrate, but not quantified.
(d) Environmental ROI
Genuinely environmentally neutral — a data/research commissioning action with no construction, land-use, or physical-infrastructure component; none needed to support that finding. Confidence: high — a research-commissioning action has no plausible direct environmental pathway to assess.
Evidence
- Backgrounder §"Ontario's disputed internal estimate, and the extrapolation methods built from it" · carried-forward, an earlier synthesis document Parts 3–6 · the 30,000–45,000 range and its unresolved gap
- Backgrounder §"International context" · carried-forward, an earlier synthesis document · Chicago Coalition for the Homeless doubled-up-homelessness methodology, named as a replicable method
Confidence & uncertainties
Medium confidence in the problem statement (directly evidenced, and the specific named gap is already identified in the page’s own substrate as the single most important unresolved question); low confidence in the cost range (anchored to a general comparator, not a specific priced study) and in all ROI dimensions except the directional Social ROI case. No specific funder, academic partner, or existing commitment to conduct this study was confirmed this review — this is a card proposing an action, not describing one already underway.
Status
DRAFT — blocked on: identifying a specific academic partner and funding source; fairness and legal review; confirmation of what StatCan microdata access would actually require and cost.
---
a recommendation card — Standardized Public Labeling of Which Count Is Being Cited, Across City Reporting
Card id: a recommendation card · Issue: homelessness-population-estimate-methodology · Backgrounder: our research file for that page · Trust: New load-bearing findings (built from this review's direct primary-source fetches)
Problem
The City itself publishes at least four structurally different homelessness counts — the SNA (single-night, visible + service-engaged), Shelter System Flow Data (rolling three-month distinct-persons measure), nightly shelter occupancy/capacity (a bed/room snapshot), and the By-Name List (an individually tracked coordinated-access roster) — each answering a different question, on different public City pages, without a single unified page stating which figure to use for which purpose [backgrounder, "Key tensions/tradeoffs"]. The City's own Shelter System Flow Data page already states the SERVICE_USER_COUNT ≠ beds distinction explicitly in its own methodology notes (nightly occupancy of 9,010 in 2023 vs. ~22,000 unique individuals served the same year) [backgrounder, "Shelter System Flow Data..."; NEW, fetched 2026-07-14], but that clarity is not surfaced anywhere outside that one dashboard's own methodology footnotes, creating exactly the kind of conflation risk this page’s substrate documents happening repeatedly in public discourse and prior library drafts.
Action
The City publishes a single, short, plain-language "which homelessness number should I use" explainer page — linked prominently from the SNA, Shelter System Flow Data, Daily Shelter & Overnight Service Usage, and By-Name List/Coordinated Access pages — that names each measure, states in one sentence what population and time window it covers, and states explicitly that they are not interchangeable or directly comparable without adjustment.
Jurisdiction split
- City does: publishing this explainer page and cross-linking it from existing dashboards is entirely within existing City of Toronto Housing Secretariat / communications authority — no other order of government needs to act.
- City demands of Province: empty bucket — this action requires no provincial action.
- City demands of Feds: empty bucket — this action requires no federal action, though the underlying PiT-count standardization this explainer would describe is itself a federal Reaching Home requirement, described but not altered by this card.
Cost
Order-of-magnitude: low four figures to low five figures CAD as a one-time content/communications cost, anchored to the general scale of the City's existing practice of publishing similar explainer content (e.g., the Housing Stability Service System Overview / glossary page already live on the City's site) [https://www.toronto.ca/city-government/data-research-maps/research-reports/housing-and-homelessness-research-and-reports/housing-stability-service-system-map-and-terms/,fetched 2026-07-14] — this is a content-authoring and cross-linking exercise on data the City already collects and already partially explains piecemeal, not a new data-collection program.
Funding path
Existing Housing Secretariat / City communications operating budget — a content-publishing change, not a program requiring new funding.
Who benefits, and how
Journalists, researchers, advocacy organizations, and the general public, via a single authoritative reference point reducing the risk of citing one count as if it were another — directly mitigating the same failure mode this page’s own carried-forward documents document happening historically in prior reporting and prior library drafts (e.g., the CBC "92,965 people" household/person conflation documented in the sibling social-housing-waitlist-data leaf, a structurally identical error pattern in a different dataset). City communications and 311 staff, via a single reference to point inquiries to rather than re-explaining the distinction ad hoc.
Who bears the cost, and how
City taxpayers city-wide, via the Housing Secretariat/communications operating budget; City staff, via the one-time work of drafting and cross-linking the explainer (an opportunity cost against other communications priorities).
Who benefits from the status quo
No beneficiary identified in the backing backgrounder's Cui Bono table — see Card 1's identical note above; this card carries no "who benefits from the status quo" line, per the template's instruction not to manufacture one where the backgrounder found none.
ROI (four dimensions) — schema v2
(a) Financial ROI
No direct fiscal-savings case is asserted — this is a transparency/communications action, not a service-delivery change. Indirect cost avoidance is plausible (fewer future correction cycles across derivative documents and press coverage, of the kind this page’s own substrate and its sibling social-housing-waitlist-data leaf both document happening) but not quantified; the City's own existing Housing Stability Service System Overview/glossary page anchors the action's own scale and cost, not a fiscal-return figure. Confidence: low — directional only, not costed.
(b) Economic ROI
Not directly modeled for this specific explainer-page action; broad open-government-data/transparency economic literature exists (e.g., the Open Data Institute's finding that open public-sector data assets generate measurable economic value, already cited in the sibling social-housing-waitlist-data page’s Card 1) but is not sized to this narrow a scope and is not re-cited here as if it were. Confidence: low — closer to "not yet estimable" than a genuine range.
(c) Social ROI
Directly traceable to a documented, recurring failure mode across this library's own history: count/unit conflations (households vs. people in the waitlist leaf; nightly occupancy vs. distinct-persons counts in this page) that required dedicated correction passes after the fact. A single, cross-linked explainer is a direct, low-cost mitigation for the same failure mode recurring in future public reporting, for the same population of downstream readers this page’s own substrate shows have already been affected. Confidence: medium — the underlying failure-mode pattern is well-evidenced within this library; the marginal effect of one explainer page on external reporting behaviour is not independently measured.
(d) Environmental ROI
Genuinely environmentally neutral — a content-publishing action with no construction, land-use, or physical-infrastructure component; none needed to support that finding. Confidence: high.
Evidence
- Backgrounder §"Shelter System Flow Data: a different, rolling measure..." · NEW, City of Toronto pages fetched 2026-07-14 · the SERVICE_USER_COUNT ≠ beds distinction, stated in the City's own words
- Backgrounder §"Key tensions/tradeoffs" · earlier research plus this review · four structurally different counts cited interchangeably in public discourse
- that page's recommendation cards, Card 1 (a recommendation card) · carried-forward · a structurally identical proposed mitigation for a parallel conflation problem in the waitlist leaf, cited here as a precedent for this card's own design, not copied uncritically
Confidence & uncertainties
Medium-high confidence in the problem statement (directly evidenced by this review's own primary-source fetches showing the City already states the distinction in one dashboard's footnotes but not centrally); lower confidence in the specific cost estimate, anchored to a general comparator rather than a cited unit cost. No existing City commitment to produce this specific explainer was found this review — this is a proposed action, not a description of one underway.
Status
DRAFT — blocked on: fairness and legal review; confirmation of whether Toronto's Housing Secretariat communications team already has a similar item in its own backlog (which would change this card's marginal-value case).
---
Production record
Drafting record
Status: DRAFT — v1.0, 2026-07-14 Leaf slug: homelessness-population-estimate-methodology · Backgrounder: our research file for that page Two cards in this file, per this library's standard page structure (ROI schema v2, four dimensions). Provenance noted per card. Costing follows the ratified order-of-magnitude + named-comparator bar (no fake-precise line items).
FIX-6 note (W1b cards audit, 2026-08-06): headings and self-referencing anchors normalized from "Card 1"/"Card 2" to the card-ID form (a recommendation card/02) so the File: anchors resolve. No inbound repo references to the old #card-1/#card-2 anchors were found.
v2.0 restructure (2026-08-11, a recorded standing decision/PLAYBOOK conversion, Lane L3b): opened with "The honest bottom line," adapted from archive/dayone/homelessness-population-estimate-methodology.md (retired day-one memo, a recorded standing decision); each card tightened, verbose ROI Range/Comparator-source/Confidence blocks collapsed into flowing prose; all citation tokens (carried-forward, New load-bearing findings, an earlier synthesis document, an earlier synthesis document) preserved verbatim.