The Social Costs of ICE Enforcement

• In Event: Immigration Enforcement as a Public Health Crisis: Spillover Effects Across Health Systems and Communities

Abstract

Immigration enforcement in the United States entered a new phase in 2025, marked by a sharp increase in community-based arrests and the rescission of long-standing protections that had limited enforcement near schools, hospitals, and other sensitive locations. While prior research has focused mainly on targeted immigrants, much less is known about the broader spillover effects on surrounding communities. This paper provides the first causal estimates of the community-wide social and economic costs of U.S. Immigration and Customs Enforcement (ICE) operations under the post-2025 enforcement regime.

We combine a novel dataset of geocoded ICE enforcement records with weekly point-of-interest-level foot traffic data across six high-enforcement states from January 2025 through January 2026. The data include 10,018 enforcement events, ranging from direct-contact operations such as arrests and raids to non-contact activities such as visible surveillance and officer presence. We link each event to nearby establishments and exploit hyperlocal spatial and temporal variation in exposure, comparing locations very close to an enforcement event with slightly more distant locations in the same neighborhood before and after the operation. This design isolates local causal effects from broader differences across communities.

We find that ICE enforcement generates an immediate and persistent contraction in local activity. Weekly visits to establishments within 0.5 miles of an enforcement event decline by 1.69 percent, with no detectable rebound over the following seven weeks. The effect is highly localized, attenuating to zero beyond roughly half a mile. Unique visitors also decline, indicating that enforcement reduces both the number of trips and the number of individuals willing to enter affected areas.

Several findings point to ambient deterrence, rather than direct confrontation, as the primary mechanism. Non-contact operations produce declines similar to those following direct-contact actions, and low-force deployments generate effects comparable to large-scale raids. These patterns suggest that the visible presence or perceived threat of enforcement is sufficient to suppress mobility and local participation.

The effects are especially pronounced for education facilities, where visits fall by 4.82 percent, and for worker-intensive establishments, where visits decline by 2.76 percent. Consumer-facing establishments also experience significant reductions, while healthcare facilities show no detectable effect, consistent with relatively inelastic demand for urgent care. The chilling effect extends beyond directly targeted populations: visits decline in both Hispanic-concentrated and predominantly White neighborhoods, and across industries with both high and low immigrant workforce shares. We find no evidence of geographic substitution, suggesting that these declines reflect genuine withdrawal from local activity rather than relocation elsewhere.

Finally, we translate these reduced-form estimates into economic costs. Under conservative assumptions, the combined labor, education, and consumer-side losses amount to approximately $195,669 per establishment per month, more than eleven times the Department of Homeland Security’s benchmark per-removal cost. A spatial logit model further shows that restoring sensitive-location protections would recover 50.4 percent of these welfare losses, whereas converting all enforcement to non-contact operations would recover only 18.3 percent. Together, the results show that immigration enforcement imposes large, localized social costs that extend well beyond those directly targeted.

Author

• Hantao Wu, Princeton University
  Presenting Author