The Complex Sources of Immigration Control

Abstract

All governments enforce immigration laws, but we have a limited understanding of the factors that determine how much they do so. Immigration policymakers can empower or compel officials and non-state actors to enforce immigration laws. This study suggests that non-immigration policymakers may play an equally important role — albeit in complex ways. Policies designed to control human mobility in other ways (transportation, crime control, segregation, etc.) can determine the amount of resources a given country devotes to immigration enforcement and the effectiveness of enforcement efforts. When officials ban old — or invent new — forms of movement control, they can determine how much resources are available to enforce immigration laws, whether these resources are used for immigration enforcement, and how effective they are. The study demonstrates this potential power of non-immigration policy through an in-depth analysis of South Africa’s unexpected capacity to control migration. South Africa is one of the world’s most prolific deporters of foreign nationals but not a very strong state. The study shows how decisions to ban segregation and invent new policing methods impacted the South African state’s capacity to deport foreign nationals. Using these findings, the article calls for more detailed research into the complex relationships between immigration enforcement and other movement control policies and for greater attention to heretofore neglected cases in the developing world.

Introduction

Immigration enforcement has resurfaced as a prominent political issue on both sides of the Atlantic. In the United States, President Trump won office in 2016 on a promise to round up and deport foreigners. In Europe, populist politicians like Geert Wilders, Marie Le Pen, and Viktor Orbán have been gaining traction with similar pledges. These developments threaten many migrants’ futures and place theories of immigration control into stark relief. To be more specific, we do not know whether these politicians’ promises will come to fruition because we have a limited understanding of the factors that determine how much states enforce immigration laws. This article seeks to address this gap in our understanding by studying a case that sits outside the geographic focus of previous research: South Africa.

Between 2000 and 2010, South Africa was a prolific deporter of foreign nationals, despite the fact that its state was relatively weak. ¹ I argue that non-immigration movement-control policies account for South Africa’s immigration enforcement capacity, albeit in indirect, complex ways. Using a multi-method and inductive research strategy, I explore the complex linkages between, on the one hand, changes in segregation and policing policy and, on the other, observed changes in South African immigration enforcement outcomes. I demonstrate that, by banning and inventing alternate forms of movement control, South African non-immigration policymakers created resources that could be used for immigration enforcement, determined whether these resources would be used for immigration enforcement, and influenced how effective these resources would be. The upshot of this argument is that variation in immigration enforcement capacity needs to be understood in relation to complex relationships between multiple movement-control policies, past and present, in both traditional and emerging centers of migration research.

While based on findings from a single national context, the study suggests the need for two developments in the way we study the politics of international migration. First, we ought to be more attentive to the complex sources of migration control. Researchers have struggled to understand variation in immigration enforcement outcomes because they have tended to examine immigration enforcement practices as if they were either the mere “implementation” of immigration policy and law (e.g., Czaika and De Haas 2013) or reflections of emergent forms of discrimination and oppression (e.g., De Genova 2010). This study suggests that we need to further explore how states’ capacity to enforce immigration laws is itself generated by complex relationships between multiple movement-control policies and practices.

Second, we need to broaden the geographical horizons of our research. Migration scholars have struggled to understand variation in immigration enforcement capacity because of an exclusive concern with a small number of European, North American, and other Organisation for Economic Co-operation and Development (OECD) destination countries. ² As a result, the literature has been inattentive to both the historical particularity of the small number of cases studied and the variation in immigration enforcement outcomes and the causes thereof. The article concludes by using the findings of this South African case study to indicate how a globally comparative research agenda on immigration enforcement outcomes might be more fruitfully pursued.

The Complex Sources of Immigration Enforcement Capacity

This study seeks to explain “immigration enforcement capacity.” Immigration enforcement involves separating out unauthorized entrants and residents from the population, taking these persons into custody, and forcibly removing them from the state’s jurisdiction. Immigration enforcement capacity is the amount of immigration enforcement that takes place and the amount of people these practices move. There are at least two reasons why it is important to know to what extent immigration laws are actually enforced. First, arrests, detentions, and deportations constitute major disruptions in migrants’ lives and are often physically violent, economically disastrous, and emotionally traumatic experiences. Therefore, those who are interested in migrants’ well-being ought to understand what determines how much of this activity will take place. Second, volumes of immigration enforcement indicate how much a state can control immigration. Thus, those scholars interested in why some states are better able to control immigration than others need to be able to explain why — alongside the use of visa restrictions, transport carrier sanctions, agreements with “buffer” states, and border defenses — some states are more able to locate, arrest, detain, and deport unauthorized entrants and residents than others. It is important to stress here that “immigration enforcement capacity” is not the same as “immigration control capacity.” States that deport high numbers of foreign nationals are not, ipso facto, more “in control” of international migration than those that do not, and vice versa.

One determinant of the extent to which immigration enforcement occurs in a given country is the capacity of its immigration agency to train and equip officials to enforce the law. However, in most countries, immigration officials and immigration departments form a small part of a much broader immigration enforcement infrastructure (Lahav 1998; Wishnie 2004; Scholten and Minderhoud 2008; Coleman and Kocher 2011; Leerkes, Varsanyi, and Engbersen 2012). One of the earliest formulations of this idea was Aristide Zolberg’s (2005) concept of “remote control,” which refers to the manner in which governments have extended both the prerogative and responsibility to enforce immigration laws to other bureaucratic agents within the state, to officials of foreign states, and to non-state actors. While the logic of “remote control” is suggestive of a complex social relation, research has tended to emphasize more linear and uni-directional associations between policy and practice: where immigration policymakers (the principal) influence the behavior of a range of immigration policy implementers (the agents). For example, research on the United States and the Netherlands has specifically studied how immigration policymakers achieve remote control by (1) empowering non-immigration officials with discretionary authority to enforce immigration laws and/or (2) limiting their discretionary authority to not enforce immigration laws (Engbersen, Guiraudon, and Joppke 2001; Coleman 2007, 2009, 2012; Leerkes, Varsanyi, and Engbersen 2012; Van der Leun and Bouter 2015).

This understanding may be specifically contrasted with research that emphasizes the less linear, multi-directional — and hence “complex” — linkages between immigration enforcement practices and other governmental “functions.” For example, some work suggests that immigration enforcement practices form part of consistently evolving assemblages (DeLanda 2006) involving relationships between a diverse set of human actors and material infrastructures (Amoore 2006; Salter 2013). Thus, immigration enforcement resources can be consistently repurposed for different ends. The empirical portraits developed by this work have been particularly useful in detecting the complex set of relationships that determine whether immigration enforcement occurs in given — and usually highly localized — contexts (e.g., Walters 2006). While not all fitting neatly within the mantle of assemblage theory, various authors emphasize how capitalist imperatives (Heyman 2004; De Genova 2010), the interests of public-private rent-seeking coalitions (Golash-Boza 2009a, 2009b; Doty and Wheatley 2013; Gammeltoft-Hansen and Sorensen 2013), and identity (Bakewell 2008; Hyndman and Giles 2011) each help decide whether specific collections of actors and material infrastructures will contribute to immigration enforcement efforts. However, this literature has yet to devote much attention to the question of what makes immigration enforcement more voluminous or effective in one state than in another.

This article demonstrates that the complex relationships which assemblage theory has revealed in practice can powerfully enable non-immigration policymakers to shape immigration enforcement outcomes as well. There are three dimensions to this argument. Movement-control policies that are not formally concerned with immigration can influence (1) how much collective resources are available for immigration enforcement, (2) whether these resources are used for immigration enforcement, and (3) how effective enforcement practices are. The remainder of this section details each of these points.

Non-Immigration Policies Create Movement-Control Capacity

As mentioned above, enforcing immigration laws requires specific types of human and material resources. More specifically, enforcement demands resources for finding people and then moving them around. My typology, drafted in the spirit of assemblage theory, includes four main resource types that have both a human and a material component. First, “locators” are people who are capable of differentiating between unauthorized and authorized residents and the identifying documents and procedures that help them make these distinctions. Second, “enforcers” are people who are physically capable of and trained to arrest suspected unauthorized residents and the uniforms, vehicles, and arms required to make successful arrests. Third, “guards” are people who are capable of maintaining detention facilities and the prisons required to maintain custody. Fourth, “transporters” are people who are capable of moving unauthorized residents domestically and internationally and the cars, planes, and supporting transport infrastructure required to get them there.

Different states have built up different sets of locators, enforcers, guards, and transporters with varying purposes in mind (Vigneswaran and Quirk 2015). Some important contemporary examples include (1) criminal justice, which proscribes the arrest and incarceration of convicted offenders (e.g., Feeley and Simon 1992); (2) property law, which sanctions the removal of trespassers and squatters (e.g., Blomley 2003); and (3) transport policy, which excludes non-paying passengers from access to certain routes and vehicles (e.g., Lucas 2012). Other major forms of movement-control policy include slavery, serfdom, conquest, disease control, banishment, segregation, apartheid, the caste system, hukou, and public order policing. Many of these forms of movement control have been specifically outlawed or banned by policymakers, indicating that policymakers have been able not only to create but also to destroy movement-control resources. In general, non-immigration policymakers have the power to determine the total pool of potential resources, which might be devoted to immigration enforcement and thereby determine the outer limits on how much immigration enforcement can be done.

Changes in Non-Immigration Policies Re-Deploy Movement-Control Capacity

What determines whether a country’s total amount of movement-control resources end up being used for immigration enforcement? Many immigration policymakers may want to command all movement-control resources available in their jurisdiction but consistently struggle to marshal these resources to their chosen ends. Researchers have often drawn on Michael Lipsky’s ideas regarding the discretionary autonomy of street-level officials to explain why this is the case (Lipsky 2010; see also Bouchard and Carroll 2002; Engbersen and Broeders 2011; Alpes and Spire 2014; Côté-Boucher, Infantino, and Salter 2014). Lipsky’s work helps us understand how officials deploy personal and institutional resources. He argues that where there are many frameworks for legitimating official behavior, individual officials have a greater range of options and therefore discretion to choose which laws they enforce. For example, a border guard charged with responsibility to enforce immigration, customs, security, and public health laws will have greater authority to decide whether she enforces immigration laws than a border guard charged with sole responsibility for enforcing immigration law.

Non-immigration policymakers can either combat or compound this problem. Non-immigration policymakers may specifically choose to focus resources on immigration policing. However, they can also change immigration enforcement outcomes in less direct ways. More specifically, when policymakers “ban” other forms of movement control, they can reduce movement controllers’ discretionary authority to choose to devote their energies to non-immigration tasks, encouraging them to focus their energies on immigration enforcement instead. Vice versa, when policymakers “invent” new forms of movement control, they can enhance movement controllers’ discretion to choose whether they enforce non-immigration movement-control laws and thereby increase the likelihood that movement controllers will neglect immigration enforcement responsibilities.

The Legacy of Other Movement Controls Determines the Effectiveness of Immigration Enforcement

What determines how effective these efforts will be? Here, I draw on a specific vein of work in the study of policing, which suggests that enforcement efficacy rests on the practiced compliance of the governed (e.g., Bittner 1970). If every member of a population — undocumented or otherwise — were to obstruct efforts to (1) identify him/her, (2) channel him/her through ports of entry, (3) inspect his/her property, and/or (4) keep him/her in state custody, then immigration enforcement would be an exceedingly, and quite possibly prohibitively, time-consuming and labor-intensive task. Perhaps more importantly, enforcement would require the consistent deployment of brute force, as compared with the communicative and non-physical manner in which coercive potential is more commonly and efficaciously conveyed (Burawoy 1982; Levi 1989). Building on this, we should expect the effectiveness of efforts to exclude undocumented foreigners to increase when the broader population, and not just migrants, tends to comply with official efforts to identify, arrest, detain, and move them around.

Non-immigration policymakers may also help determine whether civilians comply with immigration enforcement practices. Our understanding of this dimension of immigration enforcement complexity is rudimentary, so my comments here are more tentative. I draw on Louise Shelley’s (1999) work on policing, which emphasizes the manner in which more general processes of regime change in authoritarian states have encouraged non-compliance with a variety of forms of policing. Her study of post-Soviet policing notes how perestroika helped unravel civilian traditions of unthinking complicity with policing under Soviet rule but failed to replace them with democratic notions of police legitimacy. Her work, however, does not specifically address how changes in movement-control laws might impact other forms of movement control. This connection is important to consider because it is equally possible that ingrained patterns of complying with one set of movement-control laws will be easily transferred from one form of movement control to another, regardless of changes in the law. We simply do not know how or whether decisions to ban or invent forms of movement control impact patterns of compliance with immigration enforcement. We only know that it seems plausible to expect these sorts of decisions to have implications for the efficacy of immigration enforcement practices and that we ought to seek to understand more precisely what these implications are in a given context.

To summarize, much previous research has assumed that immigration policymakers are the primary actors developing the capacity to enforce immigration laws. By contrast, an approach attentive to complexity emphasizes the way they “get a little help from their friends.” Non-immigration policymakers develop resources that are suitable for immigration enforcement purposes, can help channel these resources toward or away from immigration enforcement, and influence their ultimate impact. The point here is not that immigration policy does not work and should not be studied but that immigration policies are not necessarily the only or primary origin or determinant of a given state’s capacity to enforce immigration laws.

Weak State/Tough Territory: The South African Puzzle

The remainder of this article outlines a set of empirical findings from South Africa that develop this line of argument. Hence, it is important to offer some contextual clarification about the case. During the Apartheid era, South Africa promoted immigration from historically “white” and European countries, while strictly limiting black immigration from surrounding states (Peberdy 2000; Klotz 2013). After the transition to democratic rule and a principle of non-racialism, the African National Congress government de-racialized its immigration policy. In this context, South Africa became a major destination for international migrants. In 1995, its migrant stock of 1.16 million composed 3.2 percent of its population. ³ By 2015, those numbers had risen to 3.14 million and 5.8 percent, respectively. ⁴ South Africa now hosts more migrants than any other African country and approximately 15 percent of the continent’s total migrant stock. ^5,6

The post-Apartheid period not only saw more migrants coming to South Africa but also witnessed the growth of a virulent nationalist agenda and widespread xenophobic violence against foreign nationals (Landau 2011). In this context, policymakers limited the options for African migrants to enter and work in South Africa legally, and large numbers entered informally instead (Crush 1999). Many of these migrants took up residence and employment in the urban and peri-urban areas of the country’s primary economic hub in Gauteng province. In the early 2000s, the neighborhood of Hillbrow, Johannesburg, became infamous as a site where these migrants began to congregate in significant numbers.

During this same post-Apartheid period, South Africa became a prolific deporter of foreign nationals, particularly in its border regions. However, significant numbers have also been arrested in the country’s interior, especially Gauteng (Vigneswaran et al. 2010). After arrest, migrants are usually detained in police stations before being sent to the Lindela detention center in Gauteng, where they await deportation to their origin countries (Sutton and Vigneswaran 2011). In the 2000s, South Africa consistently deported upwards of 100,000 people per annum (Figure 1); and with the exception of 2009, when the country passed a moratorium on deportations in the wake of the 2008 xenophobic attacks, it deported more people per capita than several OECD states (Figure 2). ⁷

Figure 1.

Annual Deportations: South Africa, 1990-2014.

Figure 2.

Deportations as a Percentage of Total Populations: Selected Countries, 2000-2010.

South Africa’s comparatively high number of deportations is a puzzling observation for at least two reasons. First, South Africa is, in general, not a very strong state. The World Bank’s Governance Indicators ranked South Africa at the 76th percentile for government effectiveness in 2000. It had risen to the 66th percentile by 2010, ranking the country alongside other high-performing African states like Botswana (67th) and the Seychelles (62nd) but significantly below most OECD countries that were deporting foreign nationals at lower rates. ⁸ Furthermore, many South African public officials are corrupt, and migrants routinely evade deportation by paying bribes to police officers and immigration officials (Klaaren and Ramji 2001). How, then, did such a comparatively “weak” state produce such high deportation numbers?

South Africa’s deportation numbers are made all the more puzzling by the fact that the country was, on paper, liberalizing during the post-Apartheid period (Klotz 2013). After the passage of a new constitution in 1995, South Africa introduced a wide range of laws to ensure democratic accountability, rule of law, and the protection of human rights. These sorts of provisions were evident in the changes taking place in the country’s immigration laws, with the repeal of the Aliens Control Act (n. 96 of 1991) and its replacement by both the Refugees Act (n. 130 of 1998) and the Immigration Act (n. 13 of 2002). Each new law limited officials’ powers to arrest, detain, and deport foreign nationals and provided unauthorized residents with a variety of potential forms of protection, review, appeal, and guarantees of administrative justice. Again, between 2000 and 2010, South Africa did not establish the ingrained traditions of liberalism seen in many Northern states, but its immigration policy and law appeared to be becoming more liberal. Yet more liberal states are supposed to be less capable of deporting foreign nationals (Joppke 1998). How, in the context of a seemingly weak and liberalizing state, did South Africa increase its capacity to enforce immigration laws?

Methods

This study attempted to both explore the merits of a conventional, linear explanation of these outcomes and inductively pursue alternate lines of inquiry. The project took place across two phases. The first phase involved a survey of publicly available material on South African immigration policy, including parliamentary debates, departmental annual reports, draft and final legislation, ministry press releases, nongovernmental organization (NGO) reports, and newspaper articles. To plug holes in the documentary record and help interpret how and why key decisions were made and initiatives adopted, I conducted approximately 30 semi-structured, key-informant interviews with senior officials within the Department of Home Affairs (DHA) and the South African Police Service (SAPS), members of the Parliamentary Committee on Home Affairs, NGO workers, and local academics between 2005 and 2012. ⁹ These interviews covered the reasons South Africa introduced a new immigration bill, the process through which the law was revised and passed, and the manner in which its enforcement provisions were implemented.

The first phase of the research suggested the need for more finely grained material on immigration enforcement practice, specifically covering how the SAPS helped immigration officials with their immigration enforcement work. To understand these processes, the second phase deployed two additional data collection techniques. The first technique was attentive to the everyday decision-making processes of street-level officials and involved a multi-sited, team-implemented ethnography of policing at two border posts (Kompatipoort and Beitbridge), the Lindela detention center, and six station precincts (Alexandra, Brixton, Diepsloot, Hillbrow, Johannesburg Central, and Thembisa) in Gauteng, the province where the majority of Lindela detainees were first arrested.

While this sample of sites was not aerially representative of either Gauteng or South Africa as a whole, the method approached “holism” and “representativity” from a different angle. Here, the aim was to identify why the rather occasional and sporadic incidents of immigration policing occurred in the context of a much wider set of formal policing practices and informal policing habits and routines. Participant observation was ideally suited to develop this deeply contextualized understanding of policing practice (Madsen 2004; Steinberg 2008b; Hornberger 2011). The research was primarily observational and involved spending time in police stations, border posts, and detention facilities, accompanying police officers on patrol and on operations on city streets and along the border, observing officials’ interactions and encounters with civilians and other government officials, and occasionally conducting informal and more extended open-ended interviews about policing strategies and officials’ broader understandings of their work. This technique helped me better understand why the police consistently made use of their discretionary authority to enforce immigration laws. In total, the team spent approximately 1,200 hours in the field at these various sites. I did the observational work in Brixton and Hillbrow, while research assistants covered the other locations.

The second approach was more attentive to the potential significance of historical context. Extending the study’s chronological parameters, I explored policy documents and archival sources back to the period in which the first substantial increases in deportation numbers were reported in DHA Annual Reports (1990–1995). This study included departmental reports, parliamentary debates, minutes of Cabinet meetings, and newspaper articles. I selected these documents to understand how the SAPS as an institution had become involved in policing immigration and how responsibilities for immigration enforcement had been distributed between the SAPS and DHA. I also utilized the request mechanisms prescribed by the Promotion of Access to Information Act (n. 2 of 2000) to acquire a more representative sample of official documents from the DHA and SAPS. In doing so, I sought to detect whether there were any previous changes in policy or practice that made outcomes in the post-Apartheid period more or less likely.

The Complex Sources of Immigration Enforcement Capacity in South Africa

The Failure of Immigration Policy ¹⁰

The policy documents and interviews reviewed in this study revealed the weaknesses of a conventional and linear explanation of immigration enforcement capacity. South African immigration policymakers had sought to generate hierarchical control over immigration enforcement but failed. In the late 1990s and early 2000s, under Inkatha Freedom Party leader, Mangosuthu Buthelezi, South Africa’s Ministry of Home Affairs sought to rewrite the country’s immigration legislation. The new approach would shift “administrative and policy emphasis from border control to community and workplace inspection” (DHA 1999). This shift involved exerting direct authority over the implementation of immigration enforcement (ibid.). The Buthelezi Ministry put forward motions to reassign the DHA’s non-enforcement responsibilities to municipal governments and transform the DHA’s Migration Directorate into a regulatory agency governed by a relatively autonomous Immigration Board. ^11,12 Buthelezi also sought to establish an immigration police force under the Migration Directorate, ¹³ increase DHA capacity to administer the enforcement mechanisms of other State Departments, place the DHA in charge of interdepartmental border control, and create a series of immigration courts with exclusive jurisdiction over immigration cases. ¹⁴ These policies appeared — on the surface — to have achieved impressive impacts on the country’s deportation figures. While deportation numbers had been steadily rising since the 1990s, they began to increase markedly after the 2005 implementation of Buthelezi’s immigration enforcement overhaul of the Immigration Act (n. 13 of 2002).

Despite these surface-level indicators of “success” through rising deportation numbers, Buthelezi’s policy reform was a total failure. None of the relevant proposals put forward by the Buthelezi Ministry ended up in the final Immigration Act (n. 13 of 2002), with each being scuppered by various competing agencies and political parties along the way. More significantly, when asked about the manner in which South Africa was able to achieve its high deportation rates despite these failures, policy elites within the DHA were unable to explain how such dramatic outcomes were being achieved. ¹⁵

Senior officials’ inability to account for the increase in deportations pointed to a series of organizational problems within the DHA that were only gradually becoming public knowledge at the time of the initial elite interviews (2005–2007). Over the course of the period during which these reforms were to have been introduced (2005–2010), the capacity of DHA policy elites to command their own officials was in rapid decline. Ordinary DHA officials were failing to perform their most basic functions, consistently accepting bribes to secure the release of undocumented migrants or provide them with fake documents, and flouting a wide variety of procedures in their administration of detention and deportation processes (Amit 2010; Segatti 2011). Put simply, the DHA was falling apart. This reality made any suggestion that the senior officials in the Buthelezi Ministry had been able to compel its lower-level agents to implement its new internal enforcement policy — through legislation or fiat — seem even more implausible. So, in the absence of effective hierarchy, what explained South Africa’s high deportation rates?

A small amount of source materials from interviews suggested that a different state agency might have played an independent and impactful role in immigration enforcement processes: the police. For example, one senior DHA official noted that his staff rarely determined when, where, or how they worked. Instead, the SAPS ran such a high volume of operations in which they called on the DHA to confirm immigration detainees’ status that the DHA had little time and resources to develop and deploy its own enforcement plans:

The SAPS would hold various operations and arrest large numbers of people as suspected illegal aliens. Those people would be handed over to the department, to these immigration officials. And because the numbers coming from the police were quite high, and the number of immigration officers was quite low, this almost became the prime task. ¹⁶

The DHA was not the agency responsible for generating large amounts of immigration enforcement practices, but was it the unwitting beneficiary of an enforcement dynamic grounded in the practices of the SAPS?

Apartheid Segregation Policies Generate Immigration Enforcement Capacity

A pair of other studies taking place concurrently with this project, but directed toward other ends, lent credence to the idea that policing policy and practice held clues to South Africa’s high deportation rates. On the one hand, a survey of police officials across six stations in Gauteng province asked how officials spent their time during a typical shift (n = 464) (Vigneswaran and Duponchel 2009). Results suggested that beat-level police officials were spending approximately one-quarter of an average shift locating, arresting, detaining, and processing undocumented foreigners. A survey with detainees at the Lindela detention center (n = 444) suggested that these efforts accounted for large numbers of arrests. ¹⁷ Large proportions of detainees reported having been arrested by a police official, and almost all spent time in police detention before being sent to the DHA-run, privately operated detention facility (Amit 2010). Given the fact that SAPS officials significantly outnumbered their DHA counterparts, this finding lent support to the idea that police officials were generating a large proportion of the arrests and detentions that ultimately led to deportations.

On the other hand, a review of policy documents and interviews with elite SAPS officials disconfirmed a modified version of the conventional and linear explanation: that SAPS’ senior officials had taken over responsibility for immigration policy and ordered police officers to make immigration arrests. Instead, SAPS management at station, provincial, and national levels had been repeatedly trying to communicate to their officers that they should not spend their time enforcing immigration laws but instead policing “real crime.” ¹⁸ Police involvement in immigration enforcement directly contradicted orders from above.

Ethnographic research into immigration enforcement practice provided the kernels of an alternate explanation. This work suggested that street-level police officials in South Africa were ideologically and behaviorally predisposed to immigration enforcement work. Police officers had a peculiar way of deciding how to police space (cf. Herbert 1997). The act of defining whether an individual was in the right place constituted the primary mode of inquiry through which South African street-level police officers determined how to act toward civilians. The principal focus of their policing efforts was, thus, human bodies. The main questions that they posed about those bodies pertained to characteristics — race, ethnicity, and nationality — that might identify a body as belonging to a particular place and, by extension, not belonging to others. These identifying characteristics then suggested a particular role for police intervention, depending on where the person was situated and whether that was a place to which he or she “belonged.”

The most familiar example of this form of interpretative work involved racial and spatial profiling. In neighborhoods where white and Indian populations constituted the majority of home owners, police officers commonly presumed that black pedestrians had a criminal intent for being there and would observe and often stop and interrogate them (Steinberg 2008a; Diphoorn 2015). However, this territorial style of policing cannot be reduced to this most obvious example of racially profiling black people as criminals. The same interpretative scheme could be deployed in other ways. For example, police would also formulate their response to the pleas of white civilians for assistance by first asking themselves whether that civilian was located in an appropriate neighborhood. In my own ethnographic research on the demographically black African neighborhoods of Hillbrow and Berea, this practice meant that the constabulary could justify deferring or avoiding interventions on behalf of white civilians because they were not really responsible for providing these individuals with protection or service in areas with a majority-black population (Vigneswaran 2015). This spatial mode of inquiry was not merely a way of using racial characteristics to impute criminal intent but a deeply ingrained and habitual thought process whereby police officers formulated their assessments of appropriate policing behavior by correlating the phenotypical characteristics of both potential victims and perpetrators with a given place’s known demographic characteristics.

This policing interpretative frame was not something that existed only inside police officers’ heads but, instead, was powerfully enabled by the sorts of operations that police most commonly conducted (Autesserre 2009). To simplify, our observations of operational policing suggested that South African police “arrest first and ask questions later.” Their everyday work involved a high volume of raids, roadblocks, cordon-and-search operations, and stop-and-frisk operations. Most of the time, officers had a relatively vague understanding of the specific types of criminal behavior, enterprise, or suspect they hoped to uncover by using such means in an individual case. Instead, these strategies were initiated with the working knowledge that they would result in large numbers of apprehensions of people who could subsequently be searched for possessions of firearms and drugs and, more importantly for my purposes, interrogated about whether they were “in the right place.” A common way of initiating such an encounter at a roadblock or during a stop-and-frisk operation was the Afrikaans term “waar slaap je?” (where do you live/stay?) or English term “papers,” immediately setting the tone and content of the encounter to follow.

When taken together, police officers’ spatial mode of interpreting the world around them and apprehension-oriented working habits helped us understand why they were producing such high numbers of arrests leading to deportations. They targeted outsiders and captured populations in a manner analogous to a mobile border post. Across several sites in Johannesburg and Pretoria, our ethnographers witnessed how these practices could generate large numbers of immigration arrests (Tshabalala 2010).

Yet these findings did not fully explain why South Africa generated so many deportations from 2000 to 2010. Rather, they pointed to a potentially important causal chain. The practice of policing mobility was ingrained in the way the SAPS investigated and responded to crime. The bigger question that the observational study posed, yet could not solve, was, What factors had made South African police work in this way, and why was this skill set being deployed in the service of immigration laws?

The parallel archival research revealed answers to these questions: rescinded Apartheid-era segregation laws had created the resources for contemporary immigration enforcement. The South African police had developed their unique working culture under the aegis of an altogether different regime: influx control. The key legislation here was the Group Areas Act (n. 41 of 1950), which divided South African jurisdiction into a series of discrete racial areas and set out a series of measures — known as pass laws — for policing movement between separate residential spaces (Posel 1991). Part of the issue here is that the SAPS was the primary agency that enforced the Group Areas Act. Hence, policing internal movement, asking people to show their pass and arresting those without adequate authorization, were behaviors that the police were accustomed to. Given that the numbers of pass law removals often amounted to hundreds of thousands on an annual basis, it is not surprising that this organization was able to achieve similar outcomes in the internal enforcement of immigration laws.

More substantively, maintaining a segregated South Africa had become fundamental to the manner in which the SAPS defined crime as a problem and set about ameliorating it (Brewer 1994). To simplify, police officers were taught to see the presence of black people in white areas or white people in black areas as the primary causes of criminal behavior. Apartheid South Africa framed criminality as a problem stemming from over-population, the presence of unauthorized residents, and the proliferation of informal residences (ibid.). South Africa’s dragnet approach to policing was born during this era, when unannounced and blanket raids of residential areas, stop-and-search operations, and road blocks were conceptualized not simply as means of enforcing segregation but as means of stemming criminality and suppressing political unrest. In essence, policing in South Africa became largely devoted to policing racial mobility, which was the basic response to all evidence of increasing levels of crime and, eventually, all threats to the regime (ibid.).

Crucially, this line of interpretation explains, not simply continuity, but also change: the rise in South Africa’s deportation numbers in the post-Apartheid era. The crucial date to note here is 1986, when the National Party abandoned its influx control laws. In his Presidential Address of that year, P.W. Botha announced, “The present system is too costly and has become obsolete.” ¹⁹ With the stroke of a pen, Apartheid’s defining law, the Group Areas Act (n. 41 of 1950), was gone. The reasons Botha made this decision are not significant for the argument advanced here. The decision’s consequences are far more important. More specifically, what effects did the abandonment of influx control have on other forms of movement control, especially the enforcement of immigration laws?

What appears to be beyond dispute is that the abandonment of influx control laws preceded an increased public awareness of the amount of work officials were devoting to immigration enforcement. It was only after influx control laws were dropped in 1986 that the number of people being deported under the terms of the then Aliens Control Act (n. 30 of 1963) became an object of Parliamentary discussion (Vigneswaran 2011). What is also clear is that officials within both the South African Police Force and the precursor to the DHA – the Department of Internal Affairs - were responsible for generating these arrests. ²⁰ It is more difficult to isolate the precise reason behind these new arrests. The line of reasoning most strongly supported by the archival data is that banning the Group Areas Act narrowed the range of legitimate forms of movement control. South African street-level officials had been policing the Group Areas Act and immigration laws simultaneously for many decades, drawing little distinction between the populations concerned and purposes of each. When one legislative system was suddenly rescinded, the police continued working as they always had, with significant impacts on the outcomes recorded under the one legislation that remained.

This explanation seems to be at least partly validated by the fact that the question of whether immigration control policy needed to be reformed only really began to be discussed in elite policy circles after the increase in deportations. ²¹ It would appear that in this context, Parliamentarians and legislators began to look for a post hoc rationalization for the enforcement work already taking place on the ground. The fact that immigration policymakers could now legitimate the increasing number of deportations was important, but this is not the same as saying that immigration policy changes caused the increase. Thus, it is likely that when the government committed itself to an anti-immigration policy with the passage of a revised Aliens Control Act in 1991 (n. 96), street-level officials found themselves with greater resources and license to enforce immigration laws, and their actions contributed to the continued increase in deportation rates. However, the new Aliens Control Act (n. 96 of 1991) by no means “invented” a movement-control system. It was merely a compilation of a variety of immigration control measures that, to this point, had been distributed across a number of separate legislative and regulatory mechanisms (Peberdy and Crush 1998; Peberdy 2000). It was the more targeted discretionary behavior of everyday officials that was doing the work and producing the numbers.

This historical account provides the rudiments of an alternate explanation for South Africa’s immigration enforcement capacity. In essence, the ideological outlook and operational posture of South Africa’s security structures, particularly the police, had been designed to simultaneously enforce at least two movement-control regimes: influx and immigration controls, with a significant emphasis on the former. When policymakers abandoned influx controls in 1986, the police began to focus their resources on the one that remained — immigration enforcement — and to great effect. This finding lends support to the argument advanced above: that one of the ways in which policymakers might alter deportation outcomes is by banning other movement-control regimes.

COMPSTAT Policing Replaces Immigration Enforcement in Hillbrow

The account so far is somewhat mired in the specificities of its case (Klotz 2013). South Africa could be seen as an anomaly, possessing high deportation rates because of its rather odd history. A further round of research, however, strengthened the argument in favor of complexity over mere context specificity. During the ethnographic study, I worked in a neighborhood in which — contrary to all expectations — the police were not doing immigration policing: Hillbrow, Johannesburg.

In the 2000s, Hillbrow had a very high concentration of undocumented migrants and a public reputation for high volumes of immigration enforcement (Vigneswaran et al. 2010). Hence, during the ethnographic study, we expected to observe the Hillbrow police doing a lot of immigration policing. Contrary to these expectations, during many months of observing both formal police work and the policing activities of local civilian street patrollers, I did not witness a single arrest of an unauthorized migrant or a police officer asking to see someone’s papers. When I asked street-level police officials to reflect on this observation, they simply referred me to the orders issued by their superiors: that they should focus their resources on “real criminals” and not “illegal foreigners.” Yet in all other sites that the team was studying, beat-level police officers continued to routinely enforce immigration laws in defiance of such orders. Why were the instructions being followed in Hillbrow and not elsewhere?

One potential line of explanation began to emerge from my ethnographic observations of how ordinary officials were deploying movement-control resources in Hillbrow. More specifically, the police were using crime maps to decide how they should control people’s movement. Instead of asking how an individual’s phenotypical appearance correlated with an area’s demographic composition, police officers asked how their identity correlated with the criminal patterns associated with a given space and whether this made a civilian a potential victim requiring police assistance or protection or a potential perpetrator requiring surveillance or removal. At the same time, crime maps provided police with a new method of targeting their “arrest-first” interventions, locating “bad buildings” for raids, deciding where to situate their road blocks to increase seizure yields, and determining whom to subject to body searches. Street-level officials saw these techniques as powerful means of changing crime levels by shaping patterns of human mobility. Raids were seen as means of pushing criminal syndicates out of the neighborhood. Targeted roadblocks were used to get higher numbers of criminals off the streets. Stop-and-frisk operations were means of protecting armed civilians from entering zones of violent conflict. Crucially, when Hillbrow police officers used the same operational techniques that had been previously deployed to police both the Group Areas Act and Immigration Act for the purpose of crime control, a person’s right to be in a given place was a less useful indicator of whether he or she would be interrogated/arrested by the police than the likelihood of his or her participation — as perpetrator or victim — in a criminal act. Thus, civilians in Hillbrow were more likely to be interrogated about their intentions and motives than about their origins and destinations. Asking people for their documents was a less helpful means of reaching conclusions about their motives; thus, immigration arrests became a less common outcome of everyday policing.

While these ethnographic observations could help us describe how the South African police had redeployed movement-control resources in Hillbrow, we required additional empirical resources to explain why these patterns were observed in Hillbrow and not elsewhere. Policy documents and interviews provided a plausible line of interpretation. Crucially, rather than once again leading back to historically and locally contingent factors, in this case there were stronger connections with more general and global processes of policy diffusion and change. The crucial observation was that the Hillbrow precinct had been chosen as the focal point for South Africa’s efforts to “modernize” its policing strategy. In 2000, Hillbrow was selected as the site to pilot a new national crime prevention strategy because it was one of the precincts reporting the highest concentrations of crime (Pelser 2000; Shaw 2002; Burger 2007). The central tenet of this approach was that crime maps should define how the police distributed their operational resources across space.

This was not a specifically South African trend, but one that had been inspired by the Compare Statistics (COMPSTAT) policing philosophy (Vigneswaran 2014). COMPSTAT promotes the use of crime reporting and police performance data to plan policing interventions in a more responsive and effective manner. For our purposes, COMPSTAT’s key element is its focus on policing mobility and in particular on ensuring that places reporting high levels of criminal activity are heavily policed. The strategy was developed in the United States and piloted extensively in New York, where policing problem places became regarded as the progressive way to both respond to and reduce crime trends (Braga and Weisburd 2010). While in the United States, this policing approach has been strongly associated with the move toward the increased incarceration of criminal offenders (Feeley and Simon 1992), it appeared that when married with South Africa’s existing movement-control infrastructure and actors, it absorbed policing resources that may have otherwise been devoted to the arrest and detention of unauthorized immigrants.

This evidence of the Hillbrow police effectively rolling out their policing strategy within their own organization, and thereby reducing their capacity to make large numbers of immigration arrests, lends support to the claim that governments can negatively impact immigration enforcement by “inventing” new forms of movement control. When policymakers began to provide movement-control actors with a different way of deploying their skills, the latter acquired increased discretion to gravitate toward a new task, to the neglect of other alternatives.

When considered in the context of the other findings presented in this article, the Hillbrow case helps us see the South African material as something greater than an illustration of historical path dependence or context specificity. The study suggests that immigration enforcement outcomes may not simply be changed through “bans” but also through “inventions” — in this case, of new forms of movement control. Furthermore, in addition to looking toward the past and the local context, it suggests the need to look toward the variety of emergent and more globally prevalent forms of movement control, such as COMPSTAT, which may impact the extent to which officials and non-state actors devote themselves to immigration enforcement. These forms of movement control may be invented by hard laws and administrative structures which define how exclusionary violence can be deployed but may also originate in “softer” products of the state — statistics, manuals, strategic documents — which help constitute appropriate enforcement standards and techniques.

Civilian Compliance with Movement Control Endures the End of Apartheid

To this point, the account has focused on the amount of enforcement practices taking place in South Africa. What about their effectiveness? Post-Apartheid South Africa had introduced strong protections for the right to free movement. This is no accident. Reflecting the anti-Apartheid movement’s antipathy to segregation, post-Apartheid South Africa’s constitution explicitly enshrines people’s internal movement as a basic human right: Section 21(1) reads, “Everyone has the right to freedom of movement.” Yet to my knowledge, this provision has rarely been used to question the legality of individual immigration enforcement acts within South African territory or the legislation upon which such enforcement is based, which still empowers an exceedingly wide variety of actors to locate, arrest, and detain persons not in possession of documents authorizing their stay. In the immediate post-Apartheid period, there was considerable evidence of South Africans questioning the severity of immigration enforcement processes based upon their prior experience of the violent enforcement of influx controls (Vigneswaran 2007). In combination, the laws of South Africa and this deep antipathy for Apartheid-style policing techniques provide the sub-stratum upon which a powerful tradition of non-compliance with immigration enforcement — and policing of mobility more generally — could conceivably be built. However, such a tradition has not emerged.

Our observations of civilian interactions with the police suggest that instead, enduring traditions of compliance with movement-control practices under Apartheid have simply been transferred to compliance with immigration enforcement. While others have argued that civilians in contemporary South Africa do not consent to official policing (Steinberg 2008a), this verdict seems to underplay the routinely and seemingly obsequious compliance that many South Africans display toward the movement-control operations of their police: raids, road blocks, and stop-and-search procedures in particular. Here, it is worth noting that many of the SAPS’ arrest, roadblock, raids, and stop-and-search tactics are not only invasive in a way that is reminiscent of the Apartheid regime but also patently illegal (Vigneswaran 2010). In contemporary South Africa, police officers do not have the lawful authority to stop and search an individual without cause. Yet both South African civilians and foreign nationals routinely, and without question, allow the police to subject them to a full body search, inspect their cars and premises, and interrogate them about their documents and possessions without any pretext, rarely raising any direct physical resistance or principled objection (ibid.).

Ultimately, this last finding poses questions, as well as answers. In particular, why did the ban against influx control and traditions of resistance against the policing of pass laws not combine to foster greater resistance to the policing of internal movement? However, it does help us better understand some of the peculiarities of the South African case. South Africa not only inherited an active movement-control agency, its police, but also a civil society, which, while violently resistant to many forms of policing and authority, has been remarkably compliant with state-sanctioned movement controls. It seems plausible to suggest that this ingrained tradition considerably amplified movement controllers’ capacity to produce large deportation numbers. In the absence of any tradition of consistent resistance, police officials were able to patrol streets, stop cars, and enter homes almost at will, sifting through the population and extracting those without authorization to stay.

Concluding Remarks

This account has offered two reasons to pay more attention to the complex ways in which movement-control policies shape immigration enforcement outcomes. First, the article offers evidence to suggest that non-immigration policymakers can substantially increase and decrease a country’s immigration enforcement capacity. The amount of arrests leading to deportations in South Africa, for example, changed as a result of significant changes to non-immigration movement control policies. Abandoning influx control and rolling out the COMPSTAT policing strategy both generated tangible effects on immigration enforcement outcomes. Second, the study generates data to suggest that policies concerning other forms of movement control may shape patterns of compliance. South Africa’s tradition of complying with street-level enforcement of the pass laws made contemporary policing of immigration control more impactful. Although it does not provide clarity regarding the precise nature of the relationship between these two factors, these findings suggest the need to go beyond assemblage theorists’ efforts to demonstrate the complex connections between mobility control practices and to further consider the complex manner in which non-immigration policies can impact immigration enforcement outcomes.

The article also suggests that if we aim to better understand variation in immigration control outcomes, we need more representative and comparative research. Research on immigration control outcomes remains focused on an isolated set of national contexts — primarily in Europe, North America, and other OECD countries. Theoretical models developed in these contexts are sometimes used to explore what other countries lack (Klotz 2013) or how these countries fail (Vigneswaran 2008). However, rarely does knowledge flow in the other direction, with non-OECD countries being used as exemplars of specific modes of migration governance (cf. Vigneswaran and Quirk 2015). My effort to use the principle of complexity to explain South Africa’s capacity to “out-perform” its OECD counterparts in the field of immigration enforcement suggests that we may simply have been theoretically ill-equipped to adequately comprehend the capacities of states in the Global South to control immigration. The implication here is not that we ought to look to the South for answers about how to stop global migration but that these neglected cases may offer new insights into the complex way in which migration control works.

The effort to push the comparative study of immigration enforcement beyond the Global North could pose different sorts of questions. This study has suggested two possible lines of inquiry. To begin, we might attempt to explore how different combinations of movement control policies shape contemporary immigration enforcement outcomes. There are a wide variety of places with histories of internal movement control: other racially defined internal control regimes of Southern Africa, the politically oppressive internal control regimes of Eastern Europe, or the hukou system in contemporary China. We might further explore how the continuation, or legacies, of these internal movement-control regimes impact immigration enforcement outcomes in countries that are becoming major destinations for international migrants. Will we, for example, see a dramatic uptick in deportations from emerging non-OECD economies, which may have shorter histories of large-scale immigration but comparatively large internal movement control capacities?

This study also suggests that comparative research on immigration enforcement could benefit from greater attention to the more recent diffusion of movement-control models. Here, the Hillbrow component of the study was crucial in pointing to the power of “invention.” This evidence that the roll-out of a COMPSTAT policing strategy undermined immigration enforcement outcomes suggests that we ought to more closely examine the manner in which emerging and global logics of movement control add or subtract resources and legitimacy from the way states enforce immigration laws. In the COMPSTAT example, the emergent logic was invented in New York and slowly expanded across the globe. However, we clearly need to be open to the possibility that processes of “learning” or “diffusion” can work in the opposite direction, from South to North and authoritarian to liberal state. As European and North American leaders with interests in controlling immigration — ranging from Donald Trump to Victor Orbán — seek to reinvigorate traditions of authoritarian rule, there is a pressing need to search beyond our usual sample of cases for lessons about how immigration enforcement capacity is competing with and being complimented by authoritarian traditions of movement control in the non-Western world.

Supplemental Material

Supplemental Material, MRX823191_Supplemental_Appendix_1 - The Complex Sources of Immigration Control

Supplemental Material, MRX823191_Supplemental_Appendix_1 for The Complex Sources of Immigration Control by Darshan Vigneswaran in International Migration Review