The expansion of U.S. immigration enforcement from the borders into the interior of the country and the fivefold increase in immigration detentions and deportations since 1995 raise important questions about how the enforcement of immigration law is spatially patterned across American communities. Focusing on the practice of immigration detention, the present study analyzes the records of all 717,160 noncitizens detained by Immigration and Customs Enforcement (ICE) in 2008 and 2009—a period when interior enforcement was at its peak—to estimate states’ detention rates and examine geographic variation in detention outcomes, net of individual characteristics. Findings reveal substantial state heterogeneity in immigration detention rates, which range from approximately 350 detentions per 100,000 noncitizens in Connecticut to more than 6,700 detentions per 100,000 noncitizens in Wyoming. After detainment, individuals’ detention outcomes are geographically stratified, especially for detainees eligible for pretrial release. These disparities indicate the important role that geography plays in shaping individuals’ chances of experiencing immigration detention and deportation.
Over the past two decades, the federal government has expanded its practice of detaining immigrants awaiting possible deportation from the United States. Approximately 400,000 immigrants are detained annually compared with 85,000 just 20 years ago (Meissner et al. 2013). This nearly fivefold increase in immigration detentions has occurred alongside a concerted effort by the federal government to expand immigration enforcement from the borders into the interior of the country, an effort that took on new force following September 11, 2001 (Coleman 2007). In this new era of immigration enforcement, noncitizens with long-standing community and family ties in the United States have been detained and deported at higher numbers than ever before (Kanstroom 2012). More than 1.4 million immigrants have been removed from the U.S. interior in the past decade alone (Department of Homeland Security (DHS) 2015).1
Although the precipitous rise in detentions and removals has drawn much interest by scholars, politicians, and the media alike, it remains unclear whether the recent growth in interior enforcement has been more pronounced in some states and regions than others. This study begins to fill this gap by focusing on immigration detention—the fastest growing form of incarceration in the United States (Gavett 2011)—and examining inequalities in detention rates and outcomes across U.S. states. In so doing, it advances our understanding of the immigration detention system in two key ways. First, it provides rigorous estimates of states’ immigration detention rates in 2008 and 2009—a period when interior enforcement was at its peak, with more than 470,000 removals (DHS 2015).2 Findings indicate substantial state heterogeneity in these rates, which range from an average across the two years of approximately 350 detentions per 100,000 noncitizens in Connecticut to more than 6,700 detentions per 100,000 noncitizens in Wyoming. By capturing all apprehensions leading to immigration detention, these estimates provide the most comprehensive picture to date of subnational variation in interior immigration enforcement and reveal several states with high rates of detention largely overlooked in previous research.
Second, this study demonstrates that geography influences detention outcomes, independent of individual characteristics. Individuals’ odds of winning their case, receiving pretrial release through bond or an order of recognizance, or being removed through either deportation or voluntary departure all vary by state of apprehension. Although research on immigration court proceedings has found variation across judges and courts in defendants’ outcomes (e.g., Government Accountability Office (GAO) 2008; Ramji-Nogales et al. 2007; Transactional Records Access Clearinghouse (TRAC) 2017b), this study is the first to examine detention outcomes and to link these outcomes to apprehension location. In so doing, it uncovers a set of previously unidentified disparities in the immigration enforcement system and shows how individuals apprehended in different states experience distinct trajectories through the immigration detention system, with implications for their ultimate chances of removal from the country.
The need to document subnational variation in immigration enforcement is especially pressing given the potential social, economic, and health consequences associated with detention and removal, both for the detained and removed as well as for the families and communities they leave behind. For the detained, immigration detention is essentially criminal detention without the constitutional protections afforded to criminal suspects (Torrey 2015). Investigations into the hundreds of facilities operated by local and state law enforcement agencies and private corporations that make up the immigration detention system lay bare the harsh conditions immigrant detainees face while confined, including inadequate access to medical care, environmental health and safety concerns, and exploitative labor practices (DHS 2006; Dow 2005; Stevens 2015; Venters et al. 2009). After removal, individuals often find themselves in places with poor labor market prospects and where they face substantial discrimination (Brotherton and Barrios 2011; Golash-Boza 2015; Kanstroom 2012). For those who remain in the United States, a family member’s detention and removal experience creates profound financial and emotional strain (Baum et al. 2010; Capps et al. 2007). Research has shown that even the credible fear of such an event negatively affects individuals’ health and well-being and leads immigrants to withdraw from key institutions in American life, including forgoing medical and educational services (Hagan et al. 2009; Satinsky et al. 2013). Such “system avoidance” (Brayne 2014), coupled with other adverse consequences of expanded interior enforcement on immigrants’ well-being, may hinder adaptive assimilation and curtail mobility for the estimated 4.5 million children in the United States who live in mixed-status homes, many of whom are U.S. citizens (Satinsky et al. 2013).
Place-Based Variation in Discretion and Immigrant Characteristics
Although the policies governing immigration detention specifically—and immigration enforcement more generally—are set at the federal level, there are two broad lines of explanations for why we might expect immigration detention rates and outcomes to vary across U.S. states. On the one hand, the many actors involved in immigration enforcement have discretion, which they exercise in spatially patterned ways. On the other hand, immigrants’ composition varies across states, and immigrants in some states possess characteristics that may make them more vulnerable to detection, arrest, detention, and removal than immigrants in other states. These dynamics—place-based variation in discretion use and place-based variation in immigrant characteristics—may also interact to contribute to state-level variation in immigration detention rates.
With insufficient resources to remove everyone in the country without authorization or otherwise removable,3 immigration officers have discretion when deciding whom to stop, question, and arrest; whom to detain or release; and whether to pursue removal or grant relief. Immigration and Customs Enforcement (ICE), the interior enforcement arm of the U.S. DHS, and its predecessor agency, the Immigration and Naturalization Service (INS), have released a series of memorandums enumerating the agency’s enforcement priorities and providing guidelines for when officers should favorably exercise discretion and not assert the full scope of their enforcement authority (e.g., Cooper 2000; Howard 2005; Meissner 2000; Morton 2011; Myers 2007). ICE directs these memorandums to the 24 Enforcement and Removal Operations (ERO) Field Offices tasked with overseeing immigration apprehension, supervision, and removal within their jurisdictions.4
These memorandums are intended to create consistency across offices in meeting ICE’s stated priorities. However, comparative studies of immigration enforcement have found geographic variation in the extent to which immigration officers (whether ICE officers or local police engaging in immigration enforcement) favorably exercise their discretion and follow ICE priorities. These inconsistencies exist both with respect to immigration officers’ initial decision to apprehend an individual (Capps et al. 2011; Coleman 2012; GAO 2009) and their final decision to continue prosecuting a case in immigration court or grant relief (Ellermann 2005; GAO 2008, 2016; Ramji-Nogales et al. 2007; TRAC 2017a; Wadhia 2012). Research has attributed this geographic variation in immigration officers’ discretion use to three main factors: (1) differences in local policing norms and culture (Coleman 2012); (2) differences in jurisdictions’ enforcement capacity, both with respect to personnel and detention space (Capps et al. 2011); and (3) differences in the broader social and political context in which officers work (Ellermann 2005, 2009).
Furthermore, immigration officers are not the only actors exercising discretion in ways that may lead to state-level variation in immigration detention rates and outcomes. For one, states and localities exercise discretion when choosing whether to participate in immigration enforcement programs, such as the 287(g) program and Secure Communities.5,6 These programs increase coordination between the federal government and states and localities on immigration enforcement matters and expand localities’ enforcement capacity. As such, they may lead to higher numbers of apprehensions and detentions in participating jurisdictions. In 2008 and 2009, 16 states had at least one jurisdiction participating in the Secure Communities program, and 72 law enforcement agencies in 24 states had active 287(g) agreements with the U.S. Department of Justice. Nineteen states had neither a 287(g) nor Secure Communities partnership in place (Capps et al. 2011; ICE 2013).
States and localities also exercise discretion when enacting immigration-related laws. The National Council of State Legislatures reported more than 9,000 state-initiated immigration laws and resolutions since 2005, 1,800 of which were passed (Johnston and Morse 2013). These laws vary in the level of protections and benefits provided to immigrants and range from the domain of law enforcement to public benefits. With respect to enforcement, some states and localities have passed laws to further criminalize unauthorized immigration and enact new enforcement schemes, whereas others have resisted the federal government’s initiatives, passing noncompliance ordinances (i.e., “sanctuary ordinances”) to curtail police officers’ cooperation in federal immigration enforcement efforts (Chen 2016; Eagly 2011; Walker and Leitner 2011). These laws, which are unevenly distributed across the country, may either bolster or hinder a state’s capacity to detect, apprehend, and detain potentially removable immigrants, thus affecting states’ detention rates.
Additionally, how immigration judges exercise discretion may shape states’ detention outcomes. Nationwide, more than 330 immigration judges conduct removal proceedings in 58 immigration courts.7 And although research on immigration adjudication is relatively new and has focused largely on asylum cases, it has offered preliminary evidence of variation in judicial decision-making (e.g., Miller et al. 2014; Ramji-Nogales et al. 2007; Ryo 2016). This work has also shown that in the aggregate, the extent to which immigration judges favorably exercise discretion and offer relief to immigrants facing possible deportation varies geographically (GAO 2008, 2016; Ramji-Nogales et al. 2007). These intercourt differences in judicial discretion use may lead to geographic variation in detention outcomes, with detainees in some areas of the country more frequently bonding out of detention or winning their case compared with detainees in other areas.
Immigrant Population Characteristics
Immigration detention rates and outcomes may also vary across states because immigrants in some states possess characteristics that make them more vulnerable to detection, arrest, detention, and removal than immigrants in other states. For example, the foreign-born population—and especially the undocumented population—was once concentrated in urban gateways, such as New York City, Los Angeles, and Miami, but has become more geographically dispersed since the mid-1990s (Gozdziak and Martin 2005; Massey and Capoferro 2008). New destination states now have a higher proportion of unauthorized immigrants among their foreign-born population than traditional gateways (Passel 2005), and immigrants residing in these states tend to have arrived in the United States more recently and have lower English proficiency than immigrants residing in traditional gateways (Singer 2004). These characteristics of immigrants in new destinations may lead to heightened vulnerability of detection and apprehension.
Immigrants are also unevenly distributed across industrial sectors and occupations (Passel and Cohn 2016), and these distributions vary across states (Pew Charitable Trusts 2015). As such, immigrants—and especially undocumented immigrants—may be more vulnerable to detection in certain states because they are concentrated in industries or occupations more likely to be targeted by ICE in worksite raids. A prominent example is the meatpacking industry, which is concentrated in rural areas in the Midwest and South and relies on Mexican immigrants for labor (Champlin and Hake 2006). Federal immigration officers have regularly targeted this industry, from Operation Vanguard in 1999, to the large-scale raids of six Swift & Company meatpacking plants in 2006 and of the Agriprocessors Inc. meatpacking plant in 2008. These enforcement efforts and raids all took place in Midwestern states—most notably, Iowa and Nebraska. Immigrants working in meatpacking plants and other geographically concentrated industries may be more vulnerable to detection and arrest than immigrants working in other industries and places less often targeted by ICE.
Interaction of Discretion and Composition
Place-based variation in discretion use and immigrant characteristics may also interact in ways that shape states’ immigration detention rates and outcomes. For example, in her study of migrants’ experiences living in southwest Montana, Schmalzbauer (2014:51) documented racial profiling by police, which she argued is possible because of the very “impossibility of anonymity” for Latino migrants residing in this predominantly white state. Migrants in southwest Montana appear at greater risk of profiling and subsequent arrest because the region’s demographic context makes it difficult for them to blend in and because local police regularly exercise their discretion as law enforcement officers to stop and question individuals whom they suspect are unauthorized. This combination of factors—if occurring in other parts of the country, too—may lead to higher rates of immigration apprehensions and detentions in predominately white areas and, in turn, contribute to variation in states’ detention rates.
With respect to detention outcomes, both immigration officers and judges are instructed to consider a range of factors when deciding whether to release a detainee while his case is pending in immigration court and whether to remove him. These factors are enumerated in the DHS memorandums as well as in the Immigration Judge Benchbook and include a noncitizen’s criminal record, immigration record, U.S. family ties, length of residence in the United States, and employment length and stability in the United States (Executive Office of Immigration Review (EOIR) 2017; Meissner 2000; Myers 2007).8 Given the uneven settlement patterns of noncitizens across states, immigrant communities may differ along these dimensions in ways that lead to geographic variation in enforcement outcomes. GAO (2008, 2016) analyses of U.S. asylum cases are the only studies to date to systematically account for these potential compositional differences when examining intercourt variation in enforcement outcomes. The findings suggest that judges exercised their discretion in spatially patterned ways, independent of individual and case characteristics, including nationality, family ties in the United States, detention history, and the presence of representation. It remains an open question whether immigration judges’ decisions outside the asylum context or ICE officers’ decisions to offer detainees pretrial release are also spatially patterned after accounting for detainee characteristics.
State Detention Rates and Outcomes
Although the dynamics of place-based variation in discretion use and immigrant characteristics may lead to geographic differences in interior immigration enforcement, no study has systematically examined how actual rates of interior enforcement vary at the subnational level. By looking at the practice of immigration detention—a key feature of today’s immigration enforcement system—this study begins to fill this gap. Specifically, by drawing on a novel data set of U.S. immigration detention records, I provide the first estimates of states’ detention rates and examine whether the growth in detentions has been more pronounced in particular areas of the country. Through a series of scatterplots, I also provide an exploratory analysis of how interstate differences in discretion use and immigrant characteristics relate to states’ immigration detention rates.
After detainment, immigrants’ trajectories through the detention system vary. Some detainees receive pretrial release via bond, parole, or an order of recognizance, and others are held while their case is pending and afterward while awaiting removal. This study builds on previous research on geographic variation in ICE agents’ and immigration judges’ decision-making (e.g., GAO 2008, 2016; Ramji-Nogales et al. 2007; Wadhia 2012) to look specifically at how detention outcomes vary across U.S. states, independent of detainee characteristics. Discretion at this stage—particularly with respect to pretrial release—is critical because detained immigrants are less likely than nondetained immigrants to receive legal counsel and are more likely to lose their case and be deported (Eagly and Shafer 2015). Thus, the following analysis considers both intermediate outcomes (i.e., pretrial release through bond and order of recognizance) and final outcomes (i.e., deportation, voluntary departure, and being able to stay in the United States).
I examine how detention outcomes are influenced by the state of apprehension rather than the state of detention given the mismatch in the demand and supply of detention beds in parts of the country (Schriro 2009), leading noncitizens apprehended in certain states to be detained out of state. For instance, ICE has no detention contracts in Mississippi, so during the highly publicized factory raid in Laurel, Mississippi in 2008, 475 of the nearly 600 arrested immigrants were immediately taken by bus to a detention center in Louisiana (Nossiter 2008). By examining the state of apprehension, I account for potentially systematic differences across states in such movement, which is important given research suggesting that being detained far from home impedes immigrants’ ability to successfully fight their case, rendering attorney-client relationships unworkable and separating immigrants from the evidence they need to present in court (Parker 2011).
Data and Methods
Immigration Detention Records
Research on U.S. immigration enforcement has largely relied on three sources of data—the Yearbook of Immigration Statistics, reports on enforcement programs published annually by the DHS, and collated records of local immigration ordinances—all of which provide a limited picture of subnational variation in enforcement. For one, the annual yearbook aggregates apprehensions by administrative area, which often span several states. Second, although reports published by the DHS—such as metrics on Secure Communities—report county-level apprehensions, they capture only a subset of all immigration apprehensions. Last, local immigration policies serve only as an indirect measure of enforcement and may not correlate strongly with actual enforcement. These data sources also provide no information on detentions. By drawing on records of all immigration detentions in the United States over several years, this study overcomes these data limitations and provides a direct and more comprehensive measure of subnational enforcement—the location of all apprehensions leading to detention and those detention outcomes.
Specifically, this study draws on two sets of administrative records obtained through Freedom of Information Act (FOIA) requests by the Human Rights Watch and the Transactional Records Access Clearinghouse (TRAC). The first captures the detention history of each individual detained by ICE or its predecessor (the INS) between October 1, 1998 and May 25, 2010. The files contain records for each booking of a detainee into or out of a facility. Fields include the first-ever detention date, apprehension date, apprehension location, facility name and location, entry and exit dates from the facility, a code for the reason for exit, and the detainee’s gender and citizenship country. Each record thus refers to a single detention event, such as a deportation or transfer, and together such records make up individual detention episodes. Although the data do not include an individual identifier, I use the order of records and consistencies between several variables to identify unique detention episodes.9 The second data set is similarly structured but, in lieu of apprehension location, includes information about the charge leading to detention, whether the detention is mandated under the mandatory detention statute, whether a final order of removal is issued, the detainees’ legal status at the time of entry into the United States,10 and the date the detainee first entered the United States. The second data set covers all detentions between October 1, 2006 and December 31, 2011. Combined, the two data sets include 3,308,091 detention episodes, with an average of two detention booking events per episode.
I restrict the analysis to the 717,160 detention episodes beginning in 2008 and 2009 because these years capture interior enforcement at its peak, when more than 60 % of removals originated in the interior before dropping to approximately 30 % in 2015 (DHS 2015). These years also contain the most complete information on apprehension locations. The appendix details the merging process across the two data sets and other coding procedures through which precise matches were found for 643,425 episodes, 89.72 % of all detentions. Of these matched records, 90.0 % contain complete data on all variables used in the regression analysis.11
States’ immigration detention rates are estimated as the annual number of apprehensions leading to detention in a given state over the state’s noncitizen population that year. Data on the noncitizen population are from the American Community Survey (ACS). Noncitizens include legal permanent residents, temporary migrants, unauthorized immigrants, and other resident statuses. I use estimates of the noncitizen population rather than the undocumented population because legislative changes in the 1980s and 1990s increased the risk of detention and removal for legally present noncitizens with criminal convictions (Kanstroom 2010). As such, calculating detention rates using the undocumented population underestimates the population at risk of apprehension and detention and inflates the detention rate estimates.12 To quantify apprehensions in each state, I geocode the apprehension location associated with each detention episode. The data contain more than 9,000 unique “apprehension landmarks” ranging from cities and counties to border monuments and mountain ranges.
I exclude detentions of recent border crossers (22 % of all detentions) from the estimates because these individuals are not counted in the ACS estimates and would incorrectly inflate detention rates. The reduced records consist of 562,626 detentions. To identify apprehensions of border crossers, I use both the apprehension landmark field and the field specifying individuals’ entry date into the United States. Figure 1 details this process. Because the apprehension location could not be determined for all detention episodes, I construct lower- and higher-bound estimates of apprehensions leading to detention in a given state in each year. The lower-bound estimate excludes all records in which I could not verify the apprehension location, drawing on 82.3 % of the 562,626 detention episodes. The higher-bound estimate includes the 3.8 % of records for which the apprehension state is known but the border status is unclear, and uses the state of first detention for the 13.9 % of detentions for which the apprehension state cannot be identified.13 The results section discusses rates using the higher-bound estimates, but all estimates can be found in Table 2 in the appendix.14
To descriptively explore possible explanations for state-level variation in immigration detention rates, I construct a series of graphs that plot the relationship between a states’ average higher-bound immigration detention rate in 2008 and 2009 and various discretionary and compositional factors. Four of these graphs use the 2007–2009 ACS three-year estimates data and display the relationship between a state’s detention rate and the following compositional factors: (1) percentage of foreign-born residents who arrived in the United States since 2000 (i.e., recent arrivals); (2) percentage of foreign-born residents with limited English proficiency; (3) percentage of foreign-born residents with less than a high school diploma; and (4) percentage of state population that is non-Hispanic white.
To examine the role of place-based variation in discretion, I plot a state’s detention rate against the percentage of detentions from that state mandated under the law. In states where a higher percentage of detentions fall under the mandatory detention statute, ICE officers might favorably exercise discretion more either by deciding not to pursue removal for individuals falling outside ICE’s stated priorities or by allowing these individuals to remain home while their case is pending in court. Furthermore, because research has attributed geographic variation in immigration officers’ discretion use to differences in the broader social and political context in which officers work (Ellermann 2005, 2009) and to differences in jurisdictions’ enforcement capacity (Capps et al. 2011), I construct three additional scatterplots. The first examines the relationship between a state’s detention rate and the political leanings of its population, operationalized as the percentage of people in the state who voted for John McCain for president in 2008. The second and third scatterplots explore the relationship between detention rates and personnel and detention capacity, respectively. Because no available data contain information on the number of individuals in a given state working on immigration enforcement, I operationalize personnel capacity as the per capita undocumented population in each state using Warren and Warren’s (2013) estimates of states’ undocumented population in 2008 and 2009. One could imagine that in states with lower per capita undocumented immigrant populations, fewer personnel are required to find and apprehend potentially removable individuals. Additionally, following Schriro’s (2009) methodology, I estimate detention capacity as the average daily population of detainees in each detention facility from January 1, 2008 to December 31, 2009. I aggregate these calculations and detentions to the field office level rather than the state level given that field offices manage detention space.
Last, given that states’ and localities’ decisions to pass immigration-related laws or enter into enforcement partnerships with the federal government may shape immigration enforcement capacity within their jurisdiction, I construct two additional plots that display the relationships between a state’s immigration detention rate and the following factors: (1) the presence of restrictive laws and federal-local enforcement partnerships; and (2) the share of a state’s noncitizen population residing in a jurisdiction with a sanctuary ordinance.15
To explore how apprehension location affects detention outcomes independent of individual factors, I develop a set of sequential binary logistic regression models. This technique enables me to account for the fact that certain detention outcomes can precede others. I assume that frontline immigration officers first decide whether to offer a detainee pretrial release through bond or an order of recognizance, and I model this choice. Bonding out occurs when individuals post bond and are released while awaiting a final decision on their case. If offered at all, ICE officers first set the bond amount, but detainees can appeal the decision to an immigration judge who makes the final determination about whether to grant bond and at what amount. Detainees released by ICE on orders of recognizance are allowed to return home without posting bond while their case is pending.
I next examine detainees who receive pretrial release and model whether they receive bond versus an order of recognizance. Last, for those who do not receive pretrial release, I model in three separate logistic regressions the likelihood that a detainee is deported, a detainee voluntarily departs, or the case is terminated.16 In the case of deportation, detainees are removed from the country and barred from reentry temporarily or permanently, depending on whether they have been removed previously or been convicted of certain crimes. Being apprehended again following deportation is considered a federal felony offense and can carry serious consequences, including imprisonment. Individuals taking voluntary departure waive their right to hear their case in immigration court and agree to leave the country. These individuals are not automatically barred from reentering and do not face criminal consequences if apprehended again.17
Last, I estimate a separate logistic regression on the 20 % of detentions in these years that are nonmandatory at the time the detainee exits detention. I look only at nonmandatory detentions because this group can experience all possible detention outcomes, thus providing a cleaner test of how apprehension location influences detention outcomes.18 In this regression, I combine bond and order of recognizance into a pretrial release variable and combine deportation and voluntary departure into a removal variable. This model provides estimates of the likelihood of receiving pretrial release compared with leaving the country for nonmandatorily detained individuals who do not win their case in detention.19 Because mandatory detention status can change during one’s time in detention and ICE’s coding of mandatory detention captures whether an individual is mandatorily detained at the time of exit rather than entry into detention, this analysis does not reflect the outcomes of the entire population of detainees who at some point in their detention stay were not mandatorily detained. Missing from the analysis are individuals who were initially not mandatorily detained, remained in detention for the duration of their case, and were ordered deported by a judge. These individuals are missing because their detention status switched from nonmandatory to mandatory after the judge ordered the individual deported. Because we do not know how selection into being defined as mandatorily detained might differ across states in a way that is correlated with the distribution of detention outcomes across states, we have to be cautious in interpreting these results.
Independent variables in the analyses include the detainee’s gender, place of origin, legal status, and the charge that led to detention. In the sequential logistic regressions, I also include whether the detention was mandatory; the final logistic regression conditions on not being mandatorily detained. For place of origin, the home country was used for individuals hailing from the top four countries represented in the data—Mexico, Guatemala, Honduras, and El Salvador—which account for 88 % of all detention episodes in the observed period. For detainees from a country other than these four, their geographic region (as defined by the United Nations Statistics Division) was coded as place of origin. Legal status at time of entry is grouped into eight categories: (1) false claims of citizenship, (2) legal permanent residents (i.e., green card holders), (3) present with admission (i.e., entered the country with a visa), (4) present without admission (i.e., entered the country without documentation), (5) seeking humanitarian relief, (6) smuggler, (7) U.S. citizen,20 and (8) other.21
Immigration violations are outlined in three sections of the Immigration and Nationality Act (INA). Section 212 pertains to individuals seeking admission to the United States and to individuals who already entered the United States without authorization (i.e., inadmissible persons). Section 237 applies to persons currently living in the United States after being lawfully admitted into the country (i.e., deportable persons). Section 241 applies to individuals ordered deported but released from detention because deportation did not occur within 90 days of a judge’s removal order. I examine charges connected to only Section 212 and Section 237 because these outcomes are still pending. Within these two sections, I code offenses into two groups—noncriminal and criminal—thus creating four total groups of charges.
Summary statistics for all variables used in the sequential logistic models are presented in Table 1, and summary statistics for the final logistic model are presented in Table 3 in the appendix. Across years and the full data set, nearly 72 % of detentions ended with deportation. Voluntary departure is the second most frequent outcome, accounting for 15 % of all detention outcomes. Approximately 90 % of detainees are male, and although more than 150 are represented in the data, roughly 60 % of the detainees are Mexican nationals. Approximately 76 % of the detained population are unauthorized and have no criminal record, and 80 % of detentions end with the individual mandatorily detained. When the analysis is restricted to individuals not mandatorily detained at the end of their detention, the percentage Mexican increases to 80 %, and the percentage of detainees who are unauthorized and without a criminal record increases to 88 %. Additionally, although all individuals in this group were eligible for some form of pretrial release, only 32 % of them received it. Of those receiving pretrial release, approximately 70 % bonded out. Among the 68 % of nonmandatory detentions ending in a removal from the United States, 87 % resulted from the detainee voluntarily departing.
From 1999 through 2011, annual counts of detention more than doubled from 151,253 to 340,166. As illustrated in Fig. 2, the practice of immigration detention has become more divergent across states over time. States’ rates of immigration detention were most similar to one another in 2001 and most divergent in 2010 (plot a). As shown in plot b, this growth is unevenly distributed across regions. Variation in states’ immigration detention rates is greatest in the Midwest and South and least in the Northeast. Together, these graphs show that as interior immigration enforcement intensified, immigration detention rates grew, as did interstate variation in these rates.
Focusing on 2008 and 2009 provides a fuller picture of subnational variation in immigration detention. The mean rate of detention across the two years is 2,000 detentions per 100,000 noncitizens, and the standard deviation is 1,528. Looking across years in Fig. 3, states’ relative position in the distribution of states’ immigration detention rates remains largely stable. The primary exception is Rhode Island, where the immigration detention rate dropped by more than one-half between 2008 and 2009.22 Other states with sizable fluctuations in their immigration detention rates across years are Maine, Missouri, and South Carolina.
In both years, Arizona, California, and Texas have the highest counts of detention, with a total of 54 % of all detentions occurring because of apprehensions in these states. However, after I adjust for the size of the noncitizen population in each state, a different picture emerges. Although states like Montana, North Dakota, South Dakota, and Wyoming account for less than 0.5 % of total detentions, their detention rates fall within the top quintile of states. In contrast, states like California, Florida, and New York have a high number of detentions but significantly lower detention rates after their large noncitizen populations are adjusted for. On average, across the two years, Connecticut has the lowest detention rate: 357 detentions per 100,000 noncitizens. Wyoming has the highest, with an average detention rate across the two years that exceeds 6,700 detentions per 100,000 noncitizens. With an approximately 19-fold difference between the state with lowest detention rate and that with the highest, these results indicate considerable interstate variation in immigration detention rates. States’ immigration detention rates also far exceed state incarceration rates in the United States, which at the end of 2009 ranged from 150 per 100,000 U.S. residents in Maine to 881 per 100,000 U.S. residents in Louisiana (West et al. 2010). Table 2 in the appendix presents the detention rate estimates.
What happens to detained individuals after detainment, and are these outcomes spatially patterned? The sequential logistic regression analysis assesses the probability that a detainee experiences five potential outcomes: (1) pretrial release; (2) conditional on pretrial release, bonds out; and, conditional on not receiving pretrial release; (3) deportation; (4) voluntary departure; or (5) termination. Results are displayed in Table 4 in the appendix. These models include the full set of individual covariates summarized in Table 1. Coefficients are reported as odds ratios. Thus, the first entry shows that compared with individuals apprehended in Texas (omitted category), the odds that an individual apprehended in Alaska experiences pretrial release from detention, conditional on being detained, is 2.56 times higher, or 156 %, with all other covariates held equal. Overall, ceteris paribus, the odds of all outcomes vary considerably across state of apprehension.
Furthermore, the odds of receiving pretrial release (column 1) are greater for legal permanent residents than for unauthorized immigrants (omitted category). The odds are also greater for women than for men (omitted category) and for noncitizens who legally entered the country and have no criminal record than for inadmissible noncitizens without a criminal record (omitted category). The odds of pretrial release are lowest for detainees from Mexico (omitted category). Conditional on receiving pretrial release, women are more likely than men to receive an order of recognizance rather than bond (column 2), as are detainees from the Caribbean, El Salvador, Guatemala, and Honduras compared with Mexican detainees.
Conditional on not receiving pretrial release, Mexican detainees have the lowest odds of their case being terminated (column 5) but have higher odds than all but Central Asian detainees of taking voluntary departure (column 4). With respect to deportation (column 3), detainees from El Salvador, Guatemala, Honduras, and other Latin American countries all have higher odds of deportation than detainees from Mexico. Results also indicate that detained unauthorized immigrants have higher odds than most of receiving voluntary departure and are the least likely to have their case terminated. Last, both inadmissible and deportable noncitizens with criminal convictions have lower odds of voluntary departure than inadmissible immigrants without a criminal conviction.
Looking only at individuals not mandatorily detained at the time of exiting detention, Table 5 in the appendix shows the odds ratios for experiencing pretrial release through bond or order of recognizance compared with removal through deportation or voluntary departure. Overall, the trends largely match those shown in Model 1 in Table 4, although the magnitude of these effects change. To better illustrate how detention outcomes vary spatially, I use this model to calculate the average predicted probability of pretrial release from detention for a Mexican unauthorized male conditional on the state of apprehension (Fig. 4). Unauthorized Mexican men who are nonmandatorily detained when they exit detention account for 14 % of all detentions and 70 % of nonmandatory detentions in 2008 and 2009. For these individuals, I find significant variation across states in the predicted probability of pretrial release, which ranges from .97 in North Dakota and Wisconsin to .03 in states like North Carolina and West Virginia. Also evident is a regional pattern whereby the predicted probability is low in Southeastern states and high in Midwestern states. The predicted probability of pretrial release is also high for individuals apprehended in some Northeast states, Alaska, and Hawaii.23 Because the data preclude an analysis of the entire population of detained individuals who at some point were eligible for pretrial release, the true probability of pretrial release is likely to be lower in most, if not all, states. What can be concluded from these results, however, is that although individuals apprehended in certain parts of the country in 2008 and 2009 had a chance of pretrial release through bond or an order of recognizance, individuals apprehended in many Southeast states had nearly no chance. This nearly nonexistent chance of pretrial release is because ICE officers were not issuing orders of recognizance and because bonds either were not issued or were issued at such high amounts that individuals could not pay them. Also clear from these results is that issuing voluntary departures was a common practice in many Southeast states for nonmandatorily detained individuals but nearly never used in states where results indicate a high predicted probability of pretrial release for individuals nonmandatorily detained when they exit detention.
Possible Explanations for Interstate Variation in Detention Rates and Outcomes
The scatterplots in Fig. 5 provide an exploratory analysis of possible explanations for interstate variation in detention rates, plotting the relationships between various discretionary and compositional factors (discussed earlier) and states’ detention rates. Results indicate that although both composition and discretion-related factors are associated with states’ detention rates, the factors relating to immigration officers’ discretion use as well as localities’ decisions to pass sanctuary ordinances emerge as a somewhat stronger story.24
With respect to immigration officers’ discretion use, I examine officers’ actual discretion use, measured as the percentage of detentions mandated by law, and factors identified in previous scholarship as affecting whether immigration officers favorably exercise their discretion (i.e., the political context and personnel and detention capacity). Plot a shows a weak, negative correlation between the percentage mandatorily detained in a given state and its detention rate, indicating that immigration detention rates tend to be higher in states where immigration officers detain a greater proportion of individuals not mandated by law to be detained.25 Plot b shows a moderate, positive correlation between the percentage of voters in a given state who voted for John McCain for president in 2008 and a state’s detention rate, suggesting that enforcement is higher in states with a larger share of Republican constituents. This result is in line with Ellerman’s (2005) finding that immigration officers face less resistance to tough immigration controls in more politically conservative areas.
In terms of enforcement capacity, plot c shows a weak, negative correlation between a state’s per capita undocumented immigrant population and its detention rate, suggesting that states likely to have greater personnel capacity to detect and apprehend potentially removable immigrants tend to have higher detention rates.26 Looking at the field office level, plot d shows a weak, positive relationship between a field office’s detention capacity and its detention rate. This result, however, is largely driven by Texas, which in this measure combines four field offices that oversee enforcement in Texas, Oklahoma, and New Mexico. After it is dropped, the correlation between field offices’ detention capacity and detention rates increases to .44, indicating that areas with greater detention capacity tend to have higher rates of enforcement. Together, these findings lend some support to Capps et al.’s (2011) finding that enforcement is greater in places with more enforcement capacity.
Given that discretion entails both individual decisions to proceed against identified individuals and systemic choices to commit resources and set enforcement priorities (Motomura 2011), I also examine how states’ and localities’ decisions to pass immigration-related legislation and enter into enforcement partnerships with the federal government relate to states’ detention rates. Of particular note is the very weak, positive correlation between the presence of restrictive immigration policies and enforcement partnerships in a given state and its detention rate (plot e). Although research on interior immigration enforcement has focused on these policies and programs, none of the states—with the exception of Arizona—in the top quintile with respect to their immigration detention rate fall within the top quintile of scores on the interior enforcement scale.27 Rather, states like Louisiana, Mississippi, Wyoming, and North and South Dakota—which have largely been overlooked in previous research on immigration enforcement—have some of the highest rates of immigration detention in the country. In contrast, however, plot f shows that states that have a larger share of the noncitizen population living in a jurisdiction with a sanctuary ordinance tend to have lower immigration detention rates, indicating that these laws—which have increased more than fivefold in number since 2009 (Federation for American Immigration Reform (FAIR) 2017)—are likely protective.
The remaining two plots explore the relationship between states’ immigration detention rates and characteristics of their immigrant population. Plot g shows a weak, positive relationship between the percentage of the foreign-born population in a given state arriving after 2000 and the state’s immigration detention rate, suggesting that new destination states with a higher proportion of recent arrivals among their foreign-born population tend to have higher rates of immigration detention. Results also show a weak, positive correlation between the percentage of a state’s population that is non-Hispanic white and its immigration detention rate (plot h). As Schmalzbauer (2014) showed in her study of Mexican migrants in southwest Montana, immigration enforcement may be higher in predominately white areas because the demographic context makes it is easier for law enforcement officers to detect potentially removable immigrants. In contrast, there appears to be no discernable relationship between the percentage of the foreign-born population with limited English or the percentage of the foreign-born population with less than a high school degree and a state’s detention rate, suggesting that interstate differences in the immigrant population along these dimensions are likely not contributing to immigrants’ susceptibility to immigration apprehension and detention.28
With respect to detention outcomes, a number of factors could be driving the geographic variation documented here. First, this variation is likely due at least in part to geographic differences in the extent to which ICE officers and immigration judges favorably exercise their discretion by offering detainees the option of pretrial release and setting bond at affordable amounts. I find a moderate positive correlation of .5 between the predicted probability results in Fig. 4 and the measure capturing ICE officers’ discretion use (percentage of detainees in a given state mandatorily detained), suggesting that states where immigration officers are less often favorably exercising their discretion when making detention decisions tend to have fewer detainees receiving pretrial release, despite being eligible.29 It is not possible with these data to identify why eligible individuals in some states are not receiving pretrial release, but given ICE officers’ broad discretion at this stage and the fact that they are the only individuals able to grant orders of recognizance, they are likely playing an important role.
Intercourt variation in immigration judges’ decision-making also likely affects these results. Detainees apprehended in states with the lowest predicted probabilities of pretrial release tend to have their cases heard in courts where immigration judges favorably exercise their discretion and offer respondents relief infrequently. Take, for example, the cases of North Carolina and South Carolina, where the chance a detainee apprehended in either state receives pretrial release is quite low (less than .03). Of the more than 25,000 individuals apprehended between 2008 and 2009 in these states and then detained, 98.1 % were held in the Stewart Detention facility in Lumpkin, Georgia, where the associated immigration court has been known to be the toughest in the country (Thompson 2016; TRAC 2017b).30 Similar to Stewart, judges in the Oakdale, Louisiana immigration court also rarely offer respondents relief from deportation (TRAC 2017b). This court hears the vast majority of cases for detained individuals apprehended in Alabama, Louisiana, and Mississippi, all of which have very low predicted probabilities of pretrial release. Although these data can not be used to determine whether bond is granted by an ICE officer or an immigration judge or in which court a detainee’s case is heard, it is clear from these results that for individuals detained in the Stewart Detention Center or detention centers associated with the Oakdale immigration court, the likelihood of receiving pretrial release from either an ICE officer or an immigration judge is close to zero.
Variation in detention outcomes could also be partly due to interstate variation in detainees’ composition along dimensions not accounted for in this study. Although I control for gender, nationality, legal status, and criminal convictions, because of data restrictions, I cannot account for all legally relevant factors that ICE officials and immigration judges are instructed to consider when deciding whether and how to proceed with a case, such as length of time in the United States, family ties in the United States, and employment history (EOIR 2017; Johnson 2014). With that said, however, the one study that has examined how immigration judges weigh these legally relevant factors when making decisions (in bond hearings) found that prior criminal history, which is controlled for in this study, was the only factor judges considered in both grant/deny and bond amount decisions (Ryo 2016).
Last, some interstate variation in detention outcomes may result from geographic differences in access to legal counsel, which I also could not account for in this study. Research has shown that individuals with legal counsel fare better in immigration court and that the availability of legal counsel is unevenly distributed across the United States (Eagly and Shafer 2015; Ramji-Nogales et al. 2007; Ryo 2016). Results from studies that controlled for the presence of legal counsel, however, still found significant intercourt disparities in case outcomes (GAO 2008, 2016). So although legal representation is clearly an important factor in the immigration adjudication process, it likely cannot fully explain the preceding results.
The findings from this study have multiple implications for how we think about U.S. interior immigration enforcement. First, this study provides rigorous estimates of states’ immigration detention rates in 2008 and 2009, showing that individuals’ risks of experiencing immigration detention are highly stratified by geography. Rates range from an average across the two years of approximately 350 detentions per 100,000 noncitizens in Connecticut to more than 6,700 detentions per 100,000 noncitizens in Wyoming. Although scholarship on contemporary trends in immigration enforcement has largely focused on state and local enforcement partnerships with the federal government and restrictive immigration ordinances (e.g., Amuedo-Dorantes et al. 2013; Armenta 2012; Gulasekaram and Ramakrishnan 2015; Miles and Cox 2014; Rugh and Hall 2016; Treyger et al. 2014; Varsanyi 2010), this study shows that the pattern of states’ detention rates is not clearly related to either trend. Rather, it is associated with factors relating to the composition of states’ immigrant populations, immigration officers’ discretion use, and the presence of sanctuary ordinances, although the discretion-related factors and presence of sanctuary ordinances emerge as a somewhat stronger story. That the pattern of states’ detention rates is not clearly related to the presence of restrictive enforcement programs and policies does not mean that these programs and policies are inconsequential for immigrant communities. Rather, this finding suggests that looking exclusively at places with these programs and policies obscures from view several states that achieved high rates of enforcement without establishing 287(g) and Secure Communities programs and without passing restrictive immigration policies.
Second, this study provides evidence that the location of apprehension influences detention outcomes, independent of individual characteristics. By looking specifically at detention outcomes, this analysis uncovers a set of disparities obfuscated in research using immigration court data given that bonds and orders of recognizance can occur prior to a detainee’s case being adjudicated in immigration court. Results indicate that the likelihood of receiving pretrial release from detention varies across U.S. states, as do the likelihoods of deportation, voluntary departure, and termination for those who do not receive pretrial release. Looking only at unauthorized Mexican men who exit detention as nonmandatorily detained reveals that the predicted probability of pretrial release is more than 90 % higher for individuals apprehended in certain states. Further analysis suggests that geographic differences in immigration officers’ and judges’ discretion use likely contribute to this spatial variation in detention outcomes. Given that detained immigrants are significantly more likely than nondetained immigrants to lose their case and be deported (Eagly and Shafer 2015), the findings from this study highlight the important role geography plays in shaping detainees’ chances of remaining in the United States.
The fact that immigration detention rates are so high in some states and that the outcomes of detention are so geographically stratified inspires the need for future research along several dimensions. For one, this analysis looks specifically at detention and not immigration enforcement more broadly. It is possible to be apprehended, be provided a written notice to appear in court, lose the case, and be deported without ever being detained. Future research should seek out new data sets to capture these individuals and to more comprehensively examine how enforcement outcomes are conditioned by geography. Furthermore, this study focuses on immigration enforcement when interior enforcement was at its peak. Future research should examine the geographic patterning of enforcement over a longer period and capture the post-2011 decline in interior removals as well as recent increase. Such analysis would help clarify not only the broader trends in interior immigration enforcement but also the potential mechanisms driving heightened interior enforcement in certain parts of the country. Although this study provides an exploratory analysis of how place-based variation in discretion and some immigrant characteristics affect states’ detention rates and outcomes, these and other possible explanations need be interrogated more fully.
Last, research should further document the ramifications of interior enforcement. Although scholars have examined the individual toll detention and deportation takes on the detained and deported and the families they leave behind (e.g., Capps et al. 2007; Hagan et al. 2009, 2011), little is known about how interior immigration enforcement affects schools, neighborhoods, and the economic and social well-being of communities. Equally little is known about how interior immigration enforcement affects broader processes of immigration assimilation, inequality, and mobility. With the federal government’s renewed investment in interior immigration enforcement efforts and the more than 30 % rise in immigration arrests in the first three months of the Trump administration (Sacchetti 2017), the need to answer these questions grows all the more pressing.
This research was supported by the Inequality and Social Policy Program and the Weatherhead Center for International Affairs at Harvard University and completed while I was a fellow at the Transactional Records Access Clearinghouse (TRAC) at Syracuse University. I thank TRAC for sharing their immigration detention data and Arjen Leerkes for sharing data on interior immigration enforcement programs and immigration policies across U.S. states. I also thank Jason Beckfield, Brielle Bryan, Kareem Carr, Filiz Garip, Simo Goshev, Sasha Killewald, Barbara Kiviat, Charlotte Lloyd, Collin Payne, Jessica Simes, Philip Torrey, Mary Waters, Bruce Western; and seminar participants at the Harvard University Migration and Immigrant Incorporation Workshop; Weatherhead Center Graduate Student Associates program; Kennedy School Proseminar on Inequality and Social Policy; the III CINETS Conference; and the 2016 Annual Meeting of the American Sociological Association.
Identifying Unique Detention Episodes in the 1998–2010 Data Set
Each detention event an individual experiences represents one row in the data set, and the data are ordered chronologically from the first detention event to the last. As such, I identify detention episodes by linking subsequent rows that shared the same initial book-in date, gender, and country of citizenship. A hypothetical example illustrates this process: Guatemalan female first detained on September 1, 2006, transferred six days later to another facility, and held for 10 days before being deported. The first row of data for this woman would have an initial book-in date of 09/01/06, a book-in date of 09/01/06, and would indicate that the individual was transferred on 09/07/06 to X facility. This row would also include information about the initial detention facility and apprehension date and location. The next row would indicate that the individual was initially detained on 09/01/06, the book-in date would be 09/07/06, and other fields would indicate that she was deported on 09/17/06. This row would also include the same information as the row above with respect to the apprehension date and location.
Process of Merging the Data Sets
Because the data sets do not share a unique identifier, when merging the data, I construct a person signature and an episode signature for each detention episode consisting of the shared person characteristics in the two data sets (i.e., gender and country of origin) and the shared detention characteristic (i.e., the date and location of each stay in the detention system and the reason for release). I use Python to merge records that matched in both their person and episode signatures. For nonunique combinations of signatures, I create an algorithm that sorts through the state in which the apprehension occurred and the charge that led to detention. If at least one of these fields is consistent across the duplicate combinations of signatures, then they are merged. For example, if each data set includes two sets of records pertaining to Mexican men who were detained in the same facilities on the same dates and were released for the same reason, both sets of records are merged only if either the state in which the apprehension occurred and/or the charge are consistent across both records. So if each of these individuals was first apprehended in Texas, the duplicate pairs of records are matched at random with the records in the other data set even if the charge varies. However, if both the charge and the state of apprehension31 vary across both data sets, none of the records are merged. As such, in the case of duplicate combinations of signatures, I cannot be confident that specific individuals in each data set are correctly matched. However, I am confident that the individual characteristics used in the analysis are attributed to the correct state of apprehension, which is the central unit of analysis. Fifty-four percent of the merged records have a unique person and episode signature combination, and a total of 12,544 detention episodes (1.6 % of all episodes) are not merged because the state in which the individual was apprehended and the charge that led to detention varies across the records. The remainder of unmerged records do not have a matching record in the other data set. Of these unmerged records, approximately 90 % are from the 1998–2010 data set.
There are more than 9,000 unique landmarks in the data set for 2008 and 2009. These landmarks fit into six main categories: (1) city and county names; (2) the jail or prison facility from which an individual was first transferred into immigration custody; (3) specific enforcement programs names, such as 287(g), the Law Enforcement Agency Response Unit (LEAR), Fugitive Alien, and Criminal Alien Program (CAP) programs; (4) ports of entry; (5) landmarks located along the Mexico–U.S. border, including monument numbers, mountain ranges, bridges, ghost towns, creeks, and passes; and (6) highway and road intersections. I first geocode these landmarks using Google’s Geocoding API in R. I then independently geocode each landmark by hand twice and cross-validate these results. In this second stage, I identify both the state in which the landmark is located and whether the landmark is located along the U.S. border. For cases in which the landmark (such as a county name) could be assigned to several states, I cross-reference the landmark with the initial states of detention for all individuals apprehended at that landmark and assign the landmark to the state(s) in which the detentions occurred.
To identify apprehensions of border crossers, I use the apprehension landmark field, the field specifying the original date of entry into the United States, and counts of apprehensions at that location. Figure 1 details this process. Common border landmarks include ports of entry; border monuments; ghost towns along the border; and geographical areas, such as mountain ranges and valleys, where the U.S. Customs and Border Protection has a strong presence according to local newspaper articles and the social media feeds of border patrol stations and border vigilantes. For landmarks located within 100 miles of the border that are not clearly linked to border patrol activity, counts of apprehensions are examined in conjunction with newspaper and social media analysis. For example, there is extensive news coverage of unauthorized immigrants attempting to avoid checkpoints in Falfurrias, Texas, 70 miles north of the border, by crossing through nearby private ranches by foot. Therefore, when the apprehension landmark field includes the name of a ranch located near the Falfurrias checkpoint that had multiple apprehensions (often >50) between 2008 and 2009, I code it as a border landmark.
Excluded Detention Outcomes
Excluded from the regression analyses are detentions that ended with the following outcomes: (1) order of supervision; (2) transfer to U.S. Marshall; (3) parole; (4) withdrawal; (5) escape; (6) death; (7) transition to an alternative to detention; and (8) transfer to the Office of Refugee Resettlement. Together, records with these outcomes account for 4 % of all merged detention episodes in 2008 and 2009. Orders of Supervision account for approximately one-half of these records; U.S. Marshal transfers account for an additional 40 %; parole, 7 %; and withdrawal, 2.5 %. The remaining outcomes combined account for less than 1 % of the detentions in these years. Orders of supervision occur when a detainee is released after a final order of removal. In these cases, ICE has not met the time limits imposed for deporting the individual, often because of challenges in acquiring the needed documentation from the detainee’s home country. A detainee is turned over to the U.S. Marshal Service typically when a criminal case against the individual is outstanding or the individual is needed as a material witness in a criminal case. Parole occurs when an individual is granted temporary permission to enter the United States. Parole is most frequently granted when a detainee shows demonstrated need to enter the United States for medical or humanitarian reasons. Withdrawal occurs when an individual’s request to enter the United States is allowed to be withdrawn. Orders of supervisions and transfers to the U.S. Marshal are excluded from the analysis because they are triggered by factors that cannot reasonably be tied to one’s apprehension location. Paroles and withdrawals are excluded because they apply only to people attempting to enter the country. The other outcomes are excluded because of how infrequently they occur during the observation period.
Construction of Detention Capacity Measure
Following Schriro’s (2009) methodology, I estimate detention capacity as the average daily population of detainees in each detention facility from January 1, 2008 to December 31, 2009. These counts as well as data on the noncitizen population are then aggregated to the field office level. There are 24 field offices located in 18 states (ICE n.d.). For California, Texas, and New York, each of which has multiple field offices, I combine detention capacity calculations in order to merge in ACS data on the noncitizen population and create detention rate estimates at the field office level. For example, Texas has four field offices that oversee enforcement across Texas, New Mexico, and Oklahoma. Each field office is responsible for a region of Texas, such as North Texas and Central South Texas. Because it is unclear from the information that ICE provides which cities and communities make up these regions, estimating the noncitizen population in that area is not possible. I therefore combine the average daily population of detainees in each detention facility across all three states, as well as the number of detentions and the noncitizen population estimates of all three states. In this way, I can compare the average detention capacity across all three states to the average detention rate for the three states. New York, on the other hand, has two field offices that oversee enforcement only in New York. New York’s aggregated detention capacity measure is thus the average detention capacity of the state.
I estimate detention capacity at the field office level rather than the state level because field offices manage detention space. Mississippi provides a useful case in point. Mississippi has a large number of immigration apprehensions in this period but zero ICE detention beds. As a result, the New Orleans Field Office, which oversees enforcement in Mississippi, detains individuals from Mississippi in Louisiana, which has a number of large detention centers. Calculating detention space at the state level would fail to capture the extent to which Louisiana’s bed capacity may affect detention decisions for individuals apprehended in Mississippi.
Construction of Interior Enforcement Scale
To create the interior enforcement scale, I draw on data compiled by Leerkes et al. (2012, 2013), making several changes to the data to better fit the analysis needed for this study. First, I extend the data set to all 50 states. Their scale is restricted to the 43 states for which the Pew Hispanic Center has provided estimates of the unauthorized immigrant population. Second, because I am interested in only indirect measures of enforcement (i.e., the presence of restrictive immigration-related laws or enforcement programs), I drop the component of their scale that captures the rate of the estimated unauthorized population in a given state-year arrested through the Secure Communities program. Third, I supplement the data on the presence of state laws regarding unauthorized immigrants’ access to a driver’s license—obtained from the National Conference of State Legislatures database, which starts with laws enacted in 2005—with data from a Congressional Research Service Report (Smith 2005). The result is the inclusion of 13 states that had restrictive ID laws on the books prior to 2005 that remained in effect through the observation period. Fourth, rather than using dummy variables to indicate the presence of a city or county 287(g) or Secure Communities contract, I merge data from the 2007–2009 ACS three-year estimates of the noncitizen population in each city and county and then calculate the share of a state’s noncitizen population living in a jurisdiction with an active 287(g) or Secure Communities contract. Several of the cities and counties that participated in 287(g) and Secure Communities in 2008 and 2009 had small populations and, particularly, small noncitizen populations. A 287(g) or Secure Communities contract would likely have a greater effect on a state’s immigration detention rate in jurisdictions with larger noncitizen populations. The scale now accounts for these potential differences. Last, because data are not available on the percentage of firms in a given state participating in E-Verify—a federal program that allows employers to electronically verify the work eligibility of new hires—for the seven states not included in Leerkes et al.’s (2012, 2013) analyses, I look only at the presence of state laws requiring E-Verify usage.
This scale is thus the sum of the following measures, each of which is standardized. I look only at 2009 because no state in 2009 overturned a policy or stopped participating in an enforcement program that was active in 2008, but multiple states passed policies or initiated enforcement programs in 2009. This scale is plotted against a state’s average higher-bound detention rate from 2008–2009 (see Fig. 5).
Dummy variable indicating the presence of a state 287(g) contract.
Dummy variable indicating the presence of a state law requiring employers to participate in E-Verify.
Dummy variable indicating the presence of a state law restricting access to public benefits for undocumented immigrants residing in the state.
Dummy variable indicating the presence of a state law restricting access to IDs for undocumented immigrants residing in the state.
Dummy variable indicating the presence of a state law restricting access to employment for undocumented immigrants residing in the state.
Share of a state’s noncitizen population residing in a county with a 287(g) contract.
Share of a state’s noncitizen population residing in a city with a 287(g) contract.
Share of a state’s noncitizen population residing in a county participating in Secure Communities.
Construction of Sanctuary Ordinance Measure
This measure captures the share of a state’s noncitizen population living in a jurisdiction with a sanctuary ordinance in place prior to the end of 2009. I draw on FAIR (2017) and Seghetti et al. (2006) to identify cities and counties with sanctuary measures in place prior to the end of 2009 and then merge data from the 2007–2009 ACS three-year estimates of the noncitizen population in each of these cities and counties. To avoid double-counting individuals, I exclude from the analysis the sanctuary ordinances passed in Salem, Oregon, and San Francisco, California, because the counties in which these two cities are located also had sanctuary ordinances and are included in the analysis. The measure is thus the sum of noncitizens within a given state living in a jurisdiction with a sanctuary policy over the entire population of noncitizens in that state. By the end of 2009, only 18 states had passed some form of a sanctuary measure.
Removals include deportations and voluntary departures.
Removals from the U.S. interior began to increase in 2017 for the first time in a number of years, but have not reached the levels seen in 2008 and 2009 (ICE 2017).
Legally present noncitizens with certain criminal convictions are also removable.
These offices are located across the country and do not map onto a coherent spatial unit. Although some states have multiple field offices, others have only one or share an office with neighboring states.
Here, I draw on the legal scholar Hiroshi Motomura’s definition of discretion as including “not only decisions to proceed against identified individuals, but also systemic choices to commit resources and set priorities” (Motomura 2011:174).
Through the 287(g) program, state and local law enforcement agencies can choose to partner with ICE and designate specific officers to perform immigration enforcement functions (https://www.ice.gov/287g). With the Secure Communities program, all fingerprints collected by local and state law enforcement are automatically shared with ICE to determine whether the individual is potentially removable (https://www.ice.gov/secure-communities). Although now activated in all jurisdictions, Secure Communities was initially piloted in select places.
Like ICE’s field offices, immigration courts are distributed across the country and do not map onto a coherent spatial unit. Some states have multiple courts, and other states have only one court or share a court with neighboring states.
The version of the Benchbook that includes a discussion of these factors was discontinued in 2017.
See the appendix for details of this process.
Potential differences between legal status at the time of entry and legal status at the time of arrest are not captured in the analysis. For example, an individual who was granted a visa and overstayed would have a legal status at time of entry of “present with admission” despite being present in the country without authorization at the time of arrest. Most individuals in the data entered the country without admission and are not able to adjust their legal status from within the country. As such, in discussing results, I refer to individuals who entered the country without admission as unauthorized.
The legal status at time of entry variable accounts for 99.8 % of missingness. When the full data set is compared with records without missing data, the distribution of detention outcomes and place of origin change slightly. Percentage Mexican drops from 62.5 % to 59.8 %; percentage El Salvadorian, Guatemalan, and Honduran each increase slightly. Percentage voluntary departure drops from 20.1 % to 14.5 %, and percentage deported increases from 63.7 % to 68.1 %. Records missing information on legal status also have missing data on entry date into the United States.
In some states, however, the vast majority of apprehensions leading to detention were of undocumented immigrants. Online Resource 1 includes estimates of states’ detention rates using data on states’ undocumented population from Warren and Warren (2013) and information on the share of detentions from each state that are of undocumented immigrants.
State of first detention is likely a strong proxy for apprehension state given that only 5.2 % of geocoded apprehensions occurred in a state other than the state of first detention and apprehension landmark appeared missing at random.
I discuss results using the higher-bound estimates because they provide a more complete picture of enforcement and because the relative position of states’ detention rates remains fairly stable across the estimates. The key exception is Arizona, where border crossers are likely included in the higher-bound estimate, indicating that this estimate is biased upward.
See the appendix for details on how all measures were constructed.
Terminations occur most often because a judge dismisses the case or awards some kind of relief, but they can also occur if ICE finds a mistake in seeking to deport the individual.
For the 4 % of detention outcomes not examined, see the appendix.
Due to litigation in the First, Second, and Ninth Circuits and to INA 236(c)(2), which pertains to cooperating witnesses or informants in criminal investigations, there are extenuating circumstances in which mandatorily detained individuals can receive bond.
Episodes ending with detainees winning their case are excluded because this outcome occurs too infrequently to provide reliable estimates after I condition on nonmandatory detention.
It is unclear what legal statuses are included in the “other” category.
This drop may be due in part to the state losing its contract with ICE in February 2009 (Bernstein 2009).
These patterns hold when looking at unauthorized men from other countries and regions who exit detention as nonmandatorily detained.
Results did not substantively change when states’ detention rates were logged or when Lowess smoothing was used.
Because we do not know how selection into being defined as mandatorily detained might differ across states in a way that is correlated with states’ detention rates, we have to be cautious in interpreting these results.
Section 134 of the Illegal Immigration Reform and Immigrant Responsibility Act mandates the Attorney General to assign at least 10 full-time immigration officers in each state. In states with a small undocumented immigrant population, 10 officers may constitute substantial personnel capacity.
When Arizona is dropped from the analysis, the correlation coefficient reduces to .093.
See Online Resource 1 for a visualization of these results.
See Online Resource 1 for a visual display of this relationship. Additionally, because we do not know how selection into being defined as mandatorily detained might differ across states in a way that is correlated with the distribution of detention outcomes across states, we have to be cautious in interpreting these results.
More than 98 % of the cases heard in this court in 2008 and 2009 ended in deportation or voluntary departure (TRAC 2017b).
Approximately 18 % of the records were missing information on the location of apprehension. In this case, the state of first detention was used as a proxy for the state of apprehension given that more than 95 % of geocodable records were apprehended in the same state where they were first detained.