Structural Root Causes: Why Homelessness Persists as a Systems Problem — and the Case for Structural Change Over Symptom Management

By Common Ladder · June 26, 2026 · 30 min read

A companion to the Common Ladder Core Framework on ending homelessness. Written for policymakers and legislative staff — congressional and state housing, health, and judiciary committee staff; gubernatorial and mayoral policy advisors; USICH and state interagency council staff; and the analysts who draft the bills, appropriations, and waivers that set the structural conditions homelessness systems operate inside.

Thesis

Homelessness in America is not primarily a service-delivery failure. It is the predictable output of structural conditions — a housing market that does not produce homes affordable to the poorest households, an income floor set below the cost of shelter anywhere, generations of racialized wealth stripping, a criminal-justice system that manufactures housing barriers, behavioral-health and safety-net systems with holes large enough to fall through, and a government budget architecture that systematically underinvests in the solutions even when the evidence is settled. Symptom management — shelter, emergency response, encampment sweeps — can hold the line on suffering, but it cannot reduce a problem produced upstream of the service system. The communities that have actually reduced homelessness did it by changing structural conditions, not by managing symptoms harder. This document maps the root causes, ranks them by leverage, and identifies the structural changes that the evidence says would work.

Executive Summary

Part I — The Frame: Symptom, Trigger, and Cause

1.1 Three Levels, Not One

Most public conversation about homelessness collapses three different things into one. Untangling them is the precondition for any serious structural analysis.

A proximate trigger is the event that precipitates an individual episode: an eviction, a job loss, a hospitalization, a relationship breakdown. Triggers are visible, individual, and easy to mistake for causes. But triggers are nearly universal — most people experience a version of them and do not become homeless, because they have a financial or relational margin that absorbs the shock.

A structural cause is the upstream condition that determines how many people have no margin when the trigger hits, and whether any path back to housing exists. A housing market with no affordable units, an income floor below the cost of rent, a wealth gap that leaves a household with no savings and no family to fall back on — these are the conditions that convert ordinary triggers into homelessness at scale.

The service system sits below both: shelter, outreach, coordinated entry, rapid rehousing, permanent supportive housing. It catches people after the trigger has fired and the structural conditions have done their work. A well-run service system matters enormously to the people it reaches. But it operates on the back end of a process whose volume is set upstream. This is the core insight policymakers need: you can run the best service system in the country and still lose ground, because the inflow is governed by conditions the service system does not control.

1.2 The Root-Cause Hierarchy

The structural causes operate at three levels:

Level 1 — Foundational market and policy conditions (the substrate): housing supply and affordability; income and wealth inequality; structural racism. These set the size of the at-risk pool.

Level 2 — System failures that increase vulnerability (the amplifiers): healthcare and behavioral-health access gaps; criminal-justice exposure and reentry barriers; safety-net erosion and benefit cliffs; the child-welfare-to-homelessness pipeline. These determine who in the at-risk pool actually falls.

Level 3 — Structural triggers that appear individual (the precipitants): eviction, domestic violence, disaster displacement. These look like personal crises but are shaped by structural exposure — who can leave an abuser depends on whether safe affordable housing exists; who survives an eviction depends on whether any margin remained.

The homelessness a community sees is the product of how many people fall through the gaps at all three levels, against a housing supply that may or may not be able to absorb whoever remains.

1.3 The 2025 Data Is the Argument

The clearest recent demonstration of the structural thesis is the January 2025 national count. For the first time since 2016, overall homelessness fell — down 3.4 percent to 745,652.^[w13] But the decline was narrow and bifurcated. Family homelessness dropped 11 percent, accounting for the entire national reversal, while homelessness among individuals rose 0.6 percent to a record 515,286 and chronic homelessness rose to a record 155,750.^[w13]

Read structurally, this is not a paradox. Families benefited from the structural supports that were actually deployed at scale during and after the pandemic — rapid rehousing, family-targeted prevention, and the tail of the $46 billion federal emergency rental assistance program.^[f13] Single adults and the chronically homeless depend on permanent supportive housing in the tightest-supply markets, and that supply did not grow to meet them. The same year, veteran homelessness fell to a record-low 32,495 — the population with a dedicated federal resource stack and bipartisan political will.^[w13][f10]

Three populations, three different structural-support levels, three different outcomes in a single count. Where the structure changed, the number fell. The conclusion is not that homelessness is intractable. It is that homelessness responds, precisely and measurably, to structural inputs — and that the populations still rising are the ones whose structural supports were never built to scale.

Part II — The Substrate (1): Housing Supply and Affordability

2.1 The Threshold Effect

The most rigorously established community-level cause of homelessness is the relationship between rent and income. The peer-reviewed Glynn-Byrne-Culhane analysis (Annals of Applied Statistics, 2021), built on 386 Continuums of Care, found a nonlinear acceleration: once median rent crosses roughly 30 percent of area median income, marginal rent increases produce disproportionate homelessness increases.^[f16] The relationship is relatively flat below the threshold and sharply rising above it. Communities above the threshold hold 15 percent of the U.S. population but 47 percent of people experiencing homelessness — a more than threefold overrepresentation.^[f16]

The mechanism is not complicated. As housing costs consume more of a household's income, the margin available to absorb a shock shrinks toward zero. When the margin is zero, any shock produces housing loss. The structural question is never whether a particular person had a crisis. It is whether the housing market left them any room to survive one.

A note on precision: the specific threshold percentages are pre-pandemic. U.S. rents rose 20 to 30 percent in many markets between 2021 and 2023, so the number of communities above the line has almost certainly grown. The 30 percent figure should be read as directionally robust, not as a current precise benchmark.^[f16]

2.2 The Supply Gap and "Deep Affordability"

The National Low Income Housing Coalition estimates a shortage of roughly 7.3 million rental homes affordable and available to extremely low-income households (those at or below 30 percent of area median income) — about 37 such units for every 100 households that need one.^[nlihc] That 63-unit-per-100 gap is the arithmetic of homelessness at scale. (The figure is PROVISIONAL: NLIHC compiles it annually from federal data with a stable methodology, but it is an advocacy-aligned aggregation rather than a single peer-reviewed estimate.)

A distinction that policymakers must hold precisely: "affordable housing" in the federal Low-Income Housing Tax Credit sense means housing affordable to households at 50 to 60 percent of area median income — not to the homeless or near-homeless population, who typically earn 10 to 30 percent. Deep affordability — housing genuinely affordable to someone on SSI or minimum wage without additional subsidy — requires a project-based voucher layered on tax credits, a direct operating subsidy, or a construction subsidy at a scale the financing system has never produced in sufficient quantity. The practical consequence is that advocacy for "more affordable housing" routinely produces units the homeless population cannot access. Legislation aimed at homelessness must specify deep affordability, which requires a different and more expensive toolkit.

2.3 The Policy Lever Cuts Both Ways

The threshold finding has a direct implication: a community can attack either side of the rent-to-income ratio. It can lower rents through supply-side policy — zoning and entitlement reform, production subsidy, density — or it can raise effective incomes through demand-side policy — vouchers, minimum wage, refundable tax credits. Both work. What the evidence rules out is the belief that a very tight market can program its way out of homelessness through services alone, without touching supply or income.^[f16] This is why the veteran reduction was accomplished primarily through a dedicated voucher program — a demand-side income support — rather than through new construction: vouchers solve the affordability problem for the household without waiting years for supply to grow, provided landlords will accept them and the voucher pool is large enough to reach the need.

Part III — The Substrate (2): The Income Floor

3.1 Homelessness Is Concentrated at the Bottom

Homelessness is concentrated not merely among the low-income but among the extremely low-income. A single person on SSI receives about $11,300 a year — 16 to 25 percent of area median income in most markets. SSDI averages roughly $18,000. Full-time work at the federal minimum wage of $7.25 yields about $15,080 — below the poverty line for one person. At these incomes, even the lowest-cost rental markets in the country are structurally unaffordable without subsidy. The housing-affordability problem and the income-poverty problem are the same problem viewed from two ends.

The arithmetic is worth stating plainly because it is the crux. Federal policy defines housing as affordable at 30 percent of income. At the SSI income level, that is $283 a month. There is essentially no unsubsidized one-bedroom in the United States at that price. Housing a person on a fixed disability income is not hard without subsidy — it is impossible. That single fact is the basis for the permanent need for housing subsidy for the disabled poor, and it is not a fact that better case management can change.

3.2 The Low-Wage Economy and Benefit Cliffs

Two structural features of the income landscape compound the floor problem. First, real wages at the bottom of the distribution have grown far more slowly than productivity or median wages for four decades, while middle-skill jobs that once paid a living wage without a degree have thinned out (PROVISIONAL — this is aggregate policy literature rather than a single canonical study). The implication for homelessness is that a meaningful share of the homeless population is not homeless because of a behavioral-health condition; they are poor in a high-cost market, and the wages available to them cannot sustain housing without help.

Second, the means-tested safety net produces benefit cliffs — points where a modest income gain triggers a larger benefit loss. The working poor can face effective marginal tax rates above 50 to 70 percent once SSI reductions, SNAP phase-outs, and subsidy losses are counted. The rational response — declining income growth that would reduce total resources — is a feature of the system's design, not a failure of individual motivation. For housing case managers, helping a client earn more requires careful benefit navigation to avoid net financial harm, which is itself a structural absurdity.

Part IV — The Substrate (3): Structural Racism

4.1 The Disparity Is Structural

Black Americans are 14 percent of the U.S. population and 33 percent of people experiencing homelessness in the 2025 count — and more than 37 percent of homeless families with children.^[w13][f9] The disparity is present across every population segment — chronic adults, families, youth, veterans — and is not explained by differences in individual risk factors. Peer-reviewed historical and sociological research traces it to structural conditions that systematically disadvantaged Black Americans in housing, wealth, and opportunity across generations.^[f9]

4.2 The Mechanisms

Four interlocking mechanisms recur in the literature.^[f9] Redlining and exclusionary zoning denied Black households the federally insured mortgages that built White middle-class wealth during the greatest home-appreciation period in U.S. history, while keeping them out of the suburbs that appreciation enriched. Segregation and concentrated poverty, legally enforced through 1968 and perpetuated afterward through lending discrimination and steering, compounded disadvantage across generations. Mass incarceration, which expanded dramatically from the 1980s and fell disproportionately on Black men and communities, created a direct pipeline to homelessness through the housing barriers a record imposes. And the cumulative intergenerational wealth gap — a persistent 5-to-8x ratio between median White and median Black family wealth — means less cushion against shocks and less family network to draw on, the informal resource base that quietly prevents many White households from ever entering the homeless system.

Ongoing mechanisms perpetuate the disparity independent of history: rental discrimination documented in HUD audit studies; criminal-history and credit screens that are formally race-neutral but fall disproportionately on Black applicants because criminal-justice exposure is racially skewed; and asset tests that penalize the savings of households already disadvantaged by the wealth gap.^[f9]

4.3 The Disparity Extends Into the System

Structural racism reaches into the homeless system itself, not only the pathways into it. Some coordinated-entry assessment tools have shown differential scoring for Black versus White clients with comparable needs; Black individuals and families wait longer for placement in some analyses. The policy implication is decisive and CANONICAL: racial equity cannot be achieved by neutrally implementing evidence-based interventions. It must be actively designed in — assessment audited, criminal-history and credit barriers removed or mitigated, outcomes disaggregated by race and reported.^[p4] Equity is a design requirement, not a separate initiative.

Part V — The Amplifiers

The substrate sets the size of the at-risk pool. A second tier of system failures determines who in that pool actually falls. Each is well-established in direction, with magnitudes that are mostly PROVISIONAL.

5.1 Criminal Justice and Reentry

The criminal-justice system manufactures homelessness through two mechanisms. The direct pipeline: roughly 650,000 to 700,000 people leave state and federal prisons annually, and millions more exit jails, frequently with gate money in the tens or low hundreds of dollars, no ID, and no housing plan. The best-evidenced figures show formerly incarcerated people are roughly ten times more likely to be homeless than the general public, with more than 10 percent experiencing homelessness in the period surrounding release (PROVISIONAL; Prison Policy Initiative 2018, Urban Institute).^[srrc1] The collateral consequences then create lasting barriers: federal rules permit and sometimes require public-housing providers to screen out applicants with certain records; private landlords run background checks that systematically exclude people with convictions, most acutely in tight markets; and license suspensions and ID barriers block housing applications, employment, and benefits.

The jail-homelessness cycle is the most wasteful expression of this. Homeless people are cited or jailed for status-related offenses — camping, public intoxication — disrupting whatever service connections they had, then released to the same street, then cycled again. Jail beds cost $80 to $200 a day, producing no housing outcome. Because mass incarceration fell disproportionately on Black communities, the housing barriers a record creates are inseparable from the racial-equity root cause; they are the same disparate outcome produced through a different institution.^[f9]

5.2 Healthcare and Behavioral Health

The U.S. mental-health system's history is braided into homelessness. Deinstitutionalization emptied state psychiatric hospitals from the 1950s onward without building the community infrastructure to replace them, leaving a population with serious mental illness in communities with insufficient support. The conditions that produced persist: chronically underfunded community mental-health centers with long waits and geographic gaps; medication discontinuity driven by unstable housing, which produces the crisis-housing-loss-crisis cycle behind much chronicity; and a Medicaid coverage gap in the states that have not expanded, where adults without dependent children are too poor for marketplace subsidies and ineligible for Medicaid — cutting off treatment for a population with elevated mental-illness prevalence.

Substance-use disorder intersects similarly (PROVISIONAL). Effective medications for opioid use disorder exist and reduce mortality, but access is constrained by prescriber shortages, clinic geography, prior authorization, and stigma. Abstinence-contingent treatment models exclude the active users that Housing First serves, producing a gap in the care continuum precisely where the chronically homeless population sits. The dual-diagnosis population — co-occurring mental illness and substance use — falls through both systems, and is heavily concentrated in chronic homelessness.

5.3 Safety-Net Erosion

The cash and in-kind safety net for very low-income households has thinned in ways that reduce the margin against shocks. SSI's maximum benefit has eroded in real terms even as housing costs rose, widening the gap between what the net provides and what housing costs. TANF — the primary cash program for families with children — has decayed since 1996: in 1996 it reached 68 of every 100 families in poverty; by 2020, 21 (PROVISIONAL; CBPP). One CBPP analysis associates every 1,000 fewer TANF families with roughly 149 additional homeless children, controlling for poverty trends.^[srrc3] SNAP provides modest support but is subject to cyclical work-requirement debates; the March 2023 end of pandemic emergency allotments produced the largest single benefit drop in the program's history at a moment of elevated food prices.

A connected gap runs the other way: a significant share of the chronically homeless population qualifies for SSI/SSDI but has never obtained it, because the application and appeals process is daunting without stable housing and consistent support. Communities with dedicated SOAR (SSI/SSDI Outreach, Access, and Recovery) workers achieve far higher approval rates — roughly 65 percent versus a 10-to-15 percent baseline — which is both a service practice and a housing-stability practice, since SSI plus a subsidy is enough to sustain a tenancy that SSI alone cannot.

5.4 The Child-Welfare Pipeline

The foster-care-to-homelessness pipeline is among the most clearly documented structural mechanisms, and among the most clearly a design problem. Roughly 22 to 30 percent of youth who age out of foster care experience homelessness during the transition (PROVISIONAL; Chapin Hall Midwest Study and multiple replications, confirmed in Chapin Hall's 2024 update).^[srrc4] About 20,000 youth age out each year — reaching the eligibility limit without adoption or family reunification. The mechanism is straightforward: foster care provides housing as a component of placement, and when the placement ends, at an age when most have not developed independent-living capacity, there is no housing plan, no income, and often no family to return to. Support does not taper; it stops. Extended foster care to age 21 reduces but does not remove the cliff — it buys time without changing the underlying design. LGBTQ+ youth, overrepresented in foster care through family-rejection pathways, carry elevated risk through this transition.

5.5 The Triggers That Look Individual

Domestic violence is a leading cause of homelessness for women and families (PROVISIONAL; prevalence estimates range 20 to 50 percent across program studies). Its structural character is the economic and housing trap that makes leaving difficult: abusers controlling finances and credit; full DV shelters with waitlists; mainstream shelter that is unsafe for survivors; and the parallel-systems problem in which DV and mainstream homelessness operate with different philosophies, privacy protocols, and physical spaces, producing poor handoffs and families cycling between them. Immigration status creates a separate structural exclusion (PROVISIONAL; poorly quantified by design): undocumented and many mixed-status households are ineligible for federally funded housing and most safety-net programs, and fear of enforcement deters access to what is available.

Part VI — The Meta-Cause: Budget Architecture and Political Economy

The preceding causes drive homelessness. This one prevents the solutions from working even when they are known — which is why it is the most important section for a legislative audience.

6.1 The Cross-Budget Problem

Permanent supportive housing produces documented cost offsets across Medicaid, corrections, mental health, and shelter budgets. The foundational analysis — a 2002 study of nearly 5,000 New York City placements — found about $12,146 per placement per year in service-cost reductions, covering most of the annual housing cost.^[f17] The structural problem is in the distribution of those savings: they accrued to the state mental-health agency, Medicaid inpatient, the shelter system, city hospitals, the VA, and corrections — not to the housing budget that paid for the housing.^[f17] No automatic mechanism recycles the savings back. Each budget manager has a rational incentive to free-ride on housing investment made by another agency.

This is why twenty-five years of strengthening cost-offset evidence has not produced proportionate investment. The evidence problem is solved. The budget-structure problem is not. A community that cannot solve the cross-budget misalignment will perpetually underfund supportive housing relative to demonstrable need, regardless of how strong its evidence or how well-run its programs. The barrier is not ignorance. It is the architecture of government finance — and architecture is something legislators can change.

6.2 The Partial Fixes Already in Motion

The most promising structural response routes the costs and the savings to a single payer who bears both. Medicaid is that payer for the health-related share. As of early 2024, eight states — Arizona, Arkansas, California, Massachusetts, New Jersey, New York, Oregon, and Washington — had CMS-approved Section 1115 demonstrations covering health-related social needs: housing navigation, tenancy-sustaining services, and one-time move-in costs, with four states adding short-term rent assistance.^[w12] This is a genuine structural innovation — it lets Medicaid, which captures much of the cost offset, fund the housing services that generate it.

The honest caveats matter for legislation. HRSN spending is capped at 3 percent of a state's Medicaid spending and limited to clinically eligible populations; it complements rather than replaces the CoC and ESG funding stacks. And the federal posture reversed in 2025: the administration rescinded the guidance framework, ended the financing mechanism several states used, and froze the presumptive approval pathway for new states.^[w12] The structural tool exists and works where it was adopted, but its expansion is now a contested political question rather than an administrative default — which makes it a state-legislative and congressional matter, not a regulatory one.

6.3 The Political-Salience Gap

Layered over the budget architecture is a political-economy asymmetry. The populations most affected — extremely low-income adults with behavioral-health conditions, Black and Hispanic families, people with criminal records — have the least electoral power and the least ability to organize sustained pressure. The predictable result is a spending bias: reactive emergency response over proactive prevention; visible shelter and enforcement over slow, invisible housing and supply; and prevention programs cut first in downturns precisely because cutting them produces no visible crisis.

The veteran reduction is the exception that proves the rule. Veterans have a politically powerful, bipartisan constituency and a clear claim on public resources — and veteran homelessness fell to a record-low 32,495 in 2025 through a dedicated federal resource stack.^[w13][f10] The lesson is not that veterans were uniquely deserving. It is that sustained structural investment produces sustained structural results, and that for populations without a comparable constituency, the durable path is to design financing mechanisms that create automatic investment — Medicaid integration, cross-agency budgeting — rather than requiring political mobilization in every budget cycle.

6.4 The Enforcement Detour

There is a structural response that is politically attractive and demonstrably counterproductive: criminalization. Encampment sweeps and enforcement-only approaches show no peer-reviewed evidence of sustained community-level homelessness reduction, and substantial evidence of harm — destroyed medications, IDs, and documents essential to housing applications; severed service relationships; worsened physical and mental health over time; and degraded count accuracy as people disperse to less visible locations.^[pit6] Sweeps move the problem; they do not reduce it. For a legislative audience, this is the cleanest case of spending against the evidence: enforcement budgets purchase displacement and harm at a cost that, redirected to prevention or housing, would purchase reductions.

Part VII — What Follows: Leverage and Sequencing

Root-cause analysis is only useful if it ranks interventions by where structural investment produces the greatest downstream reduction. The hierarchy below distinguishes scale of effect from speed of effect — because high leverage is not the same as first priority.

The sequencing logic for a legislator: fund the fast, high-return lever now while building the slow, high-leverage one in parallel. Prevention is the lever that is both structural and fast. Emergency financial assistance directed to genuinely imminent-risk households cut shelter entry by 76 percent in a Science natural experiment and 81 percent in a Review of Economics and Statistics randomized trial, with a marginal value of public funds of 2.47 — each public dollar returning $2.47 in social benefit, among the highest returns documented in the field.^[f23] The catch, equally well-established, is that the large effects depend on accurate targeting: empirically derived risk models substantially outperform worker judgment at identifying who will actually enter shelter, and the same grants spread across less-imminent households would show smaller effects.^[f22] Effective prevention is accurate targeting times effective intervention; legislation should fund both the assistance and the data infrastructure that aims it.

The slower levers are not optional — supply expansion is the highest long-run leverage and the only durable fix for tight markets — but they cannot be the only track. A community in crisis today gets no relief from a building that opens in 2032. The two-track structure (long-horizon supply and income, short-horizon prevention and acquisition) is the only configuration that serves both the people already in crisis and the inflow producing more of them.

Part VIII — The Asks

This analysis is written for the people who set structural conditions. The asks are directed accordingly.

For federal policymakers and congressional staff. Fund deep-affordability production and project-based subsidy at a scale matched to the 7.3-million-unit gap, not the 50-to-60-percent-AMI band that current tax-credit production mostly reaches.^[nlihc] Fund targeted homelessness prevention at structural scale — the evidence puts it among the highest-return public investments available, and it currently operates at marginal scale relative to need.^[f23] Close the Medicaid coverage gap by completing expansion, and restore a stable national pathway for the Section 1115 HRSN financing that lets Medicaid fund the housing services it benefits from.^[w12] And build cross-budget cost-accounting into housing and health appropriations, so that the savings a housing investment generates in Medicaid, corrections, and hospital budgets can be measured and routed back toward the investment that produced them.^[f17] Protect the veteran resource stack that produced the field's strongest structural result, and study what made it work for replication.^[f10]

For state policymakers and legislative staff. Use the cross-agency financing tools that are already working: the eight HRSN-demonstration states have built Medicaid-funded housing-services infrastructure that other states can pursue, and renewals due in 2026–2027 are now political decisions worth defending.^[w12] Reform the criminal-history screening rules within state authority that bar people from subsidized housing for records that are racially skewed and often unrelated to tenancy.^[f9][srrc1] Address benefit cliffs in state-administered programs so that income growth does not produce net financial harm. Implement and fully fund extended foster care, and require housing-transition planning before youth age out.^[srrc4] Fund SOAR-style benefits-enrollment capacity, which converts a population's existing entitlement eligibility into housing stability at high return.

For local policymakers and elected officials. Stop spending on enforcement the evidence says does not work. Encampment sweeps and criminalization purchase displacement and health harm, not reduction; the same dollars redirected to prevention and housing purchase results.^[pit6] Require a cross-budget cost analysis of the local homeless population before the next housing vote, so that the cost of housing is compared to the cost of not housing rather than to zero.^[f17] And disaggregate family from individual trends before claiming progress or failure — a flat topline can hide a family-side win offset by an individual-side loss, as the 2025 national data shows.^[w13]

The knowledge of what works has existed for two decades. The levers are identified and ranked. What remains is structural change, and structural change is not a discovery. It is a decision that policymakers make.

Part IX — What This Analysis Is Not Saying

Precision about the argument's limits protects its credibility with a sophisticated reader.

This is not an argument that individual factors do not matter. Serious mental illness, substance-use disorder, and trauma are real and shape individual trajectories; this document brackets them deliberately because its question is why the United States produces homelessness at this scale, which is a structural question, not because individual factors are irrelevant to any given person's path. The companion Common Ladder analysis of population segments treats individual-level risk in full.

This is not an argument that the service system does not matter. A well-run system is the difference between brief, non-recurring homelessness and chronic homelessness for the people it reaches, and operational excellence — matching intervention to acuity, maximizing exits, minimizing returns — is essential. The point is narrower: operational excellence cannot, by itself, overcome structural headwinds that guarantee a continuous inflow. A well-run system losing ground to structural conditions is a different problem from a poorly run system, and the two require different responses — policy support for the first, management improvement for the second.

This is not an argument that nothing works or that the problem is hopeless. The opposite: the 2025 family-side reduction, the veteran result, the prevention evidence, and the cost-offset literature all show that homelessness responds to structural inputs. The argument is that the response requires changing structures, not managing symptoms harder — and that the structures are within the reach of the policymakers reading this.

And this is not an argument that structural change is fast or cheap. The highest-leverage levers are slow and expensive. But the relevant comparison is never structural investment versus nothing. It is structural investment versus the recurring, distributed, larger cost of leaving the conditions in place — the cost the public is already paying, in budgets where no one is accountable for the total.

Appendix A — Evidence Index

All findings cite the Common Ladder homelessness knowledge base by ID. CANONICAL findings are externally citable; PROVISIONAL findings are flagged in-body and usable with the caveat noted.

Medicaid expansion and homelessness: coverage gains in expansion states are reliable (+22 percentage points); the direct effect on homelessness rates is contested and is not cited as a structural claim in this document's body.

Appendix B — Definitions

Defined on first use in the text; collected here for the reader who has not spent a career inside the field.

AMI (Area Median Income)

The midpoint household income for a metropolitan area, used to set housing-program eligibility. "Extremely low income" (ELI) is at or below 30% of AMI.

CoC (Continuum of Care)

The HUD-defined regional planning body that coordinates homelessness funding and services for a geography.

Deep affordability

Housing affordable to households at 10–30% of AMI (e.g., people on SSI), as distinct from "affordable housing" pegged to 50–60% of AMI.

HRSN (Health-Related Social Needs)

The category under which Medicaid Section 1115 demonstrations can cover housing navigation, tenancy-sustaining services, and limited move-in or short-term rent costs.

MVPF (Marginal Value of Public Funds)

The ratio of social benefit to net government cost for a policy; an MVPF above 1.0 means the policy returns more than it costs.

PSH (Permanent Supportive Housing)

Long-term subsidized housing paired with voluntary support services, the primary intervention for chronically homeless adults with disabilities.

PIT (Point-in-Time count)

HUD's annual single-night census of sheltered and unsheltered homelessness.

SSI / SSDI

Supplemental Security Income (means-tested) and Social Security Disability Insurance (work-history-based) — the two federal disability income programs.

SOAR

SSI/SSDI Outreach, Access, and Recovery — a model that raises disability-benefit approval rates for homeless applicants.

Read next: Intervention Matching — Which Interventions Work for Whom →