AN EMPIRICAL EXAMINATION OF IMMIGRATION DETENTION

In our nation’s immigration system, a noncitizen charged with deportability may be detained pending the outcome of removal proceedings. These individuals are housed in remote facilities closely resembling prisons, with severe restrictions on access to counsel and contact with family members. Many of the noncitizens initially detained by enforcement officials have the opportunity to request a bond hearing before an administrative adjudicator called an Immigration Judge (IJ). Although these IJs preside over relatively formal on-the-record hearings and are understood to exercise “independent judgement,” concerns have been raised that they have been subject to control by political superiors in the Executive Branch.

This Article analyzes roughly 630,000 individual custody decisions by IJs from 2001 through June 2019 to explore this question. Using statistical analysis to control for over a dozen variables, this Article explores the impact that political, economic, legal, and institutional variables play in impacting the IJ’s decision to release a noncitizen on bond. The Article also offers a descriptive assessment of how border court judges differ in their detention practices than judges hearing cases in other cities.

2 IMMIGRANT DETENTION [28-Jan-20INTRODUCTION

In our nation’s immigration system, a noncitizen charged with deportability may be detained pending the outcome of removal proceedings. The immigrant detention system constitutes the largest single system for confinement in our nation. Almost half a million individuals were detained in fiscal year 2018,1 and on any given day, up to 47,000 noncitizens will be held.2 Although immigrant detention has been characterized as civil rather than penological in nature,3 immigrant detention facilities are virtually indistinguishable from jails and prisons.4 Many are operated 1

U.S. IMMIGRATION AND CUSTOMS ENFORCEMENT, FISCAL YEAR 2018 ICE ENFORCEMENT AND REMOVAL OPERATIONS REPORT, at https://www.ice.gov/doclib/about/offices/ero/pdf/eroFY2018Report.pdf; see also Dep’t of Homeland Security, Office of Immigration Statistics, ANNUAL REPORT FOR IMMIGRATION ENFORCEMENT ACTIONS: 2017, at 10 (reporting 323,591 initial admissions into ICE Detention facilities in 2017, from a high of 464,190 in 2012). Detentions of individuals from Mexico and “Northern Triangle” countries of Honduras, El Salvador and Guatemala constituted 83% of the total. Id.

2 Congress through the appropriations process imposed a 47,000-bed quota for immigrant detention, including 44,500 adult beds and 2,500 family beds. DEP’T OF HOMELAND SECURITY, U.S. IMMIGRATION AND CUSTOMS ENFORCEMENT BUDGE OVERVIEW: FISCAL YEAR 2019 CONGRESSIONAL JUSTIFICATION, at https://www.dhs.gov/sites/default/files/publications/U.S.%20Immigration%20and%20Customs%20Enforcement.pdf.

3 Wong Wing v. United States, 163 U.S. 228, 235 (1896) (invalidating criminal punishment for unlawfully present aliens without trial, but noting, “We think it clear that detention or temporary confinement, as part of the means necessary to give effect to the provisions for the exclusion or expulsion of aliens, would be valid.”). But see Cesar Cuauhtemoc Garcia Hernandez, Immigration Detention as Punishment, 61 U.C.L.A. L. REV. 1346, 1349-50 (2014) (identifying similarities between immigrant detention and criminal incarceration); Jennifer M. Chacon, Immigration Detention: No Turning Back?, 113 S. ATLANTIC Q. ___ (2014) (“The glaring problem with the legal doctrine that constructs immigration detention as nonpunitive is that it is a fiction. Detention is punitive, and it is experienced as such by immigrants.”); Anil Kalhan, Rethinking Immigration Detention, 110 COLUM. L. REV. SIDEBAR 42 (2010) (suggesting that immigrant detention has evolved into a “quasi-punitive system of incarceration”).

4 Dora Schriro, DEP’T OF HOMELAND SECURITY, IMMIGRATION DETENTION OVERVIEW AND RECOMMENDATIONS (Oct. 6, 2009) (hereinafter SCHRIRO REPORT) (“With only a few exceptions, the facilities ICE uses to detain aliens were build and operate, as jails and prisons to confine pre-trial and sentenced felons” and “ICE relies primarily on correctional

28-Jan-20] IMMIGRANT DETENTION 3by private prison corporations and are in remote locations far from detainees’ communities.5 There are significant restrictions on access to counsel6 and contact with family members.7 Sub-standard conditions in these facilities have been documented extensively.8 The government and media outlets have reported the lack of access to medical services,9 lack of hygiene in bathrooms and food,10 and the extensive use of solitary confinement,11 for example. Given the severe backlog in the adjudication of removal proceedings, the detention of an immigrant may last months or even years.12

Many of the individuals initially detained by immigration

incarceration standards … for care, custody and control”).5 Emily Ryo & Ian Peacock, A National Study of Immigration

Detention in the United States, 92 S. CAL. L. REV. 1, 28-29 (2018).6 See Ingrid V. Eagly & Steven Shafer, A National Study of Access to

Counsel in Immigration Court, 164 U. PENN. L. REV. 1, 32 (2015) (finding that non-detained respondents in removal proceedings are five times more likely to secure representation by counsel than detained respondents).

7 Unfortunately, in many cases the noncitizen is detained with family members, including children. See generally Ingrid Eagly, Steven Shafer, & Jana Whalley, Detaining Families: A Study of Asylum Adjudication in Family Detention, 106 CAL. L. REV. 785 (2018) (examining asylum adjudication for families in detention).

8 See e.g., DEP’T HOMELAND SECURITY, OFFICE OF THE INSPECTOR GENERAL, CONCERNS ABOUT ICE DETAINEE TREATMENT AND CARE AT FOUR DETENTION FACILITIES (June 3, 2019) (documenting results from unannounced site visits at four detention facilities and reporting unhealthy conditions, absence of outside recreation facilities, unjustified strip searches, among other violations); SCHRIRO REPORT, supra note 4, at __.

9 SCHRIRO REPORT, supra note 4, at _; see also GOVT. ACCOUNTABILITY OFFICE, Immigrant Detention: Additional Actions Needed to Strengthen Management and Oversight of Detainee Medical Care, GAO-16-231; GAO-14-38.

10 Immigrant Detention: Additional Actions Could Strengthen DHS Efforts to Address Sexual Abuse; DHS OIG, Concerns About ICE Detainee Treatment and Care at Four Detention Facilities, OIG-19-47 (June 3, 2019).

11 Ian Urbina, The Capricious Use of Solitary Confinement Against Detained Immigrants, THE ATLANTIC (Sept. 6, 2019).

12 Ryo & Peacock, supra note 5, at __ (finding that in fiscal year 2015, the average length of immigrant detention was 38 days, but that “tens of thousands were detained for many months or even years.”).

4 IMMIGRANT DETENTION [28-Jan-20enforcement officers,13 though by no means all,14 have the right to a bond hearing before an immigration judge (IJ) to argue for their release.15 IJs, the same officials who ultimately determine whether the noncitizen will be deported, are adjudicatory officials who preside over formal on-the-record hearings.16 Immigration judges are understood to exercise “independent judgment” in their decision making.17 Indeed, due process arguably mandates such independence, given the length and conditions of immigrant detention.18

Notwithstanding norms of adjudicatory independence, however, immigration judges may be vulnerable to influence by their political superiors in the Executive Branch. After all, they are housed within the Department of Justice’s Office for Immigration Review (EOIR) and explicitly subordinate to the Attorney General.19 They enjoy only those civil service

13 Immigration enforcement officials within the Department of Homeland Security’s Immigration and Customs Enforcement (ICE) agency who apprehend a noncitizen suspected of removability make the initial determination of whether the noncitizen will be detained pending the outcome of removal proceedings. For aliens who are not subject to mandatory detention, see infra note 14, the ICE official may release on conditional parole (release on recognizance) or on bond of at least $1,500. 8 C.F.R. §§ 236.1; 1236.1(d)(1); 1003.19.

14 Congress mandates the detention of certain categories of noncitizens in removal proceedings, denying any opportunity for bond altogether. These aliens include aliens apprehended at the border, as well as those removable on certain criminal and national security grounds, 8 U.S.C. §§ 1225(b); 1226(a), (c); see also Demore v. Kim, 538 U.S. 510 (2003) (affirming constitutionality of mandatory detention under section 1226(c)). The GAO reported that from 2011 to 2013, 77% to 80% of aliens in detention facilities were subject to mandatory detention. GOVT. ACCOUNTABILITY OFFICE, Alternatives to Detention: Improved Data Collection and Analyses Needed to Better Assess Program Effectiveness 28, GAO 15-26.

15 8 C.F.R. § 1003.19.16 8 U.S.C. §1129a.17 8 C.F.R. § 1003.10.18 See generally, Kent H. Barnett, Some Kind of Hearing Officer, 94

WASH. L. REV. ___ (forthcoming 2019) (discussing due process interest in impartiality); Gerald Neuman, Federal Courts Issues in Immigration Law, 78 TEX. L. REV. 1661 (2000); Martin H. Redish & Lawrence C. Marshall, Adjudicatory Independence and the Values of Procedural Due Process, 95 YALE L.J. 455 (1986).

19 8 U.S.C. § 1101(b)(4) (defining “immigration judge” as “an attorney whom the Attorney General appoints” and who “shall be

28-Jan-20] IMMIGRANT DETENTION 5protections extended to ordinary federal government employees.20 They do not even enjoy the statutory tenure protections of Administrative Law Judges (ALJs),21 much less those of an Article III federal judge.22

It is commonly accepted that the decisions of enforcement officials may and perhaps should comply with the policy preferences of the President. But whether adjudicatory decisions, albeit by administrative officials, should be made on the basis of the President’s political agenda is far more controversial.23 At present, however, little is known about whether, as an empirical matter, such compliance exists, or whether adjudicators preserve decision-making independence.

In earlier work, we evaluated the extent to which IJs operate independently from their executive branch political superiors in rendering the final decision to remove.24 Using logistical regression and controlling for over a dozen

subject to such supervision and shall perform such duties as the Attorney General shall prescribe”).

20 U.S. DEP’T OF JUSTICE, OFFICE OF PROF’L RESPONSIBILITY AND OFFICE OF THE INSPECTOR GEN., AN INVESTIGATION INTO ALLEGATIONS OF POLITICIZED HIRING BY MONICA GOODLING AND OTHER STAFF IN THE OFFICE OF THE ATTORNEY GENERAL (2008) (hereinafter DOJ INVESTIGATION INTO POLITICIZED HIRING) (outlining civil service protections for IJs).

21 See generally, KENT BARNETT, ET AL., ADMIN. CONF. OF THE U.S., FINAL REPORT: NON-ALJ ADJUDICATORS IN FEDERAL AGENCIES: STATUS, SELECTION, OVERSIGHT, AND REMOVAL (May 11, 2018) (comparing tenure protections of ALJs and other agency adjudicators including immigration judges).

22 U.S. Const. Art. III, sec. 1. 23 See, e.g., Henry J. Friendly, The Federal Administrative Agencies:

The Need for Better Definition of Standards, 75 HARV. L. REV. 1263, 1300 (1962) (“Everyone, including presidential activists, seems to agree that ‘the outcome of any particular adjudicatory matter is . . . as much beyond . . . [the President’s] concern . . . as the outcome of any cause pending in the courts . . . .”); Elena Kagan, Presidential Administration, 114 HARV. L. REV. 2245, 2363 (2001) (generally celebrating presidential control over agencies but conceding that in the context of administrative adjudication, “presidential participation…, of whatever form, would contravene procedural norms and inject an inappropriate influence into the resolution of controversies”); Adrian Vermeule, Conventions of Agency Independence, 113 COLUM. L. REV. 1163, 1211 (2013) (asserting existence of a “network of tacit unwritten conventions” protecting agency adjudications from political interference).

24 Catherine Y. Kim & Amy Semet, An Empirical Study of Political Control Over Immigration Adjudication, 108 GEO. L.J. __ (forthcoming 2019).

6 IMMIGRANT DETENTION [28-Jan-20variables associated with deportation outcomes, we found that the identity of the Presidential Administration that appointed the IJ was not a statistically significantly factor in predicting whether or not that judge would order a noncitizen deported from the country. We did find, however, that the identity of the Presidential Administration at the time of the removal decision was a statistically significant predictor of removal outcomes.25 For example, Obama appointees were 7% more likely to order removal during the current administration than during the former one.26 These results suggest that the current Executive-in-Chief may exert some measure of direct or indirect influence over IJ decisions to deport.

This project shifts our attention from the final outcome of removal proceedings to the decision to detain pending removal proceedings. Different Presidential Administrations have been explicit in their policy preferences with respect to this question. For example, the Obama Administration expressed a clear preference for releasing noncitizens, particularly those who are ill, disabled, pregnant, nursing, or who are primary caretakers of children or the infirm.27 The Trump Administration, by contrast, has made clear its policy preference for maximizing the number of noncitizens detained.28

Anecdotal reports suggest that IJs have been responsive 25 Id.26 Id.27 MEMORANDUM FROM JEH JOHNSON, SEC’Y OF HOMELAND SECURITY, POLICIES

FOR THE APPREHENSION, DETENTION AND REMOVAL OF UNDOCUMENTED IMMIGRANTS, Nov. 20, 2014.

28 Exec. Order No. 13767, 82 FED. REG. 8793 (Jan. 25, 2017) (announcing policy of detaining all individuals suspected of violating immigration laws). President Trump and his former Attorney General Jeff Sessions announced the end of a policy they termed “catch and release,” whereby noncitizens are apprehended and then released while their removal proceedings are pending. See, e.g., WHITE HOUSE OFFICE OF THE PRESS SEC’Y, Trump Administration Immigration Policy Priorities (Oct. 8, 2017); MEMORANDUM FROM THE PRESIDENT TO SECY’S OF STATE, DEFENSE, ATTORNEY GENERAL, ET AL., Ending “Catch and Release” at the Border of the United States and Directing Other Enhancements to Immigration Enforcement, 83 FED. REG. 16179 (Apr. 13, 2018); DEP’T OF JUSTICE PRESS RELEASE, Attorney General Sessions Delivers Remarks on Immigration Enforcement (Apr. 11, 2018); DEP’T OF JUSTICE NEWS RELEASE, Attorney General Sessions Delivers Remarks to the Exec. Off. For Imm. Rev. Legal Training Program (Jun. 11, 2018).

28-Jan-20] IMMIGRANT DETENTION 7to those preferences. One media report quotes immigration practitioners as stating, “The Obama Administration directed IJs to use their discretion to release eligible immigrants on low-cost bonds or without any bond at all, a form of parole known as ‘release on recognizance.’ That is no longer the case under President Donald Trump…. Instead, immigration court judges … are increasingly denying bond requests altogether, or setting them at amounts in excess of $10,000.”29

This project seeks to examine IJs’ responsiveness to the policy preferences of their political superiors within the Executive Branch. We analyze government data from approximately 630,000 bond proceedings held in immigration courts from January 2001 to June 2019.30 The data generally provide detailed information on the number of bond hearings requested; whether the individual was released on recognizance, granted bond, or denied bond; and the identity of the IJ who made the decision. The data also provide the amount at which bond was lowered. We analyze these data to identify descriptive political trends in bond decisions across different Presidential Eras. For example, are noncitizens less likely to be released on recognizance during the Trump Era as compared to preceding Administrations? Are they more likely to be denied bond altogether? Where bond is granted, are bond

29 PBS News Hour, Under Trump, Higher Immigration Bonds Mean Longer Family Separations (Jun. 28, 2018).

30 EOIR CASE DATA, at https://www.justice.gov/eoir/frequently-requested-agency-records (last visited July 1, 2019). TRAC researchers at Syracuse University in 2008 successfully filed a lawsuit under the Freedom of Information Act (FOIA) to force EOIR to release the data, and EOIR published these data on its website pursuant to reporting standards under the FOIA Improvement Act of 2016, Public Law No. 114-185 (2016). We limited the analysis to cases involving removal, detention, or exclusion proceedings. Overall, 99% of the bond proceedings fell within these categories. See EOIR, IMMIGRATION COURT PRACTICE MANUAL (Dec. 2016), available at https://www.justice.gov/eoir/office-chief-immigration-judge-0. We eliminated the several hundred observations of bond proceedings for the following case types: (1) “credible fear” cases, 8 C.F.R. § 1003.42; (2) “withholding only” cases, 8 U.S.C. § 1231(b)(3); “reasonable fear” cases involving aliens with a reinstated order of removal, 8 C.F.R. § 208.31; (4) “asylum-only” cases; (5) “claimed status review”; and (6) claims under the Nicaraguan Adjustment and Central American Relief Act (NACARA). Id.

8 IMMIGRANT DETENTION [28-Jan-20rates higher today than they were during the Obama or Bush II Eras? And finally, when the IJ lowers bond, do bond amounts diverge further from the initial amount set by ICE enforcement officials during certain Administrations than others? We further analyze how different appointee cohorts behave during each Era. For example, do IJs appointed by Clinton differ in their behavior across different Presidential Eras? These descriptive statistics help differentiate between mechanisms of political influence exercised through the power to appoint, and mechanisms exercised through the power to supervise.

We find that on every metric of bond hearings, noncitizens fare considerably worse during the Trump Era than they did during either the Bush II or Obama Eras. While rates of release on recognizance were extremely low throughout the period of study, they started at 1.92% of all cases decided during the Bush II Era, dropping to 0.23% during the Obama Era and then to 0.18% during the Trump Era. Similarly, while only 7% of custody hearings during the Bush II Era resulted in an outright denial of bond, that figure rose to 14% during the Obama Era and 19% during the Trump Era. Perhaps more telling, overall win rates (release on recognizance and reduced bond amount) indicate that all appointee cohorts except Obama appointees were considerably less likely to award relief to the noncitizen during the Trump Administration than during prior Eras. For example, while IJs appointed during the Bush I Administration granted a favorable outcome to the noncitizen in 42% of all cases between 2001 and 2019, they awarded such relief in only 16% of cases during the Trump Era. These results suggest that the Trump Administration is exercising control over bond decisions not only through its power to appoint more like-minded IJs, but also its power to supervise earlier appointees.

An examination of bond amounts when the IJ reduces the amount set ICE reveals a similar picture. Bond medians grew from $5,000 during the Bush II Era, to $5,500 during the Obama Era, and then jumped to $7,500 during the Trump Era. Indeed, 41% of the bonds set by IJs during the Trump Era were $10,000 or higher, as compared to only 23% and 22% for the Obama and Bush Eras, respectively. Again, breaking down these results by appointee cohort

28-Jan-20] IMMIGRANT DETENTION 9indicates that earlier-appointed IJs have behaved differently during the Trump Era than during preceding Administrations.

Finally, an analysis of the difference in bond amounts set by IJs when reducing those initially set by ICE enforcement officers shows that these amounts have been converging through time. One interpretation of these data might be that IJs are acting more similarly to ICE enforcement officials than in the past.

Our study is necessarily limited and does not seek to make any causal statements. We use simple bivariate analyses to see whether bond decisions have changed during different Presidential Eras, and for different appointee cohorts. We do not control for the multitude of other variables that might otherwise impact an IJs bond decision, such as the individual circumstances of the noncitizen, changes in migration patterns, or the socio- political- or economic contexts in which bond decisions are made, for example. As such our conclusions are purely descriptive. Nonetheless, our analysis provides a rich picture of political trends in decisions to detain in immigration courts.

This Essay proceeds as follows. Part I sets forth the legal framework for immigrant custody decisions in immigration courts and then summarizes some of the relevant scholarly literature. Part II sets forth our analyses. Part III considers avenues for further research. We conclude with some thoughts on the appropriate role of political actors in immigrant detention decisions.

I. BACKGROUNDA. The Legal Framework for Immigrant Detention

Decisions in Immigration CourtThis subsection sets forth the legal framework in which

IJs render immigrant custody and bond determinations. It begins with an overview of the bureaucratic process for deciding whether a noncitizen will be detained or released. It then identifies the formal legal factors to be considered by IJs in rendering custody decisions. It closes with a description of the current Administration’s stated policy preference for maximizing the number of immigrant detentions.

10 IMMIGRANT DETENTION [28-Jan-201. The Bureaucratic Process

Immigrant detention occurs in the context of removal proceedings that determine whether a noncitizen is removable and whether he or she warrants a discretionary grant of relief from removal. Noncitizens within the United States are removable where, for example, their presence is unauthorized (e.g., entered without inspection or overstayed a visa31) or where they are lawfully present but engaged in conduct that renders them deportable (e.g., criminal activity).32 These individuals generally are entitled to a relatively formal hearing to determine whether they will in fact be removed.33 At these hearings, the noncitizen is entitled to be represented by counsel (but not at government expense), present evidence and witnesses, cross-examine evidence and witnesses, and a formal record of the proceedings.34

Noncitizens often do not contest the grounds for removal, and their individual merits hearings typically focus on whether relief from removal will be granted. Congress has legislated various forms of discretionary relief, including not only “asylum” where the individual establishes a well-founded fear of persecution on one of five protected grounds,35 but also “waivers” of various grounds for

31 Visa overstays account for approximately 40% of the undocumented population. In FY2016, 628,000 noncitizens overstayed their visas. WHITE HOUSE FACT SHEET, ENFORCE IMMIGRATION LAWS ACROSS THE UNITED STATES (Oct. 8, 2017).

32 See generally 8 U.S.C. §1226 (defining categories of “inadmissible” aliens subject to removal); 1227 (defining categories of “deportable” aliens subject to removal).

33 8 U.S.C. § 1229a (describing removal proceedings). Some aliens are not entitled to such formal removal proceedings and are instead subject to “expedited removal.” Such aliens include those who are intercepted at the border without proper documentation or who have previously committed immigration fraud, INA § 235(b)(1), and certain classes of aliens who pose national security threats, INA § 235(c).

34 8 U.S.C. § 1229a.35 8 U.S.C. §§ 1158; 1101(a)(42). Claims for asylum have been a

growing segment of removal proceedings but still constitute only 20% of cases in master calendar hearings. Kim & Semet, supra note 24. Individuals may file for asylum affirmatively, before removal proceedings have been initiated, or defensively, after removal proceedings have been initiated.

28-Jan-20] IMMIGRANT DETENTION 11removability36 and “cancellation of removal”37 where the noncitizen satisfies a list of statutory eligibility factors including significant hardship to family members if removal were effectuated. For decades, noncitizens could also seek a form of relief called “administrative closure,” which removed a case from the immigration court’s active docket for example where the noncitizen would soon qualify for legal residence through a family members or was in the process of litigating a direct challenge to a criminal conviction that formed the basis for removal.38 Given the significant evidentiary burdens on noncitizens in seeking these forms of relief,39 removal proceedings often take years to resolve.40 A crucial question, then, for noncitizens is whether they will be detained pending the completion of those proceedings.41

When an individual is initially apprehended on suspicion of removability,42 enforcement officials within the Department of Homeland Security’s (DHS) Immigration and Customs Enforcement (ICE) agency make the initial determination as to whether the individual will be

36 See, e.g., 8 U.S.C. §§ 1182(h); 1227(a)(1)(E)(iii), (a)(1)(H), (a)(7). 37 8 U.S.C. § 1229b. 38 The Trump administration curtailed the use of administrative

closure as an option for IJs. In Matter of Castro-Tum, 27 I.&N. DEC. 271 (A.G. 2018), the Attorney General exercised his refer-and-review authority to overturn BIA precedent acknowledging the propriety of “administrative closure.” The Fourth Circuit overturned Matter of Castro-Tum in Zuniga Romero v. Barr, No. 18-1850 (4th Cir. 2019), preserving the availability of this form of discretionary relief in that circuit.

39 8 U.S.C. § 1229a(c)(4). 40 See GENERAL ACCOUNTING OFFICE, IMMIGRATION COURTS: ACTIONS NEEDED

TO REDUCE CASE BACKLOG AND ADDRESS LONG-STANDING MANAGEMENT AND OPERATIONAL CHALLENGES 22, 25-26, GAO 17-438.

41 For an overview of the statutory framework for immigrant detention, see CONG. RES. SERV., Immigration Detention: A Legal Overview (Sept. 16, 2019), at https://fas.org/sgp/crs/homesec/R45915.pdf.

42 Individuals apprehended in the nation’s interior are apprehended by enforcement officials within DHS’s ICE agency. Noncitizens apprehended at the border (either at a port of entry or between ports of entry) are apprehended by enforcement officials within DHS’s Customs and Border Protection (CBP) agency. In both circumstances, ICE officials make the initial custody determination. 8 C.F.R. §§ 236.1; 1236.1(d)(1); 1003.19. ICE officials are also responsible for representing the government in prosecuting the alien during removal proceedings.

12 IMMIGRANT DETENTION [28-Jan-20detained.43 Congress has imposed mandatory detention for certain categories of aliens, including aliens apprehended at the border44 and those who are removable on certain criminal and national security grounds.45 For those who are not subject to mandatory detention, ICE exercises discretion to release the alien on conditional parole (also known as release on recognizance), set a bond of at least $1,500, or deny bond altogether.46

The detained individual, however, has a right to appeal ICE’s initial custody determination by seeking a bond hearing before the IJ.47 Noncitizens subject to mandatory

43 8 C.F.R. §§ 236.1; 1236.a(d)(1); 1003.19.44 8 U.S.C. § 1225(b). The enforcement officials of DHS retain

authority to release these aliens through a grant of humanitarian parole “for urgent humanitarian reasons or significant public benefit” pursuant to 8 U.S.C. § 1182(d)(5).

For many years, the Board of Immigration Appeals (BIA) concluded that arriving aliens who lack proper documentation (and are thus “subject to expedited removal”) and establish a credible fear of persecution to form an asylum claim were not subject to mandatory detention and remained eligible for release on bond if they were apprehended at a port of entry, but not if they were apprehended between ports of entry. Matter of X-K-, 23 I.&N. DEC. 731 (BIA 2005). Attorney General Bill Barr overruled that decision in. Matter of M-S, 27 I.&N. DEC. 509 (A.G. 2019). The Western District of Washington, however, has concluded that individuals apprehended in the nation’s interior and subject to expedited removal are constitutionally entitled to a bond hearing before an immigration judge if they establish a credible fear of persecution. Padilla v. U.S. Immigr. & Customs Enforcement, 387 F. Supp.3d 1219 (W.D. Wash. 2019).

45 8 U.S.C. § 1226(c)(1). These individuals may be released only for witness protection purposes. § 1226(c)(2).

46 8 U.S.C. § 1226(a); 8 C.F.R. §§ 236.1, 1236.1. Conditional parole pursuant to § 1226(a)(2)(B) differs from humanitarian parole pursuant to § 1182(d)(5). Conditional parole allows the release of an alien who is subject to discretionary detention provisions and may impose conditions on release such as the wearing of an ankle bracelet or periodic check-ins with ICE. Humanitarian parole allows the release of any arriving alien, including those subject to mandatory detention, but only where such release is for an “urgent humanitarian reason or significant benefit.” Conditional parole may be granted by either DHS officials or an IJ; humanitarian parole is only available to DHS officials.

47 8 C.F.R. § 1003.19; see also 8 C.F.R. 1236.1(d). Such hearings are sometimes referred to as “bond redetermination hearings” or “custody redetermination hearings.”

The IJ’s determination may be appealed to the BIA, an intermediate administrative appeals body. Id.; see also § 1003.38.

28-Jan-20] IMMIGRANT DETENTION 13detention generally are not entitled to a bond hearing before an IJ.48 Like the ICE officials before them, IJs have authority to release the alien on conditional parole (on recognizance), set a bond amount, or deny bond altogether.49 The IJ’s custody and bond determination may be appealed by either the noncitizen or ICE to the Board of Immigration Appeals (BIA), the administrative appellate tribunal charged with reviewing the decisions of IJs.50 No further appeal, however, lies in the federal courts.51 Children are subject to different detention rules.52

Although IJs also are responsible for adjudicating the question of whether the noncitizen ultimately will be

48 8 C.F.R. § 1003.19(h)(1)(i). A noncitizen may seek a “Joseph” hearing before the IJ, seeking to determine that he or she does not fall within one of the categories for mandatory detention. Matter of Joseph, 22 I.&N. DEC. 799 (BIA 1999).

49 8 U.S.C. § 1226(a); 8 C.F.R. §§ 1003.19(a), 236.1(d).50 8 C.F.R. § 1003.38.51 8 U.S.C. § 1226(e). The Supreme Court has ruled, however, that

noncitizens retain the right to habeas review to challenge their detention as a violation of the U.S. Constitution or federal statute. INS v. St. Cyr, 533 U.S. 289, 311 (2001); see also Demore v. Kim, 538 U.S. 510, 517 (2003).

52 In a class action lawsuit over juvenile detention conditions, the government entered into a settlement agreement in 1997 known as the Flores Settlement. See Flores v. Sessions, 862 F.3d 8693, 866, 869 (9th

Cir. 2017) (discussing Flores litigation and settlement). Pursuant to that settlement, the government agreed to release all alien minors to a qualifying adult or state-licensed residential facility within five days or up to 20 days in cases of emergency or an influx of alien children. Congress partially codified this agreement in 2008, providing that children who cross the border unaccompanied by a parent (unaccompanied minors or “UACs”), may be detained for a maximum of 72 hours, after which they must be released to the Department of Health and Human Service’s Office for Refugee Resettlement (ORR), which is directed to place them in the least restrictive environment consistent with the child’s best interests. 8 U.S.C. § 1232(c)(2).

The Obama Administration subsequently took the position that accompanied minors, i.e., those who crossed the border with a parent, were not subject to the 20-day limit to detention articulated in the Flores settlement. The district court rejected that position, concluding that both UACs and children apprehended with a parent were subject to the 20-day limit. In August of 2019, the Trump Administration published a final rule reinstituting the different treatment for UACs and accompanied children, adopting the Obama Administration’s stance that children apprehended with a parent may be detained longer than 20 days.

14 IMMIGRANT DETENTION [28-Jan-20removed,53 regulations provide that bond proceedings must be “separate and apart from, and shall form no part of,” the removal proceeding.54 The IJ may consider any information available to it in rendering the custody decision.55 An alien may subsequently request an additional bond hearing after the first, but only upon showing that circumstances have materially changed.56

As of July 2019, there were 430 IJs57 presiding over 64 immigration courts58 across the nation. IJs possess many of the powers associated with ordinary judges; for example, they are authorized to administer oaths, receive evidence, examine and cross-examine witnesses, issue subpoenas, and hold individuals in contempt.59 IJs do not possess enforcement or prosecutorial responsibilities. Rather, they are designed to be independent and apolitical. Indeed, regulations provide, “In deciding individual cases before them, and subject to the applicable standards, immigration judges shall exercise their independent judgement and discretion.”60 Due process may well mandate such independence given the liberty interests at stake in detention decisions.61

Despite these norms of adjudicatory independence, IJs are executive branch officials subordinate to the Attorney General. The Immigration and Nationality Act (INA) explicitly provides they shall be appointed by the Attorney General and “subject to such supervision and shall perform such duties as the Attorney General shall prescribe.”62 IJs do not enjoy the tenure protections of Administrative Law Judges (ALJs) under the Administrative Procedure Act (APA).63 Instead, their independence is protected only to the extent of ordinary civil service laws, although such laws do protect

53 8 U.S.C. § 1229a(a)(1), (4)(C). 54 8 C.F.R. § 1003.19(d). 55 Id. 56 § 1003.19(e).57 https://www.justice.gov/eoir/page/file/1104846/download.58 https://www.justice.gov/eoir/eoir-immigration-court-listing.59 8 U.S.C. § 1229a(b)(1).60 8 C.F.R. § 1003.10.61 See, e.g., sources, supra note 18.62 8 C.F.R. § 1101(b)(4). 63 See supra note 21.

28-Jan-20] IMMIGRANT DETENTION 15IJs from adverse employment actions on political grounds.64

Figure 1 shows the locations of immigration courts across the United States. Immigration courts collocated with a detention facility are represented by a light dot, while those collocated with a correctional facility65 are indicated by a darker dot. In our dataset, over 90% of noncitizens had their bond hearings at the facility where they were held. Some bond hearings, however, occur in an immigration court located away from a detention facility. Bond hearings can also occur by video.

Figure 1

64 See supra note 20.65 DHS sometimes contracts with state or local correctional facilities

to house detained noncitizens. See SCHRIRO REPORT, supra note 4.

16 IMMIGRANT DETENTION [28-Jan-20

2. Grounds for immigrant detention The Supreme Court has authorized the detention of a

noncitizen pending removal proceedings (1) to ensure the noncitizen appears for removal proceedings66 or (2) for public safety reasons.67 It has emphasized, however, that immigrant detention may not be used as punishment.68 Circuit courts frequently affirm that immigrant detention is warranted only where the alien poses a flight risk or danger to the community.69

IJs may consider a wide range of factors in determining flight risk or public safety risk, including: whether the alien has a fixed address in the United States, length of U.S.

66 Demore v. Kim, 538 U.S. 510, 527-28 (2003) (noting that detention “necessarily serves the purpose of preventing deportable aliens from fleeing prior to or during their removal proceedings.”).

67 See Carlson v. Landon, 342 U.S. 524, 544 (1952) (sustaining detention of Communists in removal proceedings for public safety reasons).

68 Wong Wing v. United States, 136 U.S. 228, 237 (1896).69 See, e.g., Guerrero-Sanchez v. Warden York County Prison, 905

F.3d 208 (3d Cir. 2018) (concluding detention permitted only if alien “poses risk of flight or danger to community, or alien’s release or removal is imminent”); Hernandez v. Sessions, 872 F.3d 976 (9 th Cir. 2017) (noting immigrant detention depends on whether alien “present[s] a danger to persons or property, is not a threat to the national security, and does not pose a flight risk”); Sopo v. U.S. Attorney General, 825 F.3d 1199 (11th Cir. 2016) (concluding Congress intended for immigrant detention to “prevent[] flight and recidivism”); see also 8 C.F.R. § 236.1(c)(8); 1236(c)(8); Matter of Adeniji, 22 I.&N. Dec. 1102, 1112 (BIA 1999).

28-Jan-20] IMMIGRANT DETENTION 17residence, family ties, employment history, record of court appearances, criminal record, history of immigration violations, prior attempts to flee, and manner of entry into the U.S.70 IJs enjoy “wide discretion” in determining which factors to consider.71 The discretion of the IJ is further enhanced by the lack of federal court review over detention decisions. The IJ’s decision is subject to review by the BIA, but the INA provides that decisions “regarding the detention or release of any alien or the grant, revocation or denial of bond or parole” shall not be subject to judicial review.72 3. Expressed Policy Preferences of Political Superiors Within

the Executive BranchThe current Administration has made clear its political

preference for detaining all noncitizens.73 Within one week of his inauguration, President Trump issued an Executive Order explicitly stating his Administration’s policy of detaining all aliens suspected of violating immigration laws.74 In February of that year, the Administration issued a directive to all ICE employees (but not IJs) to detain all noncitizens pending removal proceedings except in narrow circumstances.75 It continues that “there is no presumption that an individual alien’s release would not pose a danger or flight risk.”

70 In re Guerra, 24 I.&N. Dec. 37, 40 (2006); overruled by Pensamiento v. McDonald, 315 F. Supp.3d 684 (D. Mass. 2018) (holding that burden of proving flight risk or dangerousness lies with the government, not the noncitizen).

71 Id.; see also Carlson v. Landon, 342 U.S. 524 (1952). 72 8 U.S.C. § 1226(e).73 FOIA records indicate that the Obama Administration imposed a

policy of requiring high bonds for certain categories of noncitizens in detention. Denise Gilman reports that in 2013, ICE imposed a policy of setting bonds uniformly at $7,500 for all women detained at the Hutto detention facility in Texas. It subsequently issued a new policy raising that amount to $10,000. When available bed space at the facility became limited, however, a new policy was issued of granting bond in the amount of $3,000. See Denise Gilman, To Loose the Bonds: The Deceptive Promise of Freedom from Pretrial Immigration Detention, 92 Indiana L.J. 157, 211-212 & nn. 238-240 (2016). It is not clear, however, that these policy preferences were directed at or even communicated to the IJs who review the custody decisions of IJs.

74 Exec. Order 13767, 82 FED. REG. 8793.75 MEMORANDUM FROM ICE EXECUTIVE ASSISTANT DIRECTOR MATTHEW

ALBENCE, “IMPLEMENTING THE PRESIDENT’S BORDER SECURITY AND INTERIOR ENFORCEMENT POLICIES” (Feb. 21, 2017).

18 IMMIGRANT DETENTION [28-Jan-20In April of 2017, then Attorney General Jeff Sessions

addressed a group of DHS CBP officers and stated, “Pursuant to the President’s order, we will be detaining all adults who are apprehended at the border.”76 While the address was to CBP officials rather than IJs, it is plausible that the latter group felt constrained by their political superior to keep these individuals in detention after ICE had refused to release them. Moreover, on October 12, 2017, Attorney General Sessions addressed IJs directly, complaining about the prior Administration’s policy of releasing aliens who demonstrated a fear of persecution.77 He asserted, “Not surprisingly, many of those who are released into the United States … simply disappear and never show up at immigration hearings.”

The extent to which IJs, as opposed to enforcement officers, are responsive to the policy preferences of political appointees in the Executive Branch remains to be seen. Media reports based on anecdotal evidence suggest that they are.78 This study seeks to examine the political independence of IJs in their bond decision making.

B. Prior ScholarshipIn earlier work, we examined the potential for political

interference in IJs’ decisions to enter a removal order.79 We found that after controlling for over a dozen variables—including attorney representation, whether detained, whether asylum claimed, noncitizen’s nationality—IJs were significantly more likely to order removal under the Trump Administration than either the Obama or Bush II Administrations.80 In this project, we shift our focus from the

76 ATTORNEY GENERAL JEFF SESSIONS REMARKS TO ANNOUNCING THE DEPARTMENT OF JUSTICE’S RENEWED COMMITMENT TO CRIMINAL IMMIGRATION ENFORCEMENT (Apr. 11, 2017).

77 ATTORNEY GENERAL JEFF SESSIONS DELIVERS REMARKS TO THE EXECUTIVE OFFICE FOR IMMIGRATION REVIEW (Oct. 12, 2017).

78 PBS News Hour, Under Trump, Higher Immigration Bonds Mean Longer Family Separations (Jun. 28, 2018); Mica Rosenberg & Reade Levinson, Trump’s Catch and Release Policy Snares Many Who Have Long Called U.S. Home, Reuters (July 20, 2018); Alejandro-Fernandez Sanabria, Costly Bonds: For Undocumented Immigrants, Bail Depends on a Judge’s Subjectivity, Univision (Feb. 28, 2018).

79 Kim & Semet, supra note 24, at _.80 Other studies have also looked at the factors impacting IJs’

removal decisions, particularly in the asylum context. See, e.g., Jaya Ramji-Nogales, Andrew I. Schoenholtz, & Philip G. Schrag, Refugee

28-Jan-20] IMMIGRANT DETENTION 19IJs ultimate decision on removal, to the IJs preliminary decision on custody pending removal proceedings.

A handful of studies have looked at custody determinations in immigration courts.81 In the earliest of these, Janet Gilboy examined the bond decisions of immigration judges among a sample of cases in the Chicago immigration court from April and May 1983.82 She found that among the 126 aliens in her sample who were initially detained by the INS, 95% of those who requested a bond redetermination hearing obtained a reduction in the bond amount, and 16% were released entirely (on recognizance). Bonds were reduced by IJs by an average of 68%. The Gilboy study pre-dates important changes in the law of immigrant detention.83 Indeed, at the time publication, the BIA Roulette: Disparities in Asylum Adjudication, 60 STAN. L. REV. 295, 339-49 (2007); Daniel Chen & Jess Eagel, Can Machine Learning Help Predict the Outcome of Asylum Adjudications?, Working Paper, at 1, 6; Daniel E. Chand, William D. Schreckhise, & Marianne L. Bowers, The Dynamics of State and Local Contexts and Immigration Asylum Hearing Decisions, 2017 J. PUB. ADMIN. RES. & THEORY 182, 189 (2017); Banks Miller, Linda Keith, & Jennifer Holmes, IMMIGRATION JUDGES AND U.S. ASYLUM POLICY 100; Linda Camp Keith, Jennifer S. Holmes, and Banks P. Miller, Explaining the Divergence in Asylum Grant Rates among Immigration Judges: An Attitudinal and Cognitive Approach, 35 LAW & POL’Y REV. 261, 266-67 (2013); Ingrid V. Eagly, Remote Adjudication in Immigration, 109 NORTHWESTERN L. REV. 933, 933 (2015); Eagly & Shafer, supra note 6; Eagly, Shafer, & Whalley, supra note 7, at 933; Ingrid Eagly & Steven Shafer, Measuring In Absentia Removal in Immigration Court, U. PA. L. REV. (2020) (forthcoming) (analyzing in absentia removal); see also Dane Thorley & Joshua Mitts, Trial by Skype: A Causality-Oriented Replication Exploring the Use of Remote Video Adjudication in Immigration Removal Proceedings, 59 INT’L REV. OF LAW & ECONOMICS 82-83 (2019); U.S. GOV’T ACCOUNTABILITY OFFICE, GAO-08-940, U.S. ASYLUM SYSTEM: SIGNIFICANT VARIATION EXISTED IN ASYLUM OUTCOMES ACROSS IMMIGRATION COURTS AND JUDGES (2008); U.S. GOV’T ACCOUNTABILITY OFFICE, GAO-17-72, ASYLUM: VARIATION EXISTS IN OUTCOMES OF APPLICATIONS ACROSS IMMIGRATION COURTS AND JUDGES (2016).

81 A few studies have also examined ICE’s initial decision to detain. See Mark Noferi & Robert Koulish, The Immigration Detention Risk Assessment, 29 GEO. IMM. L.J. 45 (2014) (examining DHS ICE’s implementation of risk classification assessment tool); DEP’T HOMELAND SECURITY, OFFICE OF THE INSPECTOR GENERAL, U.S. IMMIGRATION AND CUSTOMS ENFORCEMENT’S ALTERNATIVES TO DETENTION (Feb. 4, 2015) (assessing ICE’s use of alternatives to detention).

82 Janet A. Gilboy, Setting Bail in Deportation Cases: The Role of Immigration Judges, 24 SAN DIEGO L. REV. 347 (1987).

83 For example, it pre-dates the enactment of statutes imposing mandatory detention on aliens. See Anti-Drug Abuse Act of 1988, Pub. L.

20 IMMIGRANT DETENTION [28-Jan-20standard was to release all aliens on recognizance unless there was a national security issue or the alien presented a “poor bail risk.”84

Richard Sanders statistically analyzed the factors impacting bond determinations by INS enforcement officials (rather than IJs) in the Miami region in the late 1980s.85 He found that noncitizens with a criminal background, with many assets, or from communist countries were more likely to be detained while males, those with a legal spouse living in the U.S., minors, and those traveling with minors were less likely to be detained.86 These same factors affected the bond amount; those with criminal backgrounds or with a foreign spouse had higher bond amounts while students, minors, males, and those traveling with minors had lower bonds.87 He also found that those of non-Anglo ethnicity had higher bonds while those leaving communist countries had lower bonds.88 The Sanders study identifies the importance of criminal, family and employment factors in custody decisions. Indeed, he found that these factors explained over half of the variation in INS custody decisions, including bond amounts.89

Ingrid Eagly and Steven Shafer’s more recent study of access to counsel in immigration courts includes an analysis of whether attorney representation impacts the likelihood that a noncitizen in removal proceedings before immigration judges will be released from detention.90 They find that aliens with legal representation are seven times more likely to be released from detention than pro se litigants.91

In another study, Eagly, Shafer, and Jana Whalley No. 100-690; 7343(a)(4), 102 Stat. 4181 (1988); Antiterrorism and Effective Death Penalty Act of 1996, Pub. L. No. 104-132, 440(c), (e), 110 Stat. 1214 (1996); Illegal Immigration Reform and Immigrant Responsibility Act of 1996, Pub. L. No. 104-208, 303(a) 110 Stat. 3009-546 (1996).

84 In re Patel, Interim Dec. No. 2491 (BIA 1976). 85 Robert M. Sanders, Immigration Bonds: An Analysis of the

Determinants of Individual Decisions, 20 CRIME, LAW, & SOC. CHANGE 139 (1993).

86 Id. at 152.87 Id. at 154.88 Id. at 155.89 Id. at 152, 154.90 Eagly & Shafter, supra note 6, at 70.91 Id.

28-Jan-20] IMMIGRANT DETENTION 21examine processes for adjudicating asylum claims for detained families.92 With respect to bond hearings, they find that 19% of detained families seeking asylum were released on recognizance by immigration judges, while an additional 38% were granted money bond.93 They found that individuals who were not part of family units had a lower rate of success in immigration court, with only 1% granted release on recognizance and 44% receiving a cash bond.94 They noted, however, wide disparities in grant rates depending on the detention facility in which the families were held.95

Emily Ryo conducted in-person surveys of 565 noncitizens subject to long-term detention in the Central District of California between May 2013 and March 2014.96

She finds that IJs’ custody decisions are shaped in part by the individual’s criminal history, but that that was the only legally relevant factor that predicted whether bond was granted or the bond amount; factors relating to flight risk such as family ties or employment were not significant.97 In another study, Ryo reviewed audio recordings of the hearings in a subset of these cases and concluded that a noncitizen was more likely to be detained on the ground of dangerousness if he or she was Central American, proceeded pro se, or had a history of felony and violent convictions.98

While the immigrant detention context differs from the criminal pre-trial detention context in important respects,99 scholarship examining the factors that play a role in a judge’s decision to release a criminal defendant from pre-trial detention provide useful guidance as well. Criminal justice scholars have found that bond decisions vary

92 Eagly, Shafer, & Whalley, supra note 7, at 833. 93 Id. at 838.94 Id. 95 Id. 96 Emily Ryo, Detained: A Study of Immigration Bond Hearings, 50

LAW & SOC’Y REV. 117 (2016). 97 Id.98 Emily Ryo, Predicting Danger in Immigration Courts, 44 LAW & SOC.

INQUIRY 227 (2019). 99 See Gilman, supra note 73, at 203-209 (identifying important

differences between pre-trial criminal detention and detention pending immigration removal proceedings).

22 IMMIGRANT DETENTION [28-Jan-20significantly based on a host of factors, including individual case characteristics such as the severity of the charges and criminal history,100 community ties,101 and, more troubling, the race and sex of the defendant;102 organization factors such as caseload pressures103 and the availability of bed space in local jails;104 and the larger political, socioeconomic, and demographic environment in which these decisions are made.105 In a recent study of pretrial practices from 1990 to 2009 in state courts around the nation, Katherine Hood and Daniel Schneider find that defendants are more likely to be granted nonfinancial release from pretrial detention where jail occupancy rates are higher, where the district attorney is a Democrat, when Republicans are governors, in areas with lower levels of socioeconomic inequality, and in counties with a smaller black population.106

II. ANALYSISWe reviewed EOIR records to examine political trends in

IJs bond decisions. This section begins with an overview of how we constructed our dataset. It then sets forth our findings from three bivariate analyses: changes in “win rates” for noncitizens across Presidential Eras, the percentages of cases that resulted in low, medium, and high bonds being set; and the percentage of cases that reduced the bond amount initially set by ICE by $10,000 or more.

A. Construction of DatasetWe obtained from the EOIR’s website107 records of every

bond hearing in immigration court from January 20, 2001 100 Hood & Schneider citing to Goldkamp & Gottfredson 1985; Cohen

& Reaves 2007.101 Hood & Schneider citing to Pettee 1994; Spohn 2009. 102 Hood & Schneider citing to Spohn; Cohen & Reaves 2007103 Hood & Schneider citing to Ulmer & Johnson 2004104 Hood & Schneider citing to Flemming 1982. 105 Katherine Hood & Daniel Schneider, Bail and Pretrial Detention:

Contours and Causes of Temporal and County Variation, 5 RSF: RUSSELL SAGE FOUNDATION JOURNAL OF THE SOCIAL SCIENCES 126 (2019).

106 Id. 107 EOIR CASE DATA, at https://www.justice.gov/eoir/frequently-

requested-agency-records (last visited July 1, 2019). TRAC researchers at Syracuse University in 2008 successfully filed a lawsuit under the Freedom of Information Act (FOIA) to force EOIR to release the data, and EOIR published these data on its website pursuant to reporting standards under the FOIA Improvement Act of 2016, Public Law No. 114-185 (2016).

28-Jan-20] IMMIGRANT DETENTION 23through June 30, 2019, identifying 633,637 bond decisions for study.108 These records identify the judge before whom the request was made, the date of the IJ’s decision, and the judge’s decision.109

We used three approaches to code for the IJ’s bond decision. First, we determined whether the decision was favorable to the noncitizen. In this specification, which we call “Custody Hearing Outcomes,” we coded the decision as a “win” if the IJ released the noncitizen on recognizance or

108 We used the tables downloaded as of July 2, 2019 and eliminated the few cases decided after June 30, 2019. To determine the bond decision, we relied on the column “dec” in the Table D: Associated Bond Table.

The EOIR data code three types of bond hearings: BD, BB, or SB. Prior to 2005, most bond hearings are coded as “BD” or “bond re-determinations” proceedings. After 2005, most proceedings are “BB” or “custody re-determinations.” We treated BB and BD proceedings the same. See TRAC, What Happens When Individuals are Released on Bond in Immigration Court Proceedings, available at https://trac.syr.edu/immigration/reports/438/ (noting shift in coding twenty years ago). “SB” proceedings indicate a subsequent bond hearing after one was already heard. We deleted SB cases except where the IJ found no jurisdiction in the first proceeding but rendered a bond decision in the second proceeding.

In addition to deleting most SB cases, we narrowed the dataset further in several ways. First, we included only bond hearings that occurred in the context of removal proceedings, deleting the 2,958 cases that were associated with non-removal cases. Second, we eliminated the 96,829 hearings in which the IJ concluded that they lacked jurisdiction. A lack of jurisdiction would occur where the noncitizen is ineligible for release because they are subject to mandatory detention statutes. Third, we deleted approximately 2,000 cases heard by Nixon and Carter appointees to simplify the analysis. Fourth, we eliminated approximately 3,300 proceedings heard by IJs who decided less than 50 bond proceedings. We also eliminated proceedings heard by an individual who had not yet been formally appointed as an IJ, as well as those heard by two IJs for whom we were not able to obtain reliable biographical information. Fifth, we eliminated 5,396 proceedings in which custody status was listed as “never detained.” It is possible that in those cases, the noncitizen had never been detained but sought review over the conditions for their release; it also possible that those cases were erroneously coded. Sixth, we eliminated all cases identifiable as involving a juvenile or family on the ground that IJ decisions on bond will differ systematically when a child or family is involved. These deletions included 16,865 cases involving families, and 6,617 cases identified as involving juveniles not associated with a family unit. To identify families, we relied on the Lead and Rider Tables and assume that case numbers on those tables were family units.

24 IMMIGRANT DETENTION [28-Jan-20set a bond amount.110 By contrast, we coded decisions in which the IJ denied bond, took no action, or made no change in bond, as unfavorable to the noncitizen. In our dataset, IJs granted release on recognizance in less than one percent of cases (0.74%), but they set bond in 46% of cases. On the loss side, IJs denied bond altogether in 13% of the cases, took no action in 29% of the cases, and indicated no change in bond amount in 12% of the cases.111 We also measured the number of cases in which the IJ denied bond altogether.

In the second specification, which we call “Bond Amounts

EOIR’s coding identifying juvenile cases is unclear so we only eliminated the cases with a juvenile id from the Juvenile Table or who had a case id from either the lead or rider Tables. Some analysis has suggested this is underinclusive of all juveniles. See Nina Siulc, Zhifen Cheng, Arnold Song, and Olga Byrne, Vera Institute of Justice, LEGAL ORIENTATION PROGRAM EVALUATION AND PERFORMANCE OUTCOME MEASUREMENT REPORT, PHASE II, A REPORT PREPARED AT THE REQUEST OF THE DEPARTMENT OF JUSTICE, EXECUTIVE OFFICE FOR IMMIGRATION REVIEW, May 2008. Finally, we eliminated all cases in which the original bond amount set by ICE was lower than the bond amount set by the IJ when the IJ sets a new bond amount. We found approximately 1,500 cases in this category (or 0.48% of our dataset). We assumed these cases were erroneously coded. If these cases were accurately coded, it would mean that the IJ set a bond amount over and above the amount set by ICE. A bond hearing may occur only at the request of the noncitizen; ICE is free to raise the bond amount at any time without going through the immigration courts.

The reliability of the data is unclear. While most observations included a bond decision, many did not record a bond amount. In addition, the bond amount was sometimes inconsistent with other data fields. We assumed the bond decision code was accurate. We limited out analysis of bond amounts to those coded with a decision as “New Amount,” restricted to the vast majority of cases in which the IJ reduced the bond amount set by ICE. The incompleteness of the dataset could further be resolved with an imputation analysis.

109 Table D: Associated Bond had tens of thousands of entries lacking data on the presiding judge. To identify the judge, we merged in the Schedule Table, which codes “CY” or “Custody” to signify the custody proceeding. If the judge was still missing even after merging in the Schedule Table, we used the judge listed in the Table B: Proceeding. In alternative specifications, if there was no information on the IJ, we used the Proceeding Table.

110 As noted in supra note 115, we deleted the approximately 1,500 cases in which the data suggest the IJ set a bond amount higher than that set by ICE. We assume, as explained above, that such data were erroneously coded.

111 EOIR’s coding do not explain the distinctions between these categories.

28-Jan-20] IMMIGRANT DETENTION 25Set by IJ,” we examined the specific bond amounts set by IJs, categorizing them as “low” if they were $2,500 and under and “high” if they were $10,000 or above.112

In the final specification, which we call “Differences in Bond Amounts Set by IJ Compared to Those Set by ICE,” we evaluated the extent to which IJ decisions on bond amount departed from those of the initial ICE enforcement officers. To do so, we measured differences in the bond amounts set by IJs as compared to those initially set by ICE.113 We also included in this measure cases in which ICE initially denied bond altogether. For those cases, we set the ICE bond amount at the 99% of bond amounts for the relevant year.114 That is, if ICE denied a bond, and the highest 99% of bond amounts for that year was $10,000, then we calculated the difference between the amount of bond set by the IJ, and $10,000.

For each of these three specifications, we examine changes through time, i.e., during different Presidential “Eras.” Custody decisions rendered from January 20, 2001 to January 19, 2009 were coded as “Bush II Era.” Custody proceedings decided between January 20, 2009 to January 19, 2017 were coded as “Obama Era.” Cases decided from January 20, 2017 until June 30, 2019 (the latest date for which data are available) were coded as “Trump Era.” This analysis thus shows differences across Eras in rates at which IJs as a whole granted a favorable custody decision to noncitizens, denied bond altogether, set a low bond or high

112 We found many internal inconsistencies in EOIR’s coding on the old and new bond amounts. For example, some cases coded as a “No Bond” decision actually had an amount listed in the “New Bond Amount” column. We treated such decisions as No Bond cases. When a case was identified as a “New Amount” decision and there was no entry in the “Initial ICE Bond Amount” column, we assumed that ICE had denied bond altogether. For our analyses of bond amounts, we excluded all cases coded as a “No Change” or “No Action” decision. Thus, our analysis of bond amounts and differences from ICE amounts is limited to only cases in which the IJ granted a bond for the first time or reduced the bond amount.

113 The amount of bond set by ICE is not always clear. We concluded that a 0 or blank in the “Initial Bond” column indicated that bond was denied.

114 Specifically, because of data coding issues, these figures are based on only the cases in which the bond decision is coded as “New Amount.”

26 IMMIGRANT DETENTION [28-Jan-20bond amount, and the extent to which bond amounts set by IJs departed from those initially set by ICE enforcement officials.

For each Era, we further analyzed the behavior of different cohorts of appointees. For example, during the Trump Era, we compared the behavior of Trump appointees, Obama appointees, Bush II appointees, Clinton appointees, Bush I appointees, and Reagan appointees. This exercise allows us to examine the extent to which changes in behavior during a given Era might be attributable to the appointment of a new cohort of IJs (political power to appoint), rather than change in control of the Administration at the time the decision was rendered (political power to supervise).

Importantly, our study is limited to using cross-tabulations to conduct bivariate analyses. Unlike our prior analysis on politicization in the removal process,115 we do not conduct a multivariate regression analysis, nor do we control for other variables that might impact an IJ’s custody decision.116 As such, this study does not attempt to identify predictive values or causal relationships. Rather, our more modest goal is to provide a descriptive picture of IJ custody decisions, examining potential differences in the decision making of IJs during different Presidential Eras, as well as between different appointment cohorts of IJs.

B. Custody Hearing OutcomesWe first measured “custody hearing outcomes,” i.e.,

rates at which the noncitizen was released on recognizance or granted a new bond amount, or, alternatively, was denied bond altogether. Table 1 shows changes in these rates across Presidential Eras, as well as changes based on appointment cohort.

As an initial matter, the rates at which IJs granted release on recognizance were notably low; only 0.74% of custody decisions between 2001 and mid-2019 fell into this category.117 IJs were far more likely (46%) to issue a favorable ruling in the form of a grant of new bond amount.

115 Kim & Semet, supra note 24.116 See infra notes 117-126 and accompanying text (describing other

variables that might influence IJs’ custody decisions).117 Scholars have reported that some IJs conclude they lack authority

to grant release on recognizance. See Ryo; Gilman; Kalhan.

28-Jan-20] IMMIGRANT DETENTION 27IJs denied bond altogether in 13% of custody proceedings, and denied relief (i.e., “No Action” or “No Change”) in an additional 41% of cases.

Table 1: Custody Hearing Outcome, by Presidential Era and Appointee Cohort

All Appointees

Trump Obama Bush 2

Clinton Bush I Reagan

All ErasBond Dec.Release .74 .14 .13 .24 1.33 1.19 1.03New Amount

45.69 44.11 49.58 46.38 46.38 40.81 35.13

Win Rate 46.43 44.25 49.71 46.62 47.41 42.00 36.16No Action 28.72 31.28 27.28 27.23 28.78 31.89 33.09No Change

11.84 8.01 6.46 13.20 12.03 15.83 22.21

No Bond 13.00 16.45 16.56 12.95 11.78 10.29 8.54

Trump EraBond Dec.Release .18 .14 .13 .13 .44 0 .16New Amount

45.25 44.11 48.67 44.70 38.48 15.89 24.57

Win Rate 45.43 44.25 48.79 44.82 38.93 15.89 24.73No Action 28.20 31.28 25.90 27.91 33.26 53.43 17.12No Change

6.94 8.01 5.80 6.55 6.07 3.27 45.47

No Bond 19.43 16.45 19.50 20.72 21.74 27.41 12.67

Obama EraBond Dec.Release .23 .12 .34 .39 .41 .16New Amount

47.55 50.16 49.82 46.38 45.39 34.68

Win Rate 47.78 50.29 49.95 46.77 45.81 34.83No Action 28.60 28.15 26.49 29.61 31.06 31.66No Change

9.68 6.87 10.80 8.09 14.62 24.10

No Bond 13.95 14.68 12.75 15.52 8.52 9.41

Bush EraBond Dec.Release 1.92 .60 2.38 1.72 1.57New Amount

42.96 38.10 46.99 38.88 35.92

28 IMMIGRANT DETENTION [28-Jan-20Win Rate 44.88 38.69 49.37 40.61 37.49No Action 29.27 28.78 27.27 31.55 34.75No Change

18.47 24.46 16.75 17.11 19.94

No Bond 7.37 8.07 6.60 10.73 7.82

1. Differences between appointee cohorts We first examine differences between groups of

appointees across all Eras (from 2001 to 2019), reflected in the top section of Table 1. For example, did IJs appointed by Trump behave differently than those appointed by Obama or Bush II? This analysis examines the extent to which a Presidential Administration might influence custody decisions through exercise of the power to appoint.

IJs appointed by earlier Administrations were more likely to grant release on recognizance than more recently appointed IJs. Throughout the time period of study, Clinton-, Bush I-, and Reagan-appointees granted release on recognizance in more than 1% of cases, while IJs appointed by Trump. Obama, or Bush II granted release in less than a quarter of a percent of cases. Perhaps surprisingly, Obama appointees were no more likely to grant release on recognizance than their Trump-appointed counterparts. Throughout the years of study, Obama appointees granted release in only 0.13% of cases, as compared to 0.14% for Trump appointees.

But an examination of overall win rates yields a slightly different picture as shown in Figure 2. Obama appointees were more likely than any other group of IJs to grant a favorable outcome to the noncitizen (either release or new bond amount), granting relief in 50% of their custody hearings. Trump appointees were less likely than IJs overall to grant relief, at only 44% of the time. But they were not the least likely to grant relief among appointee cohorts across the time frame of the study: IJs appointed by Bush I and Reagan were less likely than Trump appointees to grant relief to the requesting noncitizen.

28-Jan-20] IMMIGRANT DETENTION 29

Figure 2: Win Rates by Appointee Cohort

At the opposite end of the spectrum, we looked at the likelihood of denying bond outright. Again, both Trump appointees and Obama appointees were far harsher to noncitizens than any other group of appointees. About 16% of cases decided by Trump appointees, and 17% of cases decided by Obama appointees, resulted in no bond being

30 IMMIGRANT DETENTION [28-Jan-20issued at all, as compared to an average of 13% for IJs overall.118 These findings are reflected in Figure 3.

Figure 3: No Bond by Appointee Cohort

These findings suggest that IJs appointed by Trump are among the strictest in their decision making. Perhaps

118 These figures are across the entire range of the study. As such, the figures for Obama appointees is from 2009-2019, while the figures for Trump are only from 2017 forward.

28-Jan-20] IMMIGRANT DETENTION 31surprisingly, Obama appointees are comparably strict, at least in terms of the rates at which they granted release, and the rates at which they denied bond altogether. 2. Differences across Presidential Eras

Next we look at differences in custody outcomes across different Presidential Eras for all appointees, reflected in the first column of the bottom three sections of Table 1. Were IJs as a whole (regardless of who appointed them) less likely to grant relief to noncitizens during the Trump Era than prior Eras? This analysis examines the extent to which a Presidential Administration may exercise control over IJs through exercise of the power to supervise.

Rates of release on recognizance have declined through time and were significantly lower during the Obama and Trump Eras. During the Bush II Era, 1.92% of custody hearings resulted in release on recognizance, dropping precipitously to 0.23% during the Obama Era and then 0.18% during the current Administration. But win rates were significantly higher during the Obama Era than during either the Trump or Bush II Eras (48% compared to 45% and 45%). The overall win rates by Presidential Era are reflected in Figure 4.

Figure 4: Win Rates by Presidential Era

32 IMMIGRANT DETENTION [28-Jan-20

In terms of the percentage of cases that resulted in bond being denied outright, noncitizens fared worse as time progressed. Only 7% of custody hearings resulted in a denial of bond during the Bush II Era, rising to 14% during the Obama Era, and to 19% during the Trump Era. These results are reflected in Figure 5.

28-Jan-20] IMMIGRANT DETENTION 33

Figure 5: No Bond by Presidential Era

3. Differences across Presidential Eras and Appointees Finally, we examine how different groups of appointees

behaved during different Eras. These data are reflected in the remainder of Table 1.

During the Trump Era, Obama appointees were far more favorable to noncitizen detainees than any other group of appointees. They granted relief to noncitizens in 49% of cases, as compared to the 45% win rate for IJs overall during the Trump Era. Interestingly, however, the appointees of prior Administrations were far harsher to noncitizens during the Trump Era than during the preceding Eras. For example, IJs appointed by the Clinton Administration granted relief to noncitizens in 39% of the cases they heard during the Trump Era, while they granted relief in 47% of cases during the period of study. Even more stark, appointees of Bush I granted relief to detainees in only 16% of cases during the Trump Era, even though they granted such relief in 42% of cases between 2001 to 2019. Indeed, IJs appointed by Clinton, Bush I, and Reagan all exhibited lower win rates for

34 IMMIGRANT DETENTION [28-Jan-20noncitizens during the Trump Era than Trump appointees. These findings, showing significant changes in behavior during different Presidential Eras amongst the same cohort of judges, suggest that the current Administration may be exercising strong influence over IJs through their supervisory authority. Obama appointees are the exception, showing some degree of independence in their decision making even during the current Administration. These findings are reflected in Figure 6.

Figure 6: Win Rates by Presidential Era and Appointee Cohort

C. Bond Amounts Set by IJs Next we examine changes in the bond amounts set by

IJs. As an initial matter, we analyze median bond amounts by Presidential Era as well as group of appointees. These data are reflected in Table 2. Bond medians grew throughout the three Eras in our study, starting at $5,000 during the Bush Era, rising to $5,500 during the Obama Era, and rising again to $7,500 during the Trump Era. Trump appointees have a median bond amount of $10,000 during the Trump Era, but each of the appointment cohorts exhibited a higher median bond amount during the Trump Era than either the Obama or Bush Eras. IJs appointed by

28-Jan-20] IMMIGRANT DETENTION 35Bush II, for example, set a median bond amount of $5,000 during the Bush II Era; the median rose to $6,000 during the Obama Era but then jumped to $10,000 during the Trump Era. Similarly, Clinton appointees set a median bond of $5,000 during both the Bush II and Obama Eras, which rose to $7,500 during the current Administration. These findings suggest that while bond amounts overall grew as time marched on, they grew considerably faster under the Trump Administration.119

Table 2: Median Bond Amount by Era and Appointee CohortAll

appointees

Trump appoint

ees

Obama appoint

ees

Bush II appoint

ees

Clinton appointe

es

Bush I appoint

ees

Reagan appointee

s

Trump Era

7500 10000 7500 10000 7500 7500 10000

Obama Era

5500 -- 6000 6000 5000 5000 6000

BushEra

5000 -- -- 5000 5000 5000 5000

All Eras

6000 10000 7000 7000 5000 5000 5000

Data for the rates at which IJs set low, medium, or high bond amounts are set forth in Table 3.

Table 3: Bond Amounts by Presidential Era and Appointee Cohort

_All Appointees

Trump Obama Bush 2

Clinton Bush I Reagan

All ErasBond Amount$2500 & <

14.37 5.28 9.16 106.9 20.28 16.33 15.39

$2500-9999

59.72 43.93 60.75 59.17 59.66 60.53 64.02

$10000 & >

25.90 50.79 30.09 29.87 20.06 23.14 20.59

Trump EraBond Amount

119 These changes do not account for changes in inflation over the nearly 20 year period of study, nor do they account for shifts in the bond amounts set by ICE over time and region.

36 IMMIGRANT DETENTION [28-Jan-20$2500 & <

7.24 5.28 7.81 5.26 10.05 9.80 2.32

$2500-9999

51.87 43.93 55.26 44.68 56.61 58.82 46.52

$10000 & >

40.89 50.79 36.92 50.06 33.34 31.37 51.37

Obama EraBond Amount$2500 & <

13.47 9.98 10.87 19.61 12.35 9.34

$2500-9999

63.91 64.13 63.63 62.04 69.91 72.60

$10000 & >

22.61 25.88 25.49 18.35 17.74 18.06

Bush EraBond Amount$2500 & <

20.77 15.96 22.24 19.43 19.21

$2500-9999

57.49 54.98 57.80 53.50 59.86

$10000 & >

21.74 29.06 19.95 27.07 20.93

Looking only at differences between cohorts of appointees (reflected in the top section of Table 3), we see that Trump appointees were likelier than any other cohort to set a bond of $10,000 or more (high bond). About 51% of the custody decisions by Trump appointees, as compared to 30% for both Obama and Bush II appointees, resulted in high bonds. The rates of high bonds were also considerably lower for earlier appointed IJs. The inverse was also true: Trump appointees were far less likely to set bond at $2,500 or lower (low bond) than any other cohort. Only 5% of custody decisions rendered by Trump appointees resulted in a low bond amount, as compared to 14% for all cohorts. Notably, the Obama appointees were also considerably less likely to set a low bond as compared to earlier cohorts. They set a bond of $2,500 or less in 9% of cases across the Obama and Trump Eras, while IJs appointed by earlier Administrations set a low bond in 15%-20% of cases.

Next, we examine differences in bond amounts across Presidential Eras. These data are reflected in Figure 7. Consistent with the findings on median bond amounts, we find that high bonds of $10,000 or more were considerably more likely to be set during the Trump Era than the preceding Eras. About 41% of bonds set by IJs during the

28-Jan-20] IMMIGRANT DETENTION 37Trump Era were $10,000 or more, as compared to only 23% and 22% for the Obama and Bush II Eras, respectively. Conversely, the percentage of cases in which a low bond ($2,500 or less) was set declined through the three Presidential Eras, constituting 21% of cases during the Bush II Era, 13% of cases during the Obama Era, and only 7% of cases during the Trump Era.

Figure 7: Bond Amounts by Presidential Era

We then examine low bond amounts and high bond amount by appointee cohorts across the three Presidential Eras. These data are presented in Figures 8 and 9.

Figure 8: Low Bond Amounts by Era and Appointee Cohort

38 IMMIGRANT DETENTION [28-Jan-20

Figure 9: High Bond Amounts by Era and Appointee

During the Trump Era, surprisingly, it is the Reagan appointees who were least likely to grant a low bond amount (2% of cases), and the most likely to set a high bond amount (51% of cases). During the same time period, Bush II appointees behaved similarly to Trump appointees, in the percentage of cases in which a low bond were set

28-Jan-20] IMMIGRANT DETENTION 39(5%), and the percentages in which a high bond was set (50% and 51% respectively). Each appointee cohort was significantly less likely to set a bond of $2,500 or less during the Trump Era than during the Obama Era. Each cohort was also significantly more likely to set a bond of $10,000 or more during the Trump Era than during the preceding Administration. These findings show that bond amounts set by IJs have risen considerably during the Trump Administration, and the same cohorts of judges have behaved more harshly during the Trump Administration than during prior Eras.

D. Differences Between Bond Amount Set by IJ Versus Bond Amount Set by ICE

Finally, we examine differences in the bond amounts set by IJ, as compared to the initial amounts set by ICE. We include in this analysis cases in which ICE initially denied bond altogether to enable a meaningful comparison of the extent to which IJs were more lenient to noncitizens than the ICE agents who preceded them. To incorporate those cases, we were required to input a bond amount for the original ICE amount even though there was no number. For those cases, we calculated the ICE bond amount to be the 99% for ICE bond amounts for that year.120 These data are reflected in Table 4.

120 This analysis was limited to the cases in which the IJ lowered bond (coded as dec=New Amount). For example, the 99th percentile for “initial bond” in 2003 was $35,000. If ICE set no bond for a case in 2003, we subtracted the IJ’s new bond amount from $35,000 to calculate the difference. If the new bond amount was $10,000 for example, the difference would be $25,000. The 99th percentile jumped considerably in 2019 to $73,500.

40 IMMIGRANT DETENTION [28-Jan-20

Table 4: Difference from Bond Amount Set by ICE All Appointees

Trump Obama Bush 2

Clinton Bush I Reagan

All ErasDifference$0 3.49 3.45 5.61 4.33 2.26 1.81 1.07$2500 & < 6.28 4.50 7.00 5.70 5.87 4.05 9.25$2501-9999 18.24 14.97 23.13 16.17 16.68 16.91 16.18$10000 & > 71.98 77.08 64.16 73.80 75.19 77.23 73.50

Trump EraDifference$0 7.48 3.45 7.94 9.93 8.26 19.39 4.20$2500 & < 8.91 4.50 9.53 9.93 10.21 12.24 23.46$2501-9999 21.53 14.97 24.11 16.39 25.28 55.10 19.26$10000 & > 62.09 77.08 58.43 63.75 56.25 13.28 53.09

Obama EraDifference$0 3.39 4.25 3.86 2.69 1.96 1.02$2500 & < 5.79 5.53 5.70 5.24 5.75 11.49$2501-9999 19.22 22.57 18.47 16.28 23.97 18.09$10000 & > 71.60 67.75 71.97 75.79 68.71 69.39

Bush EraDifference$0 1.21 1.84 1.12 1.42 1.00$2500 & < 5.52 2.55 5.93 2.66 7.56$2501-9999 14.49 7.98 16.00 11.36 15.00$10000 & > 78.77 87.63 76.95 84.56 76.44

Looking only at appointee-cohorts, the Obama appointees appear to be the outlier. Trump-, Bush II-, Clinton-, Bush I-, and Reagan appointees reduced bonds initially set by ICE by $10,000 or more in 73%-77% of cases. Obama appointees, by contrast, reduced bonds by that amount in only 64% of their cases. Looking only at changes in bond amounts by Era, we see that IJs’ bond amounts converge with ICE’s initial bond amount more through time. During the Bush II Era, 79% of IJ decisions lowered bond by $10,000 or more, as compared to 71% of decisions during

28-Jan-20] IMMIGRANT DETENTION 41the Obama Era, and then only 62% during the Trump Era. Interestingly, Trump-appointed IJs during the Trump Era diverged from ICE bond amounts to a higher degree than any other cohort in the Era. Trump appointees reduced bond by at least $10,000 in 77% of their cases, as compared to 58% and 64% for Obama-appointed and Bush II-appointed IJs, respectively.

At least on a purely descriptive level, these data suggest that IJs’ bond amounts during the Trump Era diverge from initial amounts set by ICE more than during preceding Eras. This finding is particularly surprising as ICE has been reported as setting higher bond amounts or denying bond outright more frequently during the current Administration than in prior years. Although one interpretation of these findings is that IJs may be asserting less independence from enforcement officials during the current Administration, the absence of controls for other potential contributing factors prevents a definitive causal relationship.

***Our findings show that along every metric of bond

hearings, noncitizens fare considerably worse during the Trump Era than they did during either the Bush II or Obama Eras. While rates of release on recognizance were extremely low throughout the period of study, they dropped considerably through time. Similarly, rates of bond denials increased through each Era. Perhaps more telling, overall win rates (release on recognizance and new bond amount) indicate that all appointee cohorts except Obama appointees were considerably less likely to award relief to the noncitizen during the Trump Administration than during prior Eras. These results suggest that the Trump Administration is exercising control over bond decisions not necessarily through its power to appoint more like-minded IJs, but rather its power to supervise earlier appointees.

An examination of bond amounts set by IJs reveals a similar picture. Without controlling for inflation, bond medians grew from $5,000 during the Bush II Era, to $5,500 during the Obama Era, and then jumped to $7,500 during the Trump Era. Indeed, 41% of the bonds set by IJs during the Trump Era were $10,000 or higher, as compared to only 23% and 22% for the Obama and Bush Eras, respectively. Again, breaking down these results by appointee cohort

42 IMMIGRANT DETENTION [28-Jan-20indicates that earlier-appointed IJs have behaved differently during the Trump Era than during preceding Administrations.

Finally, an analysis of the difference in bond amounts set by IJs compared to those set by the initial ICE enforcement officer show that these amounts have been diverging through time, opening up the possibility that IJs have acted more in concert with ICE enforcement officials than they have in the past.

IV. MAPPING FUTURE RESEARCHIn this section, we identify two directions for future

research. First, further study is warranted to see if the trends we identify persist even after controlling for other variables that might impact the IJ’s custody decision. A predictive study seeking to identify causal factors would need to control for a variety of variables including those on which the IJ decision is supposed to be based such as family ties, length of U.S. residence, employment background, financial situation, and criminal history.121 EOIR currently does not reliably code for these factors,122 but prior studies indicate that at least some of these variables are statistically significant predictors of IJs’ custody decisions.123

It would also need to control for a host of other factors that are not supposed to be legally relevant but may nonetheless influence IJ custody decisions. Extra-legal factors that have been shown to influence IJ decisions in other contexts include case-specific factors such as whether the alien was represented by counsel;124 and the alien’s nationality, race, and language.125 They also include factors

121 See In re Guerra, 24 I.&N. DEC. 37, 40 (BIA 2006) (providing non-exhaustive list of factors which may be considered by IJ in rendering custody decisions).

122 For example, the EOIR dataset provides a column for “criminal,” but the entries for this column were incomplete and inconsistent with other data; many entries were listed as “not criminal” even though the noncitizen was charged with a crime-based ground for removal.

123 See Ryo, supra note __.124 See Ryo, supra note __ (finding presence of counsel to be a

statistically significant predictor of IJ custody decisions); Eagly & Shafer, supra note 80 (finding that noncitizen represented by counsel seven times more likely to be released than pro se counterpart).

125 See Ryo, supra note __ ; Ryo, supra note 96, at 239-40 (finding in a sample of immigrant detention decisions from 2013 to 2015, Central Americans were 68% more likely to be detained because they pose a

28-Jan-20] IMMIGRANT DETENTION 43related to the IJ, such as gender, prior work experience, or tenure on the bench.126 The political and economic climate of the city in which the bond proceeding is heard may also be relevant.127

Changes in bond decision making may also result from factors entirely exogenous to the immigration courts such as shifts in migration patterns. For example, prior to 2014, the majority of noncitizens in detention were from Mexico; the years since have witnessed a surge in detainees seeking asylum from the “Northern Triangle” countries of El Salvador, Guatemala, and Honduras. These shifts in migration patterns may impact IJ custody decision making.

Similarly the behavior of ICE enforcement officials may shift IJ custody decisions. ICE in the past, for instance, set uniformly high bond rates for women detained in particular facilities. Those high bond amounts may have an anchoring effect on the IJs’ subsequent review of the bond amount. Changes in ICE enforcement patterns may also impact the types of cases heard by IJs. For example, cases decided during the Obama Administration, which sought to prioritize the removal of criminal aliens,128 would likely have a higher

danger to the community than those from other countries).126 Some scholars have found these factors to be statistically

significant predictors of whether an IJ will grant asylum to a noncitizen. GAO ASYLUM 2008, supra note 80, at 7, 36, 119 (gender); Miller et al., supra note 80 (gender and prior work history); Ramji-Nogales et al., supra note 80, at 47-48 (gender and prior work history); Chand et al., supra note 80, at 192 (tenure).

127 Daniel J. Hopkins, Politicalized Places: Explaining Where and When Immigrants Provoke Local Opposition, 104 AM. POL. SCI. REV. 40, 42 (2010) (describing “politicalized place” hypothesis and noting that “sudden demographic changes generate uncertainty and attention” and can “politicalize those changes in people’s minds”); Miller et al., supra note 80, at 84 (describing “threat” hypothesis); Benjamin Newman, Acculturating Contexts and Anglo Opposition to Immigration in the United States, 57 AM. J. OF POL. SCI. 374 (2013) (showing states with rapid growth in foreign-born population more likely to participate in immigration enforcement programs such as E-Verify); Chand et al., supra note 80, at 187 (finding share Democratic control of state government to be statistically significant predictor of asylum grant rates); Daniel E. Chand & William D. Schrechkise, Secure Communities and Community Values: Local Context and Discretionary Immigration Law Enforcement, 41 J. OF ETHNIC AND MIGRATION STUDIES 1621 (2015) (ICE agents in conservative agents deported more migrants).

128 See MEMORANDUM FROM DHS SEC’Y JEH JOHNSON, “Policies for the

44 IMMIGRANT DETENTION [28-Jan-20proportion of criminal noncitizens in the pool of immigrants seeking bond hearings. By contrast, we might expect to see a lower proportion of criminal aliens in bond hearings during the Trump Administration given its policy of enforcing against all noncitizens who may be removable.129 These case selection effects could alter custody outcomes even if IJs (as opposed to enforcement officials) were entirely independent from their political superiors in the Administration.

Changes in caseload volume could also impact custody determinations, as the IJ may have less time for an individualized consideration of the legally relevant factors and default to categorical thinking, whether based on her own predilections and policy preferences or those of her political superiors in the agency. As of June 2019, there were over 930,000 cases pending on the courts’ dockets, up from 430,000 in 2014.130 Future research should evaluate the extent to which these other factors influence IJs’ custody decisions, and whether the findings set forth in this Essay survive after controlling for those other variables.

A second avenue for future research would be to identify the factors which should be used in immigrant custody hearings. Such findings could produce more accurate detention decisions, ensuring the detention only of those noncitizens who actually pose a flight risk. They would also aid the government’s efforts to establish a “risk classification assessment” to systematize the factors rendering a noncitizen a high flight risk.131 Currently, judges Apprehension, Detention, and Removal of Undocumented Immigrants,” (Nov. 20, 2014) (prioritizing for removal criminal aliens and those apprehended at the border). Aliens apprehended at the border generally would not be eligible for bond hearings before IJs. See supra note ___.

129 Exec. Order 13,768, “Enhancing Public Safety in the Interior of the United States,” 82 FED. REG. 8799, Jan. 30, 2017 (stating policy of enforcing law “against all removable aliens”).

130 EXEC. OFF. IMM. REV., ADJUDICATION STATISTICS (July 24, 2019), at https://www.justice.gov/eoir/page/file/1060836/download; see also GAO-17-438, ACTIONS NEEDED TO REDUCE CASE BACKLOG AND ADDRESS LONG-STANDING MANAGEMENT AND OPERATIONAL CHALLENGES (June 2017).

131 Cf. Mark Noferi & Robert Koulish, The Immigration Detention Risk Assessment, 29 GEO. IMM. L.J. (2014) (criticizing current assessment tool for failing to accurately predict bail risks). We recognize that in the pre-trial criminal detention context, such assessments have been shown to perpetuate structural bias and inequality. Nonetheless, as Sandra G.

28-Jan-20] IMMIGRANT DETENTION 45are instructed to consider flight risk based on the noncitizen’s length of residence in the United States, family ties, employment background, and prior efforts to abscond from law enforcement. But none of these factors has been empirically proven to predict the likelihood that a noncitizen will appear for removal proceedings.

It is worth noting here that the necessity of using detention at all, at least in the vast majority of cases, remains unclear. While it is true that detaining an alien guarantees his or her later appearance, recent scholarship shows that noncitizens’ appearance rate is high even without detention. For example, scholars have shown that 96% of families seeking asylum attended all of their hearings after being released from detention.132 Other mechanisms such as electronic monitoring and periodic check-in requirements may also be effective in increasing the likelihood of appearance in immigration court.133 The danger of erroneously detaining absent a flight risk is particularly acute in the immigration context, where a noncitizen with a valid claim to remain in the United States might nonetheless opt to abandon that claim simply to avoid detention.

It is also possible that additional factors should be considered in assessing flight risk. For example, a noncitizen’s likelihood of ultimately obtaining relief from removal may be relevant in calculating flight risk. It stands to reason that if a noncitizen has no colorable claim to relief from removal, then he or she is more likely to abscond to avoid inevitable removal. By contrast, a noncitizen with a strong claim to relief from removal is likely to attend removal proceedings that will lead to lawful presence in the United States. Indeed, the current system, which does not generally consider likelihood of ultimate relief,134 creates a Mayson points out, such assessments could be used to provide additional support for, rather than additional detention of, individuals found to be high risk. Sandra G. Mayson, Bias In, Bias Out, 128 YALE L.J. 2218 (2019).

132 Eagly, Shafer, & Whalley, supra note 80, at __. 133 CONG. RES. SERV., CRS REPORT R45804, Immigration: Alternatives to

Detention (ATD) Programs; GAO, ALTERNATIVES TO DETENTION, supra note __

134 In re Guerra; cf. Potash v. District Director of INS, 169 F.2d 747, 751 (2d Cir. 1948) (identifying likelihood of ultimate removal as

46 IMMIGRANT DETENTION [28-Jan-20perversion in the immigrant detention system. Those with the strongest legal claims to remain in the United States are the ones most likely to remain detained. While an individual with a weak legal claim may well decide to abandon it in the face of detention, individuals would likely tolerate lengthy detentions in oppressive conditions if they truly and reasonably feared persecution if repatriated, for example.

Another factor that might be considered is the noncitizen’s ability to pay a bond amount. Policies that impose the same bond amount to entire categories noncitizens make little sense because a wealthier noncitizen may be able to post the bond amount easily (and care little for losing the bond if she absconds), while a poorer noncitizen may be detained simply because of his inability to pay.135 Pursuant to a class action lawsuit in the Ninth Circuit, IJs in that circuit (but only that circuit) are required to consider ability to pay in setting bond amounts.136

From a more normative perspective, we question whether concerns that the noncitizen would endanger public safety if released constitute a valid ground for detention pending removal proceedings. Unlike in the criminal pre-trial context, noncitizens in removal proceedings need not have been accused of any crime at all—they may simply be removable because they entered the United States without inspection or overstayed a visa. And even for those who are removable on the basis of criminal conduct, detention pending the outcome of removal proceedings occurs only after the alien has already served the full criminal sentence deemed appropriate for the crime. Detaining them further due to their immigration status under these circumstances begins to look punitive.

Justice Black’s dissenting opinion in Carlson v. Landon is instructive in this regard. In that case, noncitizens were charged with removal on the basis of their membership in the Communist party.137 In a 5-4 decision, the Supreme

permissible factor in immigrant detention decisions).135 ACLU Study showing rates at which noncitizens remain detained

due to inability to post bond amount. https://www.aclu.org/issues/immigrants-rights/immigrants-rights-and-detention/discretionary-detention

136 Hernandez v. Sessions, 872 F.3d 976 (9th Cir. 2017).137 Carlson v. Landon, 342 U.S. 524, 544 (1952).

28-Jan-20] IMMIGRANT DETENTION 47Court sustained their detention for the purpose of preventing them from “aid[ing] in carrying out the objectives of the world communist movement.” Justice Black dissented, reasoning “Since it is not necessary to keep them in jail to assure their compliance with a deportation order, their imprisonment cannot possibly be intended as an aid to deportation…. A power to put in jail because of dangerousness cannot be derived from a power to deport.”138 Additionally, in our view, deterrence should play no role in immigrant custody decisions. Deterrence is a rationale for penological incarceration; it cannot be a ground for detention pending removal. But political Administrations have periodically defended immigrant detention precisely on such grounds, asserting that such detention is necessary to deter others from seeking to enter the United States. The District Court for the District of Columbia has cast doubt on such reasoning, rejecting the notion that “one particular individual may be civilly detained for the sake of sending a message of deterrence to other Central American individuals who may be considering immigration” as “out of line with analogous Supreme Court decisions” reserving deterrence goals for the criminal justice system.139

CONCLUSIONThe prospect of politicized custody decisions challenges

the very core of our notions of due process. Even those who view the ultimate decision to deport noncitizens as being vested exclusively in the political branches should chafe at the suggestion that the decision of whom to detain—for spans of months or even years—should be directed by an individual President’s political agenda.

We do not deny, however, any role for Executive Branch political officials, at least under the current system in which IJs are housed in the Executive Branch.140 Political

138 Id. at 551.139 R.I.L.R-R v. Johnson, 80 F. Supp.3d 164 (D.D.C. 2015); see also

Aracely v. Nielsen, 319 F. Supp.3d 110 (D.D.C. 2018) (rejecting deterrence rationale for immigrant detention).

140 A growing number of scholars and organizations have endorsed moving immigration proceedings into a new Article I court. Jill Family, INJECTING INDEPENDENCE AND PROPORTIONALITY INTO IMMIGRATION ADJUDICATION, IN RETHINKING ADMIN LAW FROM APA TO Z (American Constitution Society 2019), at

48 IMMIGRANT DETENTION [28-Jan-20supervisors might legitimately act to reduce arbitrariness and disparities in custody decisions. After all, such disparities arguably compromise rule of law norms. Political actors within the Executive Branch might properly engage in notice-and-comment rulemaking to promulgate regulations specifying the types of factors IJs should consider in their detention decisions. Or, the Attorney General might identify such factors through his power to refer BIA cases to himself and formally review them.141 Where such actions do not exceed statutory bounds and do not compromise individual due process interests, they, in our view, may legitimately limit the adjudicatory discretion of IJs.

By contrast, efforts by the President, the AG, or any other political subordinates to sway IJ decisions by simply directing IJs to detain more aliens or set uniformly high bond amounts would in our view raise due process concerns. They would also result in the pointless expenditure of considerable funds to detain someone who poses little to no flight risk. A solution to mitigating decisional disparities which opts to uniformly deny bond or consistently set unrealistically high bond amounts, without any individualized assessment of the person’s flight risk or dangerousness, raises significant due process concerns. A better approach would be to impose stronger guidelines on the factors that should be considered in determining https://www.acslaw.org/wp-content/uploads/2019/06/Rethinking-Admin-Law-From-APA-to-Z.pdf; AM. BAR ASS’N, COMM’N ON IMMIGRATION, 2019 UPDATE REPORT REFORMING THE IMMIGRATION SYSTEM: PROPOSALS TO PROMOTE INDEPENDENCE, FAIRNESS, EFFICIENCY AND PROFESSIONALISM IN THE ADJUDICATION OF REMOVAL CASES 6-3 (Mar. 2019), at https://www.americanbar.org/content/dam/aba/publications/commission_on_immigration/2019_reforming_the_immigration_system_volume_2.pdf (endorsing creation of Article I court to handle removal adjudication); STATEMENT OF THE AMERICAN IMMIGRATION LAWYERS ASS’N SUBMITTED TO THE SENATE JUDICIARY SUBCOMMITTEE ON BORDER SECURITY AND IMMIGRATION HEARING ON “STRENGTHENING AND REFORMING AMERICA’S IMMIGRATION COURT SYSTEM,” Apr. 18, 2018; NATIONAL ASS’N OF IMMIGRATION JUDGES, An Article I Immigration Court – Why Now is the Time to Act, at https://www.naij-usa.org/images/uploads/publications/Article_I_-_summary-of-salient-facts-and-argumentsOct-2016-FINAL_1.pdf; FEDERAL BAR ASS’N, Congress Should Establish an Article I Immigration Court, at http://www.fedbar.org/Advocacy/Article-1-Immigration-Court.aspx (proposing new legislation).

141 8 C.F.R. § 1003.1(h).

28-Jan-20] IMMIGRANT DETENTION 49whether the noncitizen should be detained or released. Uniformity need not bend toward more detention.

* * *