On July 1, 2015, Kathryn Steinle was shot and killed in San Francisco by Juan Francisco Lopez-Sanchez, an undocumented immigrant who had been convicted of seven felonies and deported seven times. Lopez-Sanchez had most recently been arrested for an outstanding drug warrant and served time briefly in a San Francisco jail before the U.S. Immigration and Customs Enforcement (ICE) asked that the city release him into its custody for deportation. Because San Francisco is a designated “sanctuary city,” it declined to prosecute Lopez-Sanchez’s drug charge, and he was released rather than deported.
The shooting of Steinle ignited a firestorm over San Francisco’s sanctuary policy. Republican presidential candidate Donald Trump has made opposition to sanctuary policies a major theme of his campaign. A bill titled the Enforce the Law for Sanctuary Cities Act, and nicknamed the Donald Trump Act by Democrats, passed the House on July 23, 2015, and would have blocked sanctuary cities from receiving federal law enforcement funding in response to the shooting.
Since then, public scrutiny of sanctuary cities has grown. For example, Louisiana Attorney General Jeff Landry stated that sanctuary cities “allow illegals to commit crimes, then roam free in our communities,” citing Los Angeles’s spike in crime in 2015 as evidence.
But is really true? Despite popular accounts, decades of research actually shows that immigrants – whether legal or illegal – tend to have lower crime rates.
Now, our new research shows that designating a city as a sanctuary has no statistically significant effect on crime.
We examined all the sanctuary cities listed by the National Immigration Law Center (NILC). All sanctuary cities included in the study passed sanctuary laws during or after 2002. We define a sanctuary as a city that expressly forbids city officials or police departments from inquiring about immigration status.
For each city, we drew on city-level crime data compiled by the FBI by year, as well as a host of demographic and political features that may affect crime levels and/or sanctuary status, such as racial and partisan composition, the unemployment rate, average income levels, the poverty rate, education levels and the percent of the city that is foreign-born.
First, we assessed changes in crime rates at the city level immediately following the passage of a sanctuary policy. Figure 1 reports the results for changes to violent crime in the year immediately following a change in laws for sanctuary cities that have available crime data in the year after the city became sanctuary.
Some cities – such as San Francisco and St. Louis — did see increases in crime immediately following passage. Other cities, such as San Jose, saw no change in crime. Still others — such as Baltimore and Washington — saw a reduction in crime. Taken together, the average change in crime is not statistically significant. The same results hold for property crime and rape crime.
Our second method is more comprehensive. We matched each sanctuary city to a similarly situated non-sanctuary city based on relevant census and political variables. This creates a scenario in which two cities are as similar as possible – with the exception of sanctuary policy – on a variety of features associated with crime rates. If the two comparable cities systematically differ in crime rates, then we can be quite confident that the difference is attributable to the sanctuary designation.
The graph below presents our results for violent crime. It compares violent crime rates among sanctuary and non-sanctuary cities from when no cities in our data set were classified as sanctuaries to after all cities had passed their sanctuary policies.
Violent crime is slightly higher in sanctuary cities than non-sanctuary cities. However, judging by the error bands — which capture the uncertainty underlying these estimates — the relationship is not statistically significant before or after a sanctuary policy is passed. We find similar results for property crime and rape.
That is, a sanctuary policy itself has no statistically meaningful effect on crime.
These results make sense if sanctuary city policies have countervailing effects. Given lower crime rates among immigrants, crime rates in sanctuary cities should drop, if those cities do attract new immigrants.
At the same time, sanctuary policies are typically designed to increase trust between immigrant communities and law enforcement. Thus, crime reporting — but not crime itself — might actually increase in these locations if undocumented immigrants are more likely to work with police and local authorities.
Taken together, these explanations may explain what we observe in these data: a sanctuary city designation does not produce a significantly higher crime rate.
Loren Collingwood is an assistant professor of political science at University of California at Riverside whose research interests include American politics, political behavior, and race and ethnic politics.
Benjamin Gonzalez-O’Brien is an assistant professor of political science at Highline College whose research interests include American politics, immigration policy, racial and ethnic politics and American political development.
Stephen El-Khatib is a graduate student at University of California at Riverside whose research interests include political conflict, immigration, and race and ethnic politics.