- The Legal Workshop - http://legalworkshop.org -
Learning to Live with Unequal Justice: Asylum and the Limits to Consistency
Posted By Stephen H. Legomsky On December 14, 2009 @ 1:01 am In Administrative Law, Civil Procedure, Law Review Article, Stanford Law Review | No Comments
This Article is about consistency in adjudication. I explore why consistency matters, what its determinants are, and whether it can be substantially achieved at a price that is worth paying.
This Article is also about the United States asylum adjudication system. Asylum challenges the national conscience in distinctive ways. It generates hard questions about our moral responsibilities to fellow humans in distress; the recognition of human rights and our willingness to give them practical effect; the extent of our obligations to those who are not U.S. citizens; U.S. legal and moral obligations to the international community; the roles of state sovereignty and borders; foreign relations; allocation of finite national resources; and racial, religious, linguistic, and ideological pluralism.
Into this emotional and political fray, one often better known for polemic than for hard data, recently ventured Professors Jaya Ramji-Nogales, Andrew Schoenholtz, and Philip Schrag. Through painstaking and thoughtful empirical research, they collected massive data from several different federal bureaucracies and shed important light on the results asylum adjudicators reach. Their impressive study, Refugee Roulette: Disparities in Asylum Adjudication (Asylum Study), highlights the striking disparities in asylum approval rates from one adjudicator to another at various stages of the process.1 As the authors convincingly demonstrate, asylum outcomes often depend as much on the luck of the draw as on the merits of the case.
The present Article has two aims. The first, which is asylum-specific, addresses the “so what” question. What are the normative implications of the findings reached in the Asylum Study? What problems have the sharp disparities in asylum approval rates caused, and what, if anything, should we do about them? To answer those questions, the Article sets a second objective—to examine, more generically, the role that consistency should play in any justice system. What, exactly, is the relationship between consistency and justice? What forces influence consistency? What instruments might enhance it? And what trade-offs do those instruments present?
Many readers will find the patterns revealed by the Asylum Study shocking. One’s visceral reaction might be that we need to “rein in” the adjudicators. Perhaps, one might think, the answers lie in terminating or demoting the outliers, or subjecting all adjudicators to performance evaluations, or making vastly increased use of agency head review of adjudicators’ decisions, or even imposing mandatory minimum and maximum approval rates.
I argue here that these impulses should be resisted. There are times when we simply have to learn to live with unequal justice because the alternatives are worse. Disparities in asylum approval rates just might be one of those instances. As long as adjudicators are flesh-and-blood human beings, as long as the subject matter is ideologically and emotionally volatile, and as long as limits to the human imagination constrain the capacity of legislatures to prescribe specific results for every conceivable fact situation, there will be large disparities in adjudicative outcomes and justice will depend, in substantial part, on the luck of the draw.
This is not to suggest that inconsistent outcomes are harmless; they impede justice in several ways. Consistency is a reasonably good proxy for accuracy. If, for example, sixty percent of a group of decisions go one way and the remaining forty percent the opposite way, and the facts are similar enough that the two sets of outcomes cannot be reconciled, it seems likely that at least forty percent of the decisions—and perhaps sixty percent—were wrong. That conclusion is not inevitable, because not all issues lend themselves to uniquely correct results, and because, even for those issues that do, sometimes the forty percent are right and the sixty percent are wrong. Still, rational human choice is more likely than random selection to produce correct outcomes. On that assumption, a high degree of consensus makes the hypothesis of the majority being right more likely than the hypothesis of the majority being wrong. There is some reason, therefore, to assume that consistency correlates positively with accuracy.
Probably the most intuitive benefit of consistency, however, is the principle of equal treatment—the notion that inconsistent outcomes are substantively unfair. When two people are situated identically in all legally relevant respects, the law should treat them the same. To the extent reasonably avoidable, the outcomes should not hinge on the biases of whichever adjudicator the individual had the good or bad luck to draw.
Certainty, and the predictability that it brings, are commonly cited as a third set of reasons to strive for consistent adjudication. Conflicting results breed uncertainty in two ways. They do so directly, by preventing the parties from predicting how their dispute is likely to be resolved. Consistency might also contribute to certainty and predictability indirectly, by enhancing the stability of the law. Conflicts among equally authoritative bodies have ways of being reconciled eventually, either by gradual evolution or by pronouncements from above. The mere presence of a momentary conflict, therefore, can create at least the perception of imminent change, leaving affected sectors of the population uncertain how to plan for the future. Consistency reduces this uncertainty.
Inconsistency can also impair efficiency. The very fact that two decisions are inconsistent means that the second adjudicator had to revisit the analytical efforts of the first one rather than simply adopt the first adjudicator’s reasoning and result. It also means that, at some point, some government actor will have to step in to resolve the issue definitively. Moreover, the resulting uncertainty leaves the parties with less incentive to accept the first ruling in their case and more incentive to appeal it. The fact that they cannot predict the result might also discourage future parties from settling. Apart from conserving judicial and administrative resources, encouraging litigation and appeals rather than settlements and acceptance of initial decisions prolongs the waiting times—a key consideration for both the applicant and the government in asylum cases.
A final benefit of consistency is acceptability to both the parties and the general public, a central concern of every adjudication process. The public has a direct interest in consistency, since uncertainty can be problematic for the reasons already given. In addition, there is ample evidence that the public simply perceives inconsistent outcomes to be unfair. As the authors of the Asylum Study observe, we inscribe the equal justice admonition at the entrance to the Supreme Court building, follow stare decisis, promulgate uniform federal sentencing guidelines, employ pattern jury instructions, and allow judges to modify civil verdicts that veer too far from the norm.
All else equal, therefore, it is hard to be against consistency. Indeed, fidelity to the rule of law demands attention to consistency. But all else is seldom equal. Since strategies that enhance consistency can have costs, the real question is how much cost should be accepted in return for whatever amount of increased consistency it will purchase. For one thing, conflicts can have positive effects of their own. As others have observed, a judicial conversation that includes differing views expressed over a reasonable time period can be part of a healthy maturation process that ultimately aids the thoughtful resolution of a difficult issue. In addition, even when the net impact of conflicts is negative (as I assume to be the norm), some solutions might be too costly. Strategies like reductions in adjudicators’ decisional independence, broader or more frequent agency head review of adjudicators’ decisions, heightened judicial deference to administrative tribunals, or even elimination of judicial review of the decisions by centralized tribunals, for example, might well enhance consistency, but at a price that I argue would be excessive.
I offer two caveats: First, balance is not the same as, and does not promote, consistency. At best, balance prevents asymmetric inconsistency. An immigration judge corps that comprises one hundred anti-immigrant zealots and one hundred pro-immigrant zealots would be “balanced” in some sense, but in such a corps the outcomes would be more likely to diverge, not less. Second, inconsistency is a two-edged sword. It can result in an outcome favorable to the asylum seeker when another adjudicator would have reached a different result, or vice-versa. Consequently, neither one’s general ideology nor one’s specific preferences on immigration or asylum should drive one’s degree of tolerance for inconsistent outcomes.
It is useful, therefore to identify the determinants of consistency. I suggest there are at least fifteen, which the full version of this article considers in more detail. Some of those determinants relate to numbers—the number of people who decide each case, the total number of adjudicators or panels in the entire system, and the number of cases. Some of the other determinants relate to the attributes of the adjudicators, including the criteria and procedures for appointing them and their post-appointment training and guidance. Still others relate to the adjudicators’ roles—their degree of independence, the level of deference they are expected to give to other decision-makers, and their obligations with respect to the preparation of reasoned opinions and the use of stare decisis. Finally, I suggest that the level of consistency reflects the nature of the subject matter—in particular, how specialized, complex, dynamic, ideologically charged, and determinate the concepts are.
Of these variables, which ones might account for the disparate outcomes observed in the asylum setting? Given the persistence of large variances at all levels, the elimination of country of origin as an explanatory factor, and the adjudicator-specific patterns revealed by the authors of the Asylum Study, it seems easy to identify the principal contributors. They include the adjudicators’ differing ideologies and attitudes, which affect their preexisting inclinations to grant or deny asylum, and the subject matter, which is indeterminate enough, complex enough, and dynamic enough to give adjudicators relatively broad freedom to reach the outcomes they desire. The attitudes that asylum adjudicators inevitably bring to their work include not only their general philosophies about asylum or immigration, but also their normative conceptions of the adjudicative role, their levels of suspicion about the credibility of the applicants, and the weights they attach to erring on the side of either the individual or the government.
That is unequal justice to be sure, but my basic thesis is that for the most part we shall have to live with it. Unless the adjudicators can be made ideologically homogeneous—a goal I find neither desirable nor achievable—there will always be substantial asylum approval rate disparities and many outcomes will reflect the luck of the draw.
This is not, however, a call for complacency. Consistency is a positive virtue for all the reasons I have acknowledged, and I suggest steps that might enhance it at the margins. The key is to aim low and to settle for treating the symptoms.
Some worthwhile steps could achieve modest gains at the margins. More detail could be added to the statutes, regulations, and subordinate legislation on such issues as what constitutes “persecution,” though it would be hard to anticipate every conceivable means of persecution from which the law should afford protection. More adjudicators might complicate the task of achieving consistency, but the increased attention that an enlarged judge corps would allow an adjudicator to devote to each case might well generate a higher degree of consistency. Larger decisional units—for example through the restoration of three-member Board of Immigration Appeals panels in greater numbers of cases and increased use of en banc decisions on recurring issues—could enhance the Board’s internal consistency. Strengthening the support staff could have the same effect. So, too, could the provision of counsel to indigent asylum seekers at government expense—at least in cases in which the applicant has cleared some specified threshold requirement of meritoriousness. New quality controls for the hiring of asylum adjudicators seem warranted as well. While personal ideology will always be part of what an asylum adjudicator brings to the job or at least soon develops, the process might more consciously avoid affirmatively factoring a candidate’s ideology into the hiring decision. Enhancement of the various programs for professional development, the internal and perhaps even the public dissemination of each adjudicator’s asylum approval rates, expanding the Board of Immigration Appeals’ (BIA) scope of review over fact questions (from “clearly erroneous” review to de novo review, for example), and the broader use of reasoned and binding opinions carry additional potential for modest gains in consistency.
Other possible policy responses, while potentially enhancing consistency, are bad ideas nonetheless. The Attorney General, for example, may review any BIA decision that he or she wishes. This is not an unusual arrangement; Congress often authorizes agency heads to review adjudicative decisions that fall within their domains. To reduce the approval rate disparities identified in the Asylum Study, one might be tempted to urge more frequent Attorney General review of BIA decisions. Agency head review is often extolled as a means for agency heads to assure inter-decisional consistency and to maintain control over basic policy at the same time. But agency head review is not essential to either goal. When there is a designated appellate authority such as the BIA, an en banc decision of that tribunal can yield the same consistency as agency head review. Congress could even authorize the agency head to require the appellate tribunal to go en banc in a particular case if there is a concern that an overworked adjudicative tribunal would not do so on its own.
The need for agency primacy over policy matters can be conceded, but again, agency head review is not essential to agency policy primacy. Rulemaking and other policy mechanisms are also available. The multiple experts from whom the agency head can distill advice and perspectives will be just as available in a rulemaking proceeding as they are in agency head review of adjudication. The agency head will be just as capable of asserting agency policy primacy via rulemaking as he or she would be via review of adjudication. And rulemaking will be just as effective in promoting agency policy coherence as review of an adjudicative decision would have been—more so, if anything, since the facts of a particular case will not constrain the reach of the rule. In the asylum context, the arguments based on agency policy coherence are particularly inapt, since the immigration judges and the BIA are within the Department of Justice while the analogous policymaking agencies are now located within the Department of Homeland Security. Agency policy coherence, therefore, is simply not an issue in this context. Moreover, as Jeffrey Lubbers has observed, there is normally a lessened need for political control in “high-volume, fact-based” adjudication processes and those in which benefits are sought. Asylum fits both descriptions.
While I acknowledge that even adjudicative decisions will often require policy judgments—particularly if the decisions are designated as precedential—the basic functions of the adjudicators are, after all, to find facts, interpret law, and exercise specific statutory discretionary authority. Even when a case presents an important policy question, the agency head can supersede the decision by issuing a generally applicable regulation if he or she wishes—provided, of course, that Congress has delegated the relevant rulemaking authority to the agency head. If Congress has not done so, then Congress’s inaction is itself a policy decision that requires respect.
Further, as the consultants’ report for a leading Administrative Conference study acknowledges, rulemaking has tremendous advantages over adjudication as a vehicle for policy formation.2 These advantages include broader public input, notice to Congress, avoidance of adjudicative hearings to resolve issues of legislative fact, avoidance of litigating the same issues repeatedly, more enforceable rules, clearer advance notice of allowable and prohibited conduct, fairer applicability of the rules to similarly situated individuals at different points in time, and the opportunity for affected individuals to make policy submissions before the rule is adopted. To be sure, notice-and-comment rulemaking can be slow and cumbersome. But if on a given issue the agency feels that the notice-and-comment procedure would be too onerous, interpretative rules might be an alternative method of influencing adjudicative outcomes in ways that promote the agency’s policy agenda. Interpretative rules do not bind the public, and it is not clear whether they can bind the adjudicators, but they can be issued quickly and without the fiscal cost of notice-and-comment machinery. And when, “for good cause,” an agency feels that the notice and comment procedure would be “impracticable, unnecessary, or contrary to the public interest,” such as when the timing is urgent, the agency can issue an interim regulation. For all these reasons, the need for agency head review is seldom pressing.
Moreover, the central rationale for agency head review—the agency’s political accountability—is also precisely what makes agency head review affirmatively troublesome. The agency head and any subordinates to whom he or she delegates the review function are subject to popular and political pressures. On matters of policy that reality is not problematic; consideration of the public’s preferences is at home in democratic theory. But the essence of the adjudicative function is to find facts and interpret the law, not to please the public. While policy admittedly is implicated in a certain number of cases, the adjudicative function generally requires independence, not political accountability, as discussed below.
Agency head review has other costs as well. It permits a dangerous concentration of power in the hands of a single individual. When the decision being reviewed was rendered by a multi-member panel, agency head review entails the substitution of one person’s judgment for the collective judgment of several adjudicators. And the probability that a strong ideological bias will influence the result is greater when one person is deciding than when the decision is rendered by a randomly selected multi-member panel.
To sum up: There is little need for agency head review. Decisional consistency can be achieved through a combination of the administrative appellate process, legislative rules (including interim rules when necessary), and interpretative rules. Rulemaking and other powers can also preserve agency policy primacy and agency policy coherence. Moreover, agency head review poses inherent dangers to the dispensation of justice, including especially the substitution of a political outcome for one based on an independent adjudicative tribunal’s honest reading of the evidence and the law. All of these considerations have special force in the asylum context, where the stakes are high and the potential for inappropriate political and ideological influence has been amply demonstrated.
Restrictions on judicial review might also be perceived as a way to increase consistency, since the twelve general courts of appeals collectively have far more decisional units than the one BIA. But the courts are already forbidden to review at least two important categories of asylum denials—those reached in expedited removal proceedings and those based on findings that failure to file the claim within the one-year deadline was not attributable to changed or extraordinary circumstances.3 Still, for the ostensible purpose of reducing disparities in the asylum approval rates, some might be tempted to advocate further restrictions on judicial review of asylum denials. Those restrictions could conceivably include barring judicial review of other selected subcategories of asylum cases, making judicial review discretionary, or narrowing the scope of review.
Admittedly too, there are other costs of judicial review of administrative decisions. They include judges’ lack of political accountability, the risk of error when nonexperts review the decisions of experts, the fiscal expense, and the delays.
In my view, however, the benefits of judicial review overwhelm its costs, particularly in the asylum context. Those benefits too have been explored elsewhere and need only be summarized here.4 Probably the most obvious are the independence that judges bring to their work and the corresponding appearance of justice. Judicial independence, in turn, is beneficial for several reasons that are explored below and assumes special importance in asylum cases because of the recent threats to the independence of the immigration judges and the BIA members. Judicial review also adds the perspective of generalist judges to the existing perspectives of the specialists whose decisions are being reviewed. It provides a regime in which legal doctrine can evolve gradually, step by step, informed by the judicial conversation that multiple courts of appeals can supply. And the mere prospect of judicial review should add an incentive for the original decision makers to reach their conclusions thoughtfully and explain them carefully. Given all the recent criticism of the haste with which asylum claims are denied, any sobering effect of judicial review on the administrative adjudicators should be savored.
Despite the inconsistencies that judicial review of asylum claims inevitably introduces, and despite its other costs, therefore, any calls for further restrictions on judicial review of asylum claims should be vigorously resisted. To the contrary, the existing restrictions should be repealed. The Asylum Study demonstrates beyond doubt that ideology explains a large part, if not most, of the striking disparities in asylum adjudication. There is simply no reason to assume that the same biases are strangely absent when the asylum decisions are rendered in expedited removal proceedings or on the basis of no “changed circumstances” or no “extraordinary circumstances.” Nor is there any reason to assume that in these cases the consequences of error are any less grave.
By the same token, there is no convincing reason to narrow the scope of review. It is narrow enough already. As in other removal cases, the court may set aside a finding of fact only if “any reasonable adjudicator would be compelled to conclude to the contrary.”5
Replacing review by the general courts of appeals with review by a specialized tribunal might possibly enhance the consistency of the ultimate outcomes, since specialization reduces the total number of different adjudicators who will be needed to handle a given caseload. Specialization also heightens the adjudicators’ familiarity with analogous decisions. But that strategy too would come at a price—the loss of the generalist perspective, which enables judges to draw guidance from other subject areas and to approach the specialty area with fewer preconceptions or biases. A diet of specialized cases might also make the positions less attractive to potential adjudicators and staff, thus hampering both recruitment and retention of the most talented personnel. Specialization might render the appointment process more susceptible to lobbying pressures, and it might cause the adjudicators to become too cozy with the litigators who appear before them regularly.
Still other policy prescriptions carry the potential for greater gains in consistency but in my view would be especially bad ideas. Quotas or other direct controls on outcomes would undoubtedly reduce the disparities in asylum approval rates, but as the authors of the Asylum Study point out, there is no way to locate either the “right” percentages or the “right” range; any figures would be arbitrary. Besides, they observe, rapid changes in human rights conditions would render the announced percentages continually obsolete. In addition, most source countries have too few asylum applicants to provide a statistically significant sample.
Other objections might be added. The argument that statistically reliable percentages would be too hard to fashion for many countries assumes a system in which each source country is allotted a different approval rate range. That feature would itself be problematic, reminiscent of the discredited national origins quota system in place from 1921 to 1965.6 Yet, without such differentiation, the combination of drastically different human rights conditions from one source country to another and different mixes of cases by source country from one asylum office or immigration court to another would cause outcomes to hinge needlessly on the particular office or court in which the cases are filed. The end result would be less consistency, not more. Finally, numerical controls would require adjudicators to rank asylum claims. There is no uniquely correct way to do so. Some might base their rankings on the probability of persecution, others on the severity of the alleged persecution, still others on the quality of the nexus between the persecution and one of the protected groups. The range of choices would introduce another element of inconsistency, as different adjudicators would attach different weights to different factors and might even misuse that freedom to reintroduce ideology.
I have saved my most serious worry for last. Given the glaring disparities in the asylum approval rates from one adjudicator to another, one temptation might be to “rein them in.” This could be done by taking wayward adjudicators aside, quietly “encouraging” them to increase or decrease their approval rates, and then, after a decent interval, terminating or reassigning those who remain recalcitrant. Performance reviews that take approval rates into account and serve as a criterion for retention or promotion might be another device for eliminating adjudicators who veer too far from the mean.
Any of these strategies might well reduce the disparities in asylum approval rates. But threats to adjudicators’ job security inherently compromise their decisional independence. The actions of attorneys general in the recent past have already dangerously sapped the independence of the immigration judges and the BIA.
In a previous article I explored the implications of decisional independence more generally;7 here, they will be just briefly recounted. Decisional independence has costs that have to be acknowledged. Probably the most controversial cost is that, by definition, decisional independence eliminates the adjudicator’s political accountability. When the decision has broader policy implications, as is especially likely when it is designated as binding precedent, that consequence can be viewed as a cost to the democratic process. It is a cost that we readily accept when courts interpret an entrenched Constitution, use judgment in interpreting ambiguous statutory language, or make common law. It is a cost nonetheless.
Some might feel that decisional independence erodes agency policy primacy. The earlier discussion on agency head review of adjudicative decisions, however, showed how agency policy primacy can be preserved through rulemaking and other devices. The admittedly substantial logistical constraints can be minimized. But whether or not one shares that assessment, the point here is that even a passionate advocate of agency head review can applaud decisional independence. Precisely that combination—adjudicator independence in reaching the decision but agency head authority to reverse it—lies at the heart of the compromise philosophy enshrined in the Administrative Procedure Act.8
In the administrative context, a further cost, many would argue, is the kind of decisional inconsistency exposed by the Asylum Study. Earlier discussion suggested that decisional independence might have mixed effects on decisional consistency, but let us assume arguendo that the net effect is negative. There is also the related problem of assuring that adjudicative decisions cohere with other expressions of agency policy.
Decisional independence might also impair good faith measures to boost adjudicators’ productivity. While there might be ways for agencies to impart productivity expectations to adjudicators without threatening their independence, the key variable is the consequence of failure to meet those expectations. If the consequences are significant enough to alter the adjudicators’ behavior—and communicating expectations would be useless if they are not—then they will necessarily give adjudicators an incentive to trade off care and quality for quantity, since only the latter can be statistically compiled. For that reason, independence and productivity will always be in tension.
In an adjudicative setting, my view is that decisional independence, despite these potential costs, is critical to the rule of law and to the dispensation of justice. The most familiar benefit of decisional independence is procedural fairness—minimizing adjudicative bias. An adjudicator should decide each case based on his or her honest reading of the evidence, interpretation of relevant legal sources, and exercise of any delegated discretion—not by choosing whichever outcome seems most likely to please the officials who will control his or her professional future. Decisional independence can also discourage what I have called “defensive judging”—playing it safe by avoiding rulings that might prove controversial. Decisional independence can be a vital safeguard for unpopular individuals, minorities, and political viewpoints, and it is crucial to safeguarding constitutional rights against transient majoritarian preferences. And decisional independence is integral to at least the U.S. version of separation of powers.
Apart from those rationales, which I have argued all derive ultimately from fidelity to the rule of law, decisional independence has important side benefits. They include maintaining public confidence in the integrity of the justice system, avoiding “reverse social Darwinism” in which the weakest adjudicators are the ones most likely to survive ideological purges, attracting and retaining adjudicator candidates, and facilitating the continuity of adjudicative outcomes from one administration to its successor.
For all those reasons, further assaults on the decisional independence of the immigration judges and the members of the BIA would be regrettable. To the contrary, their prior decisional independence should be restored and further safeguarded despite any possible negative effects on either decisional consistency or agency policy coherence.
The hobgoblin of little minds it might well be, but consistency matters. The moral imperative of equal justice, the needs for certainty and predictability, the benefits of efficiency, and the objective of public acceptability all demand attention to consistency in any adjudicative framework. The Asylum Study—the product of a prodigious and highly successful effort by Professors Ramji-Nogales, Schoenholtz, and Schrag—has brought home the extraordinary extent to which the outcome of an asylum claim hinges on the particular adjudicators who are assigned the case.
But the forces that generate inconsistent adjudicative outcomes are not easy to constrain, at least not without costly trade-offs. In asylum cases, the unavoidable abstractness, complexity, and dynamism of the relevant legal language make it inevitable that the human adjudicators will bring their diverse emotions and personal values to bear on their decisions. Under those circumstances, we should not expect anything but the sorts of disparate outcomes that the Asylum Study has documented.
There are ways to reduce the inconsistencies at the margins, to be sure. These have been noted. But any strategies that would shrink the inconsistencies more dramatically—and some that would not do even that—have costs that I argue are unacceptably high. These include more frequent agency head review of BIA decisions, additional restrictions on judicial review, transferring review to a specialized court, and punishing wayward adjudicators. Each of those devices would either severely compromise decisional independence or impose other excessive costs.
Copyright © 2009 Stanford Law Review.
Stephen H. Legomsky is the John S. Lehmann University Professor at Washington University School of Law.
This Legal Workshop Editorial is based on the following full-length Article: Stephen H. Legomsky, Learning to Live with Unequal Justice: Asylum and the Limits to Consistency, 60 STAN. L. REV. 413 (2007).
Article printed from The Legal Workshop: http://legalworkshop.org
URL to article: http://legalworkshop.org/2009/12/14/learning-to-live-with-unequal-justice-asylum-and-the-limits-to-consistency
Copyright © 2009 The Legal Workshop. All rights reserved.