The Shelter Deficit

Executive Summary

In January 2024, 771,480 people were counted as homeless on a single night in the United States — an 18% increase over the previous year, the largest one-year jump in the history of the federal count. Family homelessness rose 39% in that same year; the number of homeless children in families rose 33%. Against this, one line of evidence is easy to overlook: several places have cut homelessness sharply and sustained the reduction for years. Finland has reduced homelessness by roughly 75% since 2008 and is the only country in the European Union where homelessness is falling. Houston, Texas cut homelessness 63% since 2011 while spending less per capita than comparable cities. Medicine Hat, Alberta became the first city in Canada to reach "functional zero" chronic homelessness in 2021. None of these results required a technology no one else has. All three used some version of the same idea: house people first, without preconditions, and treat the wraparound services — mental health care, addiction treatment, job support — as what happens after stability, not before it.

This paper's central argument, following directly from the question that prompted it, is that homelessness is not primarily an engineering or resource problem. Humanity already produces enough housing material, food, and wealth to end it many times over. What is missing is coordination: a single, accountable system connecting prevention, emergency response, permanent housing production, and the health and labour-market services that keep people housed once they are inside — rather than dozens of disconnected programmes each solving a fragment of the problem on its own budget cycle. This paper also corrects, transparently and in detail, one of the most widely repeated success stories in this field: Utah's claimed "91% reduction" in chronic homelessness, which independent analysis has shown was substantially a statistical artifact of changing how people were counted, not a 91% reduction in the number of people sleeping outside.

The paper quantifies the present scale of the crisis, walks through that correction in full, examines four jurisdictions that achieved real, sustained reductions, reviews the cost-offset economics that make Housing First frequently cheaper than the status quo it replaces, sets out the root causes and structural barriers (land cost, zoning, the CapEx/OpEx mismatch the source question for this paper identifies directly, and misaligned incentives between public companies and public need), and closes with an eight-pillar solutions framework, a practical diagnostic methodology for any city assessing its own system, and an original composite metric — the Housing System Resilience Score — for comparing how coordinated, not merely how well-funded, a jurisdiction's response actually is.

How to Read This Paper

NGE's standard practice is to verify before publishing, name what could not be verified, and correct errors in public when found. This paper mixes three different kinds of claims, and they should be read differently.

Established, citable research and data: HUD's Annual Homeless Assessment Report figures (Part II), the Finland, Houston, Medicine Hat and Vienna case-study figures (Part IV), the Utah methodology critique (Part III), and the Los Angeles and other cost-offset study figures (Part V and the real-evidence section) are drawn from named public sources — HUD, named research institutions, city and national agencies, and contemporaneous reporting from named outlets — and are cited in the References section. Where a figure is a range, an estimate, or drawn from a single study rather than independently replicated research, this paper says so explicitly.

This paper's own framing and synthesis: the characterisation of homelessness as a "systems problem" rather than a housing-stock problem, the grouping of four jurisdictions as "what worked," and the narrative connecting separate data points into a single argument are this paper's own editorial construction, building on the original question and framework supplied to NGE, not a finding attributed to any single cited source.

This paper's own proposal: the eight-pillar solutions framework in Part VIII, the diagnostic methodology in Part IX, and the Housing System Resilience Score (HSRS) introduced in Part XI are original to this paper. None is an existing federal standard, an industry benchmark, or the output of a third-party model. The worked HSRS example is illustrative, built to demonstrate the mechanic, not a validated measurement of any real city.

Part I — Homelessness as a Systems Problem, Not a Housing Shortage

Observation: public debate about homelessness routinely frames it as a question of supply — not enough shelters, not enough cheap housing, not enough beds. Evidence: yet jurisdictions with comparable or even worse housing-affordability pressure than their neighbours have achieved dramatically different homelessness outcomes depending on how coordinated their response is, and the clearest illustration is the stock-and-flow mechanic set out directly in the question this paper answers: build 10,000 cabins for 10,000 homeless people in a city, and a year later the city can still have thousands of homeless people, because new people enter homelessness from job loss, medical emergency, eviction, domestic violence, and addiction faster than a one-time housing project can absorb them. Insight: homelessness is therefore better modelled as a stock-and-flow system, in the same sense an economist models unemployment: the number of people homeless at any moment is not primarily a function of how much housing exists, but of the rate at which people flow into homelessness (the inflow) versus the rate at which they flow out of it into stable housing and stay there (the outflow), with the "stock" simply the accumulated difference. Implication: a policy response that adds housing supply without addressing inflow (eviction prevention, income support, discharge planning from hospitals, jails and foster care) or outflow durability (the wraparound services that keep someone housed once they have a unit) will show short-term visible progress and then plateau or reverse — exactly the pattern several of this paper's case studies in Part IV document explicitly, and exactly the dynamic the Utah correction in Part III shows can be mistaken for success when only the stock, not the flow, is measured.

This reframing has a direct policy consequence, developed in full in Parts VII and VIII: a city that builds housing but does not simultaneously operate coordinated intake, prevention, and support systems is solving only the stock side of the equation, and should expect only partial, fragile progress — not because the housing was the wrong intervention, but because it was an incomplete one.

Part II — The Scale of the Crisis

2.1 The United States: A Record High

HUD's January 2024 Point-in-Time count recorded 771,480 people experiencing homelessness on a single night across the United States — an 18% increase over January 2023, and the largest single-year increase in the twenty-year history of the federal count. Chronic homelessness (defined as homelessness lasting a year or longer, or recurring repeatedly, combined with a disabling condition) reached 152,585 people, meaning roughly one in three people counted had been homeless long enough, or severely enough, to meet that higher-bar definition. Family homelessness rose 39% in a single year; the number of children in homeless families rose 33%, reaching nearly 150,000.

2.2 What Drove the Record Increase

Officials and researchers attribute the 2024 jump to a convergence of several forces rather than one cause: rising housing costs and rents outpacing income growth; the expiration of pandemic-era eviction moratoria and rental assistance; a continuing shortage of housing affordable at the bottom of the income distribution; migration pressure on shelter systems in several large cities; and, in at least one state, a natural disaster — the Maui wildfires drove an 87% single-year spike in Hawaii's count. This multi-causal pattern is itself evidence for this paper's systems framing in Part I: a crisis with this many simultaneous, independent drivers cannot be resolved by any single-lever intervention, however well-funded.

2.3 Not Only a U.S. Problem: The International Picture

The scale and direction of the crisis documented above is not uniquely American. England's Shelter charity estimates at least 382,618 people are homeless in England today — roughly 1 in every 153 people in the country — a figure that rose 8% in a single year, adding approximately 28,602 people, with rough sleeping specifically up 20% year over year to at least 4,667 individuals on any given night; 350,480 of the total are in temporary accommodation, the highest figure since records began, and the count includes 175,025 children. Australia's 2021 Census recorded 122,494 people experiencing homelessness nationally, a rate of 48 people per 10,000 population — down from 50 per 10,000 in 2016 even as the absolute number rose 5.2%, illustrating the same population-growth-versus-absolute-count distinction this paper applies carefully to every figure it cites. Set against Finland's roughly 3,800 and its falling trend, documented in Part IV, these figures show the same forces — rents outpacing incomes, insufficient supply at the bottom of the market, and fragmented institutional response — operating at comparable severity across multiple high-income countries, reinforcing this paper's central claim that the crisis is systemic rather than a single national policy failure.

Part III — The Utah Myth: What a 91% Reduction Claim Actually Shows

For much of the 2010s, a single statistic dominated American homelessness policy discussion: Utah had reportedly reduced chronic homelessness by 91% through a Housing First approach, a claim repeated by national media, advocacy organisations, and policymakers as the clearest available proof that simply giving people homes, without preconditions, was the solution. NGE's house standard requires verifying a claim this consequential before building any policy argument on top of it — and the claim does not hold up.

Kevin Corinth, a senior fellow at the American Enterprise Institute, published a detailed methodological critique showing the reported decline was driven substantially by changes in how Utah counted and classified chronic homelessness over time, not by a 91% reduction in the number of people actually experiencing it. Utah's point-in-time count of chronically homeless individuals was effectively doubled in one count year (2009) but not adjusted using the same method in a later year (2015), creating an artificial discontinuity in the trend line. Separately, the state changed its classification rules so that people living in long-term transitional housing programmes were no longer counted as "chronically homeless" at all — reclassifying people out of the statistic rather than out of homelessness. Corinth's analysis concludes that if the two periods driven by these counting and classification changes are removed, chronic homelessness in Utah did not meaningfully fall over the decade in question.

Why this correction matters for the rest of this paper: Finland, Houston, and Medicine Hat's results, examined next in Part IV, are not immune to the same measurement risks Utah illustrates — any homelessness statistic can be distorted by counting methodology, and this paper has made a deliberate effort to favour figures corroborated across multiple independent sources, and to note explicitly, as in Finland's case, where a country's counting methodology is unusually broad or narrow relative to international norms, rather than treating a single favourable statistic as self-validating.

Part IV — Geography of Success: Four Places That Actually Reduced Homelessness

4.1 Finland — The Only EU Country Where Homelessness Is Falling

Finland has reduced homelessness by approximately 75% since 2008, and is widely cited, including by the European Commission's own migration and home affairs analysis, as the only country in the European Union where homelessness is declining rather than rising. The policy foundation is the "Name on the Door" initiative, introduced to Finland's Minister for Housing in 2007 and built on three arguments: an ethical case (housing as a matter of basic human dignity), a legal case (grounded in constitutional entitlements to dignified living conditions), and an economic case (long-run cost-effectiveness relative to cycling people through shelters, hospitals, and the justice system). The Y-Foundation, Finland's fourth-largest landlord, has been central to implementation, converting former shelters directly into permanent, independent apartments rather than operating them as temporary dormitories.

4.2 Houston — Doing More With Less, Not More With More

Houston, Texas cut homelessness by 63% since 2011, a reduction that outpaced most comparably sized U.S. cities, while spending less per capita on the problem than many of them — a detail this paper considers important precisely because it undercuts the assumption that better outcomes simply require more money. The mechanism was not a single new programme but a restructuring of governance: The Way Home, Houston's HUD-designated Continuum of Care, consolidated what had been a fragmented landscape of competing nonprofits and city, county and federal funding streams into a single coordinated system with one shared intake process, one shared data system to track who needed what, and one lead agency (the Coalition for the Homeless) accountable for system-wide results rather than any single programme's enrolment numbers. Over 90% of people placed into permanent housing through the system remained housed more than two years later, according to city officials.

Houston's own officials are candid about where the model still struggles: post-pandemic apartment occupancy above 95% has made it harder to find landlords willing to lease to people with eviction histories or criminal records, and the coordinated-system approach, by design, addresses people already homeless more effectively than it addresses the upstream causes — eviction, job loss, family breakdown — that keep adding new people to the flow, the exact limitation Part I's systems framing predicts for any intervention focused on outflow without equal attention to inflow.

4.3 Medicine Hat — A Small City Reaches Functional Zero

Medicine Hat, Alberta became the first city in Canada to reach "functional zero" chronic homelessness, in June 2021 — defined, under the Built for Zero Canada methodology the city adopted, as three or fewer individuals experiencing chronic homelessness in the community, sustained for three consecutive months, verified by real-time, by-name data rather than periodic point-in-time counts alone. Since 2009, the city's Community Housing Society has helped more than 1,323 unique individuals find housing, including 358 who had experienced chronic homelessness specifically. The model's distinguishing feature, relative to Houston's larger-scale coordination, is its use of continuous, individualised, by-name tracking — the system knows, by name, exactly who is currently chronically homeless at any given time, rather than relying on an annual or periodic estimate — combined with direct inclusion of people with lived experience of homelessness in the decision-making process itself, not merely as programme beneficiaries.

4.4 Vienna — Preventing Homelessness Through Housing Supply, Not Responding to It

Vienna takes a structurally different approach from the other three cases in this Part: rather than building a better system to respond to homelessness once it occurs, the city has built a housing market in which large-scale homelessness rarely gets the chance to form. Vienna's municipal government is landlord to roughly 220,000 housing units — about a quarter of the city's total housing stock — and oversees a further 200,000 units managed by regulated private and nonprofit developers under the same system, built substantially on housing programmes dating to the first half of the twentieth century. Eligibility for new tenancies is set broadly enough that roughly 75% of Vienna's population qualifies for social housing at the point they apply, meaning the system functions as a mainstream housing option for most residents rather than a narrowly targeted safety net for the poorest — and rents are set to cover each project's development and ongoing costs rather than to maximise return, removing much of the speculative pressure that drives rent growth in a purely private market.

4.5 Denver — The Randomized Controlled Trial the Field Rarely Gets

Most of the evidence in this Part, like most homelessness research generally, comes from before-and-after comparisons or matched comparison groups rather than a true randomized controlled trial — a limitation worth naming directly, consistent with this paper's measurement caution in Part III. Denver's Supportive Housing Social Impact Bond Initiative is the exception: a genuine randomized controlled trial, running from 2016 to 2020, which assigned 724 individuals with a history of frequent jail stays and chronic homelessness into a 363-person treatment group that received immediate permanent supportive housing and a 361-person control group that did not, then tracked both groups for three years. The results, evaluated independently by the Urban Institute, are among the strongest causal evidence available anywhere in this field: the treatment group achieved 86% housing stability at one year, declining to a still-substantial 77% at three years, while receiving 560 more days of housing assistance across the study period than the control group. Participants who received supportive housing had 34% fewer police contacts, 40% fewer arrests, 27% fewer days in jail, 40% fewer shelter visits, and 65% fewer visits to detoxification facilities than the control group over the same three years, with roughly half of the annual per-person cost of the housing offset by reductions in jail, ambulance, and emergency-department spending alone.

Part V — The Economics: Why Housing First Is Often Cheaper Than Doing Nothing

5.1 The Core Cost-Offset Finding

The strongest empirical argument for Housing First is not a moral one, though NGE does not dispute the moral case either; it is an economic one, and it is better evidenced than most policy interventions this paper has examined across its publication history. Los Angeles's "Just in Reach" Pay for Success programme, studied using a propensity-score-matched comparison of 336 participants against a comparison group of 672 individuals who received permanent supportive housing between October 2017 and September 2019, found that between 42% and 113% of the first-year cost of providing permanent supportive housing was offset by reductions in other public costs — meaning that for some participants, the housing intervention was effectively cost-neutral or better once the alternative costs it avoided (jail, emergency shelter, hospital emergency visits) are counted.

5.2 Why the Offset Isn't Always 100%, and Why That's Still a Good Outcome

A cost offset of 42%–113% means that, for some individuals in the Los Angeles study, permanent supportive housing cost the public system less overall than leaving them unhoused and cycling through jail, shelters, and emergency rooms, while for others it cost more even after the offset. This paper treats the honest range, rather than the most favourable single number, as the correct way to present the finding: Housing First is frequently — not universally — a net fiscal saving, and even where it is not a full offset, it reliably improves housing stability (82% at one year in the Los Angeles cohort) at a bounded, predictable net cost, compared to the unbounded and unpredictable cost of an emergency-driven status quo that still has to pay for the jail days, hospital visits, and shelter beds anyway.

5.3 The CapEx/OpEx Distinction the Original Question Identifies Directly

The framework this paper expands on makes a structural point worth stating in explicit economic terms: building a housing unit is capital expenditure, a one-time cost; operating supportive housing — staffing, healthcare coordination, security, food, utilities, maintenance — is operating expenditure, a cost that recurs every year the unit is occupied. A funder, public or private, that pays for the capital cost of construction but not the ongoing operating cost has paid for only half of what Part I's systems framing shows is actually required for a housing placement to be durable rather than a revolving door — precisely the distinction that separates Houston's and Medicine Hat's coordinated, continuously funded systems from a one-time construction programme that quietly loses its gains within a few years once support funding lapses.

Part VI — Root Causes: Why Homelessness Forms

Consistent with this paper's systems framing in Part I, no single root cause explains the inflow into homelessness; five distinct, interacting causes do, in proportions that vary by jurisdiction.

6.1 The Housing Cost–Income Gap

In the large majority of U.S. metropolitan areas, rent growth has outpaced wage growth for low-income workers for more than a decade, narrowing or eliminating the margin a household has to absorb a job loss, a medical bill, or a reduced-hours month without falling behind on rent — the single most commonly cited proximate driver behind the 2024 record increase documented in Part II.

6.2 Untreated Mental Illness and Addiction

A substantial share of the chronically homeless population — the 152,585 people in the U.S. count who have been homeless for a year or longer, or repeatedly, combined with a disabling condition — experiences serious mental illness, substance use disorder, or both, conditions that make sustaining stable housing without coordinated clinical support exceptionally difficult, which is precisely why Housing First pairs housing with voluntary, not mandatory, access to treatment rather than treating sobriety or psychiatric stability as a precondition for housing, the design choice examined directly in Part X's discussion of pushback.

6.3 Domestic Violence

Fleeing domestic violence is a leading cause of homelessness for women and families specifically, and is also one of the populations point-in-time counts are most likely to undercount, since survivors often avoid visible shelters or public spaces out of safety concerns — a measurement gap this paper flags consistent with the honest-limitation caution in Part II.

6.4 Discharge Without a Housing Plan

People leaving institutional settings — hospitals, jails and prisons, and the foster care system at the age of majority — without a confirmed, funded housing plan already in place are at sharply elevated risk of immediate homelessness; this is a structural, administrative failure as much as an individual circumstance, and is directly addressable by the kind of discharge-planning coordination examined in Part VIII's framework.

6.5 Economic Shocks Without a Buffer

For a meaningful share of people entering homelessness for the first time, the proximate cause is a single economic shock — job loss, a medical emergency, a reduction in work hours — hitting a household with no savings buffer and no accessible emergency assistance before the shock becomes an eviction, which is why this paper treats eviction-prevention and emergency rental assistance, examined in Part VIII's Pillar One, as a distinct and high-leverage intervention from permanent housing production itself.

6.6 The National Shortage of Housing Affordable to the Lowest-Income Renters

Underlying all five causes above is a structural supply gap the National Low Income Housing Coalition quantifies annually: the United States is short 7.2 million affordable and available rental homes for extremely low-income renters — households earning at or below the poverty line or 30% of their area's median income — meaning only 35 affordable and available rental homes exist for every 100 extremely low-income renter households nationally. Of the roughly 11 million extremely low-income renter households in the country, 74% are severely cost-burdened, spending more than half of their income on rent and utilities before anything else; this single population, representing under a quarter of all renters, accounts for 68% of all severely cost-burdened renters in the United States. This is the structural floor beneath every root cause in this Part: even a household with no mental illness, no addiction, no domestic-violence risk, and no institutional discharge to manage can be one missed paycheck away from eviction simply because the supply of homes it can afford does not exist at the scale its population requires.

Part VII — The Barriers: Why Isn't This Already Done at Scale

If Housing First works, and often pays for itself, the natural question — the one the original framework behind this paper poses directly — is why it has not been scaled nationally or globally. Four structural barriers explain most of the gap.

7.1 Land Cost Dwarfs Construction Cost

In most major cities, the land under a housing unit now costs more than the structure built on top of it — a cost a modular or factory-built cabin, however cheap to manufacture, does nothing to solve on its own, which is why Vienna's model in Part IV rests as much on land policy (decades of public land-banking) as on construction technique.

7.2 Zoning and Regulatory Restrictions

Zoning codes in most U.S. and many international jurisdictions restrict or outright prohibit tiny-home clusters, micro-unit developments, and high-density supportive housing in large portions of a city's land area, frequently through the same single-family-only zoning categories that restrict ordinary affordable housing construction generally — meaning the barrier to scaling homelessness solutions and the barrier to scaling affordable housing broadly are, in large part, the same barrier.

7.3 The CapEx/OpEx Funding Mismatch

As set out in Part V.3, capital funding for construction is more politically popular and easier to raise — a ribbon-cutting is a visible, one-time achievement — than the recurring operating funding supportive housing requires indefinitely for as long as it is occupied, which is a standing annual budget commitment with no comparable ribbon to cut. Public and philanthropic funders disproportionately favour the former, producing the "built but not sustained" pattern this paper's framework names directly as a reason cabin and tiny-home projects underperform when support funding is not secured on an equally durable, multi-year basis.

7.4 Misaligned Incentives Between Public Companies and Public Need

Publicly traded companies are structurally oriented toward shareholder value over a reporting horizon far shorter than the multi-decade, patient-capital timeline large-scale homelessness solutions require, and large-scale programmes of the kind examined in Part IV require sustained government partnership, cross-agency coordination, and public funding precisely because private capital, operating on its normal return expectations, is not well suited to funding a recurring operating cost with no revenue stream attached to it — a structural, not a moral, explanation for why the private sector has not independently solved this problem at scale, directly answering the question posed in the framework this paper is built from.

7.5 The Political Economy of Local Opposition

Even where land, zoning, and funding are not the binding constraint, proposed supportive housing developments routinely face sustained local opposition — commonly grounded in concerns about property values, neighbourhood safety, or a preference that services be sited elsewhere in the jurisdiction — that can add years of delay, additional design concessions, or outright project cancellation to a development's timeline. This is a distinct barrier from the regulatory restriction identified in 7.2: a parcel can be zoned to permit supportive housing by right and still face a project-specific political fight at the permitting or community-review stage. None of the four success cases in Part IV eliminated this dynamic; Houston's and Medicine Hat's coordinated systems instead absorbed it as a predictable cost of doing business, building sustained public communication and phased siting (distributing supportive units across many smaller sites rather than concentrating them in one large, highly visible development) into their operating models rather than treating each instance of opposition as a crisis requiring a one-off political response.

What Real-World Interventions Have Actually Shown

NGE's house standard is to ground policy proposals in interventions that have actually been tried, not only in what sounds plausible. The four cases in Part IV meet that bar on the geographic, system-level question; two further cases below add evidence on specific mechanisms — coordinated by-name data, and modular construction economics — that the solutions framework in Part VIII draws on directly.

HUD-VASH — The Clearest Domestic Proof of the Systems Argument

Introduced in Part I, the HUD-Veterans Affairs Supportive Housing programme combines permanent housing vouchers with mandatory-offer, integrated VA case management and health services, and has been associated with a 55% reduction in veteran homelessness since 2009 — the single largest sustained reduction documented anywhere in this paper's research. Its design is instructive precisely because it is not a housing programme with services attached as an afterthought, nor a services programme that waits for housing separately; voucher and case manager are assigned together, from the same coordinated intake, which is the structural feature Part VIII's Pillar Three recommends generalising beyond the veteran population specifically.

Built for Zero — By-Name, Real-Time Data as the Coordination Layer

Medicine Hat's functional-zero result in Part IV was achieved as part of Built for Zero Canada, a network — with a parallel Built for Zero movement across more than 100 U.S. communities — built on a specific technical innovation: a continuously updated, by-name list of every person currently experiencing homelessness in a community, shared across every agency in the local system, replacing the periodic, anonymised point-in-time count as the operational tool for managing the problem day to day. The point-in-time count remains useful for the kind of national trend analysis in Part II; the by-name list is what let Medicine Hat's system know, on any given day, whether it was gaining or losing ground, and act immediately rather than waiting for the next annual count.

Community First! Village, Austin — Modular and Tiny-Home Construction at Real Scale

Community First! Village, a 51-acre permanent supportive housing community outside Austin, Texas built substantially from micro-homes and recreational-vehicle-style units rather than conventional apartment construction, has housed several hundred formerly chronically homeless residents since opening in 2015, combining the lower construction cost this paper's framework identifies as cabin housing's core advantage with the same operating-cost commitment (on-site medical care, job training, and a resident community structure) that Parts V and VII argue is the difference between a housing project that lasts and one that does not. It is the clearest large-scale U.S. proof that the "build cheap cabins" instinct named in this paper's own source question works when, and only when, it is funded and operated as a complete system rather than a one-time construction grant.

Part VIII — The Solutions Framework: Eight Pillars

8.1 Pillar One: Eviction Prevention and Emergency Assistance Before the Crisis

Fund rapid, low-barrier emergency rental and utility assistance, accessible before an eviction filing rather than only after, directly targeting the inflow side of Part I's stock-and-flow framing — the single highest-leverage point in the entire system, because a household prevented from losing its home never enters the far more expensive emergency-response and re-housing system at all.

8.2 Pillar Two: Housing First as the Operating Default, Not a Pilot Programme

Following Finland, Houston, and Medicine Hat directly: house people without preconditions of sobriety or psychiatric stability, treating voluntary access to mental health and addiction services as a parallel offer rather than a prerequisite, consistent with the strongest available evidence in Part V while being explicit, per Part III's correction, that this does not mean every county-level success claim built on this model should be taken at face value without checking its methodology.

8.3 Pillar Three: Coordinated Intake With Combined Housing and Services Funding

Following HUD-VASH directly: fund housing vouchers and case-management services together, from a single coordinated intake process, rather than as separately administered programmes a person must navigate and qualify for independently — the structural design feature this paper's evidence most consistently associates with durable, rather than temporary, housing stability.

8.4 Pillar Four: Guaranteed Multi-Year Operating Funding Alongside Any Capital Grant

Directly responding to the CapEx/OpEx mismatch identified in Parts V and VII: require that any capital grant for new supportive housing construction be paired with a committed, multi-year (minimum five-year, following the horizon over which Community First! Village and comparable projects have demonstrated durability) operating-funding commitment before construction is approved, rather than funding the building and leaving its ongoing operation to be solved later.

8.5 Pillar Five: Zoning Reform for Micro-Units and Supportive Housing

Following the barrier identified in Part VII.2 directly: reform zoning codes to permit micro-unit and supportive-housing development by right in a materially larger share of a city's land area, recognising that the regulatory barrier to scaling homelessness solutions and the barrier to scaling affordable housing broadly are, in most jurisdictions, the same barrier and should be addressed together.

8.6 Pillar Six: Real-Time, By-Name Coordinated Data

Following Built for Zero directly: adopt a continuously updated, by-name list shared across every agency in a community's homelessness response system, rather than relying solely on an annual or periodic count, so the system can measure whether it is gaining or losing ground against inflow in real time and respond immediately rather than retrospectively.

8.7 Pillar Seven: Discharge Planning From Institutions

Require confirmed, funded housing placements before discharge from hospitals, correctional facilities, and the foster care system at the age of majority, directly closing the root cause identified in Part VI.4 — a structural, administrative failure that is more tractable to fix than most of this paper's other root causes, because it requires coordinating existing institutions rather than building new capacity from nothing.

8.8 Pillar Eight: Patient Public and Philanthropic Capital for Land Acquisition

Following Vienna's example at a scale most jurisdictions can realistically pursue incrementally rather than replicate wholesale: dedicate sustained public or philanthropic capital to land acquisition and land trusts specifically for supportive and affordable housing, addressing the land-cost barrier identified in Part VII.1 directly, on the multi-decade timeline Part VII.4 argues private capital is structurally unsuited to funding on its own.

Part IX — A Practical Diagnostic Methodology for Cities

Every case examined in this paper, success and cautionary alike, shares a common root: outcomes tracked the degree of system coordination, not simply the dollars spent or units built. This section sets out, step by step, the due-diligence sequence a city, county, or philanthropic funder should work through to assess its own homelessness response system, translating the eight pillars above into an ordered checklist.

9.1 Step One — Measure Inflow and Outflow Separately, Not Just the Stock

Following Part I's framing directly: a city that tracks only its total homeless count, without separately tracking how many people newly entered homelessness in the period and how many exited to stable housing and stayed there, cannot tell whether a declining count reflects genuine progress or a temporary lull in inflow — precisely the diagnostic gap Part III's Utah correction illustrates at the state level.

9.2 Step Two — Audit Whether Capital and Operating Funding Are Jointly Committed

Check whether every supportive housing project in the pipeline has committed, multi-year operating funding secured at the same time as its construction funding, per Pillar Four — a project with capital funding but no secured operating funding is, on this paper's evidence, a strong leading indicator of the "built but not sustained" failure pattern.

9.3 Step Three — Check Whether Intake Is Coordinated or Fragmented

Assess whether a person experiencing homelessness in the jurisdiction can access housing and services through a single coordinated entry point, as in Houston's Way Home and HUD-VASH, or must separately navigate multiple uncoordinated agencies and waitlists — a structural predictor, on the evidence in Parts IV and the real-evidence section, of both system efficiency and housing-retention outcomes.

9.4 Step Four — Verify Real-Time Data Capability

Confirm whether the jurisdiction maintains a continuously updated, by-name record of its currently homeless population, per Pillar Six, or relies solely on periodic point-in-time counts — the latter cannot detect whether the system is gaining or losing ground between count dates, a materially slower feedback loop than the one Medicine Hat's model demonstrates is achievable.

9.5 Step Five — Map Institutional Discharge Pathways

Review discharge protocols from local hospitals, correctional facilities, and the foster care system for whether a confirmed housing placement is required before discharge, per Pillar Seven — a jurisdiction without this requirement is, on this paper's evidence, continuously re-seeding its own homeless population from its own institutions.

9.6 Step Six — Assess Zoning Capacity for Supportive Housing

Quantify what share of the jurisdiction's land area permits micro-unit or supportive-housing development by right versus only through a discretionary, case-by-case variance process, per Pillar Five — a low by-right share is a structural ceiling on how much housing the system can ever produce regardless of available funding.

9.7 Step Seven — Evaluate Land-Acquisition Strategy

Check whether the jurisdiction has a dedicated, standing mechanism (a land trust, a land-banking programme, or equivalent) for acquiring land for affordable and supportive housing ahead of need, per Pillar Eight, rather than acquiring land opportunistically, project by project, at prevailing market prices each time.

Part X — Anticipating Pushback

10.1 "Housing People Without Requiring Sobriety or Treatment First Just Enables Addiction"

This is the most common objection to Housing First, and it deserves a direct answer rather than a dismissal. The evidence in Part V and the real-evidence section does not show that Housing First eliminates addiction or mental illness; it shows that stable housing is frequently a precondition for someone to engage successfully with treatment, not a reward to be withheld until treatment succeeds — consistent with the clinical observation, widely documented in the supportive-housing literature, that it is extremely difficult to maintain psychiatric medication schedules, attend regular treatment appointments, or sustain recovery from addiction while sleeping outside or in an unstable shelter environment. Housing First pairs housing with voluntary access to services specifically because mandatory preconditions have, in the decades before Housing First became standard practice, left large numbers of people perpetually unhoused and therefore never able to access the very treatment the precondition was meant to ensure.

10.2 "This Is Too Expensive to Scale Nationally"

Part V's cost-offset evidence is the direct answer: a substantial share of Housing First's cost is already being spent, just inefficiently, through jail, emergency shelter, and emergency-room use that a coordinated system reduces. Houston's result — a 63% reduction in homelessness at lower per-capita spending than comparable cities — is this paper's clearest evidence that the binding constraint is coordination, not primarily new money, though Pillar Four's operating-funding commitment does require real, sustained budget allocation that cannot be assumed away.

10.3 "Vienna's Model Could Never Work in the United States"

This objection has real force and this paper does not dispute it directly — Vienna's scale of public land ownership and decades of sustained investment are not replicable on a short timeline in most U.S. jurisdictions, as Part IV's honest limitation states explicitly. The response is that Vienna is included as evidence a supply-side, preventive approach can work in principle, not as a claim that it is the only or primary recommendation of this paper's framework; Pillar Eight recommends patient, incremental land acquisition specifically because it is the scalable, realistic version of Vienna's underlying logic, not an attempt to replicate Vienna's specific institutional history.

10.4 "If Homelessness Were Really This Solvable, Someone Would Have Solved It Already"

Finland, Houston, and Medicine Hat are direct evidence against this objection on the evidence this paper has assembled — it has been solved, repeatedly, at the city and national level, by jurisdictions willing to fund coordination and operating costs on a sustained basis. The honest caveat, consistent with Part VII's barriers, is that none of these results came from a single policy change; each required years of sustained political commitment across multiple budget cycles, which is a real and non-trivial barrier distinct from any claim that the underlying technical or economic problem is unsolved.

10.5 "Rapid Re-Housing Is Cheaper Than Permanent Supportive Housing, So Why Not Use It for Everyone"

Rapid Re-Housing, defined in Part II.3's glossary, is genuinely lower cost per household than Permanent Supportive Housing, and this paper's framework does not recommend PSH for every case — Pillar Two's "Housing First as the operating default" is compatible with matching the intensity of the intervention to the acuity of the household, light-touch rapid re-housing for a household whose homelessness was triggered by a single economic shock (Root Cause 6.5) and no ongoing disability, and PSH's deeper, ongoing support for the chronically homeless population with disabling conditions that Part II's data shows make up roughly one in three people counted. The practical risk this paper flags is the reverse error: substituting rapid re-housing for PSH to save money on a chronically homeless caseload produces the appearance of a placement without the ongoing support that Parts V, VI and the Denver RCT in Part IV.5 show is what actually sustains housing stability for that specific population — matching intervention to need, not defaulting to whichever intervention is cheapest per unit, is the operative principle.

Part XI — An Original Metric: The Housing System Resilience Score

Every individual metric in common use — the point-in-time count, units of housing built, dollars spent — captures one dimension of a homelessness response system in isolation. This paper proposes the Housing System Resilience Score (HSRS), a composite index intended to let city governments, funders, and researchers compare how coordinated, not merely how well-resourced, a jurisdiction's response actually is, directly addressing the measurement gap Part III's Utah correction exposed.

The weighting (40/35/25) reflects this paper's own judgement, informed directly by Parts IV and VII, that coordination is the single most predictive factor in this paper's case studies (Houston achieved its result primarily through coordination, not new spending), sustainable funding is the second most common point of failure (the CapEx/OpEx mismatch), and prevention and data capacity, while important, are somewhat more achievable incrementally once the first two are in place. Jurisdictions adopting an HSRS-style tool would set their own weights and thresholds; the framework's purpose is to make the trade-off explicit and comparable across cities, not to prescribe a single correct weighting.

11.2 What the HSRS Would Not Have Caught

In the interest of the same honesty this paper applies to Utah's statistic in Part III, the HSRS has real limitations that a user should weigh before treating a composite score as a complete answer. First, it scores system design, not outcomes directly — a jurisdiction could score well on coordination, funding, and data capacity while still carrying a large existing stock of unhoused people if those mechanisms were only recently put in place; the score is a leading indicator of future trajectory, not a current head count. Second, all three sub-indices as presented here rely on a jurisdiction's own self-reported assessment of its systems unless an external auditor verifies them, the same verification gap Part III shows can let a flattering but inaccurate statistic stand unchallenged for years. Third, the 40/35/25 weighting is this paper's own judgement call, built from the four geographic cases and the barriers analysis, not a weighting derived from a large cross-jurisdictional statistical study — a legitimate critique the policy-implications section below addresses by recommending verification rather than blind adoption.

Part XII — Policy Implications

• For city and county governments: adopt Houston's coordinated-intake model and Medicine Hat's by-name data system as the two most immediately transferable, already-implemented mechanisms, rather than starting system design from first principles.
• For federal and national housing agencies: treat HUD-VASH's combined housing-plus-services funding structure as the template for expanding coordinated funding beyond the veteran population specifically, given it is this paper's single strongest evidence of a durable, national-scale reduction.
• For philanthropic and impact investors: prioritise multi-year operating-funding commitments over one-time capital grants for new construction, per Pillar Four, given this paper's evidence that uncommitted operating funding is the most common point of failure for otherwise well-designed projects.
• For researchers and journalists covering this field: verify a jurisdiction's claimed reduction against its underlying counting methodology before repeating it, per Part III's Utah correction — a dramatic headline statistic is, on this paper's evidence, no less likely to be a measurement artifact than a genuine result, and the two are not distinguishable without checking.
• For local zoning and planning authorities: treat zoning reform for micro-units and supportive housing as a homelessness intervention in its own right, per Pillar Five, not merely an adjacent housing-policy issue, given that the regulatory barrier to both is, in most jurisdictions, the same one.

Part XIII — Conclusion

The question this paper was built to answer — if humanity can spend trillions on wars, luxury goods, and technology, why can't we end homelessness — has an answer the evidence in this paper supports directly: the limiting factor is not engineering, and it is not aggregate resources. Humanity already knows how to build affordable housing, deliver coordinated healthcare, and organise communities, and Finland, Houston, and Medicine Hat have each demonstrated, at national, metropolitan, and small-city scale respectively, that sustained reductions in homelessness are achievable with tools that already exist. What is missing, consistently across every case this paper examined, is coordination sustained across multiple budget cycles — combined housing and services funding, real-time data, discharge planning, and zoning capacity operating as one connected system rather than as dozens of separately funded, separately measured fragments.

What distinguishes this paper's argument from a simple call for more spending is the evidence, assembled across Parts IV, V and the real-evidence section, that specific, already-implemented mechanisms — HUD-VASH's combined funding model, Built for Zero's by-name data systems, Houston's coordinated intake, Vienna's patient land strategy — have already measurably worked, each at real jurisdictional scale, each with honestly documented limitations alongside their successes. The choice facing cities, funders, and national governments is therefore not between accepting homelessness as unsolvable and finding a silver-bullet fix. It is between building the coordination infrastructure that has already been shown to work, and continuing to fund disconnected fragments of a system that, as Part III's Utah correction shows, can produce statistics that look like success without producing the outcome those statistics claim to describe.

References

U.S. Department of Housing and Urban Development (2024). "The 2024 Annual Homelessness Assessment Report (AHAR) to Congress, Part 1." HUD, Washington, D.C.

Bipartisan Policy Center (2025). "Homelessness at a Record High: Key Takeaways from the 2024 PIT Count."

City Limits (2025). "Federal Point-in-Time Homeless Count Was Highest on Record in 2024."

National Low Income Housing Coalition (2024). "HUD Releases 2024 Annual Homeless Assessment Report."

National Alliance to End Homelessness (2025). "State of Homelessness: 2025 Edition."

European Commission, Migration and Home Affairs (2026). "Finland: Housing First Model Reduces Migrant Homelessness."

Good Good Good (2026). "Finland Lowered Homelessness by 75% — Here's Their Secret."

thisisFINLAND (2026). "How Finland's Housing First Model Makes Real Progress Against Homelessness."

Pathfinders, SDG16+ (2026). "Housing First Policy: Finland."

Smart Cities Dive (2023). "Houston's Housing-First Model Is Reducing Homelessness. Here's How It Works and the Obstacles It Faces."

Smart Cities Dive (2023). "How Houston's Homeless Strategy Became a Model for Other U.S. Cities."

CBS News (2023). "Inside Houston's Successful Strategy to Reduce Homelessness."

Cato Institute (2024). "In Houston, Housing Affordability Helps Reduce Homelessness."

Community Solutions (2021). "Medicine Hat Becomes First City in Canada to End Chronic Homelessness."

Canadian Alliance to End Homelessness (2021). "Medicine Hat Achieves Functional Zero Chronic Homelessness."

Built for Zero Canada (2026). "Bright Spot: Canada's First Functional Zero Chronic Homelessness Proof Point."

Corinth, K., American Enterprise Institute (2016). "On Utah's 91 Percent Decrease in Chronic Homelessness."

American Enterprise Institute (2017). "Think Utah Solved Homelessness? Think Again."

KSL News (2016). "Utah's 91% Drop in Chronic Homelessness Not Real, Economist Says."

Deseret News (2016). "Utah's Big Drop in Chronic Homelessness Is 'Fiction,' Economist Says."

Independent Institute (2025). "The Rise and Fall of 'Housing First' in Utah."

Shelterforce (2023). "How We Can Bring Vienna's Housing Model to the U.S."

IURC (2025). "Best Practice: Urban Planning — Vienna's Housing Policy."

Jacobin (2023). "We Should Look to Vienna for Answers to Our Housing Crisis."

RAND Corporation (2017). "Supportive Housing for Individuals Experiencing Homelessness in Los Angeles Saves Money During First Year."

Urban Institute. "Research Report: Costs and Offsets of Providing Supportive Housing to Break the Homelessness-Jail Cycle."

Urban Institute. "Housing First Breaks the Homelessness-Jail Cycle."

National Low Income Housing Coalition. "The Evidence Is Clear: Housing First Works."

Community First! Village, Mobile Loaves & Fishes (2026). Programme documentation and outcomes reporting, Austin, Texas.

Urban Institute (2021). "Breaking the Homelessness-Jail Cycle with Housing First: Results from the Denver Supportive Housing Social Impact Bond Initiative."

Urban Institute. "What We Learned from the Evaluation," Denver Supportive Housing Social Impact Bond Initiative project page.

City and County of Denver (2021). "Independent Evaluation Finds Denver's Supportive Housing Social Impact a 'Remarkable Success.'"

HUD User, PD&R Edge (2021). "Denver's Supportive Housing Social Impact Bond Initiative Breaks the Incarceration-Homelessness Cycle."

National Low Income Housing Coalition (2026). "The Gap: A Shortage of Affordable Homes, 2026."

LeadingAge (2026). "Nation Short 7M Affordable Homes for Lowest Income."

Housing Finance magazine (2026). "NLIHC: Only 35 Affordable Homes per 100 Extremely Low-Income Renters."

Shelter England (2025). "At Least 382,000 People Are Homeless in England Today."

Crisis (2025). "Third Successive Rise in Rough Sleeping as the Homelessness Crisis Deepens Across England."

Australian Bureau of Statistics (2022). "Estimating Homelessness: Census, 2021."

Centre for Social Impact (2022). "Homelessness in Australia: 2021 Census — Key Findings."

This white paper is offered as a contribution to the debate on ending homelessness. The authors welcome comments, critique, and correction.