When the federal government expanded interior immigration enforcement in 2025, officials offered a straightforward economic rationale: remove unauthorized workers and free up jobs for Americans. The logic has obvious appeal at a moment of anxiety over automation and rising costs. But a new analysis from the Brookings Institution reports the reverse. In the cities where enforcement intensified most, total employment fell, and the losses reached workers born in the United States.
The headline figure is large. The researchers estimate that enforcement surges in 86 metropolitan areas were linked to roughly 668,000 lost jobs over the first nine months of 2025. Of those, they estimate that between 51,000 and 297,000 would have been held by American-born workers.
The question behind the study
Marcela Escobari, vice president and director of the Global Economy and Development program at Brookings, led the work with colleagues Ian Seyal and Paul Beach. Their starting point was a testable claim made on behalf of the policy itself: that aggressive enforcement helps American workers.
The 2025 campaign, run by Immigration and Customs Enforcement (ICE), differed from earlier efforts in scale and visibility. The authors describe it as built around what administration officials called “shock and awe” tactics, meaning highly visible raids, worksite arrests, and widely shared videos of detentions. They contrast this with an earlier program, Secure Communities, which operated mostly through the criminal justice system and stayed out of public view.
That distinction matters to the study’s central argument. Past research on enforcement documented a “chilling effect,” in which immigrant households, including people with legal status and U.S. citizens, pulled back from public life out of fear of contact with officials. Visible enforcement, the authors argue, spreads that fear more widely than the arrests themselves.
How the researchers measured it
The study drew on two main datasets. Arrest records came from the Deportation Data Project, which compiles individual ICE arrest data obtained through public-records requests. Employment figures came from Lightcast, a labor-market firm whose estimates are built primarily from federal payroll-tax records that cover nearly all formal jobs in the country.
The team focused on “at-large” arrests, meaning arrests made in the community rather than handoffs from jails, because those are the ones most likely to register in local labor markets. They then sorted 341 cities by how sharply 2025 arrests rose above each city’s own 2024 baseline. The 86 cities in the top quarter were labeled “surge” cities. The remaining 255 served as a comparison group.
To estimate the effect, the researchers compared employment trends in surge cities against the comparison cities, accounting for the fact that ICE moved from place to place and surges began in different months. If the two groups moved together before enforcement and split apart afterward, the gap could reasonably be attributed to the enforcement. The authors report that employment paths were closely aligned for three years before each city’s surge and diverged only once arrests spiked.
This design also helps separate enforcement from other forces hitting the economy at the same time, such as tariffs, inflation, and the spread of artificial intelligence. Because surges arrived city by city on a staggered schedule, an unrelated national shock would have had to mirror that rollout exactly to produce the same pattern.
Jobs lost far outnumbered arrests
The most notable result, according to the authors, is the distance between arrests and job losses. Across the 86 surge cities, ICE made about 52,000 more arrests than in prior years. Employment in those cities ran an average of 0.73% below what the analysis projects it would have reached without the surge.
That works out to roughly 13 jobs lost for each additional arrest, jobs previously held by both immigrants and American-born workers. Even if every person arrested had been employed, the authors note, their removal alone cannot explain a number that large.
The losses also deepened over time. In the 51 cities the researchers could observe at least six months past the surge, the employment gap widened to 1.48%, and the ratio rose to about 30 jobs lost per excess arrest.
The authors point to three processes that, in their interpretation, drive losses beyond the people arrested. First, immigrant workers who were never targeted stopped showing up to work, avoided public spaces, and changed their routines out of fear. A 2025 survey they cite found rising worry about detention even among lawfully present immigrants and naturalized citizens. Second, businesses that lose a portion of their workforce overnight often cannot recruit and train replacements quickly, so they scale back or shut down, which costs additional jobs in connected roles. Third, fear suppressed consumer demand, as people stayed home and spent less, hurting businesses whose customers disappeared.
Which industries absorbed the damage
The losses were not spread evenly. Construction, which employs a high share of likely undocumented workers, was hit hardest, with employment averaging 2.2% below the projected level over eight months and reaching about 4% at six months. That is far steeper than its workforce composition alone would predict.
The researchers tie this to how the industry operates. Crews are specialized and projects are interdependent, so a framing crew missing half its workers cannot simply hand off to a roofing crew. When projects stall, the job losses ripple to project managers, equipment operators, electricians, and inspectors, roles the authors note are predominantly filled by American workers. They cite a separate finding from Howard and colleagues that earlier enforcement reduced construction employment for American-born workers and slowed homebuilding enough to raise new-home prices.
The harder pattern to explain through worker removal alone showed up in arts and entertainment, where employment fell about 7% at six months despite the sector employing few likely undocumented workers. The authors read this as evidence of the demand channel: when enforcement dominates the news and people stop going out, theaters, venues, and similar businesses lose customers and cut staff regardless of who works there.
As an informal check, the researchers point to sectors like professional services, information, and utilities, which employ few undocumented workers and are not sensitive to foot traffic. Those sectors showed no measurable effect. If a broader trend specific to surge cities were driving the results, the authors argue, it should have appeared there too.
Estimating the toll on American-born workers
The payroll data the team used cannot identify which missing jobs would have gone to American-born workers, so the authors built a range using results from a separate study by Cox and East. The lower estimate of about 51,000 jobs comes from a narrow group of less-educated, American-born male workers in four industries. The upper estimate of about 297,000 subtracts an estimate of undocumented job losses from the total and attributes the remainder to American-born workers. Both ends rest on assumptions the authors describe in detail, which is why they present a range rather than a single number.
Several limits apply to the findings. The estimates cover only the 86 surge cities and cannot be scaled to a national total. The payroll data capture formal jobs but miss informal, cash-based, and gig work, which means some of what looks like job loss may reflect a shift into informal work rather than people leaving the workforce entirely. The data also end in September 2025, before later operations such as the Minneapolis surge, so the analysis captures an early window when effects were still building. The authors describe their results as evidence of short-run local disruption rather than a final measure of long-run national employment.
Escobari frames the takeaway in terms of the original premise. “It’s tempting to think of immigration enforcement as a one-to-one replacement,” she says. “One person deported, one job freed up for an American worker. But that’s not backed up by the data.”
