As the Biden Administration decides whether to continue the 287(g) program (the controversial program deputizing local law enforcement officers to enforce federal immigration laws), our research shows that the program has broader negative effects on policing behavior than previously identified. To date, debate about the 287(g) program has focused exclusively on the policing behavior of law enforcement agencies like sheriff’s offices that sign the agreements, and on concerns that these signatory local enforcement agencies (“LEAs”) engage in racial profiling. Our research shows that the agreements also negatively affect the behavior of nearby, nonsignatory law enforcement agencies. Using 18 million traffic stops drawn from the Stanford Open Policing Project, we find that the agreements caused state troopers in North Carolina and South Carolina to stop Hispanic drivers more often than White drivers, in order to funnel them into the intensive immigration screening conducted by signatory LEAs at the shared jails. Because trooper agencies did not sign the agreements, statistical associations between the presence of agreements and the differential treatment of drivers by race are not contaminated by unobserved confounding factors. Our identification of these previously unnoticed spillover effects raises important policy questions about the program’s impact and the adequacy of existing legal and administrative controls.