In the words of a former INS commissioner, over the course of the last two decades, modern immigration enforcement has developed into a “formidable machinery.” While historically focused on securing the border area, robust interior enforcement aimed at identifying and deporting immigration violators is now an integral and expanding part of that machinery.  In 2008, the federal government launched Secure Communities, a program designed to improve the efficiency of interior enforcement as well as to improve the capacity for targeting “criminal aliens” – that is, deportable individuals with criminal convictions.  Secure Communities enables the automatic transmission of fingerprints taken upon arrest to the Federal Bureau of Investigation and the Department of Homeland Security, which allows federal officials to verify an arrestees’ immigration status.  Once activated nation-wide, Secure Communities will result in screening every single fingerprint taken by local law enforcement agents against databases on potential immigration violators.  As such, the program is unprecedented in its scope and mandatory involvement of local law enforcement agencies.  Proponents of the program have argued that it not only enables a more efficient identification of criminal aliens, but also augments public safety by lowering crime rates.  Critics of the program, on the other hand, have expressed concern that the program encourages local law enforcement agencies to engage in discriminatory or arbitrary policing practices, making arrests for the sole purpose of checking an individual’s immigration status.  An atmosphere of fear created by Secure Communities, according to the critics, also erodes the cooperation with the police among immigrant communities, which only threatens effective law enforcement.

In this research, we exploit the staggered activation dates of Secure Communities across counties to examine whether the program has a detectable effect on local crime rates or the arrest behavior of local law enforcement agencies. Regressions of the crime rate on the treatment indicator are run using monthly panel data on U.S. cities with a population greater than 50,000 individuals over the 2007-2011 time period.  While we find evidence of isolated effects both on crime and arrests, which call for further investigation, overall there is little evidence either for the most ambitious promises of the program, or the biggest fears behind embroiling criminal law enforcers in immigration enforcement.  In particular, the program’s activation does not appear to be associated, in general, with lower crime rates, with more arrests for less serious crimes or with a greater arrest share for white (the ethnic group in which Hispanics are classified in the FBI’s arrest statistics) versus black offenders.  Results are robust to the inclusion of month fixed effects and either city-by-year or county-by-year fixed effects each of which constitutes a powerful set of predictors and explains a great deal of unobserved heterogeneity in monthly crime and arrest rates.