This is an independent explorer built entirely from public HUD data. It is not a HUD product. It exists to answer, in one place: how much Continuum of Care (CoC) Program funding is there, where does it go, and which populations does it support, especially families? Built and maintained by Gaither Dynamic (gaitherdyn.com) as a free public tool.
1) How much CoC funding exists nationally, by state, and by CoC, and how has it changed since FY2012. 2) How much of it supports families with children (modeled). 3) How each CoC compares with peers on size, per-person funding, mix, and growth. 4) Which agencies hold the grants. 5) Where funding aligns with veterans, youth, chronically homeless people, and DV survivors.
Award dollars: HUD Exchange "CoC Awards by Program Component" reports, FY2012 to FY2024 (national and state), FY2015 to FY2024 (CoC level). FY2015 is the first year that report series breaks awards out by CoC. These are actual competition award totals, not estimates.
FY2012 CoC totals: the HUD Exchange Awards and Allocations database (project rows summed per CoC; exact duplicate rows are dropped, for example Hawaii's HI-501 project list is double-loaded in that database for $9,720,539 of phantom dollars). Validation against the state reports above: 41 of 54 states match to the dollar, every other state is within 1%, national within 0.06%. FY2012 CoC rows carry totals and project counts only; no component split was published, so component metrics read n/a for FY2012, never 0.
Homeless counts: HUD Point-in-Time (PIT) counts by CoC, 2012 to 2024, from the AHAR Part 1 files on huduser.gov.
Bed inventory: HUD Housing Inventory Count (HIC) by CoC, 2014 to 2024, from the same AHAR Part 1 page (earlier files pool bed types and cannot support the model).
Agencies: the HUD Exchange Awards and Allocations database, which publishes grantee-level rows through FY2023. Agency counts are distinct organization names exactly as HUD publishes them; a couple dozen organizations appear under slight name variants (for example with and without "Inc."), so the true entity count is marginally lower.
HUD does not tag award dollars by who they serve. The model here: for each geography and year, take each housing component's award dollars (PSH, RRH, Joint TH-RRH, TH, SH) and multiply by the share of that geography's beds of that type serving the population, from its own HIC. Example: if a CoC's RRH beds are 70% family beds, 70% of its RRH dollars count as family-supporting. Joint TH-RRH uses pooled TH plus RRH beds. If a CoC reports no beds of a type, the state share is used, then the national share. Services and infrastructure dollars (SSO, HMIS, Planning, UFA) serve systems, not beds, and are never split.
Family vs individual is a clean split: family beds serve households with children; everything else serves adult-only households. The two add up to the housing subtotal.
Dedicated-population lenses overlap. Veteran, youth, chronic, and DV figures use DEDICATED beds (beds reserved for that population). A DV-dedicated bed can also be a family bed. Never add the lenses together.
Why veterans get beds, not dollars. Most veteran-dedicated beds are funded by the VA (HUD-VASH vouchers, Grant and Per Diem), not by the CoC Program. Weighting CoC dollars by those beds would badly overstate CoC veteran spending, so the tool shows veteran-dedicated bed counts and veteran PIT counts as facts instead. Youth figures include some HHS-funded inventory and read as alignment, not spending; chronic figures cover PSH only, which is how HUD reports dedicated chronic beds.
Dollars per person homeless = total award divided by the PIT count, same year. Dollars per family household = modeled family dollars divided by PIT family households. The PIT is a one-night census in late January; it undercounts people in motels or doubled up, and the 2021 count was disrupted by covid. Per-person figures are for comparing geographies, not for estimating spending per served client.
CoC codes merge and retire over time (Maryland and Massachusetts did this repeatedly). This tool does not stitch merged histories together: every code is shown under its own code for exactly the years HUD reported it, so a merger appears as one series ending and the surviving CoC stepping up, not as organic growth. Growth figures are computed within a single code's own history and should be read with that caveat for CoCs that absorbed a neighbor. 25 retired codes have no PIT record under that code, so their per-person figures read n/a. The Kansas City CoC (MO-604) spans Missouri and Kansas: HUD publishes its people counts under the footnote code MO-604a, which this tool folds back into MO-604, and the whole CoC sits under Missouri, matching HUD's own state reports.
The data pipeline, verification scripts, and this interface were built with AI assistance (Anthropic's Claude). The numbers themselves are not AI-generated: award dollars are HUD's published figures parsed from the source reports, people and bed counts come straight from HUD's files, and the population split is the deterministic bed-share arithmetic described above, not an AI estimate. Every figure class has been verified against fresh downloads of the HUD sources, including independent re-parsing of all 713 award reports behind this tool (every published state report FY2012 to FY2024 and every national report FY2012 to FY2024).
Three official sources were tested for pre-2015 CoC-level awards, each validated against the verified state totals from the Awards by Component reports. The Awards and Allocations database reconciles almost perfectly for FY2012 (see sources above) but is missing roughly 36% of dollars for FY2013 and FY2014: the combined FY2013 to FY2014 competition is only partially loaded there. HUD's per-CoC Dashboard Reports for 2012 to 2014 cover only a subset of awards, with state sums 12% to 29% short and individual states off by as much as 99%. The original award announcement lists are no longer hosted on HUD.gov or HUD Exchange. Publishing numbers that cannot be reconciled against HUD's own state totals is not something this tool does, so FY2013 and FY2014 remain state-level only. If HUD republishes complete project lists for those two years, they will be added.
CoC-level detail covers FY2012 (totals only, from the grants database) and FY2015 to FY2024 (full component detail, from the Awards by Component reports). Virginia and Wyoming FY2023 state reports in the Awards by Component series are missing from HUD Exchange, so those trends skip a year. HUD's 2017 national report omits its UFA component row even though its own printed grand total includes those dollars; the missing $1,032,382 of UFA is restored here from the five state reports that carry it, so the FY2017 national total matches both HUD's printed grand total and the sum of states. HUD publishes the Kansas City (MO&KS) CoC's people counts under the footnote code MO-604a; they are folded into MO-604 so that CoC keeps its per-person figures. Population modeling starts FY2014; veteran, youth, and chronic bed detail starts FY2017; DV starts FY2014. Agency detail is FY2023 and comes from the live grants database, so it can differ slightly from announced totals (about half a percent nationally). Guam and the small territories have no boundary layer on the CoC map but appear in every table. CoC and state boundaries are a snapshot of HUD's own GIS layers served from this site for speed, with HUD's live GIS endpoint as the automatic fallback.
CoC: Continuum of Care, the regional planning body HUD funds. PSH: Permanent Supportive Housing. RRH: Rapid Re-housing. Joint TH-RRH: crisis housing plus rapid re-housing in one project. TH: Transitional Housing. SSO: Supportive Services Only (includes street outreach and coordinated entry). SH: Safe Haven. HMIS: the CoC's client data system. Planning / UFA: CoC administration. PIT: Point-in-Time count. HIC: Housing Inventory Count. Percentile: share of CoCs this one exceeds. Peer group: CoCs of similar award size, optionally same state or urban/rural category. Quantile map colors: equal-count bins, because funding is so skewed a linear scale would leave most of the map blank.
Is the family number HUD-reported? No. It is modeled as described above and always labeled modeled. The award totals ARE HUD-reported.
Where do I find a specific grant or agency? The Agencies tab covers every FY2023 grantee; for other years use the HUD Exchange Awards and Allocations database.
Why does my CoC show n/a per person? It has no PIT record under that code for that year (usually a merger).
Can I cite this? Yes, with the caveat that modeled figures are estimates. Every underlying source is public HUD data; the methodology is on this page.