Excerpted From: Huyen Pham and Pham Hoang Van, Sheriffs, State Troopers, and the Spillover Effects of Immigration Policing, 64 Arizona Law Review 463 (Summer, 2022) (154 Footnotes) (Full Document)


PhamAndVanAs the oldest and most visible form of local--federal cooperation in immigration enforcement, the program created by § 287(g) of the Illegal Immigration Reform and Immigrant Responsibility Act of 1996 has long been the focus of heated debate. The 287(g) program deputizes local enforcement agencies (“LEAs”) like sheriff's offices and police departments to enforce federal immigration laws. Proponents praise the program's “force multiplier” effect, increasing the manpower available to enforce immigration laws. Opponents criticize the program for targeting immigrants without criminal records for deportation (despite press releases to the contrary), for engaging in racial profiling of Hispanics and other groups associated with immigrants, and, as a result, destroying crucial trust between LEAs and immigrant communities. Disagreement about the program's efficacy has led to wide swings in both the size and scope of the program during different presidential administrations. The first 287(g) agreement was signed during President George W. Bush's Administration, and the program was slowly expanded during President Obama's first term, reaching 78 agreements, before being whittled down to 40 agreements at the end of his second term. Consistent with his restrictionist immigration priorities, President Trump greatly increased the 287(g) program to a historical high of 152 agreements and introduced a new model of agreement to appeal to more LEA partners. The number of active 287(g) agreements over time is detailed in Figure 1 below.

Now, President Biden is at a crossroads: he could continue to expand the program like the Trump Administration did, continue the program with reduced agreements and more restrictions like the Obama Administration did, or eliminate the program altogether. He has been pushed by different interest groups toward all these options. To make a good policy decision, the Biden Administration needs to have full and complete information about the program's effects, but, to date, the debates and analysis--including the arguments about force multiplication, overpolicing, and racial profiling--have focused solely on the policing behavior of signatory LEAs and its consequences.

Our research demonstrates that the agreements also affect the policing behavior of nonsignatory LEAs, causing them to engage in the same racial profiling behavior attributed to signatory LEAs. Using the North Carolina State Highway Patrol and the South Carolina Highway Patrol as case studies, we analyzed approximately 18 million traffic stops made by these state troopers from 2000-2016; for this time period, neither state agency signed a 287(g) agreement but, crucially, used jail facilities operated by signatory LEAs, where intensive immigration screening occurred. Specifically, we found that the presence of the agreements caused the state troopers to stop Hispanics more often. An example of this effect is shown in Figure 2. The Mecklenburg County Sheriff's Office (North Carolina) signed a 287(g) agreement in February 2007; before this date, the number of trooper stops of Hispanics in Mecklenburg was similar to trooper stops of Hispanics in counties in North Carolina and South Carolina that never signed 287(g) agreements. After February 2007, however, the two trends diverged, with stops of Hispanic drivers in Mecklenburg increasing each month to all time highs. A similar pattern emerged in Charleston County (South Carolina), where the sheriff's office signed a 287(g) agreement in November 2009. Before this date, trooper stops of Hispanics in Charleston paralleled the number of stops in the counties that never signed 287(g) agreements, but shortly after the 287(g) was signed, active stops of Hispanics increased each month, even when they fell in “Never-287(g)” counties.

While stops increased, we also saw that arrest rates of Hispanics fell. These findings, taken together, support the conclusion that the state troopers engaged in racial profiling: the high incidence of initial stops of Hispanics are critical to funneling Hispanics into the intensive immigration screening processes provided by the 287(g) program, and the lower “hit rates” between stops and arrests are widely accepted as evidence of racial profiling by law enforcement agencies.

Our project makes important methodological and policy contributions. From a methodological perspective, by observing the policing behavior of state troopers who are not parties to the 287(g) agreements, our project strengthens the causal connection between 287(g) agreements and racial profiling. Other studies that looked only at the policing behavior of signatory LEAs cannot make similarly strong claims of causality because of possible confounding factors. For example, if the leadership of a county LEA--e.g., the sheriff--favors anti-immigrant policies, that leadership would likely push the county to sign a 287(g) agreement and also support any observed racial profiling by agents of that LEA. Thus, the underlying anti- immigrant policy direction confounds the analysis and makes it difficult to conclude that any observed racial profiling is in fact caused by the 287(g) agreement. In our study, because the state troopers were not parties to the 287(g) agreements but would know whether such an agreement was active in the county where they patrolled, the observed statistical associations between the presence of 287(g) agreements and the differential treatment of drivers by race are, in fact, causal.

From a policy perspective, our findings demonstrate that 287(g) agreements have important and heretofore unexplored spillover effects on the policing behavior of nonsignatory LEAs. First, the force multiplier effect of the program may be even stronger than advocates estimated, as nonsignatory LEAs are joining signatory LEAs to enforce federal immigration laws. Like signatory LEAs, these nonsignatory LEAs are reallocating their resources--e.g., trooper time--to join in immigration enforcement efforts, but unlike signatory LEAs, their reallocation is happening without the controls of written agreements or formal reviews. Second, with nonsignatory LEAs also engaging in racial profiling, the pernicious effects of racial profiling created by the 287(g) program are likely more extensive than critics feared. From a civil rights perspective, the specter of racial profiling is particularly troubling because it threatens the constitutional rights of all those who are profiled, without regard to their actual immigration status.

Finally, our observed spillover effects suggest that, due to the interconnectedness of LEA operations, there is less political control over the 287(g) program than its formalized structure indicates. Formally, LEAs that want to participate in immigration enforcement efforts must apply, be approved by federal authorities, and sign a written agreement detailing their immigration-related duties. In this formal model, the 287(g) agreement is seen by advocates and critics as the gateway to perceived benefits and harms. Critics of the program understandably focus their efforts on lobbying signatory LEAs to terminate their agreements or lobbying potential signatory LEAs not to sign agreements in the first place. When reports of abuse within the program have emerged, reform efforts have focused exclusively on the form of the 287(g) agreement and the signatory LEAs. For example, after investigations by the Department of Justice (“DOJ”) found widespread racial profiling by the Maricopa County Sheriff's Office (Arizona) and the Alamance County Sheriff's Office (North Carolina), the Obama Administration responded by revoking those LEAs' agreements. The baseline assumption then is that the 287(g) program is subject to political controls, exercised by elected officials who control entry into and exit from the program (at the federal level, by the President of the United States, and, at the local level, by LEA heads, often sheriffs who are elected in most jurisdictions). But our findings of spillover effects suggest that existing political controls may be inadequate to counter abuses within the program. When nonsignatory LEAs engage in racial profiling, their bad behavior can't be curbed by simply revoking an agreement, changing agreement terms, or even by voting out a political official who makes decisions within the official 287(g) program. The inadequacies of existing political controls raise important questions about accountability within the 287(g) program structure.

Relatedly, the leadership of nonsignatory LEAs needs to be aware of these spillover effects. An LEA that chooses not to enter into a 287(g) agreement may do so for a variety of reasons: disagreeing with the mission of local--federal immigration cooperation, wanting to remain neutral on a very divisive issue, wanting to preserve trust with their immigrant communities, or having concerns about the fiscal and other costs of immigration cooperation. But given our observed spillover effects, an LEA's reasons for not signing a 287(g) agreement may be undermined if its rank-and-file officers are informally participating in immigration law enforcement. This discretionary participation, without the benefit of training or regulation, has important implications for LEA policies and for public safety. By highlighting this spillover effect, our project makes an important and novel contribution to the literature showing divergence between the 287(g) program as designed and as implemented.

This Article proceeds in five parts. In Part I, we analyze the origins and operation of the 287(g) program, comparing the program with other forms of local--federal immigration cooperation. In Part II, we discuss our case studies of North Carolina and South Carolina, including information about the operation of the 287(g) program in these two states. In Part III, we explain our methodology, including our research design. Part IV analyzes our results, linking the results to relevant explanatory literature in the fields of policing and criminal procedure.

[. . .]

Our project identifies important spillover effects of the 287(g) program. To sum up, we analyzed approximately 18 million traffic stops made by the North Carolina State Highway Patrol and the South Carolina Highway Patrol, from 2000 to 2016, and found evidence that 287(g) agreements caused state troopers to engage in racial profiling. Though neither state trooper agency signed a 287(g) agreement, individual state troopers patrolling in 287(g) jurisdictions knew that the county jails where they booked their arrestees employed immigration screening processes. That knowledge and the shared jail facilities, we argue, incentivized troopers to engage in racial profiling. In counties with active agreements, we found that troopers increased stops of Hispanics relative to Whites but decreased the rate of arrest of Hispanics after a stop, relative to Whites. Both of these findings taken together are evidence of racial profiling, because the increase in stops is necessary to funnel Hispanics into the intensive immigration screening processes provided by the 287(g) program and the lower “hit rates” between stops and arrests are widely accepted as evidence of racial profiling by LEAs.

What are the implications of these observed spillover effects? On one level, the spillover effects could just represent additional divergences between the 287(g) program as implemented and the program as designed. Existing analyses of the 287(g) program have noted different divergences in the program's history; as explained in Part I, those divergences include the targeting of immigrants without serious criminal records, despite stated federal priorities that limited the program to deporting dangerous immigrants. Another divergence that is particularly relevant to our project is the evidence of racial profiling committed by signatory LEAs, again contravening the agreements that they signed with the federal government.

Zooming out however, we see that our observed spillover effects are qualitatively different. Our analysis highlights new players and a new dynamic to the 287(g) program, in which nearby, nonsignatory LEAs informally participate in immigration law enforcement without formal training in the nuances of immigration law and without federal regulation or control. Our project studied state trooper agencies in two states, but our observed effects are likely occurring in other contexts as well--e.g., a nonsignatory police department that uses the jails of a signatory sheriff's office. If these spillover effects are widespread, they raise important policy questions for the Biden Administration as it determines its course of action. First, what are the true impacts of the 287(g) program? If nonsignatory LEAs are joining immigration enforcement efforts through the 287(g) program, albeit in an unofficial manner, then the force multiplier effect of the program could be considerably larger than even advocates of the program estimate. Similarly, the instances of racial profiling that can be attributed to the program (even if indirectly) could also be considerably higher than what critics charge is being committed by signatory LEAs. Because of the civil rights implications, a higher incidence of racial profiling attributable to the 287(g) program is particularly troubling in any assessment of the program's efficacy.

Our observed spillover effects also raise important questions about the political controls that are supposed to exist in the 287(g) program. As detailed in Part I, the program was created by Congress with a formalized structure: LEAs that want to participate in the program must apply, be approved by federal authorities for participation, and then sign a written, public agreement that details their immigration-related duties and their federally imposed obligations under the program. And as we saw with the Maricopa County Sheriff's Office and the Alamance County Sheriff's office, the agreements can be revoked by the federal government when the signatory LEA violates their provisions. But with the spillover effect that we observed, nearby nonsignatory LEAs can join this federal mission of immigration enforcement without having to go through any formal screening process or worry about revocation of that participation. To be sure, the nonsignatory participation that we observed is far less extensive than even the most basic form of 287(g) model offered; yet even this limited participation is nowhere accounted for in the 287(g) program's formal operation. Instead of focusing solely on signatory LEAs and whether they sign or terminate agreements, opponents and proponents of the 287(g) program would do well to consider these possible spillover effects and to focus attention on the actions of nearby nonsignatory LEAs.

The leadership of nonsignatory LEAs also needs to be aware of possible spillover effects. A nonsignatory LEA may choose not to participate in the 287(g) program for various reasons: disagreement with the mission of local-- federal immigration enforcement, desire to remain neutral on a very divisive issue, desire to preserve trust with their immigrant communities, or concerns about the fiscal and other costs of immigration enforcement. But if the LEA's officers are informally participating in immigration law enforcement, those reasons for not signing may be undermined. For example, an LEA may have decided against signing a 287(g) agreement so that immigrants do not associate the LEA with immigration enforcement, thus protecting its relationship with local immigrant communities. But that message of community trust could be undermined if its officers are participating, even informally, in immigration enforcement. Furthermore, the participation by officers may also be reallocating the LEA's resources in terms of diverted officer time without approval or oversight by LEA supervisors, arguably weakening the LEA's ability to control its own resources. This discretionary participation, without the benefit of training or regulation, has important implications for the LEA's effectiveness and ultimately for public safety, implications that the Biden Administration must consider as it formulates its 287(g) policies.

Huyen Pham, Professor of Law and Associate Dean for Faculty Research & Development, Texas A&M University School of Law (J.D. Harvard Law School, A.B. Harvard College).

Pham Hoang Van, Professor of Economics, Hankamer School of Business, Baylor University (PhD, Economics, Cornell; SM, SB, Mechanical Engineering, MIT).