ICE Arrests across Trump's First and Second Terms: Variation in Targeting, Method, and Geography
Deportation is often framed as a necessary tool to protect public safety by removing people who commit crimes. We use newly available, and externally validated, administrative data containing all US Immigration and Customs Enforcement (ICE) arrests from September 2015-October 2025. Beyond demonstrating national trends in immigration arrests by method and composition over time, we are also able to compare, for the first time, apprehensions spanning the start of the two Trump administrations, both of which focused on mass immigration enforcement. Our results reveal that the reality of immigration enforcement diverges sharply from the public narrative: while arrests spiked at the outset of both Trump presidencies, there were significant declines in the percentage of arrested individuals with criminal convictions, with especially marked declines in 2025. Examining potential mechanisms reveals that this is driven by a change in ICE tactics, but even conditional on tactic, as arrests rose, the percent with a criminal record declined. Moreover, we find substantial heterogeneity over time and across ICE Areas of Responsibility. Taken together, our results highlight a substantial gap between political rhetoric and reality.
-
-
Copy CitationChloe N. East, Caitlin Patler, and Elizabeth Cox, "ICE Arrests across Trump's First and Second Terms: Variation in Targeting, Method, and Geography," NBER Working Paper 34794 (2026), https://doi.org/10.3386/w34794.Download Citation