Coalition for the Homeless's 2026 NYC shelter-census claims cannot be checked against this government-data corpus — no matching table exists

Summary

The Coalition for the Homeless's State of the Homeless 2026: A Crisis Inherited, A Choice Ahead (published June 2026) makes two central headline claims about NYC homelessness, both sourced to NYC Department of Homeless Services (DHS) shelter-census administrative data: (1) 194,531 unique individuals used the NYC DHS shelter system in 2025 — the most in the system's 40+ year history, and (2) the nightly shelter census of longer-term New Yorkers (excluding new arrivals) grew 27 percent, or by 12,443 people, from January 2022 through December 2025 under Mayor Adams. Neither figure can be independently checked against this government-data corpus — there is no table anywhere in the warehouse (housing schema or otherwise) that carries NYC DHS shelter-census data, HUD Point-in-Time (PIT) counts, or any other homelessness/shelter-population measure. This is a genuine sourcing gap, confirmed live via a full catalog search and a full listing of every table in the housing schema, and has been filed as a new issue (kenstott/govdata-ops#755). Independent government data (HUD's own 2024 AHAR Point-in-Time report, fetched directly) corroborates the general direction — a large, rapid rise in NYC sheltered homelessness — but uses a different methodology and cannot verify the exact figures.

Claim-by-claim verdicts

2 not checkable here · 2 assertions checked

Verdicts: true, mostly true, partially true, mostly false, false, not checkable here (no table carries the measure), stale vintage (the article cites a release this corpus has not loaded — a freshness gap, not a falsehood).

Claim detail

What the report actually says (parsed directly from the PDF)

Read directly from the report's own downloadable PDF, not a secondary summary or the webpage alone — both were fetched and text-searched directly. Key figures verbatim:

• 'In 2025, 194,531 unique individuals utilized the NYC DHS shelter system over the course of the year — the most in the history of the shelter system.' (appears identically in the Executive Summary, Section I body text, and Figure 1.1's caption)
• 'The number of longer-term New Yorkers sleeping each night in the NYC DHS shelter system increased by 27 percent — or by 12,443 people — during the four years of the Adams administration' (Jan 2022–Dec 2025). Note: the report's own website introduction states '12,442,' a minor inconsistency with the PDF's consistent '12,443' — likely a typo, not a data disagreement.
• Within that 27% increase: single adults +35% (5,753 people), members of families with children +21% (5,395), members of adult families +43% (1,295).
• As of January 1, 2026, 'more than 350,000 people in New York City were without homes' (a broader figure including sheltered, unsheltered, and doubled-up populations, not the DHS shelter census alone).

The report's data source is explicitly NYC DHS's own shelter-census administrative records — an annual unique-user count and a nightly/monthly census — not HUD's national Point-in-Time (PIT) count or Annual Homeless Assessment Report (AHAR), which use a single night in January each year as their measurement window. This distinction matters: it is the reason no federal dataset carries the exact same measure the Coalition report cites.

Why this corpus cannot check it

Investigated as an open question, not an assumption in either direction:

1. search_catalog('homeless shelter census New York City') — matched only unrelated Census/ACS/housing tables (acs_population, building_permits, etc.); the tool's own unmatched_terms flag confirmed 'homeless', 'shelter', and 'york' appear in no table or column description in the corpus.
2. search_catalog('point in time count HUD sheltered unsheltered persons') — no match for HUD PIT/CoC data; unmatched_terms flagged 'sheltered' and 'unsheltered'.
3. search_catalog('homelessness') (bare noun) — zero matches.
4. list_tables(schema='housing') — full listing of all 20 tables/views reviewed (no SQL run against any, since none are topically relevant): building_permits, building_permits_place, fair_market_rents, hmda_applicant_demographics, hmda_lar, hmda_lending_by_tract, hmda_loans, house_price_index, housing_affordability_by_state, housing_hmda_denial_rate_by_race, housing_hpi_state_annual, housing_mortgage_originations, housing_opportunity_zone_counts, housing_permits_by_county, housing_subsidized_by_county, housing_subsidized_profile_by_county, hud_subsidized_county, hud_subsidized_housing, income_limits, income_limits_county, opportunity_zones. The closest tables (hud_subsidized_county / hud_subsidized_housing) measure occupied, subsidized housing units (Section 8, public housing, vouchers) — a materially different population from people in homeless shelters or unsheltered — and would not answer this claim even as a proxy.

No other schema in the corpus (census, health, fiscal, etc.) carries homelessness or shelter-census data either. This is a confirmed, permanent gap as of this session, not a matter of the wrong search terms.

Independent corroboration attempted

Since the corpus has nothing to offer, external primary sources were fetched directly (not just search-result summaries) to check the claim's plausibility against independent government data: NYC Open Data's DHS Daily Report dataset page (confirmed to exist, as the exact live series the report's figures are drawn from, though its raw data table itself was not extractable via this session's fetch tool), and HUD's own 2024 Annual Homeless Assessment Report (AHAR) Part 1 PDF, fetched and text-searched directly. The AHAR states: 'Between 2023 and 2024, New York saw a 53 percent increase [in sheltered PIT homelessness]... New York City... accounted for almost 88 percent of the increase in sheltered homelessness in New York City.' This corroborates the DIRECTION of the Coalition's claim — a large, NYC-driven surge in sheltered homelessness — from an independent federal source, but it measures a single January night's PIT count across a different one-year window (2023→2024) than the Coalition's DHS annual/nightly-census figure (Jan 2022–Dec 2025), so it cannot verify the specific 194,531 or 27%/12,443 figures.

Verdict

Both graded claims are rated not checkable here against the AskAmerica corpus — the reason is a confirmed absence of any matching table (no NYC DHS shelter-census data, no HUD PIT/AHAR data, no city-level homelessness breakdown of any kind), not a contradiction or confirmation of the reported figures. Directly parsing the report's own PDF confirms the figures are quoted faithfully and consistently (fidelity count 1, for one minor internal 12,442-vs-12,443 inconsistency between the Coalition's own webpage and PDF). Independent government data (HUD's 2024 AHAR) corroborates the general direction but not the precise magnitude, since it uses a different methodology and window (claims_accuracy count 0, with that caveat stated explicitly). A sourcing-gap issue (kenstott/govdata-ops#755) has been filed naming HUD PIT/AHAR data and NYC Open Data's DHS Daily Shelter Census as candidate sources for a follow-on ingest.

What This Report Does Not Answer

• Query the data directly and compute the actual current figure for NYC homelessness/shelter population: No table exists to query — this is a genuine corpus coverage gap, not a computation that was skipped. Filed as kenstott/govdata-ops#755.