Deporting “Bad Hombres”? The Profile of Deportees under Widespread Versus Prioritized Enforcement

Abstract

US President Trump’s administration has vowed to boost immigration enforcement to get rid of “bad hombres,” or undocumented immigrants with criminal records. Using past data on the alleged detention motives for a representative sample of Mexican deportees, we evaluate how prior widespread and prioritized enforcement has fared in that regard. We find that while the early sweeping approach to enforcement raised deportees’ propensity of being detained for minor offenses, the trend reversed with prioritized enforcement. These findings inform policy tactics that, aside from proving effective in prioritizing serious criminal offenses, can also lead to significant savings to taxpayers.

Keywords

K37 J15 J61

Introduction

Since 9/11, the United States has witnessed an unprecedented escalation of immigration enforcement at all levels: federal, state, and local (Chishti and Bergeron 2011; Mittelstadt et al. 2011). Federal initiatives, such as the Criminal Alien Program (CAP), along with border operations (e.g., Operation Streamline) have been accompanied by a number of immigration enforcement initiatives that have delegated the federal authority to carry out immigration enforcement tasks to local and state police personnel (Kandel 2016). Specifically, 287(g) agreements¹ between Immigration Customs Enforcement (ICE) and local/state law enforcement agencies flourished after 2002, contributing considerably to an increase in the overall number of removals (see Figure 1). Although popular, these agreements were heavily disputed in a Department of Homeland Security (DHS) task force evaluation in 2011, following allegations of racial profiling and discrimination (Provine and Sanchez 2011). Likewise, complaints that the vast majority of those arrested were charged with minor offenses or misdemeanors,² such as traffic offenses and public drunkenness (Hagan, Castro, and Rodriguez 2010; Koper et al. 2013), led DHS to progressively phase out the 287(g) agreements in favor of the Secure Communities program.³ Begun in 2008, Secure Communities was a cheaper local enforcement than 287(g) and emphasized community safety. Between 2009 and 2012, it resulted, on average, in about one million arrests and close to 400,000 removals of noncitizens on a yearly basis (Venturella 2010). The Secure Communities program expanded quickly, practically covering the entire United States by late 2014. In 2015, however, it was replaced by the Priority Enforcement Program, which targeted individuals convicted of significant criminal offenses or who otherwise posed a threat to public safety.⁴ In January 2017, under President Donald Trump’s immigration regime, Secure Communities was reactivated. Across these different programs and despite their fine-tuning to target serious offenders, as well as the overall criminalization of unlawful entries, the vast majority of removals continued to be immigrants without a criminal status (Rosenblum and McCabe 2014).

Figure 1.

Deportations by US Department of Homeland Security (in thousands, by fiscal year). Sources. Data for 2001–2003 come from US Department of Homeland Security, Yearbook of Immigration Statistics: 2010. Data for 2004–2013 come from US Department of Homeland Security, Yearbook of Immigration Statistics: 2013. Data for 2014 and 2015 come from Yearbook of Immigration Statistics: 2014 and 2015 , respectively.

To alleviate an increasingly overwhelmed detention and judicial system, DHS began to prioritize the apprehension, detention, and removal of individuals convicted of serious criminal offenses.⁵ However, the implementation of prosecutorial discretion⁶ occurred at a slow pace. It was not until June 2012, following then DHS Secretary Janet Napolitano’s suspension of deportations of “DREAMers” (National Immigration Forum 2013),⁷ that ICE officers were instructed not to place undocumented immigrants in removal proceedings or pursue a deportation order if they were considered “low priority.”⁸ Nevertheless, the average immigrant detainee continued to be a much less serious offender than the average prisoner in the criminal justice system, leading to the announcement in November 2014 of the extended Deferred Action for Childhood Arrivals (DACA) and the Deferred Action for Parents of Americans (DAPA) programs.⁹

These rapid changes in detention practices created a complicated legal landscape for undocumented immigrants and their allies. To date, however, we lack a systematic analysis of how the intensification of immigration enforcement starting in 2002 and subsequent DHS prioritization of serious criminal offenses altered the profile of deportees. Gaining a better understanding of the role played by different immigration enforcement approaches on the profile of deportees, however, is essential in the current policy environment. Curtailing unauthorized immigration was at the forefront of President Trump’s campaign. As a candidate, he promised to increase border security through the construction of a wall along the US-Mexico border, initiate the deportation of two to three million undocumented immigrants with criminal records, discontinue the DACA program, and intensify immigration enforcement.¹⁰ On January 25, 2017, President Trump signed an executive order vastly expanding the pool of undocumented immigrants to be considered priorities for deportation, reversing the 2012 priority list.¹¹ President Trump’s administration has also sought the collaboration and cooperation of state and local enforcement agencies with DHS via threats to withhold federal funding to states and cities that attempt to shield undocumented immigrants.¹² Therefore, learning about the success of refocusing apprehension, detention, and deportation efforts on serious criminal offenders, as opposed to implementing a far-reaching immigration enforcement approach, is critical given the costs imposed by deportations on families, the communities in which they reside, and US taxpayers. How effective were past immigration enforcement approaches at targeting serious criminal offenders? Did DHS’s focused enforcement priorities after 2012 lower the share of deported immigrants arrested for minor offenses? Additionally, did the change in enforcement priorities affect detention times and operational costs borne by US taxpayers?

We address these questions using a rich data set on deported Mexican immigrants that inquires about the circumstances surrounding their detention while in the United States, including the location, date, and rationale for their arrest (www.migrante.weebly.com).¹³ The MIGRANTE Project involves a series of cross-sectional, probability surveys of migrants traveling through the Tijuana–San Diego border region. Funded by the US National Institutes of Health, the MIGRANTE surveys include, although are not restricted to, circular and undocumented Mexican migrants to and from the United States.¹⁴ They contain a wealth of information on migration, health, and socioeconomic factors, including the reasons and circumstances surrounding deported migrants’ arrest, which can be used to increase understanding of this hard-to-reach population. Given our focus on the motives behind immigrant apprehensions leading to deportations, we use the MIGRANTE probability survey of deported Mexican migrants conducted from March 2014 to June 2015 in Tijuana, Mexico. We then merge detailed data on the implementation of various immigration enforcement initiatives at the state and county levels to gauge their impact as well as that of DHS’s prioritization of serious criminal offenses on the profile of deportees as captured by the share detained for minor offenses.

We rely on a quasi-experimental approach that compares changes in the propensity of being arrested for a minor offense among deportees in counties that adopted tougher enforcement measures (treated counties) to those in counties that did not (control counties), before and after their implementation, as well as before and after DHS’s later prioritization of serious criminal offenses. Our results reveal that while the intensification of immigration enforcement raised deportees’ likelihood of being arrested for minor offenses, including traffic, disorderly conduct, and drug use violations, such an event became less likely following DHS’s prioritization of serious criminal offenses. Furthermore, the change in enforcement priorities from a widespread to a targeted strategy reduced deportees’ length of time in detention, potentially saving US taxpayers $81 million.

Overall, the findings, which prove robust to a number of identification and robustness checks, underscore the legitimacy of complaints directed at DHS that the vast majority of those arrested were charged with minor offenses or misdemeanors as immigration enforcement intensified. In particular, they are consistent with qualitative reports and anecdotal evidence suggesting increased pressure on immigrant communities, heightened fear of deportation, and destabilization of immigrant families following the strengthening of enforcement efforts.¹⁵ Additionally, the results uncover the value of prioritizing the apprehension and deportation of serious criminal offenders to decrease the costs imposed on many mixed-status families, their communities, and ultimately, US taxpayers. Our hope is that these results refocus future discussions on immigration enforcement and provide evidence of the benefits of prioritizing serious criminal offenses at a time of aggressive immigration enforcement.

Literature Review

As noted earlier, the past two decades have witnessed many changes in immigration policies in the United States, with the implementation of tougher immigration enforcement initiatives contributing to a vast increase in the number of removals since 2001 (Bipartisan Policy Center 2014). The expansion of immigration enforcement from the federal purview to a more fragmented local level through programs like 287(g) and Secure Communities not only considerably contributed to those statistics (Venturella 2010) but also changed the characteristics of those detained, according to a number of studies. For example, Donato and Rodriguez (2014) examine changes in arrest narratives before and after the adoption of a 287(g) agreement in Nashville’s Davidson County. They find that the reasons for stops changed substantially before and after the program’s adoption, with minor violations, such as speeding, careless driving, running a stop sign or red light, or music/muffler/tint violations, representing a significantly greater share of police stops after 287(g) went into effect. They also document how foreign-born residents in particular experienced an increase in pretextual stops — that is, stops involving greater officer discretion. In a similar vein, Koper et al. (2013) examine the effects that 287(g)’s implementation had on crime and disorderly conduct in Prince William County,¹⁶ Virginia. They find that although the number of arrests of undocumented immigrants rose substantially in 2008 and 2009, there were no significant changes in the types of crimes attributed to undocumented immigrants. Furthermore, the vast majority of undocumented immigrants who were arrested were charged with traffic offenses or misdemeanors.

Not only were most arrests for minor offenses, but a number of authors found a generalized practice of racial profiling nationwide. For example, Provine and Sanchez (2011, 474) document how the implementation of 287(g) agreements and, later, SB1070 in Arizona led to more arrests based on appearance. Specifically, reasons for suspicion statements included “smelling like an illegal alien,” “speaking only Spanish,” or “looking dirty and soiled.” Other reasons leading to stops involved the color of the skin, denim clothes, and playing Mexican music. While DHS responded to ongoing racial profiling and discrimination complaints that arose with the 287(g) program by phasing out 287(g) agreements in favor of the Secure Communities program, the latter did not help reduce the Federal Bureau of Investigation’s overall index crime rate, as was initially advertised (Miles and Cox 2014).¹⁷ Despite its goal of keeping communities safer, the Secure Communities program soon triggered critiques from immigrant advocates and other social sectors, which noted that immigrant communities felt more insecure under this program (e.g., Sládková, Garcia Mangado, and Quinteros 2012; Valdez, Padilla, and Valentine 2013). A 2011 American Immigration Lawyers Association report found that many people entered deportation proceedings under Secure Communities despite lacking a serious criminal history. The Transactional Records Access Clearinghouse (2014) examined millions of deportation records since the launch of Secure Communities and found that the program did not increase the removal of its primary targets, noncitizens who committed crimes other than minor violations. Instead, reports found communities were being ripped apart, with impacts extending to businesses, parental school involvement, and overall community cohesiveness (Hagan, Castro, and Rodriguez 2010). Heightened fear of the police among undocumented immigrants and their mixed-status families were also thought to lead to reductions in crime reporting (Sládková, Garcia Mangado, and Quinteros 2012). Moreover, in some instances, a private prison industry seeking to profit from the mass detention of immigrants was the greatest beneficiary of the interplay between immigration enforcement and the criminal justice system (Ackerman and Furman 2013; Martinez and Slack 2013).

In what follows, we exploit the temporal and geographic variation in the intensification of immigration enforcement at the local level to assess the extent to which the distinct approaches to immigration enforcement implemented by DHS from the early 2000s through 2015 shaped the profile of deportees. Specifically, for the first time, we assess how DHS’s prioritization of serious criminal offenses following the intensification of immigration enforcement helped steer apprehension, detention, and deportation efforts away from minor offenders and toward serious criminal offenses after 2012. In addition, we explore if DHS’s prioritization also impacted detention times and operational costs paid for by US taxpayers.

Data and Descriptive Statistics

MIGRANTE Data

Our main data come from the MIGRANTE project,¹⁸ which surveys migrants as they arrive or depart from the border city of Tijuana, Mexico. Migrants eligible for the survey include (1) circular migrants returning from the United States to Mexican sending communities, (2) migrants traveling north from their communities of origin in Mexico, (3) migrants returning to their communities after a stay in the border region, and (4) migrants returning to Mexico forcedly, via deportation. Our focus is on this last group of migrants forcedly removed from the United States with or without a formal deportation order. Even though the survey asks about arrests, it is worth keeping in mind that these are deportees. While the survey does not capture all Mexican deportees, it includes a probability sample of all deportees in Tijuana over the period of our study.

The MIGRANTE probability survey of deported Mexican migrants that we use was conducted between March 2014 and June 2015 in Tijuana, Mexico. In 2015, the San Diego–Tijuana border region accounted for 8 percent of all US Border Patrol apprehensions¹⁹ and 14 percent of all deportations to the Mexican border region.²⁰ Approximately 28,743 migrants were deported through Tijuana in 2015, of whom a target sample size of 1,800 was determined, based on power estimates.²¹ A total of 2,195 individuals were approached, and 2,145 were identified as eligible based on the screening survey. Of them, 2,136 agreed to participate and started the survey, with 2,038 completing the entire survey. The response rate for the survey component, from which data for this study originated, was 95 percent. Survey weights were computed for each observation, following the general principles to estimate survey weights with multistage sampling procedures and account for the survey design, eligibility rates, and response rates.²² The weighted sample represented an estimated population of 18,394 deported migrants. According to data from the Mexican Institute of Migration (Instituto Nacional de Migración 2010), the actual number of migrants deported through Tijuana during the survey period was 17,589.

Immigration Enforcement Data

Our interest is in the role that changes in interior immigration enforcement had on deportees’ propensity of being arrested for a minor offense. To that end, we first gathered data on state- and county-level measures of immigration enforcement adopted over time in the locality in which the migrant resided while in the United States. Specifically, data on the enactment of state-level employment verification (E-Verify) mandates, a key element in several Omnibus Immigration Laws (OIL), and data on OIL themselves were gathered from the National Conference of State Legislatures’ website.²³ Data on the implementation of 287(g) agreements and Secure Communities at the state and county levels were collected from the ICE 287(g) Fact Sheet website²⁴; Kostandini, Mykerezi, and Escalante (2014); and ICE’s Activated Jurisdictions document, respectively (ICE 2013).

We then constructed an index intended to serve as a proxy for the intensity of interior immigration enforcement at the time and location where detention occurred. Our index is the sum of five dichotomous variables, each signaling the existence of an E-Verify mandate at the state level, a state-level OIL, a county-level 287(g) agreement, a state-level 287(g) agreement, and county-level participation in the Secure Communities program, respectively. Each of those five dummy variables equals “1” if the county in which the migrant was detained had adopted the measure in question (month, year) and “0” otherwise. The index is the sum of all five dummies at the time (month, year) and US county where the deportee was detained. Therefore, it can fluctuate between 0 and 5, the number of immigration enforcement initiatives under consideration. Over the time period during which surveyed deportees were in the United States (1989–2015), this index averaged 0.6 (see Table 1).²⁵

Table 1.

Descriptive Statistics.

		High Level of Enforcement Index
	All	Treated Areas	Control Areas
Dependent variables
Detention for minor offense	0.105	0.142	0.068
Detention for minor offense	(0.307)	(0.349)	(0.252)
Detention due to traffic violation	0.074	0.102	0.046
Detention due to traffic violation	(0.262)	(0.303)	(0.209)
Detention due to disorderly conduct	0.025	0.036	0.014
Detention due to disorderly conduct	(0.157)	(0.187)	(0.118)
Detention for drug use	0.013	0.016	0.011
Detention for drug use	(0.114)	(0.124)	(0.102)
Duration in detention (days)	262.267	248.137	276.674
Duration in detention (days)	(740.327)	(504.172)	(921.035)
Regressors
Enforcement Index	0.598	1.191	0
Enforcement Index	(0.695)	(0.505)	(0)
Age	34.172	35.016	33.319
Age	(10.207)	(10.095)	(10.257)
Male	0.807	0.856	0.757
Male	(0.395)	(0.351)	(0.430)
Married	0.421	0.411	0.431
Married	(0.494)	(0.492)	(0.496)
Less than high school education	0.777	0.763	0.792
Less than high school education	(0.416)	(0.426)	(0.407)
High school graduate	0.152	0.172	0.131
High school graduate	(0.359)	(0.377)	(0.338)
More than high school education	0.041	0.040	0.042
More than high school education	(0.198)	(0.196)	(0.201)
Years in the United States	11.914	14.063	9.742
Years in the United States	(10.815)	(11.147)	(10.020)
Number of prior deportations	1.819	1.659	1.981
Number of prior deportations	(2.975)	(2.607)	(3.300)
Indigenous ethnic	0.071	0.064	0.077
Indigenous ethnic	(0.256)	(0.245)	(0.267)
Observations	1,148	577	571

Note. Standard deviations are displayed in parentheses.

We merged the constructed immigration enforcement index with the MIGRANTE data using information on the month and year when the migrant was arrested in the United States and the county in which he or she resided at the time of detention. As can be seen from Table A in the Appendix, the enforcement to which deportees in our sample were exposed rose steadily during the period examined. While the first immigration enforcement measure being considered was implemented in 2002, the first year when deportees in our sample were exposed to intensified enforcement is 2005. The immigration enforcement index grew from an average of 0.03 in 2005 to 0.97 by 2015. Additionally, there was a significant degree of regional variation in immigration enforcement, with the latter reaching higher levels in states like Arizona and lower levels in New York or Washington (see Table A).

Descriptive evidence

Table 1 presents descriptive statistics for our sample of 1,148 deportees.²⁶ We are interested in detentions for reasons we label as minor as they were more likely to be classified as misdemeanors. Our goal is to assess how the likelihood of being arrested and subsequently deported for such motives fared when compared to deportations for serious crimes following the initial intensification of immigration enforcement and DHS’s subsequent prioritization of serious criminal offenses. About 11 percent of deportees reported a minor offense,²⁷ which includes (a) a traffic offense, (b) disorderly conduct, and (c) drug use, as the trigger for their deportation. Traffic offenses were the most common ones, with 7.4 percent of deportees reporting them as the main motive for their apprehension and detention. Detentions for disorderly conduct and those related to drug use followed. Finally, for survey participants who were detained, the average duration of detention was about 262 days, with the average duration of migrant detentions being somewhat shorter in areas with intensified immigration enforcement than in their counterparts without.

Table 1 also shows some general characteristics of deportees in our sample. The vast majority are male (81%) and relatively young (on average, 34 years old). Approximately 42 percent are married, and for the most part, their educational attainment is low. Only 15 percent are high school graduates, and only 4 percent have more than a high school education. On average, they resided in the United States for about 12 years and were deported twice before. When distinguishing based on where they were arrested, we find that deportees who were arrested in areas with tougher enforcement were about 10 percentage points more likely to be male, 4 percentage points more likely to have graduated from high school, and had been in the United States four years longer than their counterparts in areas without such measures. However, they were about the same age, similarly likely to be married, and reported a similar number of previous removals.

Methodology

Our aim is to gain a better understanding of how DHS’s prioritization of criminal offenses following the escalation of interior immigration enforcement shaped the detention profile of deportees after 2012. To that end, we consider the following benchmark regression model:

Y_{i c t} = α + β_{1} E I_{c t} + β_{2} A f t e r 2012_{t} + β_{3} E I_{c t} \times A f t e r 2012_{t} + X_{i c t} γ + ϑ_{c} + φ_{t} + ϑ_{c} t + ε_{i c t},

where $Y_{i c t}$ equals 1 if the ith deportee reported being arrested in county c and year t for a minor offense like a traffic, disorderly conduct, or drug use violation. Otherwise, $Y_{i c t}$ equals 0. The vector $E I_{c t}$ is the enforcement index in county c where the deportee resided in the United States at time (month, year) t. $A f t e r 2012_{t}$ is a dummy indicative of whether the time period refers to the years after DHS’s prioritization of serious criminal offenses, whereas the interaction term that follows allows us to assess how the latter change in policy altered the reasons for which deportees indicated being arrested.

Equation (1) includes the vector X, which contains information on a number of individual-level characteristics, such as age, gender, indigenous ethnicity,²⁸ marital status, educational attainment, years in the United States, and number of prior deportations for each deportee. We also incorporate a number of county fixed effects ( $ϑ_{c}$ ) and time fixed effects ( $φ_{t}$ ) in our model to account for unobserved time-invariant characteristics possibly impacting detentions in a particular county and/or point in time.²⁹ The aforementioned time-invariant characteristics allow us to address, respectively, circumstances such as whether the county has a sheriff particularly tough on immigration, as in the case of Arizona’s notorious Maricopa County with Sheriff Arpaio (sheriff tenure 1993–2016), or whether it was an election year in the United States, during which politicians try to court voters with enforcement measures or initiatives. Finally, equation (1) also includes county-specific ( $ϑ_{c} t$ ) time trends to capture other unobserved time-varying characteristics that might be playing a role but are not explicitly controlled for, including the local specific time trends for changing socioeconomic conditions.³⁰ Equation (1) is estimated as a linear probability model, and standard errors are clustered at the county level.

We are particularly interested in $β_{1}$ and $β_{3}$ , which capture the role that intensified immigration enforcement and DHS’s prioritization of serious criminal offenses played on the profile of deportees as captured by their propensity of being arrested for a minor offense.³¹ Specifically, we hypothesize that:

Hypothesis 1: Intensified immigration enforcement raised the propensity of being arrested for a minor offense, that is, $β_{1} > 0$ , whereas:

Hypothesis 2: DHS’s later prioritization of serious criminal offenses helped lower the propensity of being arrested for a minor offense after 2012, that is, β₃ < 0.³²

Arrests under Widespread Versus Prioritized Enforcement

We estimate a number of model specifications that progressively add controls. Specification (1) in our tables of results is our baseline. Specification (2) adds a number of individual-level characteristics as control variables. Specification (3) further includes county fixed effects to capture unobserved geographic traits potentially driving our estimates. Finally, specification (4) concludes by adding temporal controls, including time fixed effects and county-specific time trends. We focus our discussion on the most complete model specification. In all instances, our reference category is deportees detained for nonminor offenses.

Our first set of results is displayed in Table 2, which assesses the likelihood of being arrested for a minor offense. We would expect the latter to increase as immigration enforcement intensified, as hypothesized in Hypothesis 1. Indeed, according to the estimates in Table 2, the two were positively related until the change in DHS’s apprehension and deportation priorities in 2012. As it turns out, after DHS’s prioritization of serious criminal offenses, the propensity of being arrested for a minor offense versus a nonminor offense became lower as immigration enforcement intensified.³³ Specifically, a one standard deviation increase in the intensity of immigration enforcement (roughly 50% of its average over the period under examination) raised the propensity of being arrested for a minor offense by 10 percentage points (68%) prior to 2012.³⁴ However, that propensity became negative following ICE’s targeted efforts on criminal offenses. The same one standard deviation increase in immigration enforcement now lowered the propensity of being arrested for a minor offense relative to a nonminor offense by 2.1 percentage points (15%).³⁵

Table 2.

Likelihood of Being Arrested for Minor Offenses.

Key Regressors	(1)	(2)	(3)	(4)
Key Regressors	Baseline Model	Plus Controls	Plus County Fixed Effects (FE)	Plus County and Time Trends
Enforcement Index	0.260***	0.204***	0.164***	0.191***
	(0.023)	(0.020)	(0.014)	(0.003)
Enforcement Index × after 2012	–0.211***	–0.180***	–0.164***	–0.233***
	(0.008)	(0.010)	(0.008)	(0.008)
Age		–0.002***	–0.002***	–0.001***
		(4.60e–04)	(2.21e–04)	(8.04e–05)
Male		0.077***	0.061***	0.072***
		(0.007)	(0.001)	(4.27e–04)
Married		–0.023***	–0.033***	–0.029***
		(0.005)	(0.005)	(0.005)
High school graduate		–0.037***	–0.027***	–0.025***
		(0.010)	(0.002)	(0.004)
More than high school education		0.059*	0.026	0.008
		(0.032)	(0.025)	(0.016)
Years in the United States		0.006***	0.004***	0.005***
		(2.24e–04)	(3.18e–04)	(1.14e–04)
Number of prior deportations		–0.007***	–0.005***	–0.007***
		(0.002)	(0.002)	(0.001)
Indigenous ethnic		–0.142***	–0.034***	–0.017***
		(0.010)	(0.002)	(3.38e–04)
County FE	N	N	Y	Y
Year FE	N	N	N	Y
FE and time trends	N	N	N	Y
R ²	0.039	0.099	0.169	0.216
Observations	1,148	1,148	1,148	1,148

Note. All regressions include a constant term. Standard errors are displayed in parentheses.

***p = .01.

We next distinguish according to the type of minor offense: traffic, disorderly conduct, or drug use violations. As shown in Table 3, prior to 2012, a one standard deviation increase in immigration enforcement raised the propensity of being arrested for a traffic, disorderly conduct, or drug use offense by seven, four, and four percentage points, respectively. These were very large increases, ranging from a 69 percent higher propensity following a traffic violation to a 2.5 times upsurge in the propensity of being arrested for a drug use offense. This finding confirms qualitative reports in the literature (e.g., American Immigration Lawyers Association 2011; Frankel 2011). However, after DHS’s prioritization of serious criminal offenses, the likelihood of being arrested for traffic, disorderly conduct, or drug use violations dropped by one percentage point (8%), one percentage point (34%), and half a percentage point (34%), respectively. In sum, refocusing immigration enforcement efforts contributed to reducing detentions for minor offenses as opposed to nonminor ones after 2012, thus changing the prior trend in that regard.

Table 3.

Likelihood of Being Arrested for Minor Offenses.

Key Regressors	(1)	(2)	(3)	(4)
Key Regressors	Baseline	Plus Controls	Plus County Fixed Effects (FE)	Plus County and Time Trends
Panel A: Detained for traffic violation
Enforcement Index	0.154***	0.113***	0.084***	0.139***
	(0.016)	(0.013)	(0.011)	(0.004)
Enforcement Index × after 2012	–0.107***	–0.087***	–0.073***	–0.156***
	(0.004)	(0.004)	(0.005)	(0.008)
R ²	0.025	0.072	0.142	0.180
Panel B: Detained for disorderly conduct
Enforcement Index	0.119***	0.105***	0.096***	0.076***
	(0.007)	(0.009)	(0.004)	(4.90e–04)
Enforcement Index × after 2012	–0.113***	–0.105***	–0.103***	–0.100***
	(0.004)	(0.006)	(0.004)	(1.27e–04)
R ²	0.028	0.042	0.068	0.128
Panel C: Detained for drug use
Enforcement Index	0.068***	0.062***	0.059***	0.078***
	(0.003)	(0.003)	(0.001)	(0.001)
Enforcement Index × after 2012	–0.068***	–0.065***	–0.063***	–0.089***
	(0.004)	(0.004)	(0.002)	(0.001)
R ²	0.018	0.024	0.138	0.216
County FE	N	N	Y	Y
Year FE	N	N	N	Y
FE and time trends	N	N	N	Y
Observations	1,148	1,148	1,148	1,148

Note. All regressions include a constant term and other regressors as in Table 2. Standard errors are displayed in parentheses.

***p = .01.

Identification Tests

A main concern when examining any policy’s impact using a quasi-experimental difference-in-difference approach is whether the apparent impacts predated the policy change. To assess whether that was the case here, we construct an indicator for the year before the immigration enforcement index for a particular county turned positive, which we include, along with the immigration enforcement index, in equation (1). If indeed the policy impact was anticipated or preexisting, we should be able to find a statistically significant coefficient with the same sign on the term indicative of the year before the index turned positive in a given county. Panel A of Table 4 displays the result from such an exercise. It is evident that the observed changes in detention patterns did not take place prior to the adoption of tougher immigration enforcement measures in the county as the coefficient for the preceding year is not statistically different from zero. Furthermore, the point estimates on the enforcement index continue to be statistically different from zero and of similar magnitude to the estimates from the most complete model specification in column (4) of Table 2 and Table 3.

Table 4.

Identification Checks.

	Dependent Variables
Panel A: Parallel Trend Test Key Regressors	(1)	(2)	(3)	(4)
Panel A: Parallel Trend Test Key Regressors	Minor Offenses	Traffic Violation	Disorderly Conduct	Drug Use
Enforcement Index 1 year prior	–0.018	–0.060	0.018	–0.009
	(0.081)	(0.120)	(0.014)	(0.008)
Enforcement Index	0.201***	0.146***	0.080***	0.080***
	(0.006)	(0.006)	(4.82e–05)	(3.08–05)
County fixed effects (FE)	Y	Y	Y	Y
Year FE	Y	Y	Y	Y
FE and time trends	Y	Y	Y	Y
R ²	0.212	0.178	0.125	0.215
Observations	1,148	1,148	1,148	1,148
	Key Regressors
Panel B: Endogeneity Test Dependent Variable	(1)	(2)	(3)	(4)
Panel B: Endogeneity Test Dependent Variable	Share of Detentions for Minor Offenses	Share of Detentions for Traffic Violations	Share of Detentions for Disorderly Conduct	Share of Detentions for Drug Use Violations
Year when the Enforcement Index > 0	–0.628	0.851	0.511	–0.959
	(0.837)	(1.486)	(1.185)	(0.446)
Constant	2,023.943***	2,021.636***	2,022.670***	2,017.222***
	(3.539)	(1.526)	(2.647)	(3.300)
State FE	Y	Y	Y	Y
R ²	0.390	0.432	0.388	0.306
Observations	28	28	28	28

Note. Each column in Panel A shows results from the complete specification (4), and all the regressions include a constant term and other regressors as in Table 2. In Panel B, the dependent variable is the first year when immigration index turned positive, and independent variables are share of deportees declaring being arrested for a reason specified in each column and other controls used in specification 4 of Table 2. Only state fixed effects are added in regressions in Panel B. Standard errors are displayed in parentheses.

***p = .01.

Another concern with policy analyses is the possibility of reverse causality — in our case, whether the timing of the adoption of tougher immigration enforcement measures depended on the share of deportees being arrested for minor offenses. Therefore, we next collapse the data at the county level for the years prior to the time in which the immigration index turned positive in that particular county and model the year the county adopted any of the interior immigration measures under consideration as follows:

E I Y e a r_{c} = α + Y_{c}^{0} μ + X_{c}^{0} π + ε_{c},

where EI Year_c is the year in which the enforcement index turned positive for county c. The vector $Y_{c}^{0}$ represents the average share of deportees indicating being arrested for a minor offense or for any of the specific minor offenses being studied prior to the enforcement index turning positive in that particular county (at point 0). The vector $X_{c}^{0}$ is the vector of controls in equation (1), now aggregated at the county level, thus reflecting average county-level characteristics (e.g., the average share of male deportees, the average share married, or their average educational attainment in the county) prior to the implementation of tougher enforcement. We estimate equation (2) with state fixed effects and cluster standard errors at the state level. The results from this exercise are displayed in Panel B of Table 4. The county’s average share of deportees detained for minor offenses prior to the adoption of stricter enforcement measures does not seem to play a significant role in the adoption timing of tougher immigration enforcement by the county, suggesting the lack of reverse causality.

Robustness Checks

In addition to the aforementioned identification tests, we conducted a number of robustness checks. First, we experimented with an alternative measure of intensified immigration enforcement. Instead of relying on an index, we created a simple dichotomous variable indicative of whether the migrant resided in a county among those in the top 25th percentile in terms of its immigration enforcement intensity, what we refer to as a county with a high level of immigration enforcement. Results from estimating our models using this alternative dichotomous variable are shown in Panel A of Table 5. Overall, the key findings remain unchanged. In areas with tougher immigration enforcement, deportees detained prior to DHS’s prioritization of serious criminal apprehensions were 52 percent more likely to have been detained for a minor offense than for a nonminor offense.

Table 5.

Robustness Checks.

Panel A: Likelihood of Being Arrested for Minor Offenses Using Alternative Measure of Enforcement
Key Regressors	(1)	(2)	(3)	(4)
	Minor Offenses	Traffic Violation	Disorderly Conduct	Drug Use
High Enforcement Index	0.524***	0.329***	0.197***	0.295***
	(0.002)	(0.002)	(2.76e–04)	(7.63e–05)
High Enforcement Index × After 2012	–0.554***	–0.327***	–0.243***	–0.310***
	(0.001)	(0.001)	(0.001)	(2.51e–04)
County fixed effects (FE)	Y	Y	Y	Y
Year FE	Y	Y	Y	Y
FE and time trends	Y	Y	Y	Y
R ²	0.219	0.181	0.131	0.238
Observations	1,148	1,148	1,148	1,148
Panel B: Likelihood of Being Arrested for Minor Offenses by Whether the Locality Had a Trust Act
Key Regressors	Minor Reasons
	(1)		(2)
	Areas with a Trust Act		Areas without a Trust Act
Enforcement Index	–0.022***		0.197***
	(1.18e–04)		(0.005)
Enforcement Index × after 2012	—		–0.302***
			(0.060)
County FE	Y		Y
Year FE	Y		Y
FE and time trends	Y		Y
R ²	0.162		0.326
Observations	918		230
Panel C: Likelihood of Being Arrested by Whether the Locality Issued Driver Licenses to Undocumented Immigrants
	Traffic Violations
Key Regressors	(1)		(2)
	Areas Offering Driver’s Licenses to Immigrants		Areas That Do Not Offer Driver’s Licenses to Immigrants
Enforcement Index	3.39e–04		0.139***
	(1.56e–04)		(0.003)
Enforcement Index × after 2012	—		–0.193***
			(0.036)
Panel C: Likelihood of Being Arrested by Whether the Locality Issued Driver Licenses to Undocumented Immigrants
	Traffic Violations
Key Regressors	(1)		(2)
	Areas Offering Driver’s Licenses to Immigrants		Areas That Do Not Offer Driver’s Licenses to Immigrants
County FE	Y		Y
Year FE	Y		Y
FE and time trends	Y		Y
R ²	0.207		0.187
Observations	776		372

Note. Each column in all panels shows results from the complete specification (4), and all the regressions include a constant term and other regressors as in Table 2. Standard errors are displayed in parentheses. In Panels B and C, the interaction term Enforcement Index × after 2012 drops in columns 1 because Trust Act is activated after 2012 and majority of states allowing driver’s licenses to migrants enacted it after 2012.

***p = .01.

However, DHS’s prioritization of serious criminal offenses in 2012 marked a sudden change. The propensity of being arrested for a minor offense in counties with a high intensity of immigration enforcement effectively reversed its sign, as we saw in Tables 2 and 3 using the index. Deportees indicated being 3 percent, 4.6 percent, and 1.5 percent less likely to be detained for overall minor offenses, disorderly conduct, and drug use after 2012, respectively, hinting on the redirection of efforts toward more serious offenses. The propensity of being arrested for traffic violations dropped as well, from being an event 33 percent more likely to take place prior to 2012 to being practically no more likely (i.e., only 0.2% more likely) after 2012.

As an alternative robustness check, we next assessed the role that the simultaneous implementation of more benevolent policies might have played on our estimates. To that end, we first looked at how the likelihood of being arrested for minor offenses varied across localities depending on whether they had a Trust Act. Trust Acts, enacted at the city, county, and state levels,³⁶ limit cooperation between local law enforcement and federal immigration authorities to increase the community’s trust in the police and facilitate communication between police and community residents. If the gauged impacts in Table 2 are indeed stemming from intensified immigration enforcement, as opposed to an unobserved factor, we would expect the impact to be greater in areas without a Trust Act, where immigration enforcement is likely in full swing.

Results from such an exercise are displayed in Panel B of Table 5. Because the first Trust Act in our sample was not adopted until after 2012, the interaction term drops out. Still, prior to 2012, intensified immigration enforcement raised the likelihood of being arrested for a minor offense in non-sanctuary cities, whereas it had been decreasing in sanctuary cities. Hence, the impacts observed in Table 2 originate from intensified immigration enforcement in areas without a Trust Act.

To conclude, we performed a similar analysis looking instead at the role that the granting of driving licenses to undocumented immigrants might have had in the observed likelihood of being arrested for a traffic violation. As with the Trust Acts, the interaction term drops as most of these were granted after 2012 in our sample.³⁷ Still, it is clear from the estimates that the intensification of immigration enforcement practically had a null impact on the likelihood that deportees were arrested for a traffic violation, as opposed to for any other reasons, when they resided in localities that granted driving licenses to undocumented immigrants. In contrast, in what would be considered less permissive localities (i.e., those not granting a driver’s license to an undocumented immigrant), a one standard deviation increase in immigration enforcement raised the likelihood of being arrested for a traffic violation by seven percentage points (69%) prior to 2012. This propensity dropped after 2012, as shown in Panel A of Table 3. Overall, the impact of intensified immigration enforcement stemmed, once more, from localities that were less permissive of undocumented immigration.

Implications on the Length of Detention and DHS’s Operational Costs

Thus far, we have shown that the intensification of immigration enforcement was accompanied by a higher likelihood that deportees would be detained for minor offenses, including a traffic, disorderly conduct, or drug use offense. Yet that likelihood changed its sign with DHS’s prioritized focus on serious criminal offenders after 2012. Furthermore, these impacts did not appear to be preexisting or contaminated by potential reverse-causality linkages of policies. Rather, they stemmed from tougher immigration enforcement, as is evident by the fact that they were present in localities without a Trust Act and in the case of traffic violations, from those less likely to grant driving licenses to undocumented immigrants. They also appeared robust to the use of an alternative dichotomous immigration enforcement indicator.

A number of studies have documented the collateral damage of detentions and deportations on immigrants’ families, given that most detentions involved fathers and heads of household.³⁸ In addition to these costs, there are operational costs. Despite ICE’s more focused enforcement priorities, DHS requested $1.84 billion for Custody Operations in Fiscal Year 2014, approximately $5 million/day for immigrant detentions (National Immigrant Forum 2013). Hence, it is reasonable to ask if in addition to reducing the likelihood of such unintended consequences, DHS’s prioritization of serious criminal offenses in 2012 helped shorten detentions and decrease operational costs relative to the indiscriminate, sweeping immigration enforcement strategy that DHS is currently adopting under Trump’s administration.

Table 6 looks at this question by examining how the intensification of immigration enforcement as well as DHS’s later prioritization of serious criminal offenses impacted the length of time migrants were held prior to being deported.³⁹ According to the estimates in the most complete specification, a one standard deviation in the immigration enforcement index raises the duration of immigrant detentions by 175 days prior to 2012. At the average daily cost of $159/detainee and a capacity of 31,800 detainees (National Immigration Forum 2013), the increased duration translated into an additional $885 million cost.⁴⁰ With the announcement of DHS’s more focused approach on immigration enforcement in 2012, detention duration dropped by 16 days, lowering the operational cost imposed by intensified immigration enforcement by $81 million.⁴¹ By prioritizing the detention of serious criminal violators, the policy change might have helped lower the number of detainees held at detention centers, alleviating backlogs at immigration courts and reducing detention times. These figures, which represent a lower-bound estimated cost reduction if the overall number of detentions was also dropping, emphasize just another cost of the earlier sweeping approach to immigration enforcement. This cost, however, is borne not just by deportees, their families, and the social fabric for the communities in which they had integrated but also by US taxpayers. Since the new immigration guidelines mark a significant overhaul of deportation policies under President Obama, the aforementioned estimates can potentially be a lower-bound estimated cost that the US taxpayers now must shoulder each year.

Table 6.

Detention Duration.

Key Regressors	(1)	(2)	(3)	(4)
Key Regressors	Baseline	Plus Controls	Plus County Fixed Effects (FE)	Plus County and Time Trends
Enforcement Index	1,020.031***	823.244***	627.226***	345.336***
	(53.010)	(54.840)	(13.729)	(10.151)
Enforcement Index × after 2012	–1,087.544***	–946.784***	–859.168***	–361.303***
	(40.968)	(26.873)	(11.454)	(12.181)
County FE	N	N	Y	Y
Year FE	N	N	N	Y
FE and time trends	N	N	N	Y
R ²	0.102	0.181	0.299	0.886
Observations	1,135	1,135	1,135	1,135

Note. Detention duration is measured in days. All regressions include a constant term and other regressors as in Table 2. Standard errors are displayed in parentheses.

***p = .01.

Summary and Conclusions

Since the signing of the first 287(g) agreement between ICE and the state of Florida in 2002, the United States has witnessed an intensification of immigration enforcement at the state and county levels that has been found to be responsible for 1.8 million deportations between 2009 and 2013 (Vaughan 2013). The increased number of removals, however, was accompanied by protests regarding racial profiling and growing public awareness of the many negative impacts of intensified enforcement on undocumented immigrants’ family members, many of whom were US-born children. DHS’s prioritization of serious criminal offenses and ICE’s more focused deportation efforts after 2012 came, partially, in response to such concerns.

In this analysis, we use a rich data set on deported Mexican immigrants that inquires about circumstances surrounding their detention while in the United States, including the location, date, and rationale for their detention, and merge it with data on the implementation of a number of immigration enforcement initiatives at the state and county levels. Because of the many complaints about unsubstantiated stops, we look at how deportees’ propensity of being arrested for minor offenses was impacted by the initially tougher approach to immigration enforcement seen in the mid- to late 2000s. We find that intensified interior immigration enforcement raised the propensity that deportees would indicate being arrested for a minor offense, including a traffic, disorderly conduct, or drug use violation. Subsequently, we looked at whether the likelihood of such an event changed following DHS’s refocused efforts on criminal arrests and deportations. The established enforcement priorities did steer DHS’s focus away from minor offenses and reduced the length of time deportees were held in detention, pointing to potentially substantial savings in the order of $81 million yearly.

This analysis underscores the truthfulness of complaints directed at DHS as immigration enforcement intensified as well as the importance of prioritizing apprehensions and deportations of serious criminal offenders given the costs imposed by deportations on families, the communities in which they reside, and US taxpayers. In the absence of prioritized enforcement, local authorities appeared more likely to loosely interpret federal immigration laws, leading to more arrests and deportations for minor offenses. These findings are crucial in light of the sweeping changes in immigration policy under President Trump. The first major immigration raids under his administration manifested a clear shift in deportation strategy, one that, contrary to the president’s claims during his campaign, is targeting undocumented immigrants without serious criminal records.⁴² Although official statistics about the number of arrests and deportations over the first 12 months of the Trump administration are not yet available, ICE data from the first 100 days of the Trump administration suggest a dramatic switch from targeted to indiscriminate enforcement practices throughout the nation. Overall arrests have increased by 40 percent, and noncriminal arrests have risen by 150 percent when compared to the same period in the previous year.⁴³ Particularly worrisome is the fact that these detentions are taking place more and more often at routine traffic stops or for minor traffic infractions (e.g., failure to signal a turn)⁴⁴ and in settings until now “off limits,” such as courthouses and hospitals.⁴⁵ As our study demonstrates, such an approach is dubiously effective at deporting “bad hombres” and will raise the DHS’s operational costs to be paid for by US taxpayers. More importantly, this broad approach will come at the cost of extraordinary suffering for thousands of families who contribute in many valuable ways to our diverse social fabric and will push immigrant communities (with and without legal status) further into the shadows and away from the American dream. Given the limited resources and the damage caused on many US immigrant communities and the larger society around them, the findings underscore the value of prioritizing immigration enforcement and the need for a long-awaited comprehensive immigration reform.

Footnotes

Appendix

Table B.

Description of Dependent Variables.

Variable Name	Definition
Arrest variables	Question asked: “What was the reason you were detained this last time?”
Arrest for minor offense	Equals 1 if respondent’s answer to the question is either: traffic-related reason (e.g., driving under the influence, driving too fast, parked in wrong spot, etc.) disorderly conduct drug use Equals 0 otherwise
Arrest due to traffic violation	Equals 1 if respondent’s answer to the question is: traffic-related reason (e.g., driving under the influence, driving too fast, parked in wrong spot, etc.) Equals 0 otherwise
Arrest due to disorderly conduct	Equals 1 if respondent’s answer to the question is: disorderly conduct Equals 0 otherwise
Arrest for drug use	Equals 1 if respondent’s answer to the question is: drug use Equals 0 otherwise
Duration in detention (days)	Question asked: “For how long have you been held in a detention center, prison, or jail since the last time you were detained?” The duration is measured in number of days.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by the National Institutes of Health (NIH) and Eunice Kennedy Shriver National Institute of Child Health and Human Development (grant R01 HD046886).

Notes

References

Ackerman

Furman

. 2013. The Criminalization of Immigration: Contexts and Consequences. Durham, NC: Carolina Academic Press.

American Immigration Lawyers Association. 2011. Immigration Enforcement Off Target: Minor Offenses with Major Consequences, AILA InfoNet Doc. No. 11081609. http://www.aila.org/File/Related/11081609.pdf.

Amuedo-Dorantes

Puttitanun

Martinez-Donate

A. P

. 2013. “How Do Tougher Immigration Measures Affect Unauthorized Immigrants?” Demography 50 (3): 1067–91.

Bipartisan Policy Center. 2014. “ Issue Brief: Interior Immigration Enforcement by the Numbers.” http://bipartisanpolicy.org/wp-content/uploads/sites/default/files/files/Interior%20Immigration%20Enforcement.pdf.

Brabeck

. 2010. “ The Impact of Detention and Deportation on Latino Immigrant Children and Families: A Quantitative Exploration.” Hispanic Journal of Behavioral Sciences 32 (3): 341–61.

Brooks-Gunn

Duncan

G. J.

Maritato

. 1997. “Poor Families, Poor Outcomes: The Well-Being of Children and Youth.” In Consequences of Growing Up Poor, edited by Duncan

Brooks-Gunn

, 1–17. New York: Russell Sage Foundation.

Chishti

Bergeron

. 2011. “ Post-9/11 Policies Dramatically Alter the U.S. Immigration Landscape.” https://www.migrationpolicy.org/article/post-911-policies-dramatically-alter-us-immigration-landscape.

Delva

Horner

Martinez

Sanders

Lopez

W. D

Doering-White

. 2013. “Mental Health Problems of Children of Undocumented Parents in the United States: A Hidden Crisis.” Journal of Community Positive Practices 13 (3): 25–35.

Donato

K. M.

Rodriguez

L. A.

. 2014. “ Police Arrests in Time of Uncertainty: The Impact of 287(g) on Arrests in a New Immigrant Gateway.” American Behavioral Scientist 58 (13): 1696–722.

10.

Dreby

2012. “ How Today’s Immigration Enforcement Policies Impact Children, Families, and Communities: A View from the Ground.” https://www.americanprogress.org/wp-content/uploads/2012/08/DrebyImmigrationFamiliesFINAL.pdf

11.

Frankel

2011. “ Detention and Deportation with Inadequate Due Process: The Devastating Consequences of Juvenile Involvement with Law Enforcement for Immigrant Youth.” Duke Forum for Law and Social Change 63. http://chicagounbound.uchicago.edu/cgi/viewcontent.cgi?article=2427&context=journal_articles.

12.

Hagan

Castro

Rodriguez

. 2010. “ The Effects of U.S. Deportation Policies on Immigrant Families and Communities: Cross-Border Perspectives.” North Carolina Law Review 88: 1799–1824.

13.

Immigration and Customs Enforcement. 2013. “ Activated Jurisdictions.” http://www.ice.gov/doclib/secure-communities/pdf/sc-activated.pdf.

14.

Instituto Nacional de Migración. 2010. “ Centro de Estudios Migratorios. Estadísticas Migratorias. Boletines Estadísticos. Boletín Mensual de Estadísticas Migratorias.” http://www.inm.gob.mx/estadisticas/2010/BoletinEst2010.pdf.

15.

Kandel

W. A.

2016. “Interior Immigration Enforcement: Criminal Alien Programs,” Congressional Research Service Report R44627. Washington, DC: Congressional Research Service.

16.

Koper

C. S.

Guterbock

T. M.

Woods

D. J.

Taylor

Carter

T. J.

. 2013. “ The Effects of Local Immigration Enforcement on Crime and Disorder.” Criminology& Public Policy 12 (2): 239–76.

17.

Kostandini

Mykerezi

Escalante

. 2014. “ The Impact of Immigration Enforcement on the U.S. Farming Sector.” American Journal of Agricultural Economics 96 (1): 172–92.

18.

Martinez

Slack

. 2013. “ What Part of ‘Illegal’ Don’t You Understand? The Social Consequences of Criminalizing Unauthorized Mexican Migrants in the United States.” Social& Legal Studies 22 (4): 535–51.

19.

Menjivar

. 2006. “ Liminal Legality: Salvadoran and Guatemalan Immigrants’ Lives in the United States.” American Journal of Sociology 111 (4): 999–1037.

20.

Miles

T. J.

Cox

A. B.

. 2014. “ Does Immigration Enforcement Reduce Crime? Evidence from ‘Secure Communities.’” Journal of Law & Economics 57 (4): 937–73.

21.

Mittelstadt

Speaker

Meissner

Chishti

. 2011. “ Through the Prism of National Security: Major Immigration Policy and Program Changes in the Decade Since 9/11.” https://www.migrationpolicy.org/research/post-9-11-immigration-policy-program-changes.

22.

National Immigration Forum. 2013. The Math of Immigration Detention: Runaway Costs for Immigration Detention Do Not Add up to Sensible Policies. Washington, DC: National Immigration Forum.

23.

Passel

J. S.

Lopez

M. H.

Cohn

Rohal

. 2014. As Growth Stalls, Unauthorized Immigrant Population Becomes More Settled. Washington, DC: Pew Rsearch Center.

24.

Provine

D. M.

Sanchez

. 2011. “ Suspecting Immigrants: Exploring Links between Racialist Anxieties and Expanded Police Power in Arizona.” Policing& Society 21 (4): 468–79.

25.

Rosenblum

M. R.

McCabe

. 2014. Deportation and Discretion: Reviewing the Record and Options for Change. Washington, DC: Migration Policy Institute.

26.

Sládková

Mangado

S. M. Garcia

Quinteros

J. Reyes

. 2012. “ Lowell Immigrant Communities in the Climate of Deportations.” Analyses of Social Issues and Public Policy 12: 78–95.

27.

Transactional Records Access Clearing house. 2014. “ Secure Communities and ICE Deportation: A Failed Program?” http://trac.syr.edu/immigration/reports/349/

28.

US Department of Homeland Security. 2010. Yearbook of Immigration Statistics 2010. https://www.dhs.gov/immigration-statistics/yearbook.

29.

US Department of Homeland Security. 2013. Yearbook of Immigration Statistics 2013. https://www.dhs.gov/immigration-statistics/yearbook.

30.

US Department of Homeland Security. 2014. Yearbook of Immigration Statistics 2014. https://www.dhs.gov/immigration-statistics/yearbook.

31.

US Department of Homeland Security. 2015. Yearbook of Immigration Statistics 2015. https://www.dhs.gov/immigration-statistics/yearbook.

32.

Valdez

C. R.

Padilla

Valentine

J. Lewis

. 2013. “ Consequences of Arizona’s Immigration Policy on Social Capital among Mexican Mothers with Unauthorized Immigration Status.” Hispanic Journal of Behavioral Sciences 35 (3): 303–22.

33.

Vaughan

J. M.

2013. “ Deportation Numbers Unwrapped Raw Statistics Reveal the Real Story of ICE Enforcement in Decline.” https://cis.org/sites/default/files/Deportation-Numbers-Unwrapped.pdf.

34.

Venturella

D. J.

2010. “ Secure Communities: Identifying and Removing Criminal Aliens,” The Police Chief 77 (September): 40–49. http://www.nxtbook.com/nxtbooks/naylor/CPIM0910/index.php#/40.