Border crossings fell. City asylum costs did not.
Federal policy changes helped drive a sharp drop in border encounters and a lower New York shelter census in 2025. City spending, however, is unwinding on a slower clock.
The border numbers are finally moving in the direction city officials wanted. That much is clear. The harder truth is that local asylum costs do not fall on the same timeline as border encounters. Crossings can collapse quickly; municipal obligations rarely do.
New York City’s own oversight pages now make that mismatch visible. The Comptroller reports that the city-funded asylum shelter population had fallen to roughly 33,300 people as of September 28, 2025, down 51 percent from the January 2024 peak. The same census page says the average number of southern border encounters from February through August 2025 was down 93 percent from the same period in 2024. Those are dramatic shifts.
But the city’s fiscal page tells a different second-half story. FY 2025 actual asylum-shelter and service spending still reached $3.02 billion. The June 2025 adopted financial plan still carries $1.303 billion for FY 2026. The curve is moving down, but it is still measured in billions.
Border encounters
-93%
Average February–August 2025 southern border encounters relative to the same period in 2024, as cited on the Comptroller census page.
NYC shelter census
33,300
Approximate number of people seeking asylum in city-funded shelter as of September 28, 2025.
Peak-to-current change
-51%
Drop from the city-funded shelter peak of nearly 70,000 people in January 2024.
FY 2025 spending
$3.02B
Actual city spending on asylum seeker shelter and services, even after the census had started to recede.
Why the fiscal unwind lags the flow story
The explanation is not mysterious. Shelter spending is sticky.
Families remain in the system for longer periods than a monthly border headline suggests. Sites close in stages, not overnight. Vendor contracts, staffing, transportation, medical services, and outstanding payables all unwind on administrative rather than political timelines. Even if fewer people arrive at the front end, the back end does not clear itself at the same speed.
The Comptroller’s overview makes this plain. It describes a city that scaled up hundreds of emergency contracts across multiple agencies to provide shelter, meals, medical care, legal assistance, and other services. Those systems were built under pressure. They cannot be shrunk with the same speed as a border encounter chart.
This is exactly why the city’s own financial plans kept changing. According to the fiscal impacts page, total actual and budgeted asylum-related spending across FY 2023 through FY 2029 fell from $17.86 billion in the June 2024 plan to $11.82 billion in the June 2025 plan. That is a large revision, but it is still a massive multiyear bill. The cost story is falling, just nowhere near as abruptly as the border story.
The local effect is now composition, not only volume
There is another reason the decline feels slower than the headline numbers imply: the composition of the remaining population is changing. The Comptroller census page says families with children now make up a larger share of the city-funded population than before. That tends to mean more complex placements, more school coordination, and less opportunity to collapse shelter capacity into a single cheap congregate model.
The site-closure pattern tells the same story. The city has been able to shutter many of the highest-visibility emergency facilities, but it is doing so while shifting more of the remaining population into the Department of Homeless Services network. That is a transition, not a disappearance.
Why this matters beyond New York
New York is just the clearest illustration because it publishes the most legible local oversight materials. The same logic applies in any city that has already built a shelter and service apparatus around emergency arrivals. Once contracts are signed and families are in beds, the marginal cost of delay becomes a city budget problem even if the federal flow picture improves.
That has two policy implications.
First, local officials should resist reading better border numbers as an invitation to declare the cost crisis over. The lagging obligations are real and budgetable even if new entries are down.
Second, the fastest way to reduce those obligations is not only lower inflow. It is faster exits from shelter into lawful work, family reunification, and stable housing. A city can benefit from calmer federal flows and still remain stuck with a costly local backlog if those downstream pathways stay slow.
This is what the current data shows. The border emergency has eased. The municipal balance sheet has not caught up yet.
Sources