Why "Federal Homelessness Spending Has Failed" Gets the Numbers Wrong

By Common Ladder · May 25, 2026 · 8 min read

In spring 2025, a senior program officer at a mid-sized foundation told a CoC director something she hadn't expected: "Our board is asking why we're still funding this system when Congress is saying it doesn't work." The CoC director had heard the framing before — it had been building in appropriations hearings for two years. What she didn't have was a clean, precise answer she could give in ten minutes to someone who wasn't steeped in the field.

That framing now has a name. Call it the "spending up, outcomes down" syllogism: federal investment in homelessness programs has grown substantially over the past decade; homelessness has not decreased; therefore the spending has failed. It surfaced in Congressional testimony earlier this spring, and it will appear in virtually every major funding conversation between now and June 1 — and well beyond.

The argument is wrong in three ways simultaneously. Getting the rebuttal right matters not just for advocacy, but for how the field thinks about what it's actually measuring — and what it should be doing instead.

The metric problem

The "outcomes down" half of the syllogism rests almost entirely on Point-in-Time counts: single-night estimates conducted each January in which CoC staff and volunteers attempt to count every person experiencing homelessness in their geography. PIT counts are the most widely cited homelessness statistic in the country. They are also a genuinely poor measure of whether housing programs work.

PIT counts happen in a single overnight window. Weather conditions — temperature, precipitation, presence of warming centers — dramatically affect who is found and counted. Methodology varies by CoC and shifts year over year. HUD itself describes PIT counts as "intended to be an approximation" and is explicit that they were not designed to evaluate program effectiveness.¹ Using PIT count trends to assess whether homelessness interventions are working is roughly equivalent to measuring school quality by counting how many students are visible outside on a given Tuesday afternoon.

The more fundamental problem: even if PIT counts were methodologically robust, what they measure is the net balance of inflows and outflows across an entire geography — not the outcomes of specific programs. A CoC can be genuinely excellent at moving people from homelessness into stable housing while its PIT count rises, because the count reflects factors the CoC has no authority to control.

The housing market problem

The 2020–2024 period produced the largest sustained increase in housing costs in the United States since World War II. Median rents nationally rose more than 26% between 2020 and 2023 — in many major metros, significantly more.² Eviction moratoriums, which had temporarily suppressed entries into homelessness during the pandemic, expired across the country between 2021 and 2022, releasing a backlog of housing instability that had been accumulating for two years. Opioid-related mortality and behavioral health crises surged. The number of older Americans aging into economic precarity grew substantially — a population with sharply higher homelessness risk.

Every one of these factors drove homelessness upward with no relationship to anything a CoC does or doesn't do. A CoC with disciplined coordinated assessment, strong rapid rehousing performance, and well-targeted prevention can still see its PIT count rise if market conditions push 5,000 new households into crisis — households whose affordability crisis was created by rental markets and policy failures the CoC has no jurisdiction over.

This is not a defense of every CoC's performance. Some CoCs perform better than others, and the structural research on why is worth taking seriously.⁵ It is, however, a basic observation about what is and isn't within the system's span of control — one that the "spending up, outcomes down" framing deliberately ignores.

What the evidence actually shows

Here is where the syllogism does its most serious analytical damage: by treating all "homelessness spending" as a single undifferentiated category, it obscures the fact that the evidence base for specific interventions is strong — and that most current spending still doesn't reach them.

Research on prevention targeting — directing limited prevention resources to households statistically most likely to enter shelter within 90 days, rather than offering universal access — consistently shows dramatic results. Work by Shinn and Greer (2013), Evans and colleagues (2016), and more recent research by Phillips and Sullivan (2025) finds that targeted prevention reduces shelter entry by 76–81% compared to universal or intake-first approaches.³ The Marginal Value of Public Funds for well-targeted prevention is approximately 2.47 — meaning every dollar invested generates more than two dollars in measurable social value (the 2025 figure is PROVISIONAL and should be described as such until wider peer review is complete).⁴

Structural research on high-performing CoCs offers complementary evidence. Systems that achieve significantly lower per-capita homelessness rates — controlling for housing market conditions — share identifiable features: disciplined coordinated assessment, rapid rehousing capacity at scale, and sustained landlord engagement infrastructure. Critically, these features are distinct from higher absolute spending levels (drawing on Kim & Sullivan 2023; Jenisa & Jang 2025; Nisar et al. 2019).⁵ A system spending $50M on poorly targeted, shelter-first approaches does not outperform a system spending $20M on evidence-based interventions.

The evidence doesn't show that homelessness spending fails. It shows that most current spending doesn't reach the interventions the evidence supports. Prevention targeting, which arguably has the strongest cost-effectiveness profile in the field, is funded at a fraction of the scale the research justifies. Rapid rehousing, which consistently outperforms transitional housing on both exit rates and cost per outcome, remains underfunded relative to legacy shelter capacity in many CoCs. The problem is not the principle — it's the allocation.

What this means before June

The June 1 NOFO application cycle opens in an environment where the "spending up, outcomes down" framing is live and will be used to justify both funding reductions and structural changes to how HUD evaluates program performance. For CoC leaders, the response cannot be defensive. Defending the current system in aggregate — arguing that more money will produce better results without specifying what changes and why — feeds the framing it's trying to counter.

The productive response is precision: distinguishing PIT counts from actual program outcomes; naming the housing market and pandemic-era forces that drove the 2020–2024 numbers; and pointing clearly to the specific interventions that do work, what they cost to scale, and what the return on that investment looks like in evidence-grounded terms.

That is a harder conversation than "the system needs more funding." It requires CoC leaders to be honest about where their own systems are and aren't aligned with the evidence. It requires foundation program officers to ask more specific questions in their grant-making. And it requires anyone using PIT count trends as a proxy for program quality — on any side of this argument — to stop.

The evidence is there. The work is getting it into the right rooms before the conversation is already closed.