Delegation and Divergence: A Study of 287(g) State and Local Immigration Enforcement. The Migration Policy Institute has issued a report on the 287(g) program, under which ICE delegates to local law enforcement departments authority to arrest individuals who are in violation of immigration laws. The report shows that implementation of these programs is very diverse, with some localities targeting only serious criminals and others picking up anyone at a traffic stop. The impact of traffic stop-like programs (which MPI called “universal” programs) on the entire undocumented population appears to be very high.
Excerpts from the MPI’s executive summary:
At the national level, the program is not targeted primarily or even mostly toward serious offenders. Nationally, about half of program activity (defined by the number of immigration detainers issued) involves people who have committed felonies and other crimes that ICE deems to be serious (Priority Level 1 and 2 in ICE’s terminology). The other half of detainers issued are on people who have committed misdemeanors (usually considered Level 3) and traffic offenses.
Some jurisdictions operate “targeted” models, aimed primarily at identifying serious criminal offenders, while others pursue “universal” models, designed to identify as many unauthorized immigrants as possible. In FY 2010, Las Vegas operated the most targeted program among our sites: officers placed 70 percent of detainers on Level 1 or 2 offenders.
By contrast, Cobb County (GA) and Frederick County (MD) placed about 80 percent of their detainers on Level 3 or traffic offenders, and officers there placed detainers universally (i.e., on every unauthorized immigrant booked into jail or encountered during policing operations).
Contrary to public perception, 287(g) is almost entirely a jail program. In the first ten months of FY 2010, jail models accounted for 90 percent of detainers issued, while hybrid models accounted for 8 percent and task force models just 2 percent. Under the jail model, 287(g) officers screen for immigration status and place detainers after people have been booked into jail. Jurisdictions with jail models are not substantially more likely to operate targeted programs than jurisdictions with task forces. Except for Colorado, which apprehends a large number of recent border crossers who are in transit on the state’s highways, task forces issued about half of their detainers on Level 1 and 2 offenders — a similar proportion to jail models.
Universal models are concentrated in the Southeast. The ten sites with the largest share of detainers placed on traffic violators are all located in the Southeast. Mecklenburg County (NC) initiated the first universal program in 2006, and our Southeastern study jurisdictions have followed Mecklenburg’s model.
The 287(g) jail model does not impose federal oversight on officers who make the initial arrests. Under the jail model, 287(g) officers screen for immigration status and place detainers after people have been booked into jail. Initial arrests are generally made by police officers working for agencies without 287(g) agreements and who lack federal oversight and training in immigration laws. The lack of federal control over arresting officers opens the door to racial profiling and pretextual arrests, especially in jurisdictions that place immigration detainers universally.
The formal program changes issued by the Department of Homeland Security (DHS) in 2009 have not substantially affected implementation. Despite reduced program activity (from 60,000 detainers in FY 2009 to 48,000 in FY 2010), we found no evidence that the revised template and the new 287(g) agreements the Obama administration required participating jurisdictions to sign in 2009 have had any substantial effect on 287(g) priority setting, program operations, outcomes, or community impacts.
State and local officials operate 287(g) programs according to priorities shaped largely by political pressures. In many communities in the Southeast and Southwest, the immigrant population grew rapidly during the 1990s and 2000s, leading to a public backlash and putting pressure on sheriffs and other elected officials to pursue a set of enforcement strategies. These strategies include pursuing a 287(g) program alongside adoption of state and local laws that directly target unauthorized immigrants, such as Arizona’s Senate Bill 1070 and laws in Georgia imposing mandatory jail time for driving without a license.
ICE closely supervises 287(g) officers in our study sites, but allows state and local agencies to set enforcement priorities. ICE supervisors sign off on virtually all detainers issued by 287(g) officers, regardless of whether the detainers apply to large numbers of Level 3 and traffic offenders.
Resources drive 287(g) program goals and implementation in many jurisdictions. In all sites operating universal models, 287(g) officers and ICE supervisors report that ample ICE detention capacity, in tandem with the increase in personnel resulting from the training of state and local officials (the “force multiplier” effect), allows officials to place detainers on as many Level 3 and traffic offenders as they encounter.
Universal enforcement gives rise to substantial negative community impacts and downstream federal detention and removal costs. Our study respondents reported impacts on immigrants similar to those cited in reviews by others of the 287(g) program, including avoidance of public places, changes in driving behavior, fear and mistrust of authorities, and reluctance to report crimes. Respondents also reported racial profiling, but we did not investigate these claims due to data limitations. Downstream federal detention costs can be significant in universal enforcement jurisdictions, with ICE detainees averaging 81 days in detention at a cost of $60 per day in Georgia, for example.
Some immigrants have left jurisdictions with controversial 287(g) programs. Frederick County experienced a 61 percent drop in its Hispanic noncitizen population after it adopted the 287(g) program, and Prince William County (VA) witnessed a 23 percent decline, while neighboring counties — especially Montgomery County (MD) and Fairfax County (VA) — experienced an increase. The population drop in Cobb County was only 5 percent, and it is too early to observe population changes in Gwinnett County (GA), which implemented its 287(g) program in late 2009. Public school enrollment data show drops followed by rebounds in Hispanic student populations in Frederick, Cobb, and Prince William counties.
Some jurisdictions have targeted their enforcement and conducted outreach to explain its scope, thereby muting community criticism. Prince William County implemented its 287(g) program in a relatively targeted fashion and modified its ordinance from a requirement that police screen the immigration status of everyone encountered to a screening requirement after booking into jail. The police department issued bilingual brochures and conducted hundreds of briefings explaining the 287(g) program, restoring the Hispanic community’s trust in the police department to levels near those documented before the program and the initial local ordinance were implemented.
In sum, the statutory language of section 287(g), the Obama administration’s statements and guidelines, and ICE’s implementation practices allow jurisdictions to operate the 287(g) program in fundamentally different ways across the country. We find that the 287(g) program has had more significant negative impacts on immigrant communities in jurisdictions where a universal model is used and traffic offenders comprise a large share of persons channeled through the program than in targeted jurisdictions.
Recommendaions
We therefore recommend that ICE take additional steps to ensure that the 287(g) program functions consistently across jurisdictions according to the targeted enforcement model promoted in the 2009 revised template and in public statements by DHS and ICE leadership — and that the agency reconfirm that the program is intended to target primarily serious criminals or people who have prior criminal convictions, represent a threat to national security, or have prior removal orders. To achieve this goal, ICE should:
1. Ensure that state and local 287(g) officers place detainers on noncitizens only after they have been convicted for Level 1 and 2 offenses, unless they have committed prior immigration violations.
2. Work with the US Department of Justice to investigate accusations of racial and ethnic profiling in 287(g) jurisdictions, including in law enforcement agencies that arrest those booked into 287(g) jails but do not themselves have agreements.
3. Discontinue 287(g) agreements with jurisdictions that operate universal models and do not focus on serious criminals, as well as with jurisdictions where evidence of racial profiling or other civil-rights violations emerges.
4. Limit new agreements to jurisdictions that can demonstrate no prior history of racial profiling by police as well as an intent and likelihood to operate the program in a targeted fashion.
5. Conduct a comparative cost-benefit analysis of the 287(g) program versus Secure Communities and other interior enforcement strategies, one that includes downstream detention, adjudication, and removal costs, as well as community impacts.
6. Create a real dialogue between ICE field offices and local communities by engaging in direct outreach, holding regular meetings with key constituencies, and establishing meaningful steering committees.
Secure Communities Program
Our findings and recommendations also apply to the Secure Communities program, which ICE plans to expand to every state prison and local jail by 2013. Through Secure Communities, ICE officers remotely screen immigration status by checking inmates’ fingerprints against federal databases. Like the 287(g) jail model, Secure Communities authorizes screening after inmates are booked into jail and lacks federal oversight or training of arresting officers. Unless ICE takes steps to ensure that it is a targeted program, Secure Communities may replicate many of the controversies and negative community impacts we observed in universal enforcement 287(g) jurisdictions.