Appeals Court Pauses Order That Blocked HUD's $1.3 Billion Shift in Homelessness Grants

A Pause, Not a Verdict

On Wednesday, September 16, 2026, a three-judge panel of the U.S. Court of Appeals for the First Circuit put a lower court's order on hold^[1][2]. That order had blocked the Department of Housing and Urban Development from changing how it hands out Continuum of Care grants, the main federal money for homeless housing and services. The pause lets HUD run its $4.04 billion grant competition for fiscal 2026 while the appeal continues^[1].

Inside that pot sits about $1.3 billion HUD wants to steer toward transitional housing and service-only projects, instead of the permanent apartments that have absorbed most of the money for years^[1]. That is roughly a third of the funding moving toward a different kind of program. The panel has not ruled on whether HUD's move was lawful — only that the agency can keep competing for now while judges sort that out^[2].

Two things are true at once here, and they explain why both sides can claim momentum. HUD has now won a procedural stay letting it act. The states have won every ruling so far on the actual merits of how HUD went about it^[5][1]. Neither fact cancels the other out.

What "Housing First" Actually Means, and Why It's the Whole Fight

The dispute turns on a term that shows up in nearly every version of this story: Housing First. Under this approach, a person gets a permanent apartment with no requirement to be sober, in treatment, or enrolled in any program first. Support services are offered, but not mandatory^[1].

HUD Secretary Scott Turner wants to flip that. His plan pushes money toward shorter-term beds tied to treatment and steps toward self-sufficiency instead^[1][4]. Turner has called Housing First a failed "experiment" that "warehoused the vulnerable" and "enabled addiction"^[4].

The mechanism matters because both sides read the same evidence differently. Randomized studies show Housing First keeps people in the apartment they were given, and a HUD-commissioned review of 26 studies found it cut homelessness by 88%^[13]. Critics, including AEI economist Kevin Corinth, argue that keeping someone housed once housed is a different question from cutting total homelessness. Corinth estimates it takes about 10 permanent supportive housing units to reduce the homeless count by just one person, because new units also draw in people who would have found other arrangements anyway^[11].

Critics also point to a re-analysis of three "best-in-class" Housing First sites. It found only 43% of people stayed housed a full year, with 41% cycling in and out and 16% leaving or dying within that year^[10][11]. That is well short of the roughly 80% retention rate often cited for the model. Twenty-two mostly Democratic-led states, the District of Columbia, and a coalition of homelessness and housing groups sued to block Turner's shift^[1].

The Cheapest Place to Win Isn't the Same as Winning the Argument

Here's the tension worth sitting with: no court has ruled that Housing First is legally required. The states haven't even asked one to. Instead, they've fought this on process, arguing HUD skipped legally required steps, because that fight is winnable and fast^[5][1].

U.S. District Judge Mary McElroy in Providence has ruled for the states twice on exactly that ground. On June 29, 2026, she set aside HUD's November and December 2025 funding notices, finding they violated the Administrative Procedure Act^[5]. That law requires agencies to publish major policy changes, take public comment, and explain why they rejected objections before acting.

In August, McElroy vacated the fiscal 2026 notice too, ruling HUD skipped that public-comment process entirely^[1]. Her sharpest line, quoted across left-leaning coverage, called HUD's process "the hallmark of unreasoned decision making"^[16]. But she also declined to permanently bar HUD from ever making a change like this, calling that broader request speculative^[5].

That distinction is why a procedural win isn't necessarily a durable one. HUD can, in principle, redo the notice-and-comment process correctly and try again. The states have bought time and slowed the rollout, not settled the underlying policy question^[1].

Why a Calendar Date Beat a Judge's Ruling

If the states have won on the law so far, why did HUD get its stay? The answer is a deadline. HUD must make its Continuum of Care awards by December 1, 2026^[1][2]. Miss it, and grants freeze for thousands of projects that already exist, with tenants and staff attached to them right now.

Courts weigh "irreparable harm" when deciding whether to pause a lower court's order, and "no grants go out at all" is the kind of concrete harm judges can see clearly^[2]. That's the practical lever HUD's win rests on, separate from whether its underlying policy survives the appeal. Roughly 6,000 Continuum of Care projects sat in limbo during the litigation, and Congress had to step in separately to direct HUD to renew expired projects while the case dragged on^[19].

That's the part easy to lose in headlines calling this a "green light" for HUD's overhaul^[3][15]. The panel didn't bless Housing First's replacement. It decided that freezing the money entirely, mid-appeal, would cause more damage than letting the competition proceed. Filings from Washington state described more than 2,900 residents facing destabilization, and King County officials said roughly 4,500 households depend on about $26 million in permanent-housing money now back in play^[23].

The Argument Each Side Isn't Getting Credit For

HUD's case, stripped of the "warehousing" language, is a structural one: Congress funds an outcome, people leaving homelessness, not one specific method. For about 15 years, the agency's scoring rules pushed grantees toward permanent apartments with no treatment requirement, which HUD argues locked in a single theory and crowded out alternatives^[1][4]. On the legal question, HUD's position is that a yearly funding notice is a grant competition, not a binding rule, so it shouldn't need months of public comment to adjust priorities within a single funding cycle^[1][20].

The states and provider coalitions have a structural argument of their own that gets less airtime in right-leaning coverage. Most Continuum of Care dollars fund renewals for units and staff that already exist. A lapsed grant isn't an abstract policy loss. It's rent that stops getting paid for people already living somewhere^[19]. Providers argue transitional beds are time-limited by design: when the clock runs out with no apartment waiting at the end, the person often returns to the street, and the count HUD tracks goes back up^[18].

Both arguments are sincere, and both rest on incentives worth naming plainly. HUD wants a visible break from what it considers a failed policy consensus, in place on the ground before the December deadline, regardless of how the appeal ultimately resolves^[1][4]. States and providers want to protect existing budgets and existing tenants from a funding shift they didn't design and can't easily absorb if it lands^[1][16]. Neither motive makes the other side's legal argument wrong.

The Number Nobody's Number Settles

Whatever the courts eventually decide, one fact sits underneath the entire fight: HUD's own January 2025 count found 745,652 people homeless on a single night, including 266,320 with no shelter at all^[9]. That's up 27% since 2013. It's also down 3% from 2024.

Critics of Housing First emphasize the 27% rise. Defenders emphasize the 3% dip. Both numbers come from the same annual snapshot, taken by volunteers on one night each January, and the method isn't precise enough to settle either claim by itself^[9].

Coverage of the stay itself split along familiar lines. Reuters and The Hill described the court as "clearing the way" or giving a "green light," language that reads as a merits victory even though the panel ruled only on a temporary pause^[1][3]. CalMatters led with McElroy's harshest quote in its headline, while The Epoch Times led with Turner's^[16][15]. The panel's actual question, whether HUD followed the law when it rewrote the rules, is still pending^[2]. Both the December 1 deadline and the appeal itself are now on the clock, and which one moves first will decide what the next round of this fight even asks.