Evidence-Based Allocation in Homelessness Philanthropy and Federal Investment

By Common Ladder · May 26, 2026 · 35 min read

A companion to the Common Ladder Core Framework on ending homelessness. Written for foundation program officers, foundation trustees, federal agency leadership (HUD, USICH, HHS, VA), and the state-level grantmakers whose RFPs follow philanthropic norms.

Thesis

The country has the evidence on what reduces homelessness. It does not have the funder discipline to spend on the evidence at the scale and proportion the evidence supports. Better allocation of existing philanthropic and federal dollars — without new appropriations — could move community-level homelessness outcomes meaningfully. The leverage point is not the size of the pool. It is the allocation rules that govern how the pool is spent.

This guide is the case. It is written for the people who set those rules: foundation program officers and trustees, federal agency leadership, and the state-level grantmakers whose RFPs follow philanthropic norms.

Executive Summary

Part I — The Funder Architecture, Briefly

A working homelessness response in the United States requires at least five distinct funder categories operating in coordination. No single category covers the full stack. Understanding what each category can and cannot fund is the precondition for asking any single category to do more of what it should be doing.

The federal layer is the foundation but is structurally constrained. HUD's Continuum of Care program — approximately $3.5 billion annually — funds permanent supportive housing, transitional housing, rapid rehousing, supportive services, HMIS, and coordinated entry; it is competitive, renewal-heavy, and gated by System Performance Measures.^{[funder_landscape]} Emergency Solutions Grants — about $300 million in base funding, expanded substantially during ARP — pass through to states for shelter, outreach, rapid rehousing, and a tightly-scoped prevention authority. HOME funds capital but not services; CDBG funds capital and a limited band of public services capped at 15 percent of grant. Federal VA programs — HUD-VASH, SSVF, GPD — together approach $1 billion annually and demonstrate what dedicated, integrated, accountable federal funding produces.^[f10] Federal HHS programs — TANF (block-granted at $16.5 billion, state-discretionary), PATH ($65 million for SMI outreach), the now largely wound-down Emergency Rental Assistance — touch the system at the edges.

The state and local layer is the most flexible. General fund dollars can fund what federal programs cannot — emergency shelter at full cost, outreach without the federal program constraints, prevention with earlier eligibility windows than ESG's 14-day rule, landlord incentive programs, system infrastructure. Several major metro areas — Seattle/King County, Los Angeles, Denver — have passed dedicated local levies generating hundreds of millions annually. State housing trust funds are discretionary grant pools, varying widely in scale and flexibility by state. The state and local layer is also the most volatile — first cut in a budget downturn — which limits its reliability as the funder of last resort.

The philanthropic layer is small in absolute scale (no comprehensive national total exists; major national funders combined likely contribute well under one percent of total system spending in a typical year) but disproportionate in strategic importance. Philanthropy funds what federal programs cannot fund: innovation, equity initiatives, policy advocacy, evaluation, capacity-building, and risk-tolerant pilots that produce the evidence federal programs later operationalize. Restricted grants outperform on accountability; unrestricted general operating support outperforms on workforce retention and adaptive capacity. National funders (Arnold Ventures, MacArthur, Robert Wood Johnson, Gates, Conrad N. Hilton, JPB, JPMorgan Chase, MacKenzie Scott's vehicles) and community foundations fund different parts of the stack and play different roles.

The healthcare layer is the fastest-growing structurally important new lever in homelessness funding. Medicaid — through state Section 1115 waivers (California's CalAIM is the most-developed example), state plan amendments, managed care organization programs, and CMS's expanded Health-Related Social Needs (HRSN) guidance — now funds housing navigation, transition support, and time-limited rental assistance in several states.^[f15] The federal authority exists and is broadening. State adoption is uneven. Healthcare and Medicaid are uniquely positioned to address the cross-agency funding problem because Medicaid bears the cost of the high-utilizer hospitalizations that PSH most reliably reduces — when Medicaid funds the services and Medicaid receives the savings, the cross-agency misalignment is partially resolved at the payer level.

The private capital layer — LIHTC equity, tax-exempt bonds, CRA-driven CDFI lending, impact investing — finances housing development but not operations or services. LIHTC alone generates approximately $10 billion annually in equity investment, financing about 100,000 affordable units per year nationally. Critical for housing production; structurally inadequate at the deep-affordability tier where homelessness risk concentrates.^{[funder_landscape]}

The funding stack matters because it tells the funder where to act. No single source can fund the complete response a community needs. The leverage points are different for each funder type. The questions this guide takes up — where is the evidence, where is the money, and how do we close the gap between them — are answered differently for a federal agency, a national foundation, a state housing finance agency, and a community foundation.

Part II — Why the Evidence Has Not Translated Into Investment

Twenty-plus years of stronger evidence on what reduces homelessness have not produced proportional investment. The reason is not that funders do not know the evidence. The reason is that the structure of public budgeting routes the costs and the savings into different accounts.

The foundational cost-offset evidence comes from Culhane, Metraux, and Hadley's 2002 quasi-experimental analysis of New York City's NY/NY supportive housing program — 4,679 placements tracked across eight linked administrative databases (shelter, state mental health, city hospitals, Medicaid inpatient and outpatient, VA, state corrections, city corrections).^[f4] The regression-adjusted reductions per placement per year, in 1999 dollars:

• Shelter (DHS): −$2,819
• State mental health (OMH): −$6,162
• City hospitals (HHC): −$1,321
• Medicaid inpatient: −$2,825
• Medicaid outpatient: +$1,982 (an increase — appropriate care-seeking after housing stability, 47.2 additional visits per placement per year)
• VA: −$444
• State corrections: −$312
• City corrections: −$245
• Net: −$12,146 per placement per year

Against annual housing costs of approximately $17,277 per unit, adjusting for turnover and length of stay, the net offset reached approximately 94 percent of housing costs within two years — meaning net public cost of a PSH placement, after service-cost reductions, was approximately $995 per unit per year. The Community Guide's 2022 systematic economic review of twenty PSH studies found median program cost of $16,479 per person per year against median economic benefit of $18,247 — net positive, consistent direction.^[f4][f5]

The cost-offset evidence is canonical. The problem is its budgetary geography.

The $12,146 in savings is distributed across at least seven distinct budgets:

The housing cost — $17,277 per unit per year — is borne by the housing budget. The housing budget is not the same budget as state mental health. It is not the same budget as Medicaid. It is not the same budget as city shelter. It is not even the same level of government as several of the savings line items.

There is no automatic mechanism that routes the savings back into housing investment. The state mental health authority does not write a check to the city housing department because someone moved out of an OMH-funded clinical placement into a PSH unit. Medicaid does not capitate a portion of its inpatient savings to the city housing budget because someone reduced their ED utilization after being housed. The corrections department does not transfer corrections savings into a housing trust fund because chronic recidivism declined.

Each budget manager has an incentive to free-ride on housing investments made by another agency. The housing budget sees the cost; some other budget sees the savings; the housing budget gets pressed to fund more of what produces other agencies' savings while its own appropriation does not grow.

This is the structural paradox that two decades of stronger evidence has not unlocked. The evidence problem is not a knowledge problem. It is a budget-structure problem. More evidence will not move funding. What will move funding is one of three structural fixes:

1. Cross-system cost accounting — building the data infrastructure to show specific decision-makers that "housing budget X produces Medicaid savings Y and corrections savings Z" in their specific community. Once the data is in hand, cross-agency appropriations conversations become possible; before the data is in hand, they remain abstract.
2. Single-payer routing through Medicaid — using state Section 1115 waivers (California's CalAIM is the most-developed example) to route Medicaid dollars into housing services. When Medicaid funds the services and Medicaid receives the savings, the cross-agency misalignment is partially resolved at the payer level. Medicaid cannot fund rent itself under current federal statute, but the surrounding services can be Medicaid-financed.^[f15]
3. Healthcare-system partnership — hospitals and managed care organizations in capitated arrangements have a direct financial incentive to fund housing for high-utilizer patients whose hospitalizations they bear the cost of. The MCO-CoC and hospital-CoC partnerships that route Medicaid or hospital savings into housing are growing but uneven.

For philanthropy, the cross-agency problem creates a clear strategic role: funding the structural fixes that the federal and state systems will not fund on their own. Cross-system data infrastructure is the highest-leverage example. It is operational work — governance, data-sharing agreements, analytical capacity — that no current federal program reliably funds, and that produces the evidentiary basis for funding asks that move budgets the philanthropic dollar itself cannot move. The philanthropic dollar that funds cross-system data infrastructure unlocks the public dollars that would not otherwise flow. The leverage ratio is large; the visibility is low; the work is unglamorous; the proposal frequently loses to programs with more legible outputs. This is exactly the kind of investment philanthropy is uniquely positioned to make and frequently does not.

Part III — Where the Evidence Is and the Money Is Not

A funder seeking to allocate the next dollar to the intervention with the strongest evidence-to-investment ratio has, in approximate order of leverage, the following options. Each pairs a canonical or strongly-promising finding with a structural underfunding pattern.

3.1 Targeted Emergency Financial Assistance

The strongest underfunded finding in the homelessness evidence base. Two converging studies — Evans, Sullivan, and Wallskog's 2016 Science paper using a quasi-experimental discontinuity at the Chicago Homelessness Prevention Call Center, and Phillips and Sullivan's 2025 Review of Economics and Statistics RCT in Santa Clara County — find that targeted emergency financial assistance reduces shelter entry by 73 to 81 percent. The Phillips and Sullivan 2025 trial reports a Marginal Value of Public Funds of 2.47 — each dollar of public spending returns $2.47 in social benefit.^[f22][f23]

The convergence is significant. Two independent studies, different methodologies, different populations (single adults in Chicago; families in Santa Clara), produce effect sizes of the same order of magnitude. This kind of replication is unusual in homelessness research and elevates the policy confidence the findings should command. Combined with the prevention-targeting evidence from Shinn and Greer 2013 and the LA County Prevention Targeting Tool, the integrated finding is: statistical risk-screening tools substantially outperform unaided worker judgment in identifying imminent-risk households (top-decile risk concentration of approximately five times the general low-income rate), and targeted financial assistance to those households reduces shelter entry by 70 to 80 percent.^[f22][f23] Targeting plus intervention is the formula.

The federal investment pattern does not match the evidence. ESG's prevention authority is constrained by a 14-day-imminence eligibility rule that excludes many households for whom prevention would work. The Emergency Rental Assistance program of the pandemic era — $46 billion across 12.3 million payments, the largest federal prevention investment in U.S. history — was a crisis appropriation that has largely wound down without a steady-state successor.^[f13] No comparable ongoing federal prevention program exists at scale. State and local prevention funding is patchy, often pegged to crisis appropriations, and rarely paired with statistical targeting infrastructure.

This is exactly the kind of intervention that philanthropy can move on. Funding a community to build a statistical risk-screening tool — adapting the LA County PTT methodology to local administrative data — is in the seven-figure range over multiple years. Funding the targeted financial assistance program that the tool feeds — at the scale Evans 2016 and Phillips 2025 indicate would be cost-effective — is in the seven- to eight-figure range for a mid-sized metro. The combined ROI, at MVPF of 2.47, is among the highest in the documented U.S. homelessness evidence base. The number of major foundations operating prevention-targeting programs at this scale is small. The leverage is large.

3.2 Permanent Supportive Housing Pipeline Development for Chronically Homeless Adults

The PSH evidence is canonical. The Lancet Public Health 2020 systematic review reported rate ratios of 1.13 for moderate-need and 1.42 for high-need participants compared with usual care on long-term housing stability.^[f6] The Santa Clara County PSH RCT — n=423, four-year follow-up — found 86 percent of intervention participants were ever housed during the trial, compared with 36 percent of controls.^[f2][f6] The cost-offset evidence is canonical and discussed in Part II. The NASEM 2018 PSH evidence review reached the same directional conclusion.^[f6]

As of 2024, the United States has approximately 412,623 PSH beds, of which roughly 170,000 rely on Continuum of Care program funding. The chronically homeless population in 2024 was approximately 152,585 — approximately 1 in 5 of all people experiencing homelessness.^[f1][f6] In aggregate, existing inventory approximately matches need, but is geographically maldistributed and underdeveloped in many high-need markets.

The funding challenge for PSH is structural. The capital side — LIHTC, HOME, state housing trust funds, local bonds — is competitive and time-consuming (three to five years from predevelopment through occupancy for new construction), but is a known process with established financing tools. The operating side — CoC rental assistance, project-based vouchers — is competitive but available. The services side is where developments fail in practice. A PSH building constructed and leased without adequate services is not delivering PSH — it is delivering subsidized housing with inadequate support, which produces worse outcomes than the model evidence predicts. The capital-services split is the operational manifestation of the cross-agency funding problem.^[f17]

The high-leverage philanthropic move on PSH is not funding the construction itself — LIHTC at $10 billion annually moves orders of magnitude more capital than any foundation could deploy. The high-leverage move is funding the services and the system-level capabilities that LIHTC and HOME cannot fund: the predevelopment costs that put deals together, the operating reserves that bridge services funding gaps, the cross-system data infrastructure that enables cross-agency funding asks, the workforce capacity that retains case managers and clinicians, the Medicaid billing infrastructure that converts services into sustainable revenue. Each of these unlocks public capital that the philanthropic dollar itself does not provide.

3.3 Housing Choice Voucher Access Strategy and Expansion

The HUD Family Options Study — multi-site randomized controlled trial, 2,282 families across 12 communities — is the strongest RCT evidence in family homelessness. Families randomized to a housing voucher experienced substantially better housing stability than those randomized to rapid rehousing, transitional housing, or usual care (emergency shelter). Voucher effects extended into children's behavioral and educational outcomes at both 20-month and 37-month follow-ups.^[f8]

The Family Options Study was published in 2016. It was updated in 2022 and 2024. Family homelessness rose 39 percent in 2024 — driven in substantial part by the simple arithmetic that the Housing Choice Voucher program reaches only about one in four eligible households.^[f1][f8] The voucher was the evidence-based answer. The voucher was not available. The family entered shelter. This pattern is repeated millions of times across the system every year.

Federal voucher expansion is the highest-leverage federal investment available for family homelessness — and the most politically constrained. Universal voucher provision for income-eligible households, phased over a decade, would close the binding constraint that produces the family homelessness increase. The Congressional Budget Office and major housing research organizations have estimated the federal cost in the range of $100 billion to $410 billion over ten years depending on phase-in pace and depth of subsidy — not a small lift, but proportional to a problem that produces 770,000-plus people experiencing homelessness on any given night.

Philanthropy cannot fund voucher expansion at federal scale. But philanthropy can fund three things that the federal advocacy environment requires: the evidence infrastructure that makes the case for expansion (the Family Options Study itself was federally commissioned but the analytic and dissemination infrastructure was substantially built by philanthropy and university partnerships); the advocacy infrastructure that moves the federal political environment toward expansion (organizations like NLIHC, NAEH, and the Center on Budget and Policy Priorities); and the local voucher-access strategy capacity that makes existing vouchers usable. The last point is operationally consequential. Voucher lease-up failure rates in tight markets run 30 to 50 percent of allocated vouchers — meaning the federal investment in the voucher itself is wasted absent local landlord engagement, search assistance, and time extensions. Philanthropy that funds local voucher-access strategy converts already-appropriated federal dollars into housing outcomes. The leverage ratio is large; the work is unglamorous.

3.4 Workforce Capacity and Retention

The service-quality multiplier across the entire homelessness response is the workforce. The National Alliance to End Homelessness 2023 survey (n=5,044 across 1,154 agencies — the first comprehensive workforce survey of the field in nearly thirty years) reported that 71 percent of agencies experienced turnover challenges, 74 percent reported being understaffed, and workers of color disproportionately reported wage inadequacy.^[workforce] Annual turnover of 30 to 50 percent is common in many homelessness agencies. Relationship rupture is itself a driver of housing instability — clients lose continuity with the staff who know their story, and the next staff member starts the relationship-building from zero.

The evidence on caseload ratios at the program level is clear: Assertive Community Treatment teams require approximately 10:1 ratios; Intensive Case Management ratios should not exceed 20:1; standard case management typically operates at 35:1. Caseloads that exceed these benchmarks degrade service quality in predictable ways. Licensed clinical roles — psychiatry, therapy — are the scarcest and most under-compensated relative to peer sectors.

The workforce problem has no dedicated federal funding stream. Workforce costs are embedded in program grants at inadequate levels; CoC competitive grant scoring does not directly reward workforce investment; ESG and HUD-VASH do not include workforce-specific authorities. AmeriCorps placements are subsidized but time-limited and credentially-restricted. TANF is broadly eligible for workforce development but unevenly used for homelessness-services workforce specifically. The Substance Abuse and Mental Health Services Administration's SOAR program demonstrates what dedicated training infrastructure can produce — 65 percent SSI/SSDI approval rates versus 10 to 15 percent baseline — but SOAR is the exception, not the rule.^[workforce]

This is exactly where philanthropy's distinctive role applies. General operating support that retains experienced staff is the single highest-leverage philanthropic investment available for workforce. Restricted grants that pay for specific program staff at competitive wages — particularly for licensed clinical roles in PSH and Critical Time Intervention teams — convert the program grant from "we'll hire when we can find someone" to "we can pay above-market and retain them." Philanthropy that funds workforce capacity as a primary outcome — not as a residual of program funding — produces durable service-delivery quality that program-only funding cannot.

3.5 Cross-System Cost Accounting Infrastructure

The single most consequential operational capacity a community can build for advocacy purposes is the ability to track public service costs across systems for a defined client population, before and after housing placement. As discussed in Part II, this is the data infrastructure prerequisite for any cross-agency funding conversation. Without it, "PSH saves money in healthcare and corrections" is a national finding from a 1999-dollars NYC study. With it, the same statement becomes "in this community, the chronically homeless cohort generated $X per person per year in public service costs before housing; after housing, that figure dropped to $Y; here are the specific budgets that captured the savings."

The work required to build this is governance and data-sharing agreements, plus the analytical capacity to aggregate. Shelter data exists in HMIS. ED and hospital data exists in healthcare partner records, sometimes accessible through Medicaid claims if the CoC has data-sharing agreements with the state. Jail and corrections data exists in county records. Mental health utilization exists in state behavioral health data. The technical lift is modest; the relational and governance lift is substantial. The reward is the funding conversation that follows.

Federal funding for this infrastructure is minimal. CoC planning dollars are capped at 3 percent of the grant and largely cover compliance reporting. No HUD authority funds cross-system data infrastructure. Medicaid's Quality Improvement and integration authorities (MACRA, state innovation grants) can fund pieces of it in some states. State-level innovation funds can fund pieces in others. In aggregate, the infrastructure is built — where it is built at all — by philanthropy.

This is among the highest-leverage philanthropic investments in the homelessness field. It is also one of the least funded. The proposal does not have legible immediate outputs. The deliverable is a dashboard, not a service. The constituency for the work is technical, not emotional. The funders that have invested — Arnold Ventures has been notable in this space, as have several community foundations partnered with university research centers — have produced disproportionate downstream effects. The funders who have not invested have left on the table the dollar-for-dollar leverage ratio that comes when the philanthropic dollar funds the data that unlocks the public dollars.

3.6 Medicaid HRSN Capacity Building

State Section 1115 Medicaid waivers and CMS's expanded Health-Related Social Needs guidance now allow Medicaid to fund housing navigation, transition support, move-in costs, and time-limited rental assistance in approved states.^[f15] California (CalAIM), Oregon, Washington, Arizona, and the District of Columbia have moved most aggressively. Most other states have not. The federal authority is broadening; state adoption is the binding constraint.

The barrier is administrative capacity. Medicaid billing for housing-adjacent services requires documentation systems, billing infrastructure, compliance infrastructure, and clinical credentialing that most homelessness agencies do not have. The states moving aggressively are also having to build provider capacity to participate in the new authority. The administrative complexity is real and frequently underestimated in policy advocacy that focuses on the regulatory authority without naming the operational ramp.

Philanthropy is well-positioned to fund the capacity-building layer that enables state-level HRSN adoption: technical assistance for state Medicaid agencies pursuing waivers; provider capacity-building grants for the homelessness agencies that will bill the new authority; cross-state learning networks that accelerate diffusion; evaluation infrastructure that documents what works. The Center for Health Care Strategies, the National Academy for State Health Policy, and several university research centers operate in this space; the funding base is thin relative to the magnitude of the opportunity.

3.7 Evaluation Infrastructure

The Notre Dame / J-PAL diversion RCT — started in 2024, results expected 2026–2027 — is the kind of evaluation philanthropy is uniquely positioned to fund.^[f18] Federal evaluation capacity (HUD PD&R, HHS ASPE) operates under chronic resource constraints. State capacity is uneven. Most program evaluations are descriptive and conducted by the program operators themselves. Rigorous causal evaluation — RCT and quasi-experimental — is concentrated in a small number of academic centers (J-PAL, Wilson Sheehan Lab at Notre Dame, California Policy Lab, Chapin Hall, NYU Furman Center, the University of Pennsylvania's housing initiative under Culhane, the University of Michigan's Poverty Solutions, the University of Chicago's Harris School).

The aggregate field would benefit from more such investments. There should be one funded RCT per major intervention category — diversion (Notre Dame is underway), targeted prevention at scale (Phillips and Sullivan 2025 is a starting point that should be replicated in different market conditions), landlord engagement (Tsai and Solis 2024 documented 46 programs with essentially zero outcome evaluation evidence), LGBTQ+ youth-specific interventions (Tubertini 2025 systematic review found persistent evidence gap), Medicaid HRSN housing services (CalAIM evaluation pending), CoC governance interventions (Kim and Sullivan 2023 identifies the correlations but no causal study tests whether changing governance composition causes SPM improvement).^{[f18][f20][f19][services][f21]}

Per-RCT costs run in the seven- to low-eight-figure range over multi-year periods. The aggregate need across the categories listed is in the high-eight-figure to low-nine-figure range — meaningful but not large relative to total philanthropic homelessness allocation. The strategic return is disproportionate: the next decade of policy will be made on the evidence that exists at the time. Philanthropy that funds the evidence shapes the policy.

3.8 Policy Advocacy

Federal grant programs generally prohibit use of funds for lobbying or advocacy. General fund dollars cannot be used for advocacy. The advocacy infrastructure that moves federal voucher expansion, Medicaid HRSN diffusion, fair-chance housing legislation at the state level, right-to-counsel funding, foster-care-aging-out reform, and the long list of structural policy changes named in the Core Framework document — that infrastructure is funded almost exclusively by philanthropy.

The leverage ratio on advocacy is unusual. A modest philanthropic investment in advocacy capacity can shape billions of dollars in federal appropriations. Right-to-counsel in eviction proceedings is a paradigmatic example: New York City's program reported a 41 percent reduction in evictions and 86 percent of represented tenants remaining housed; San Francisco's reported 92 percent of represented tenants avoiding homelessness; Baltimore's cost-benefit analysis found a $5.7 million investment produced $36 million in avoided costs.^[pl3] The state and local legislative action that produced these programs was led by coalition advocacy organizations whose operating budgets are substantially philanthropic.

The under-recognized philanthropic move in advocacy is funding 501(c)(4) social welfare affiliates of established 501(c)(3) organizations, which can do the broader advocacy work the (c)(3) cannot. This is more legally complex than program funding and frequently avoided by foundations whose legal counsel default to risk-aversion. The risk is generally manageable; the leverage is large.

Part IV — Where the Money Is and the Evidence Is Not (Or Is Thin)

Aggregate funder allocation in homelessness does not consistently match the evidence pattern. Some current investment goes to interventions whose evidence is weak, thin, or contested. Naming the pattern is uncomfortable and necessary. The point is not that any single funder is making bad decisions — funders operate with good intentions, on the evidence available at the time decisions are made, under constraints that often preclude rigorous comparative analysis. The point is that the aggregate pattern can change as shared norms change, and shared norms change as the pattern is named.

Three patterns recur across the field.

4.1 Programs with Weak Outcome Evidence Receive Substantial Allocation

Some intervention categories are widely funded with evidence that does not support the funding level. Landlord engagement programs are the clearest current example. Tsai and Solis's 2024 HUD Cityscape national scan identified 46 housing programs across the U.S. operating landlord engagement activities — financial rewards (cash bonuses), financial assistance (risk mitigation funds, security deposit guarantees), and development support (training, legal guidance).^[f20] The same scan found that "few empirical studies have assessed the effectiveness of any landlord engagement initiatives." Decades of widespread deployment have produced essentially zero controlled outcome evidence. Risk mitigation funds appear to function primarily as psychological reassurance to risk-averse landlords — claim rates are very low — rather than as financial coverage for actual losses. The mechanism may matter; that's an empirical question; the field operates on a theoretical rationale without having tested it.

This is not a case for defunding landlord engagement. Lease-up failure in tight markets is a real bottleneck, and landlord engagement programs are responding to a real problem. It is a case for pairing funding with evaluation. Funders investing in landlord engagement should require outcome documentation, support comparative studies across program models, and resist the framing that "everyone does it, so it must work." The aggregate allocation pattern would be better if it were evidence-informed. Tsai and Solis 2024 is the starting point; rigorous follow-up evaluation is what philanthropy should fund.

Other examples in this category include some workforce and employment programs marketed as preventing homelessness (the homelessness-prevention effect is theoretically appealing but empirically weak in most program-design configurations); some wraparound case-management approaches whose service intensity is not matched to client acuity (well-intentioned, but without targeting, the effect on housing stability is small); and some transitional-housing programs whose evidence base has been substantially superseded by Housing First and Rapid Rehousing approaches but whose institutional inertia keeps the funding in place.

4.2 Programs Sized to Visible Crisis, Not to Highest-Leverage Intervention

Funder allocation in homelessness — both public and philanthropic — frequently follows visibility rather than leverage. Encampment response funding spikes when encampments are politically visible. Shelter expansion funding flows during cold-weather media cycles. Crisis funding follows whatever the most photogenic version of homelessness is in a given community.

The result is that visible crisis response is overfunded relative to upstream prevention. Prevention is invisible. A household that did not become homeless does not appear in the Point-in-Time count, does not appear in encampments, does not appear in media coverage. The systemic measurement gap — federal AHAR reporting captures shelter and unsheltered counts but does not count prevented homelessness — reinforces the visibility bias. Crisis-side spending is reportable; prevention-side spending is harder to claim credit for.

This is the structural reason that targeted prevention with statistical screening — the highest-MVPF intervention in the documented evidence base — receives substantially less philanthropic attention than encampment-response programming whose evidence base on outcomes is much weaker (encampment clearings without simultaneous housing offers produce displacement, not resolution; RAND's LA LEADS longitudinal data documents this directly).^[f14]

The corrective is funder discipline: allocate to leverage, not to visibility. The funder who funds the unphotogenic prevention program reaches more households at lower cost per household than the funder who funds the highly-visible encampment intervention.

4.3 The Capital-Services Split Produces Buildings Without the Model

PSH is housing plus services. The Housing First evidence base evaluated housing combined with Assertive Community Treatment or Intensive Case Management, depending on client need. Housing alone — without the services — is not what was tested for the chronic and high-acuity population. Yet the funding architecture frequently produces buildings with inadequate services because capital is funded by one stream (LIHTC, HOME, state housing trust funds, local bonds) and services are funded by another (CoC services component, Medicaid, state mental health, philanthropy) on different timelines with different decision-makers.^[f17]

The result is a recurring funding-stack failure: capital secures; building opens; services budget does not come together at the scale the evidence-based model requires; PSH delivers degraded outcomes; the funder community draws the wrong conclusion ("PSH doesn't work as well as we hoped") rather than the right one ("we built the housing without funding the services and the model is not being implemented at fidelity").

Funder discipline on PSH means refusing to fund capital allocations that do not have committed services funding at the evidence-based level. Some state housing finance agencies have begun building this into Qualified Allocation Plan scoring. Most have not. Philanthropic investment in services-side reserves and operating runway can bridge the gap; philanthropic advocacy for QAP reform can prevent the gap from existing.

Part V — What Funder Discipline Looks Like, In Practice

The aggregate allocation pattern this guide is critiquing is not produced by any individual funder making a mistake. It is produced by shared norms about what evidence is required, what outcomes count, what proposals get funded, and what gets renewed. The leverage on aggregate allocation is the explicit standards adopted by major program officers and trustees. Standards diffuse. A foundation that publishes its evidence criteria establishes a benchmark that other funders read; the benchmark migrates into RFPs at other foundations; the migration changes what grantees write and what gets funded.

Six commitments operationalize the discipline this guide is asking for.

5.1 Adopt Evidence Standards in Grant Criteria

The simplest and highest-leverage commitment. Specify in RFPs what evidence base programs must reference. Examples:

• For PSH and supportive housing grants: programs must reference Housing First fidelity and an evidence-based services continuum (CTI for transition moments; ACT or ICM for high-need residents; standard case management for stable residents). Programs operating sobriety-contingent or treatment-first models should be ineligible.
• For prevention grants: programs must reference a targeting methodology — statistical risk-screening preferred, with the LA County PTT and Shinn-Greer 2013 NYC model as benchmarks. Programs that distribute prevention dollars without targeting should require explicit justification and outcome reporting that addresses the targeting question.
• For workforce and capacity-building grants: programs must reference sector-median compensation benchmarks, evidence-based caseload ratios, and retention outcomes (not just hiring).
• For diversion grants: programs must reference the operational definition of housing-focused problem-solving (not shelter screen-out), trained staff capacity, and a flexible assistance fund.
• For Medicaid HRSN capacity-building grants: programs must reference state regulatory context, provider billing infrastructure, and clinical credentialing.

The standards do not need to be punitive. They need to be explicit. Once explicit, programs adapt to the standards; the field's average proposal quality rises; the next funder to publish standards finds a field already partly aligned with them.

5.2 Require Outcome Reporting Commensurate With Funding Scale

A six-figure grant cannot require RCT-level outcome reporting; the evaluation cost would exceed the grant. A seven-figure multi-year grant should require pre-specified outcome metrics, baseline measurement, follow-up measurement at defined intervals, and disaggregation by race and population segment. An eight-figure multi-funder initiative should require independent evaluation.

The norm in the field is significantly weaker than this. Many seven-figure homelessness grants are evaluated by the grantee using narrative reporting and aggregate outputs (people served, people placed) without follow-up, without disaggregation, and without a counterfactual. The result is that the funder cannot tell at the end of the grant whether the program produced an effect — only that activities occurred. Outcome reporting commensurate with scale is the discipline that converts grant cycles into evidence accumulation.

5.3 Pair Evaluation Funding With Program Funding

The Notre Dame / J-PAL diversion RCT is a model: philanthropy funded the evaluation in parallel with funding the underlying diversion program operations.^[f18] The pattern should be more common. Major foundation initiatives in PSH, prevention, voucher access, and workforce should include a parallel evaluation grant — typically 10 to 20 percent of the program grant cost, depending on evaluation design — to an independent academic or university-based evaluator.

The objection that "evaluation is expensive" is true and beside the point. The cost of not evaluating is higher. A multi-year initiative that ends without rigorous outcome data leaves the field without the evidence needed to scale what worked or revise what didn't. The marginal cost of evaluation is small relative to the lifetime cost of a poorly-targeted intervention category.

5.4 Fund Workforce Capacity, Not Just Program Delivery

General operating support is the most leveraged philanthropic instrument available. Within program-specific grants, the workforce line should be funded at sector-median or above, not at the minimum the program budget can sustain. Pay-above-median should be a grant condition when the funder has the leverage to require it. Funders that require evidence-based caseload ratios but do not fund the workforce capacity required to maintain those ratios are funding a program model the grantee cannot actually deliver.

A subset of foundations have shifted toward general operating support over the past decade. The shift has been documented in the broader nonprofit literature as producing better retention, better adaptive capacity, and better long-term outcomes. The homelessness field is downstream of the same dynamic and benefits from the same shift.

5.5 Fund Policy Advocacy

The federal voucher expansion, Medicaid HRSN diffusion, fair-chance housing legislation, right-to-counsel funding, foster-care-aging-out reform, and the long list of structural policy changes named in the Core Framework — each requires advocacy infrastructure that program grants cannot build. 501(c)(3) foundations can fund public education, research, and broad advocacy that is not direct lobbying; 501(c)(3) private foundations can also fund 501(c)(4) affiliates of 501(c)(3) public charities for broader advocacy. The legal architecture exists. The under-utilization is largely conservative legal advice and risk-aversion at the foundation board level, not regulatory constraint.

The high-leverage philanthropic move in advocacy is the multi-year, general-operating-style grant to a state-level coalition advocacy organization. Coalitions that can sustain professional advocacy staff across legislative cycles produce policy outcomes; coalitions funded only project-by-project do not. The pattern is well-established in other policy domains (criminal justice reform, climate, healthcare expansion) and is less developed in homelessness.

5.6 Fund Operational Infrastructure

The unphotogenic, structurally determinative work: by-name lists, HMIS data quality, coordinated entry redesign, cross-system cost accounting, racial equity audit infrastructure, Medicaid billing infrastructure. None of it has obvious immediate outputs. All of it is the operational layer that converts programs into system outcomes. Program-only funding produces excellent programs in degraded systems.

The challenge is that infrastructure grants compete in the same RFP cycles as program grants and frequently lose on legibility. A program grant produces "people served" as an output. An infrastructure grant produces "a coordinated entry redesign and pilot" or "a cross-system data dashboard and case-conferencing protocol" as outputs. The latter is harder to describe in a one-page summary to a trustee board. The discipline required is to recognize the trade-off explicitly and to fund the unphotogenic work anyway. Foundations that have built distinctive operational-infrastructure portfolios in homelessness — Arnold Ventures and several community foundations — produce outsized effects on community-level outcomes because the infrastructure compounds across programs.

Part VI — High-Leverage Federal Agency Allocation

The federal allocation question is partly distinct from the philanthropic allocation question because the federal government can move appropriations at scale and is constrained by political feasibility rather than philanthropic norms. But the underlying discipline is the same: allocate to the evidence at the proportion the evidence supports, with the operational infrastructure that converts programs into outcomes.

The federal allocation priorities, in approximate order of leverage:

6.1 Expand the Housing Choice Voucher Program

The voucher program is rationed by appropriation. The rationing is the policy choice that produces the 39 percent family homelessness increase.^[f8] Universal voucher provision for income-eligible households, phased over a decade, is the single highest-leverage federal investment available for homelessness. Interim expansion at the scale of the homeless population is achievable within current budget windows; the political question is whether it is prioritized.

The voucher expansion case is sometimes framed as a HUD-only ask. It should be framed more broadly. Voucher expansion is a Medicaid case (housed families have lower healthcare utilization), a corrections case (housing stability reduces recidivism), a child welfare case (housing instability is a major driver of foster care entries), and a workforce participation case (housing instability is a major driver of work disruption). The framing matters for the cross-agency coalition that voucher expansion requires.

6.2 Hold Housing First as the Federal Evidence Standard

The post-2024 federal political environment is creating pressure to reframe federal homelessness funding around treatment-first or sobriety-contingent models. The evidence does not support that pivot. HUD's CoC competitive grant scoring should continue to reward Housing First fidelity. The CoC NOFO, HUD technical guidance, and the broader federal posture should hold Housing First as the standard.^[f2][f3]

The structural protection here is that the evidence base for Housing First is among the strongest in the social policy literature. Treatment-first and sobriety-contingent models have been tested and produce worse housing outcomes. A federal policy pivot away from Housing First would be a pivot away from the evidence — and the consequences for community-level outcomes are predictable. The philanthropic role in this debate is funding the evidence dissemination, the practitioner training infrastructure, and the policy advocacy that holds the line during a federal environment unfavorable to it.

6.3 Fund Deep Affordability at Scale

LIHTC expansion alone does not address the deep affordability gap. The current LIHTC architecture primarily produces units affordable at 50 to 60 percent of area median income — not the 10 to 30 percent range where homelessness risk concentrates. The 7.3-million-unit ELI rental shortage is the supply-side structural condition that produces homelessness flows the system layer cannot absorb.

The national policy levers are National Housing Trust Fund expansion and capitalization, layered with operating subsidy (project-based vouchers or equivalent) to reach 30 percent AMI; LIHTC modification to deepen targeting; and federal preservation funding for naturally occurring affordable housing being lost to conversion. None is small. All are structurally necessary.

6.4 Stand Up Medicaid HRSN Authorities in All Fifty States

CMS has expanded the regulatory pathway through HRSN guidance and Section 1115 waiver authority. State adoption is uneven. A federal technical assistance program to support all fifty states in building Medicaid HRSN infrastructure — modeled on the diffusion of Medicaid managed care over the past two decades — would accelerate the cross-agency funding solution.^[f15] HHS and CMS have the regulatory and technical-assistance authority to do this. The constraint is operational capacity at the federal agency level and at state Medicaid agencies — itself fundable.

6.5 Release the 2025 AHAR and Commit to Annual Release

The single most consequential current data governance failure is the non-release of the 2025 Annual Homelessness Assessment Report. Local Point-in-Time counts were conducted in January 2025; the national synthesis has not been published. The field is operating critical funding and policy decisions without national outcome data. Release of the 2025 AHAR — and a commitment to annual release going forward — is a basic prerequisite for evidence-based federal policy. The absence is corrosive to every downstream decision.

This is not a partisan technicality. Federal data integrity is the infrastructure on which every downstream policy choice depends. Withholding the data is a policy choice with predictable effects: the system layer the data supports — SPMs, CoC competitive scoring, congressional appropriations — loses ground.

6.6 Restore and Adequately Fund HUD PD&R and HHS ASPE

The federal evidence functions — HUD's Office of Policy Development and Research, HHS's Office of the Assistant Secretary for Planning and Evaluation — produce essential evidence under chronic resource constraints. Funding them adequately is itself a policy lever, because without them, state and local actors operate without national context. The bipartisan case for federal research investment in homelessness is straightforward: it makes every other federal investment more effective.

6.7 Modernize HUD Screening Guidance

The 2016 HUD guidance on criminal background screening is the most recent meaningful update. Stronger guidance is needed to constrain background-check practices that produce racially disparate outcomes without commensurate safety benefit, particularly for non-violent offenses years past. The federal authority exists; what is required is regulatory action.

Part VII — The Asks

This guide is making concrete asks of specific audiences. The asks are not the conclusion. They are the operational implication of the argument the body of the document makes.

7.1 To Foundation Program Officers

Adopt evidence standards in your homelessness grantmaking. Specify in your RFPs what evidence base programs must reference. Require outcome reporting commensurate with funding scale. Disaggregate outcomes by race and population segment as a routine governance practice, not a separate initiative.

Fund the operational infrastructure that converts programs into system outcomes — by-name lists, HMIS data quality, coordinated entry redesign, cross-system cost accounting. The work is unphotogenic. The leverage is large.

Pair evaluation funding with program funding. There should be one rigorously funded evaluation per major intervention category. The Notre Dame / J-PAL diversion RCT is the model. Replicate it across the categories named in Part III.

Fund workforce capacity as a primary outcome, not as a residual. General operating support that retains experienced staff is the most leveraged philanthropic investment available. Pay-above-median should be a grant condition where you have the leverage to require it.

Fund policy advocacy. What federal grants restrict, philanthropy can support. Multi-year general operating support to state-level coalition advocacy organizations is the pattern that produces durable policy outcomes.

7.2 To Foundation Trustees

Hold your program staff to evidence standards. Ask in grant approvals: what is the evidence base for this intervention? What is the outcome reporting requirement? What is the evaluation plan? Is the workforce funded at sector median or above?

Adopt the discipline as policy, not as preference. Foundation evidence standards diffuse across the field; your foundation's published criteria become benchmarks that other funders read. The leverage on aggregate allocation runs through shared norms.

Resist the legibility bias. Operational infrastructure, evaluation, workforce capacity, and policy advocacy are harder to describe in a one-page board summary than program delivery. They are frequently the highest-leverage philanthropic investments available. The trustee role is to recognize the trade-off and fund the unphotogenic work anyway.

7.3 To Federal Agency Leadership

Expand the Housing Choice Voucher program toward universal provision for income-eligible households. Voucher expansion is the highest-leverage federal investment available for family homelessness.

Hold Housing First as the federal evidence standard. The CoC NOFO scoring, HUD technical guidance, and federal posture should not be reframed around treatment-first or sobriety-contingent models.

Release the 2025 AHAR. Commit to annual release. No evidence-based federal policy is possible without national outcome data.

Fund deep affordability at the scale of the supply gap. National Housing Trust Fund expansion and capitalization, layered with operating subsidy to reach 30 percent AMI.

Stand up Medicaid HRSN authorities in all fifty states. CMS technical assistance program for state implementation, with capacity-building grants for state Medicaid agencies and homelessness-services providers.

Fund HUD PD&R and HHS ASPE adequately. The federal evidence base is infrastructure. Underfunding it produces predictable downstream costs.

Modernize HUD screening guidance to constrain background-check practices that produce racially disparate outcomes without commensurate safety benefit.

7.4 To State Housing Finance Agencies and Medicaid Agencies

Reform Qualified Allocation Plans to reward PSH set-asides, deep affordability, and Housing First fidelity. The largest affordable housing production tool in the country is in state hands. QAP reform is the operational lever that translates federal LIHTC into the units the homelessness population can access.

Pursue Section 1115 Medicaid waivers for housing services. California, Oregon, Washington, Arizona, and DC are the model. Replicate.

Require services-funding commitments before approving capital allocations for PSH. The capital-services split is the operational manifestation of the cross-agency funding problem; QAP scoring is one of the few mechanisms that can prevent it.

Fund the homelessness-services provider capacity required to participate in Medicaid HRSN authority. Billing infrastructure, documentation systems, compliance infrastructure, clinical credentialing. The administrative complexity is real and frequently underestimated.

7.5 To Community Foundations and Place-Based Funders

Fund the operational infrastructure your CoC needs and cannot get from federal sources: by-name list implementation, HMIS data quality investment, coordinated entry redesign, cross-system cost accounting, Medicaid billing infrastructure for local providers.

Fund the workforce capacity your local agencies require to retain experienced staff. General operating support for the agencies in your CoC produces the relational continuity that case-management evidence depends on.

Fund the local advocacy infrastructure that moves state and local policy. State coalitions, local right-to-counsel campaigns, fair-chance housing advocacy. The leverage ratio on community foundation advocacy investment is among the highest in the philanthropic portfolio.

Pair evaluation with the programs you fund. Even at the community-foundation scale, modest evaluation grants paired with multi-year program funding produce evidence accumulation the field would not otherwise have.

Appendix A — Definitions

1115 Waiver

A Section 1115 demonstration waiver of the Social Security Act authorizing CMS to permit state-specific Medicaid program features outside standard authorities. Used by several states to fund housing-related services through Medicaid.

AHAR

Annual Homelessness Assessment Report — HUD's primary national reporting on homelessness, based on Point-in-Time counts and CoC data submissions.

CalAIM

California Advancing and Innovating Medi-Cal — California's Section 1115 waiver authority and broader Medi-Cal transformation, including Enhanced Care Management and Community Supports authorities relevant to homelessness.

CDBG

Community Development Block Grant — HUD formula grant to states and entitlement communities, eligible for a range of community development uses including some homelessness activities; 15 percent cap on public services.

CDFI

Community Development Financial Institution — mission-driven lender that provides capital to underserved communities, often using CRA-driven bank investment.

CoC

Continuum of Care — geographically defined regional planning body designated by HUD to coordinate homelessness services and receive federal CoC funding.

CRA

Community Reinvestment Act — federal law requiring banks to invest in low- and moderate-income communities; primary driver of CDFI investment and LIHTC equity flows.

CTI

Critical Time Intervention — time-limited intensive case management for people transitioning from institutional settings (jail, hospital, shelter) into community housing; evidence base includes multiple RCTs.

ELI

Extremely Low Income — household income at or below 30 percent of area median income, the income band where homelessness risk concentrates.

ERA

Emergency Rental Assistance — federal pandemic-era program providing direct rental assistance to households at risk of housing loss; $46 billion appropriation, 12.3 million payments.

ESG

Emergency Solutions Grants — HUD formula grant to states and entitlement jurisdictions for emergency shelter, outreach, rapid rehousing, and homelessness prevention.

Functional Zero

Community state in which homelessness is rare, brief, and non-recurring; operationally, monthly homelessness inflow at or below monthly housing placement capacity for a defined population.

HCV

Housing Choice Voucher — federal rental subsidy program (Section 8) administered by public housing authorities; reaches roughly one in four eligible households nationally.

HMIS

Homeless Management Information System — federally mandated data system for tracking homelessness service utilization at the CoC level.

HOME

HOME Investment Partnerships Program — HUD formula grant for affordable housing development, acquisition, and rehabilitation.

HRSN

Health-Related Social Needs — CMS-defined category of Medicaid-eligible services addressing housing instability, food insecurity, transportation, and other social determinants of health.

LIHTC

Low Income Housing Tax Credit — federal tax credit for affordable housing development; the largest affordable housing production tool in the United States, generating approximately $10 billion in annual equity investment.

MVPF

Marginal Value of Public Funds — economic ratio measuring social benefit per dollar of public spending; Phillips and Sullivan 2025 reports MVPF of 2.47 for targeted prevention.

PSH

Permanent Supportive Housing — permanent housing with affordability subsidy combined with onsite or wraparound supportive services for chronically homeless or disabled populations.

PTT

Prevention Targeting Tool — operational statistical risk-screening tool developed by California Policy Lab and University of Chicago researchers for LA County, identifying top-decile-risk households at approximately 5x general low-income homelessness rate.

QAP

Qualified Allocation Plan — state housing finance agency document governing LIHTC scoring and allocation; the operational lever through which states can prioritize PSH, deep affordability, and Housing First fidelity in LIHTC awards.

SOAR

SSI/SSDI Outreach, Access, and Recovery — SAMHSA-funded technical assistance program training providers to support SSI/SSDI applications for homeless adults; produces 65 percent approval rates against 10 to 15 percent baseline.

SPM

System Performance Measure — HUD-defined community-level metric on CoC system performance (length of time homeless, returns to homelessness, exits to permanent housing, and others).

USICH

U.S. Interagency Council on Homelessness — federal coordinating body for the federal homelessness response across nineteen agencies.

Evidence Index