Sanctuary Status and Crime in California: What’s the Connection?

Abstract

In 2017, California officially became a sanctuary state following the passage of Senate Bill 54, which limits state and local police cooperation with federal immigration authorities. Following the passage of SB54, critics worried that crime rates would rise. What impact did this policy have on crime in California? The current study, the first of its kind, addresses this question. Using a state-level panel containing violent and property offenses from 1970 through 2018, we employ a synthetic control group design to approximate California’s crime rates had SB54 not been enacted. We interpret the gap between California’s 2018 crime rate and its synthetic counterfactual as SB54’s impact. Results show that SB54’s impact on violent and property crime is neither robust nor sufficiently large to rule out a null effect. Sensitivity analyses buttress this finding. We discuss the implications of the findings for crime policy in the U.S.

Key Words:

• crime
• crime policy
• immigration
• sanctuary laws
• synthetic control

Disclosure statement

No potential conflict of interest was reported by the author(s).

Acknowledegments

We thank Mary Scully for research assistance on this paper.

Notes

1 The Secure Communities Program provides a system that automatically transmits and checks fingerprints against the Department of Homeland Security’s Automated Biometric Identification System, which contains information on known immigration violators, known and suspected terrorists, and “criminal aliens.” A fingerprint match prompts officers from Immigration and Customs Enforcement (ICE) to investigate, determine the individual’s immigration status, and forward their conclusion to the relevant field office. If ICE decides to take action, a detainer is issued to the law enforcement agency requesting that the individual be detained for up to 48 hours so that ICE can assume custody (Kubrin, 2014:1).

2 The Criminal Alien Program provides ICE-wide direction and support in the identification, arrest, and removal of priority “criminal aliens” who are incarcerated within federal, state, and local prisons and jails as well as at-large “criminal aliens” who have circumvented identification (Kubrin, 2014:3).

3 287g agreements authorize federal authorities to deputize local law enforcement agencies to perform tasks such as screening individuals for their immigration status, issuing detainers to hold potential violators, and issuing charging documents that initiate removal proceedings (Kubrin, 2014:3).

4 Although many were surprised by California’s bold move with SB54, in fact there is a previous history of sanctuary policies throughout the U.S., but especially in California, in relation to the Central American Sanctuary Movement of the 1980s (for a more detailed discussion of this history see O’Brien et al. (2019:9-11) and Martinez-Schuldt and Martinez, (2019:571-572).

5 Local law enforcement can choose to notify ICE or Border Patrol of an individual’s release or transfer if that individual has certain convictions including state prison felony convictions, most other felony convictions within 15 years, and high-level misdemeanors within 5 years. Likewise, local law enforcement can notify immigration agents of an individual’s release date if the information is already public.

6 These opinions were made famous by President Trump’s claim: “When Mexico sends its people, they’re not sending their best…They’re sending people that have a lot of problems…They’re bringing drugs. They’re bringing crime. They’re rapists. And some, I assume, are good people” (Mark, 2018).

7 For a list of variables included in the match see O’brien et al. (2019:18).

8 Or more precisely, states that have not been contaminated by a similar sanctuary state policy within the analysis time frame of 1970-2018.

9 The synthetic control group method is a special case of the quasi-experimental “difference-in-differences” (DiD) design. In general, DiD models are used to estimate the impact of an intervention on a DV time series by identifying an untreated control time series with pre-intervention trends that are parallel to the treated unit (Ashenfelter & Card, 1985). If a control unit is identified which satisfies the parallel trends assumption, a departure from the parallel relationship following the intervention can be interpreted as the policy’s impact on an outcome. The parallel trends assumption can be difficult to satisfy using only the controls available, however, this severely limits the range of applications and overall utility of the DiD design for evaluating the impact of social policies such as SB54. This limitation is particularly acute when long time series are available, the time series lack substantial serial correlation, or the pool of comparison units is relatively small (Ryan, Kontopantelis, Linden, & Burgess, 2019:3369). The synthetic case control design aims to resolve this issue by constructing a control unit which optimally recreates the treated unit’s pre-intervention DV trends as a weighted combination of untreated units.

10 Given the length of the pre-SB54 segment, this route is computationally expensive, but it invariably produces the best fitting synthetic control.

11 The weights are constrained conventionally to be non-negative and sum to one, which prohibits extrapolation as well as fitting on inverse trends (Abadie et al. 2010).