The Yale Law Journal

April 2017

Inside the Agency Class Action

Administrative LawCivil ProcedureAntidiscrimination Law

abstract. Federal agencies in the United States hear almost twice as many cases each year as all the federal courts. But agencies routinely avoid using tools that courts rely on to efficiently resolve large groups of claims: class actions and other complex litigation procedures. As a result, across the administrative state, the number of claims languishing on agency dockets has produced crippling backlogs, arbitrary outcomes, and new barriers to justice.

A handful of federal administrative programs, however, have quietly bucked this trend. The Equal Employment Opportunity Commission has created an administrative class action procedure, modeled after Rule 23 of the Federal Rules of Civil Procedure, to resolve “pattern and practice” claims of discrimination by federal employees before administrative judges. Similarly, the National Vaccine Injury Compensation Program has used “Omnibus Proceedings” resembling federal multidistrict litigation to pool common claims regarding vaccine injuries. And the Office of Medicare Hearings and Appeals—facing a backlog of hundreds of thousands of claims—recently instituted a new “Statistical Sampling Initiative,” which will resolve hundreds of common medical claims at a time by statistically extrapolating the results of a few hearing outcomes.

This Article is the first to map agencies’ nascent efforts to use class actions and other complex procedures in their own hearings. Relying on unusual access to over forty agencies—including agency policymakers, staff, and adjudicators—we take a unique look “inside” administrative tribunals that use mass adjudication in areas as diverse as employment discrimination, mass torts, and health care. In so doing, we unearth broader lessons about what aggregation procedures mean for policymaking, enforcement, and adjudication. Even as some fear that collective procedures may stretch the limits of adjudication, our study supports a very different conclusion: group procedures can form an integral part of public regulation and the adjudicatory process itself.

author. Associate Professor and Associate Dean for Research, Michigan State University College of Law, and Professor of Law, Loyola Law School, Los Angeles. This Article draws from our research as academic consultants to the Administrative Conference of the United States, which adopted Recommendation 2016-2, Aggregation of Similar Claims in Agency Adjudication. See Adoption of Recommendation, 81 Fed. Reg. 40,259 (June 21, 2016). As part of this project, we surveyed dozens of agencies and conducted in-depth interviews with high-level officials and adjudicators in the Equal Employment Opportunity Commission, the National Vaccine Injury Compensation Program, the Office of Medicare Hearings and Appeals, as well as many others. For discussion and comments, we are grateful to Michael Asimow, Andrew Bradt, Sergio Campos, Seth Davis, David Engstrom, Maria Glover, Kristin Hickman, Samuel Issacharoff, Maggie Lemos, David Marcus, Noga Morag-Levine, Sean Pager, Judith Resnik, Glen Staszewski, Joan Steinman, Mila Sohoni, Christopher Walker, Amber Williams, Matthew Weiner, Judge Jack B. Weinstein, and all of the participants of the Federal Courts Workshop, the Administrative Law New Scholarship Roundtable, and the Civil Procedure Workshop. Barbara Bean, Gabrielle Fournier, Chelsea Gumaer, Evan Hebert, Zeeshan Kabani, Michael Kreiner, Ashley Sarkozi, David Sheaffer, Daryl Thompson, and Collette Torunyan provided invaluable research assistance.


A crisis is brewing in Medicare. In 2003, Congress created the Office for Medicare Hearings and Appeals (OMHA)—a special administrative court designed to resolve billing disputes between the federal government and hospitals, nursing homes, medical providers, and others.1 But after six years of relative normalcy, case filings at OMHA spiraled out of control. By 2014, OMHA’s backlog had spiked to almost 500,000 cases.2 Worse yet, average wait times for decisions mushroomed to almost two years in 2015.3 OMHA’s workload became so heavy that at one point it took five to six months just to enter new cases onto its docket.4

Medicare’s problems are hardly unique. Across the administrative state, the number of claims languishing in bureaucratic limbo has become a new crisis—creating significant backlogs, arbitrary outcomes, and new barriers to justice.5 The Department of Veterans Affairs recently admitted that veterans face average wait times of four years to obtain their disability benefits.6 In July 2016, the Department of Education reported that, nearly eighteen months after the collapse of the Corinthian Colleges, over 20,000 students were anxiously waiting for the Department to hear their claims for loan forgiveness.7 Even as Congress tries to create administrative programs to resolve claims more quickly than federal courts, agencies often meet the same Kafkaesque fate.8

But what made OMHA unusual was its response. Last year, OMHA adopted a new pilot program dubbed the Statistical Sampling Initiative (SSI) that allows hospitals, doctors, and other medical providers with large numbers of similar claims to conduct “trials by statistics.”9 Petitioners with more than 250 similar claims have the option to try a small sampling of those claims before an Administrative Law Judge (ALJ) and extrapolate the average result to the rest.10 To do so, petitioners meet with one of Medicare’s “trained and experienced statistical expert[s]” to develop the “appropriate sampling methodology” and randomly select the sample cases to be extrapolated to the whole.11 All of the pending claims are consolidated in front of a single ALJ who hears the sample cases. The results of the sample cases are then applied to the thousands of remaining cases.

Although OMHA’s SSI is still in its initial stages, it is notable for two reasons. First, it differs from the Supreme Court’s approach to such “trials by formula” in federal courts. Six years ago, the Court warned that the “novel” use of statistical sampling could stretch hearing procedures too far under the Rules Enabling Act by “abridg[ing], enlarg[ing] or modify[ing]” the substantive rights of the parties in such a mass action.12 To the extent statistical sampling remains a problem for federal courts,13 the Supreme Court’s words do not bind federal agencies. Federal agencies often enjoy discretion under their own statutes to craft procedures they deem “necessary and appropriate” to adjudicate the claims that come before them.14 OMHA’s program thus illustrates agencies’ freedom relative to federal courts to create innovative procedures that respond to problems in mass adjudication.

Second, agencies rarely exercise this freedom. As we discussed five years ago in The Agency Class Action,15 even though federal agencies hear far more cases each year than our federal court system, they have routinely avoided tools used by courts to efficiently resolve large groups of claims, like class actions and other complex litigation procedures. Unlike federal courts—where nearly forty percent of all cases now proceed in some form of organized litigation16—most agencies and specialized courts rarely use class actions or otherwise coordinate multiparty disputes. Consequently, in a wide variety of cases, such programs risk wasting resources in repetitive adjudication, reaching inconsistent outcomes for the same kinds of claims, and denying individuals access to the affordable representation that aggregate procedures promise.

Part of the reason for agencies’ restrained, individualized approach stems from the perceived limits of adjudication. For years, the Supreme Court and scholars have said that legislative bodies are better than judges at responding to problems of mass harm.17 Policymakers can resolve cases that raise the same complex factual and legal issues more openly and effectively through the legislative process.18 Judges, by contrast, should avoid such disputes because they lack the capacity to hear and resolve diffuse claims among large groups of people.19

That same perceived line between the appropriate roles of adjudicative and legislative bodies also exists inside administrative agencies. Before the Administrative Procedure Act (APA),20 agencies combined investigation, policymaking, and adjudication in the same department.21 The APA, however, separated the practice of “adjudication” from the agencies’ rulemaking and enforcement powers, establishing distinct rules for each type of agency activity.22 Going forward, formal individualized adjudications would be conducted on a case-by-case basis by ALJs insulated from undue political influence.

A handful of federal administrative programs, however, have quietly bucked this trend—employing class action rules, collective claims handling, and even the kinds of “trials by statistics” embraced by innovative federal judges around the United States.23 The Equal Employment Opportunity Commission (EEOC), for example, created an administrative class action procedure, modeled after Rule 23 of the Federal Rules of Civil Procedure, to resolve “pattern and practice” claims of discrimination by federal employees before federal administrative judges (AJs).24 The National Vaccine Injury Compensation Program (NVICP) uses “Omnibus Proceedings,” which resemble federal multidistrict litigation, to pool together common claims alleging a large group of vaccine-injured children.25 And, as discussed above, OMHA recently began a “Statistical Sampling Initiative” that will use trained and experienced experts to resolve thousands of common medical claims at a time by statistically extrapolating the results of a few hearing outcomes.26

This Article presents the first look inside the ways that federal agencies have used class actions and other complex litigation techniques in their own hearings. Building on our prior theoretical work,27 we received unusual access from the Administrative Conference of the United States (ACUS) to survey and interview adjudicators, policymakers, and staff involved in aggregate proceedings across the administrative state. A year and a half later, ACUS adopted and published our recommendations in the Federal Register, proposing that all agencies consider the use of aggregate procedures.28This Article presents our findings, including the extent to which agencies aggregate claims and the types of cases in which aggregation has proven most useful. In so doing, we offer important practical and theoretical lessons for both administrative and class action law.

As a practical matter, class actions and other complex procedures offer agencies important new tools to respond to rising case volumes while promoting legal access. These lessons are particularly timely inasmuch as agency aggregation appears to be on the rise. Just this past year, plaintiffs petitioned the Federal Maritime Commission to hear a multi-billion-dollar antitrust class action involving price fixing;29 the federal government conceded for the first time that a veterans court could hear class action claims by veterans in “appropriate cases”;30 the Department of Education adopted a process modeled on federal court class actions for students seeking loan forgiveness from predatory colleges that commit fraud;31 and a prominent federal judge recommended the Federal Trade Commission itself aggregate thousands of consumer and municipal false advertising claims.32

On a theoretical level, our on-the-ground assessment opens up new lines of inquiry for the study of aggregate adjudication. Scholars have long feared that collective procedures push the limits of what judges can do to resolve big cases.33 Years ago, the Supreme Court even barred federal courts from aggregating certain claims because they defy judicial resolution.34 Instead, it called on Congress to establish administrative programs to provide “the most secure, fair, and efficient means of compensating victims.”35 And yet, our study of those same programs supports a very different conclusion: far from pushing the limits of adjudication, aggregate procedures form an essential part of the adjudication process in any court.

The Article proceeds in five parts. Part I sets out the legal framework for adopting aggregate litigation procedures in federal courts and administrative agencies. Federal courts have long enjoyed authority to aggregate large groups of similar cases in one of two ways. First, courts may formally aggregate claims by, for example, permitting one party to represent many others in a single lawsuit. Second, courts may informally aggregate claims. In informal aggregation, different claimants with very similar claims each retain separate counsel and advance a separate lawsuit; however, these separate lawsuits proceed in front of the same adjudicator or on the same docket in an effort to expedite cases, conserve resources, and assure consistent outcomes.36 Agencies generally enjoy even more authority than federal courts to aggregate common cases, formally and informally.

Part II surveys the use of aggregation in the administrative state. We identified more than seventy administrative agencies and Article I courts with rules permitting some form of aggregation, but found that very few of them actually use those rules. Part III then looks inside agency aggregation, developing a typology of formal and informal aggregation and presenting three case studies that illustrate different approaches to aggregation in agency adjudication. This empirical work draws on our extensive interviews and surveys with high-level officials and adjudicators in many administrative programs, including the EEOC, the NVCIP, and OMHA, as well as our own independent review of their administrative dockets.

Part IV charts the costs and benefits of such programs. Our case studies show that aggregate adjudication techniques raise unique challenges. The sheer number of claims in aggregate agency adjudication may: (1) create “diseconomies of scale”—inviting even more claims that stretch adjudicators’ capacity to administer justice to many people; (2) increase the consequence of error; and (3) impact the perceived “legitimacy” of the process and challenge due process. Nevertheless, each program has identified best practices to ameliorate these concerns. In the process, aggregate adjudication allowed each tribunal to take advantage of the benefits of aggregation—pooling information about recurring problems, achieving greater equality in outcomes than individual litigation, and securing expert assistance at a critical stage in its own decision-making process—while minimizing its potential dangers.

Part V considers broader lessons about what aggregation procedures mean for policymaking, enforcement, and adjudication. Courts and commentators frequently raise concerns about the dangers of group litigation. Among other things, they worry that class actions and other complex procedures encourage free-form policymaking; create unaccountable “private attorneys general” who interfere with public enforcement; and stretch the very limits of judicial power and legitimacy.37 However, agency adjudicators face their own legitimacy crisis when they cannot aggregate and actively manage cases. Far from undermining legitimate decision making, group procedures can form an integral part of public regulation and the adjudicatory process itself.

i. Aggregation in Judicial and Administrative Proceedings

Civil and administrative proceedings begin with the premise that every person deserves her or his own “day in court.”38 Plaintiffs in civil courts receive personalized hearings to sort out private disputes with others. Agencies similarly must provide citizens with “some kind of hearing”39 to challenge government acts that threaten their lives, property, or liberty.40

Both systems, however, have exceptions—grouping together and resolving large groups of similar claims, or what we call “aggregation.”41 In some ways, a central tenet of all legal systems is to aggregate. Policymakers and judges create and interpret substantive rules to account for recurring problems and treat “like cases in a like manner.” It is the reason why common law judges must consider the precedential impact of their decisions on similar cases42 and why legislators create agencies with specific missions to create rules for, and adjudicate, particular kinds of cases.43 One theory posits that administrative agencies represent a public counterpart to class action lawsuits—another form of aggregation—because Congress delegates them authority to pursue ends that benefit broadly defined interest groups against those who violate the law.44

But federal courts also use procedural rules to group together large numbers of cases. These aggregation rules vary based on at least three factors: (1) the degree to which people actively participate in adjudication, (2) the preclusive effect of any decision, and (3) the number of decision makers responsible for the final outcome. We chart the features of formal aggregation, informal aggregation, and individual adjudication in Table 1 below and the discussion that follows.



The most famous kind of “formal aggregate” lawsuit is the class action—a single binding lawsuit that, in one representative proceeding, resolves claims or defenses held by many different people. Other kinds of formal aggregations include lawsuits by and against organizations in bankruptcy, trustee actions commenced on behalf of many beneficiaries,45 statistical sampling and extrapolation,46 and parens patriae actions by state attorneys general.47 What all formal aggregations have in common is that a single person, or a single proceeding, may bind others to an outcome, even if those others never directly participate.

But courts also group together civil claims in far more informal ways.48 Courts frequently “informally aggregate” cases—channeling individually represented parties into the same courthouse, before the same judge, or onto a specialized docket. In civil litigation, the most well-known form of informal aggregation is the multidistrict litigation,49 where a panel of judges may assign a large number of similar claims filed around the country to the same judge to streamline discovery, manage motion practice, coordinate counsel, and, in many cases, expedite settlement.50 Thus, even though a single case in a multidistrict litigation does not formally bind other parties like a class action, it may strongly influence the resolution of thousands of similar cases. Since its creation in 1968, the Judicial Panel on Multidistrict Litigation has centralized almost half a million civil actions for pretrial proceedings.51 Other forms of informal aggregation in civil lawsuits include specialized dockets—like those designed to expedite patent claims filed in the Eastern Districts of Virginia and Texas52—or district court rules designed to ensure that a single judge hears all “related claims” in the same district.53

The traditional version of adjudication, by contrast, contemplates something very different. It imagines that each party retains a separate attorney and commences a separate case before a randomly assigned adjudicator. The end result only binds the parties (even though the decision, if published, may establish precedent for similar cases). Of course, even this model of individual adjudication obscures what happens out of court.54 Plaintiffs and defendants frequently “privately aggregate” claims—settling large numbers in bunches completely outside of court.55

A. The Costs and Benefits of Aggregate Adjudication in Court

Aggregate procedures in federal court seek to provide more access, efficiency, and consistency than individualized litigation. Legal access is promoted through aggregate litigation in federal and state courts by enabling the resolution of claims that otherwise would not be brought individually. Formal aggregate procedures enable litigation when damages are too small for individuals to justify the high costs of retaining counsel.56 Informal aggregation streamlines large-scale litigation and encourages parties to participate through bellwether trials, steering committees of plaintiffs that collect and manage claimant input, and judicial oversight of attorney conduct. In both cases, aggregation holds defendants accountable for wide and diffuse harms that are too costly to be litigated through individual adjudication.57

Aggregate procedures also seek more efficient resolutions than piecemeal individual adjudication. Aggregation hopes to avoid the duplicative expenditure of time and money associated with traditional case-by-case adjudication,58 which may entail months or years of the “same witnesses, exhibits and issues from trial to trial.”59

Finally, aggregate procedures seek more uniform application of law. Aggregate proceedings and settlements seek consistency and distributive fairness—to treat like parties in a like manner.60 Otherwise, in cases seeking injunctions or declaratory relief, a court may never hear from plaintiffs with competing interests in the final outcome, or may subject defendants to impossibly conflicting demands over time.61 In addition, in cases seeking monetary relief, the first claimants to bring lawsuits might receive astronomical awards, while other victims receive nothing.

But large cases also create new risks. Class actions require judicial review, for example, to ensure class counsel’s faithful representation of absent class members, to provide a forum to hear from dissenting interest groups, and to ensure that the final settlement adequately reflects the underlying merits and the public interest. Thus, even as they aspire to promote legal access, efficiency, and consistency, class action lawsuits struggle to (1) ensure legitimacy when clients lack input and control over the outcome and when attorneys serve disparate interests (or their own); (2) promote efficiency when processing large volumes of cases; and (3) achieve accuracy when group-wide outcomes or settlements blur characteristics or overlook the merits of many different kinds of cases.

Informally aggregated cases may also complicate legitimacy and accuracy. First, lawyers must overcome conflicts of interest when they settle individual cases in informal aggregations, particularly because the success of any one case often depends on the same lawyer or judge resolving hundreds of similar claims.62 Informally aggregated civil cases may also compromise individual parties’ control over the outcome, as a small number of lawyers, special masters, or magistrates make decisions about common questions of discovery, motion practice, or other “common benefit work.” According to the American Law Institute’s Principles of Aggregate Litigation, informal aggregations afford participants some important powers, but deny them others: “In important respects” parties go forward “at the mercy of others . . . . [T]hey must accept services from and pay fees to lawyers and other persons they have little power to control.”63

Second, informal aggregation can compromise accuracy—particularly when the same plaintiff and defense counsel settle large groups of cases in bulk. This is sometimes a result of perverse incentives created by the ways parties must organize themselves to process large volumes of claims. For example, plaintiffs and defendants have complained that multidistrict litigation favors volume over knowledge: attorneys often receive coveted and lucrative positions on steering committees based on the sheer number of clients they retain in the litigation.64 Those incentives may, in turn, delay and discourage lawyers from investing limited resources to develop the facts of individual cases before reaching a global settlement.65

In other words, like many kinds of bureaucratic systems, formal and informal aggregate litigation struggles to govern many different kinds of constituencies feasibly, legitimately, and accurately. As set forth below, agencies also enjoy power to formally and informally aggregate claims.

B. The Power of Agencies To Aggregate Cases and Claims

Since the dawn of the republic, Congress has created special administrative courts whose adjudicators do not enjoy the life tenure and salary protections provided to federal judges by Article III of the Constitution.66 When Congress vests adjudicatory power in such non-Article III courts, it usually employs one of its enumerated powers in Article I, in combination with the Necessary and Proper Clause.67 Such non-Article III courts include both administrative agencies that adjudicate cases and what are sometimes called “legislative courts.”68

Most agencies and legislative courts enjoy broad authority to craft hearing procedures to help them carry out their statutory missions.69 The Supreme Court has reasoned that agencies “should be free to fashion their own rules of procedure and to pursue methods of inquiry capable of permitting them to discharge their multitudinous duties.”70 Therefore, agencies can consolidate cases and decide “subordinate questions of procedure,” such as “the scope of the inquiry, whether [cases] should be heard contemporaneously or successively, whether parties should be allowed to intervene in one another’s proceedings, and similar questions.”71

Moreover, agencies may exercise their discretion over procedures by promulgating general rules or by tailoring ad hoc rules to specific cases as needed.72 Indeed, prohibiting aggregation mechanisms under the APA would be at odds with the substantial flexibility the Supreme Court has granted agencies when choosing the best procedural format for decisions that affect large groups of people.73

For this reason, in 2004, the Office of Legal Counsel (OLC) for the Department of Justice (DOJ) rejected a Postal Service challenge to EEOC’s class action rule and confirmed the EEOC’s broad authority to aggregate claims in its adjudicatory proceedings, even without an express statutory provision for aggregation.74 Observing that class actions were “procedural in nature,” the OLC concluded that the EEOC could properly adopt class action rules under its congressional directive to issue “such rules . . . as it deems necessary and appropriate to carry out its responsibilities.”75

Indeed, we are aware of only one non-Article III court that has said it expressly lacks authority to hear class actions under its general powers to craft rules of procedure: the Court of Appeals for Veterans Claims (CAVC) rejected class actions without more explicit authority to do so, even while recognizing the value of consolidating similar disability claims by veterans.76 Nevertheless, as noted in the Introduction, the CAVC’s position on class actions appears to be changing.77

Indeed, just as class actions fall “within the Supreme Court’s mandate to adopt rules of ‘practice and procedure’ for the district courts, . . . [t]here is no reason why [administrative agencies] cannot use the same device” in appropriate cases.78 Class actions are merely “procedural technique[s] for resolving the claims of many individuals at one time . . . , comparable to joinder of multiple parties and intervention.”79

In some ways, federal agencies enjoy more power to develop procedural rules than Article III courts. Under the Rules Enabling Act, Article III courts may only “prescribe general rules of practice and procedure” that do not “abridge, enlarge or modify any substantive right.”80 By contrast, administrative agencies generally have no such limitation: Congress creates most administrative agencies precisely because Congress wants them to make substantive law.81 Even legislative courts that most closely resemble the Article III courts generally are not subject to the same restrictions under the Rules Enabling Act.82

The procedural flexibility Congress generally vests administrative agencies with reflects a basic feature of administrative law: agencies must have the authority to shape their own rules and, when appropriate, to adapt those rules to the types of cases and claims that they hear. This means that absent an express statutory prohibition to the contrary, administrative agencies may use aggregate procedures to handle their cases more expeditiously, consistently, and fairly than would be possible with individual, case-by-case adjudication.

When neither an agency’s organic statute nor the APA requires or prohibits specific procedures, due process still limits the procedures that federal agencies may use. Nevertheless, while the particular process due varies from case to case,83 no court has suggested that due process limits what kinds of adjudicators may use class actions or other tools to aggregate cases. The aggregation of common issues by both courts and administrative agencies has long withstood due process challenges. The Supreme Court has held that due process permits the use of class actions that bind absent plaintiff class members so long as the absent class members receive adequate representation, sufficient notice, and an opportunity to either participate in the litigation or “opt out.”84 Moreover, the Supreme Court has long recognized that agencies may, consistent with due process, bind parties to common findings of law or fact without an individualized hearing.85

In fact, courts have approved the use of aggregation tools in the context of agency adjudications. For example, courts have consistently rejected claims that statistical sampling in the Medicare and Medicaid programs violates due process, explaining that if a sample is representative and statistically significant, the risk of error to a provider is fairly low and the private interest “at stake is easily outweighed by the government interest in minimizing administrative burdens.”86

ii. The Extent of Agency Aggregation

No one has surveyed the use of class actions or other aggregate procedures by administrative agencies. Part II bridges the gap by reviewing the extent to which agencies and other non-Article III courts aggregate cases in their adjudicatory proceedings. The virtual absence of aggregate practice from the administrative state makes the agencies that do aggregate all the more fascinating.

A. Identifying Agencies that Use Aggregation

In consultation with the Administrative Conference of the United States (ACUS), we identified more than forty administrative agencies and Article I courts with rules permitting some form of aggregation. We later supplemented this list using the Federal Administrative Adjudication database, a joint project of ACUS and Stanford Law School spearheaded by Professor Michael Asimow.87 The complete list of seventy-one agencies with some kind of aggregation rule is included in the Appendix.88 Some have adopted formal class action rules that resemble Rule 23 of the Federal Rules of Civil Procedure,89 while most merely permit the consolidation of cases or claims with common questions of law or fact. The number of agency rules that permit aggregation was surprising given how little attention has been devoted to aggregation by administrative agencies. Therefore, we sought to determine how often agencies made use of these aggregation rules. We did this in several ways.

First, we looked for references to each of these aggregation rules in electronic databases.90 We then reviewed these records to determine if they discussed the number of class members, cases, or claims aggregated in a class or consolidated proceeding. We also recorded cases involving more than forty class members, cases, or claims.91

Second, to support our results, we contacted staff at the administrative agencies with some kind of aggregate procedure identified by ACUS.92 We asked each agency to identify, among other things, (1) how often it used aggregation; and (2) whether it had ever aggregated more than forty cases or claims. We followed up on our email surveys by telephone when the agency’s response seemed promising—i.e., that they might make robust use of aggregation.

Third, we conducted extensive interviews with administrative judges, special masters, agency personnel, and staff at the EEOC, OMHA, and the Office of Special Masters (OSM), which hears claims from NVICP. Following these interviews, we presented our findings and solicited additional feedback from representatives of the EEOC, Medicare, and other agency officials at two open roundtables conducted by ACUS in February and March 2016.

B. Most Agencies Do Not Aggregate Claims

In general, we found that very few agencies use formal class action or other complex litigation procedures. Although there were seventy-one agencies with at least one class action, consolidation, or other aggregation rule, we found that in most cases these procedures were invoked very infrequently, if at all. Our findings are presented in Table 2 and the discussion that follows.



In sum, we identified two “Article I courts”—the Bankruptcy Court and the Court of Federal Claims—and seven agencies—the Board of Governors of the Federal Reserve, the Consumer Financial Protection Bureau, the Consumer Product Safety Commission, the Corporation for National and Community Service, the EEOC, the Government Accountability Office Personnel Appeals Board, and the Merit System Protection Bureau—that permit class actions. But five of the agencies did not have any reported decisions involving the rule’s use.93 Only the Bankruptcy Courts, the Court of Federal Claims, and the EEOC made frequent use of their class action rules.94 The EEOC class action rule yielded over 700 reported administrative decisions involving the rule, more than any other agency.

We found far more rules permitting consolidation of cases or claims in non-Article III tribunals. Sixty-nine agencies and Article I courts have a rule permitting consolidation or joinder. But forty-one of them did not have a reported administrative decision involving the rule. Many other agencies only referenced the rule infrequently—that is, fifteen or fewer times since they began including their decisions in electronic databases. Only eleven agencies and Article I courts had more than fifteen administrative decisions involving a consolidation rule.95 More importantly, most efforts to consolidate involved a very small number of cases—generally far fewer than the forty cases required to certify a class action or to justify multidistrict litigation in federal court. We identified only nine agencies or Article I courts that considered aggregating more than forty cases or claims in a single proceeding through consolidation, joinder, class action, or another form of aggregation.96 Moreover, many motions to consolidate or certify a class were denied, dismissed on other grounds, or reversed on appeal.97

Finally, we found a rule permitting the Provider Reimbursement Review Board to conduct “group appeals.”98 The agency made frequent use of this aggregation mechanism to aggregate more than forty cases at a time.

We note that Table 2 may understate the ways that agencies use class actions and other complex procedures. Some agencies do not publish the results of their adjudications or include all of them on Westlaw. In addition, several of our interviewees noted that they could not share their decisions due to the sensitive nature of the information they contained about the parties.

Moreover, agencies and other tribunals may adjudicate cases even without a formal rule. For example, counsel at the Federal Energy Regulatory Commission reported that the Commission frequently aggregates cases involving the same wholesale energy practice without granting a formal motion to consolidate.99 Indeed, two of our case studies—the NVICP and OMHA—involve aggregation in the absence of a formal aggregation rule or as part of a pilot program.

Finally, as noted above, the CAVC previously rejected class actions, citing its lack of authority; however, it may be in the midst of changing its opinion.100

Nevertheless, our findings reveal how rarely agencies aggregate cases compared to federal courts. The small number of reported cases that we identified is stunning when compared to the federal court system, where the volume of cases associated with class actions and multidistrict litigation almost exceeds forty percent of the entire federal caseload.101

Moreover, we are aware of only three agencies that have considered and invited public comment on the use of aggregation in their administrative proceedings. First, the Federal Communications Commission (FCC) considered and then rejected a proposal by private attorneys to hear class actions in its own adjudications for alleged violations of the Federal Communications Act.102 Among other things, the FCC worried that the procedure would “needlessly divert” the resources of its lone ALJ to adjudicating extremely “fact-intensive” and “complex” cases, which can just as easily be filed in federal court.103 If the federal court needed the agency’s expertise to resolve particular issues in the case, it could refer them to the FCC by invoking the doctrine of “primary jurisdiction.”104

Second, the Commodity Futures Trading Commission (CFTC) similarly considered and rejected the use of class actions in its proceedings involving broker-dealer disputes.105 It likewise questioned whether its adjudicators could handle complex class action cases, as well as whether they needed do so, given that parties could always pursue class actions in federal court.106

Third, in November 2016, the U.S. Department of Education adopted a group process loosely modeled after Rule 23 of the Federal Rules of Civil Procedure.107 The process allows the Department to hear large numbers of student claims for debt relief when they attend schools that go bankrupt or commit fraud. Under the final rule, however, only designated Department officials may commence the class proceeding, and students lack any formal right to petition the agency to do so.108

Thus, it appears that many federal agencies have not even begun to devote serious attention to whether or how they might benefit from aggregation in their adjudicatory proceedings.

iii. Inside the Agency Class Action

The relative absence of aggregate practice when compared to courts makes the agencies that do aggregate worthy of study. Like federal courts, agencies have aggregated cases in formal and informal ways. We chart these “agency class actions” in Table 3 and the discussion that follows.



First, agencies such as the EEOC and OMHA,109 discussed more fully below, have formally aggregated large groups of plaintiffs’ claims through consolidations, statistical sampling, and even class actions. In addition, the NLRB recently consolidated dozens of enforcement actions against McDonald’s franchisees for the same alleged unfair labor practices.110

Second, agencies have used different forms of informal aggregation to streamline categories of individually represented claims and centralize them before the same adjudicator, into the same courthouse or docket, or into a coordinated settlement program. The Vaccine Court’s “Omnibus Proceedings,” described in more detail below, centralize similar cases involving the same scientific questions in front of the same special master. The Executive Office for Immigration Review—which hears all cases involving detained aliens, criminal aliens, and aliens seeking asylum—offers another example of this kind of informal aggregation in a courthouse. In the past year, it has created special “surge courts” to respond to over 2,000 Central American asylum cases pending in West Texas.111

Another form of informal aggregation includes coordinated settlement programs. For example, Medicare and the EPA occasionally offer “industry-wide” settlements112—whereby the agency brokers coordinated individual deals as part of a systemic response to an ongoing policy or problem. In one well-known case, the EPA in 2005 offered qualified animal feeding operations (AFOs)—over 2,500 agribusinesses that produce pork, dairy, turkey, and eggs across the country—a global settlement to resolve their liability under the Clean Water Act.113 Each individual AFO would enter into a separate, but otherwise identical, agreement with the EPA and agree to pay a civil fine (categorically based only on the size of the AFO) to fund a nationwide study on monitoring AFO emissions. In return, the EPA agreed not to sue the participating AFOs for past and ongoing violations while the study was undertaken.114

Although we cannot address all the uses of aggregation by federal agencies, the three case studies below illustrate a range of techniques that can be used to resolve large groups of cases in administrative programs, the challenges each has faced, and potential lessons for the future.

A. Class Actions in the Equal Employment Opportunity Commission

The EEOC is the nation’s leading government enforcer of federal civil rights laws prohibiting discrimination in employment based on race, color, sex, religion, national origin, age, disability, and genetic information, as well as reprisal for protected activity.115 The EEOC’s specific role and responsibilities depend on the nature of the employer involved.116 Federal employees must first file any civil rights complaint with the EEOC office of their federal employer. When the agency’s investigation is complete, the employee may either ask for a final decision from the agency or request a hearing before an EEOC AJ.117

More than 100 AJs work in EEOC regional offices to adjudicate disputes between federal employees and their federal employers.118 After conducting an evidentiary hearing on the record, the AJ issues a decision and may order appropriate relief. Once the AJ hands down a decision, the agency has forty days to issue a final order, which either accepts or rejects the decision of the AJ. If the agency does not accept the decision or disagrees with any part of the decision, the agency may file an appeal with the EEOC’s Office of Federal Operations. Similarly, an employee who is unhappy with an agency’s final order may appeal the order to the Office of Federal Operations.119

1. EEOC Class Actions in Administrative Proceedings

The EEOC’s regulations grant EEOC AJs the power to certify and hear class actions against federal employers in administrative proceedings.120 The EEOC’s use of class action procedures—loosely modeled after Rule 23 of the Federal Rules of Civil Procedure—makes the EEOC something of an outlier in our federal administrative state.121 Some agencies are specifically empowered to hear class actions in cases involving workplace disputes—like the Merit Systems Protections Board and the Personnel Appeals Board—where employees claim that a government employer’s “pattern and practice” violates their rights.122 A number of other agencies have promulgated rules permitting the certification of class actions in their administrative proceedings, but almost never use the power.123 In contrast, the EEOC has heard petitions for class actions for over three decades. Even in the four years following the Supreme Court’s decision in Walmart v. Dukes124which some argue severely limits class actions in federal courts—the EEOC has considered over 125 class action claims.125 The EEOC has kept up its practice of hearing class action claims even though, like the FCC and CFTC, federal employees may also pursue class action claims in federal court.126

Based on our review of EEOC class actions considered over the past five years, they most commonly involve workplace discrimination claims based on race, sex, disability, and age.127 Of those cases, many follow the same pattern as class actions in federal court. A majority of cases were dismissed or remanded as untimely filed or on the merits. At least twenty-two cases were settled. Of the twenty-six actions that adjudicators considered certifying as class actions, they rejected fifteen and certified eleven.128

The design of the EEOC class action process appears to promote collaborative reform. Following an EEOC AJ’s decision on the merits, the federal employer is given time to “accept, reject, or modify” the AJ’s recommendations and final report.129 The employee then decides whether to appeal the final agency decision to the EEOC’s Office of Federal Operations.

1. EEOC Class Action Procedures: Similarities to and Differences from Federal Rules

EEOC class action procedures mostly track Rule 23(a) of the Federal Rules of Civil Procedure. Like federal courts, EEOC AJs hear class actions based on a petition, typically filed by lawyers from a highly specialized bar, demonstrating (1) that the proposed class is so numerous that a consolidated complaint of the members of the class is impractical; (2) that there are questions of fact common to the class; (3) that the claims of the agent of the proposed class are typical of the claims of the class; and (4) that the class or representative will fairly and adequately protect the interests of the class.130 Before certifying a class, AJs ensure that a class action is feasible and likely to resolve the claims more efficiently than individual adjudications.131 Thus, like their judicial counterparts, EEOC AJs may require class-wide discovery, appoint liaison counsel, or certify class actions on the condition that parties obtain more experienced counsel, hear complex statistical evidence involving company-wide practices, and sometimes certify sub-classes to ensure parties with distinct interests are adequately represented at trial, or more commonly at settlement.132 In addition, EEOC AJs conduct evidentiary hearings to screen out unreliable expert testimony.133

This means that EEOC class actions, like their federal court counterparts, can also be time consuming. They may take years of motion practice, class discovery, appeals, and fairness hearings to determine the reasonableness of settlements.

But EEOC class actions have no equivalent to Rule 23(b) of the Federal Rules of Civil Procedure. That has at least two important consequences. First, unlike federal damage class actions, federal employees cannot “opt out” of an EEOC class action.134 After the EEOC certifies a class and renders a class-wide decision, employees only retain an individual right to challenge damages in mini-trials required by federal regulations.135 Accordingly, EEOC AJs reported making extra efforts to ensure that attorneys representing a class with absent class members have sufficient experience, resources, and skill to adequately represent large groups of similar claims.136

Second, unlike Rule 23(b)(3) class actions, EEOC class actions do not require that common questions “predominate” over individual issues before certifying a class action. This “predominance” requirement is often a difficult hurdle in federal court.137 Among other things, federal courts have rejected class actions that raise too many questions of law, vexing causation questions, and in rare cases, highly individualized damages due to fear that individual issues among class members will overwhelm common ones.138 If common questions do not predominate, aggregation may not be more efficient than case-by-case adjudication.139

Nevertheless, EEOC class actions typically resemble federal class actions under Rule 23(b)(2), which permit class actions for declaratory or injunctive relief where “the party opposing the class has acted or refused to act on grounds that apply generally to the class.”140 EEOC cases involving structural reforms or declaratory relief tend to be less controversial than those seeking money damages because an injunction usually impacts all class members in the same way.141 These class actions permit the EEOC to consistently apply decisions to groups of claimants working for the same employer.142

Finally, EEOC administrative proceedings are not bound by recent Supreme Court decisions limiting the use of employment class actions in federal court. Thus, EEOC class actions remain central to implementing anti-discrimination policy in the federal government.143

B. Multiparty Consolidation in the National Vaccine Injury Compensation Program

Congress created the National Vaccine Injury Compensation Program (NVICP) in 1986 to provide people injured by vaccines with a “no-fault” alternative to lawsuits in federal court.144 Under the program, after filing claims for compensation with the Office of Special Masters (OSM),145 a petitioner is entitled to a decision within 240 days based on a showing that the vaccine caused the injury.146By mandating that people first file their vaccine injury claims with the NVICP, Congress hoped to reduce lawsuits against physicians and manufacturers, while providing those claiming vaccine injuries an expedited claims process and a reduced burden of proof. Petitioners under the NVICP, unlike those who sue, do not have to prove negligence, failure to warn, or other tort causes of action; they must only prove that a covered vaccine caused their injury.147 A seventy-five cent excise tax for each dose of vaccine sold goes to a trust, which, in turn, funds awards and the administrative costs of the Program.148

Generally, a petitioner can get compensation under the NVICP in two ways. In a “table” case, the petitioner has an initial burden to prove an injury listed in the Vaccine Injury Table.149 Upon satisfying this initial burden, the petitioner earns a “presumption” that the vaccine caused his or her injury. The burden then shifts to the Department of Health and Human Services (HHS) to prove that a factor unrelated to the vaccination actually caused the illness, disability, injury, or condition.150Petitioners can also get compensation for “off-table” cases. A petitioner in an off-table case has the burden to prove the vaccination in question “caused” a particular illness, disability, injury, or condition.151The NVICP table originally covered vaccines against seven diseases: diphtheria, tetanus, pertussis, measles, mumps, rubella (German measles), and polio.152 Coverage has since been extended to a total of seventeen vaccines.153

OSM adjudicators possess an interesting mix of powers—falling somewhere between Article I judges and agency adjudicators. On the one hand, Congress expressly considered—and then rejected—allowing HHS to create regional panels to hear claims arising out of NVICP.154 Moreover, the OSM sits in the U.S. Court of Federal Claims, to which parties may appeal OSM’s decisions.155 On the other hand, the OSM must follow special procedures created specifically for the vaccine program, rather than the Federal Rules of Civil Procedure. Similar to other agency adjudicators, OSM’s medical and scientific findings are subject to substantial weight and deference, and may only be set aside on appeal if found to be “arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law.”156

As in most benefit programs, many claims under the Vaccine Act proceed one at a time. However, sometimes, this small office of adjudicators has had little alternative but to find ways to streamline the disposition of large groups of cases—particularly those raising similar scientific questions. OSM relies on its inherent authority to exercise “specialized knowledge” and aims to resolve common scientific questions in a consistent and informed way. To meet this end, it has used procedures that loosely resemble multidistrict litigation, bellwether hearing procedures, and creative case management, in ways designed to simultaneously increase public participation and input.

Generally, NVICP uses two types of omnibus proceedings to handle claims raising similar scientific questions. The first type resembles an issue class action in federal court: a special master hears evidence on a general theory of causation, makes findings based on that evidence, and orders the parties to demonstrate consistency with the general findings. The second type proceeds like a bellwether trial: the adjudicator decides one or a few cases first with the goal of enabling other claimants to better understand the strengths and weaknesses of their claims.

The “issue class action” variant dates back to 1992, when Special Master George Hastings decided an omnibus proceeding involving 130 cases alleging that a rubella vaccine caused chronic arthritis and related problems.157 In that case, Special Master Hastings observed early on that a large number of similar claims presented the general question whether rubella could cause chronic arthropathy.158 The Special Master thus conducted an inquiry into this “general” question for the benefit of each of the related cases “with the hope that knowledge and conclusions concerning the general causation issue . . . could be applied to each individual case.”159

At the time, there was “only a very, very limited amount of data directly applicable” because “this issue really ha[d] not been scientifically studied.”160 Therefore, he sua sponte encouraged plaintiffs’ attorneys with such claims to form a steering committee to develop general causation evidence and coordinate its presentation.161 At the general causation hearing, Special Master Hastings then evaluated a range of evidence that applied to this “general causation” question—including several isolated cases of chronic arthritis following the rubella vaccination, a study that discussed several cases of chronic joint pain, certain evidence of pathological markers, and formal expert testimony. At the end of the hearing, Special Master Hastings conceded that the evidence, while “not overwhelming,” generally supported a causal link between the rubella vaccine and chronic arthritis.162 He then entered a case management order requiring individual parties to put forward evidence consistent with his findings—acute onset of arthritis, no history of pre-existing conditions, as well as other evidence—to qualify for compensation.163

The general proceeding helped expedite the evaluation of a common but still-evolving scientific question of general causation. In addition, the proceeding made otherwise “small dollar” claims for joint pain worthwhile. The Federal Claims Court later favorably recounted that “[f]ollowing the 1993 Decision, over 130 related cases were either resolved or voluntarily dismissed based upon the Special Master’s findings.”164

Moreover, by forcing the parties to pool together common scientific evidence on the issue, OSM created awareness about an issue that, up to that time, had escaped the attention of HHS as well as Congress. Shortly after the decision, the Vaccine Injury Table was administratively modified, consistent with Special Master Hastings’ decision, to include “chronic arthritis” as a table injury associated with the rubella vaccine.165 As a condition of establishing a table injury for chronic arthritis, a petitioner must demonstrate that a physician observed actual arthritis (joint swelling) in both the acute and chronic stages.166

The second, and more common, type of omnibus proceeding proceeds like a bellwether trial in federal district court. Cases raising similar issues are organized in front of the same or a small number of adjudicators. One or a small number of test cases are then adjudicated first, in expectation that the outcome in the bellwether cases will help similarly situated parties understand the strengths and weaknesses of their claims, facilitating the settlement of the remaining cases:

[B]y the agreement of the parties, the evidence adduced in the omnibus proceeding is applied to other cases, along with any additional evidence adduced in those particular cases. The parties are . . . not bound by the results in the test case, only agreeing that the expert opinions and evidence forming the basis for those opinions could be considered in additional cases presenting the same theory of causation.167

Special masters adopted this approach in the “Omnibus Autism Proceeding” (OAP), established to determine the existence of a causal link between childhood vaccines and autism. Between OAP’s adoption of the approach in July 2002 and August 2010, over 5,600 cases alleging an association between autism and the measles-mumps-rubella (MMR) vaccine—a non-thimerosal-containing vaccine, thimerosal-containing vaccines, or both were filed with the NVICP.168 Three special masters structured discovery, motion practice, and expert testimony to hear three separate “test cases” on this theory of general causation.

The special masters in each case considered a wealth of scientific evidence. As Chief Special Master Vowell observed:

The evidentiary record in this case . . . encompasses, inter alia, nearly four weeks of testimony, including that offered in the Cedillo and Hazlehurst cases; over 900 medical and scientific journal articles; 50 expert reports (including several reports of witnesses who did not testify); supplemental expert reports filed by both parties post-hearing, the testimony of fact witnesses on behalf of [the injured child], and [the child’s] medical records.169

Although non-binding, the findings in those three cases—which found no causal connection between vaccines and autism—helped the remaining claimants evaluate the strength and merits of their claims in the vaccine program.

C. Consolidation, Statistical Sampling, and Group Settlement in the Office of Medicare Hearings and Appeals

The Office of Medicare Hearings and Appeals (OMHA) operates in HHS and hears appeals involving Medicare benefits.170 OMHA was created by the Medicare Prescription Drug, Improvement, and Modernization Act of 2003 (the Medicare Modernization Act)171 to address concerns that Social Security Administration (SSA) ALJs lacked guidance to handle the distinct issues raised in Medicare appeals.172 Most of the appeals OMHA hears arise from denials of claims for reimbursement by hospitals, skilled nursing facilities, or home health care and hospice care providers, under Medicare Part A, or for doctors’ services and other medical services, equipment, or supplies that are not covered by hospital insurance, under Medicare Part B.173

In 2012, OMHA began to experience significant backlogs in appeals. The number of appeals received by OMHA grew from 59,600 in 2011 to 117,068 in 2012, 384,151 in 2013, and 473,563 in 2014.174 Put differently, the number of claims increased 700% from 2011 to 2014. Meanwhile, the number of appeals decided by OMHA only grew from 53,868 in 2011 to 61,528 in 2012, 79,377 in 2013, and 87,270 in 2014.175 Thus, even as OMHA’s ALJs increased their own productivity by 64.8% within the four-year period,176 OMHA simply could not keep pace with the huge number of new cases coming in the door. Average processing times for completed appeals grew from 121 days in 2011 to 661 days in 2015.177

The dramatic surge was caused primarily by stepped-up efforts to recover excess billing under several post-payment audit programs conducted by private contractors and “more active Medicaid State Agencies.”178 In addition, there was a “[l]arger beneficiary population” during this period.179 It is important to note, however, that appeals by individual beneficiaries receive priority processing.180 Thus, most of the parties suffering from the delays caused by the backlogs were service providers or medical suppliers—with sometimes hundreds or thousands of similar appeals on behalf of different Medicare beneficiaries.181

Facing an existential crisis, OMHA began to explore ways to reduce the backlog and process a much larger number of appeals without adding more ALJs.

1. OMHA’s Statistical Sampling Initiative

Like many other agency heads, the Secretary of HHS has broad discretion to establish “specific regulations to govern the appeals process.”182 Over the years, HHS has used that power to adopt rules authorizing OMHA ALJs to consolidate cases at the request of the appellant or “on his or her own motion,” “if one or more of the issues to be considered at the hearing are the same issues that are involved in another request for hearing or hearings pending before the same ALJ.”183 The purpose, as described in the regulations, is “administrative efficiency.”184 After the hearing, the ALJ may issue “either a consolidated decision and record or a separate decision and record on each claim.”185

Although OMHA ALJs rarely formally consolidate appeals, they often informally combine appeals in what they refer to as “big box cases,” so named for the hundreds or thousands of appeals from the same organization, often involving overlapping facts and legal issues, arriving together literally in a large box. Such informally coordinated groups of cases might involve (1) the same appellant with distinct but related factual claims;186 (2) the same appellant with a large number of claims involving common questions of law or fact;187 or (3) the same legal representative appearing on behalf of multiple appellants raising common legal or factual questions.188

In addition, the Medicare program has used statistical sampling since 1972 to estimate Medicare overpayments in light of the enormous administrative burden of auditing businesses on an individual claim-by-claim basis.189 In Chaves County Home Health Service, Inc. v. Sullivan,190 the D.C. Circuit approved the use of statistical sampling to determine Medicare overpayments. Even though the court acknowledged that the Medicare Act did not expressly authorize its use, the D.C. Circuit deferred to the Medicare program’s adoption of statistical sampling as “a judicially approved procedure that can be reconciled with existing . . . requirements.”191 In so doing, the court also pointed to longstanding uses of statistical sampling in other contexts.192

The use of statistical sampling and other aggregation techniques in Medicare appeals—as opposed to post-payment review in the above example—emerged organically in the late 1990s.193 SSA ALJs began using them to manage Medicare disputes that involved large numbers of similar claims before the same adjudicator. Both ALJs and the parties themselves proposed the use of statistical sampling to expedite such claims. Statistical sampling benefited providers who did not want to spend the time necessary to produce documentation for every claim for which they sought reimbursement.194 As a matter of policy, OMHA often required parties’ consent before performing statistical sampling.195

As the number of Medicare Part A and Part B appeals spiked, OMHA formally adopted the Statistical Sampling Initiative (SSI) as a way to formalize and systematize the process that had begun with individual ALJs. OMHA proceeded cautiously in designing the pilot program, concerned that its backlog-elimination efforts might create new backlogs, particularly given limited staff and large caseloads.196 OMHA also had to address concerns of DOJ and CMS about allowing companies with a history of fraud or wrongdoing to participate in the pilot program.

OMHA attorneys, ALJs, and statisticians developed criteria for piloting the new program on a limited basis. The pilot program was restricted to appellants with at least 250 claims on appeal currently assigned to an ALJ or filed within a three-month period in 2013, but not yet scheduled for a hearing.197

In order to identify claims appropriate for statistical sampling, OMHA used its own database to identify large numbers of appeals from the same provider.198 Based on these “data runs,” OMHA made offers to eight providers to participate in the sampling program. Seven parties agreed to participate in the program and one party declined.199

Most of the participants in the pilot program were providers of medical supplies and equipment. Notably, a single diabetic supplies proceeding accounted for 17,134 claims, dwarfing the other statistical trials, which only resolved caseloads of four hundred to six hundred cases at a time. Our interviewees suggested that these cases lend themselves to sampling because the claims involved are more similar to each other than are inpatient provider care claims, which exhibit greater individual variations.

Nine ALJs volunteered to adjudicate cases in the pilot program.200 They agreed to participate while continuing to maintain their regular workload to avoid interference with the existing appeals process. Creating a specialized pool of ALJs allows OMHA to take advantage of their expertise in handling such matters. One of the ALJs in the pool is randomly selected for each statistical sampling case.201 OMHA is guided by CMS policies on statistical sampling.202 In short, a statistician selects the sample from the universe of claims, the ALJ makes decisions based on the sample units, and the statistician then extrapolates the results to the universe of claims.

Although appellants can request statistical sampling on their own, none has done so to date. 203 Some appellants have worried that aggregate proceedings in front of the wrong adjudicator or with the wrong methodology could jeopardize their day in court.204 Others have worried that there was not enough information about the statistical sampling methodology that would be used in the SSI.205 OMHA has attempted to address these concerns and plans to do more on this front as it expands the program.206 Specifically, OMHA is weighing additional outreach efforts, increased staffing levels, and restructuring the adjudication process to make the program more appealing to medical providers who are otherwise unfamiliar with the use of sampling.

2. OMHA’s Settlement Conference Facilitation Initiative

In addition to statistical sampling, in June 2014, OMHA also piloted an aggregate settlement program—the Settlement Conference Facilitation (SCF) Initiative.207 Once again mindful of avoiding the creation of new backlogs, the pilot program was limited to groups of at least twenty appeals, or groups of less than twenty appeals comprising at least $10,000 in the aggregate, filed in 2013 for the “same” or sufficiently “similar” items or services.208 OMHA takes a common sense approach to the meaning of “same” or “similar.” For example, all wheelchairs—whether electronic or manual—or all nutritional supplies for people with digestive troubles—including both the nutritional supplements and the device to deliver them—would be the “same” or “similar” items. But wheelchairs and diabetes test strips—even if stemming from the same illness—are not related and would not be the “same” or “similar.”209

Under the pilot program, OMHA facilitates a discussion between CMS and the appellant regarding settlement. OMHA devotes one attorney trained in facilitation working full time, along with four other trained facilitators working on a rotating basis. If the parties reach an agreement, OMHA dismisses the appeal. If no agreement is reached, the appeals return to their prior positions in the appeals queue.

The initial phase of the program resolved 2,400 appeals,210 which equals the number of cases typically resolved by two ALJ teams working for one year.211 Most of the settlements resolved around two hundred appeals. A few resolved five hundred to seven hundred appeals. Phase I of the SCF Pilot was staffed by the attorney trained in facilitation, a program analyst, a management assistant, and five facilitators.212

iv. Benefits and Challenges of Aggregate Agency Adjudication

Our case studies illustrate how aggregate agency adjudication can yield many of the same benefits as aggregation in federal court discussed in Section I.A. Like federal courts, each tribunal has used aggregate adjudication to pool information about common and recurring problems, as well as to eliminate the duplicative expenditure of time and money associated with traditional one-on-one adjudication.213 They have also sought more consistent outcomes in similar cases than is possible with case-by-case adjudications. Finally, aggregation has proved to be an important method to improve access to legal and expert assistance by parties with limited resources, allowing individuals to pursue claims that would otherwise be difficult on an individual basis.214

But aggregate agency adjudication also raises some of the same challenges and costs of aggregation in federal court discussed in Section I.A: (1) creating potential diseconomies of scale—inviting even more claims that stretch agencies’ capacity to administer justice; (2) undermining the perceived legitimacy of the process and challenging due process; and (3) increasing the consequence of error. In other words, like many administrative systems, aggregate adjudication struggles to deal with many different constituencies feasibly, legitimately, and accurately.

Each program has responded to these concerns in various ways. They have cautiously piloted aggregate procedures to avoid replacing old backlogs with new ones. Where appropriate, they have relied on panels of adjudicators to reduce allegations of bias and have provided additional opportunities to assure individuals’ voluntary participation in the process. Finally, some have developed guidance to standardize the use of statistical evidence, while others require cases raising novel factual or scientific questions to mature before centralizing claims before a single decision maker. This Part summarizes the benefits of aggregate agency adjudication and the ways that these agencies have attempted to respond to their challenges.

A. Promoting Efficiency While Avoiding Diseconomies of Scale

The efficiencies afforded by aggregation can be especially helpful in the administration and review of large benefit programs, such as those reviewed by the NVICP and OMHA,215 where appellants continually file cases involving similar legal and factual issues.

When over five thousand parents claimed that a vaccine additive called thimerosal caused autism in children, the NVICP used a national Omnibus Autism Proceeding (OAP) to pool all the individual claims that raised the same highly contested scientific questions.216 In the words of one special master, omnibus proceedings were “a highly successful procedural device,” facilitating settlement of individual cases and allowing those cases that proceed to a hearing to be resolved “far more efficiently than if we had needed a full-blown trial, with multiple expert witnesses, in each case.”217 Similarly, although OMHA pilot programs are in their early stages and any conclusions must be tentative, they have already resolved thousands of similar cases, sometimes involving the same issue for the same beneficiary with only a different service date. Indeed, OMHA’s programs have been so successful that medical providers are urging OMHA to expand opportunities to aggregate and settle large numbers of claims.218

Nevertheless, as the agencies in our case studies adopted aggregate procedures, they confronted longstanding concerns with diseconomies of scale—creating more backlogs and inviting claims that were difficult to manage with limited staff and large caseloads. OMHA adjudicators and personnel acknowledged they hoped to avoid creating “a backlog to another backlog” when they developed a formal program to use statistical evidence to resolve large groups of common claims commenced by a single provider or supplier.219 AJs with the EEOC, all with decades of experience hearing class actions, observed that class action proceedings involved substantial time and resources, sometimes requiring extensive motion practice and complex statistical proofs to establish unlawful patterns of discrimination.220 Even more informal aggregation, like the NVICP’s Omnibus Proceedings, has required adjudicators to invest resources tracking and closing pending individual cases long after the court resolves common questions involving a particular vaccine.221

In each case, however, adjudicators have responded by using aggregate tools cautiously, through active case management; reliance on experienced counsel and special masters to avoid duplicative motions; and encouragement of settlement where appropriate. OMHA, for example, rolled out its pilot statistical sampling program for a very limited category of claims—those filed before 2013; actively identified appellants with large volumes of identical claims in its database; and proceeded on a voluntary basis with the consent of the parties.222 As OMHA expands its mediation program, it has kept in mind and actively dealt with the risk of uncommon, unclear, and cherry-picked cases. First, OMHA only invites appellants with appeals appropriate for the SCF program based on the claims’ similarity. Second, the claims appealed may not involve items or services billed under unlisted, unspecified, unclassified, or miscellaneous healthcare codes. These claims are difficult to settle because they do not have an approved reimbursement amount. Third, OMHA requires settlement discussions to be comprehensive. In other words, requests for mediation must include all of the party’s pending appeals for the same items or services that are eligible for SCF. If an appellant has fifty pending wheelchair appeals that meet the SCF requirements, the appellant must request SCF for all fifty wheelchair appeals. In addition, appellants may not request SCF for some but not all of the items or services included in a single appeal.223 This prevents parties from submitting their weakest appeals to the settlement process and going to hearings with their strongest appeals.

NVICP special masters and EEOC AJs rely on steering committees of experienced lawyers to organize and manage common discovery. The NVICP special masters sometimes allow evolving scientific and novel factual questions to “mature”—putting off centralizing novel cases involving a single vaccine until receiving the benefit of several opinions and conclusions from different special masters about how a case should be handled expeditiously.224 EEOC AJs exercise active judicial management to expedite cases for trial and, in many cases, settlement.225

Still, in some cases, an overly cautious approach can limit the full value of agency aggregation. For example, OMHA’s SSI is hindered in what it can achieve by both the limited pool of eligible claims and OMHA’s decision to require the parties’ affirmative consent to participate in the program.226 At this point, not enough parties have been willing to consent to statistical sampling for it to make a significant dent in the backlog. As long as it remains an entirely voluntary program, OMHA will need to build greater trust among appellants to realize the program’s full potential as an aggregation mechanism.

Thus, agencies considering aggregation must evaluate whether they hear a sufficiently large number of similar cases to warrant the potential costs. To do so, agencies need a good handle on the types of claims they receive and are likely to receive in the future. Detailed case management and tracking systems are particularly helpful in this regard, as demonstrated by OMHA’s use of its own database to identify candidates for its mediation program. In addition, agencies adopting aggregation need to utilize experienced adjudicators and leverage a pool of skilled counsel to manage complex cases. Finally, when the costs and benefits of aggregation are uncertain, agencies can pilot programs before scaling up.

B. Promoting Consistency While Ensuring the Accuracy of Agency Decisions

Aggregate procedures can also provide uniform and consistent application of the law,227 particularly in cases seeking indivisible remedies such as injunctive or declaratory relief. Absent a class action, a tribunal may never hear from plaintiffs with competing interests in the final outcome. The EEOC, for example, has long claimed its class action procedure was important to consistently resolve “policy or practice” claims of discrimination by federal employees.228The EEOC deems the process important in light of the volume of claims it processes each year and the potential for inconsistent judgments.229Agencies also may use class procedures to avoid subjecting defendants to impossibly conflicting demands.230 OMHA adjudicators, for example, have observed that aggregate procedures have been vital to ensuring that hospitals and medical suppliers with hundreds of the same claims, sometimes for the same beneficiary, are reimbursed consistently.

Nevertheless, each case study also illustrates how aggregation puts pressure on the ability of adjudicators to achieve accurate decisions, especially when many cases are concentrated before the same judge. As noted, many appellants before OMHA worried about the accuracy of any final statistical extrapolation.231 EEOC AJs observed that unlike federal judges, who benefit from the Reference Manual of Scientific Evidence, no similar guidance exists for EEOC judges tasked with deciding statistical or other technical evidentiary questions frequently raised in EEOC proceedings.232 Special masters in the NVICP exist precisely because Congress assumed that over time they would develop expertise in the complex medical and scientific questions frequently raised in the program; and yet, in proceedings where groups allege new theories of general causation for large numbers of vaccines, decision makers warned of the importance of getting the science right in a single adjudication.

The agencies have responded to these concerns by requiring sufficient similarity between aggregated claims and developing guidelines and screens to address complex statistical evidence. OMHA, for example, relies on its database of billing codes to ensure that claims are sufficiently similar to warrant aggregation, and uses statistical experts along with detailed guidelines for statistical evidence.233 Special masters in the NVICP wait for cases to mature before grouping them, which limits the adverse impact of hasty decisions on other related claims; adjudicators also afford attorneys additional time to allow their experts to better develop and understand the relationship between a vaccine and a new disease.234 EEOC AJs, like the federal courts, still carefully screen complex evidentiary issues common to the class, relying on guidelines long-established in federal court under Daubert v. Merrell Dow Pharmaceuticals, Inc.235

Thus, agencies that award indivisible relief or large volumes of similar types of claims are particularly ripe candidates for aggregation. But agencies that more typically award damages tailored to widely varying circumstances of individual cases may have less need for aggregation. If choosing to aggregate, agencies should develop threshold rules and actuarial tools to identify common cases and develop the evidentiary record.

C. Promoting Legal Access, Generating Information, and Enhancing Legitimacy

Finally, our case studies illustrate how aggregate proceedings can foster legal access, while pooling information about policies and patterns that otherwise might escape detection in individualized trials.236 The EEOC, for example, observed that its “class actions . . . are an essential mechanism for attacking broad patterns of workplace discrimination and providing relief to victims of discriminatory policies or systemic practices.”237 The class action procedure enables counsel to pool information about employers’ policies and allows EEOC AJs to assess their lawfulness—to identify patterns that otherwise might escape detection in an individual proceeding. In some cases, the scale and visibility of an EEOC class action itself attracts the attention of government agencies and leads to workplace reforms. For example, after an EEOC class of disabled applicants challenged the State Department’s “world-wide” availability requirement for foreign-service workers—a policy that rejected candidates for promotion unless they could work without accommodation—the State Department was alerted to a systematic problem in its hiring practices.238

Similarly, the NVICP’s omnibus proceedings allow any party alleging a vaccine-related injury to benefit from the record developed in test cases and general causation hearings by the most qualified experts and experienced legal counsel.239 In one of the NVICP’s first omnibus proceedings, the parties pooled common scientific evidence on whether a rubella vaccine caused chronic arthritis. As a result, the proceeding raised the profile of an issue that, up to that time, had not been the focus of HHS or Congress.240 As noted above, shortly after the decision, the Vaccine Injury Table was administratively modified, consistent with the decision, to include chronic arthritis as an injury generally associated with the rubella vaccine.241

OMHA’s statistical sampling initiative—though still in an early stage—has the potential to make it easier for the Secretary of HHS to coordinate the work of OMHA with other parts of HHS. Currently, OMHA may approve a payment on appeal from a denial of payment by the Centers for Medicare and Medicaid Services (CMS), while the next day CMS may deny the same provider’s claim on behalf of the same beneficiary for the same medical supplies with only a different date of service. Indeed, even the Medicare Appeals Council, which issues the Secretary of HHS’s final decision in these appeals, does not bind OMHA and CMS beyond the appeals that it reviews. Aggregate adjudication provides agency heads with a thoughtful first crack at important questions of law and policy by the agency’s most experienced and expert adjudicators, with the benefit of a fully developed record and competent counsel.

OMHA’s initiatives may also increase the ability of the political branches to ensure agency accountability.242 Policymakers are rarely concerned with the outcomes of individual adjudications beyond the provision of constituent services by individual representatives.243 Aggregated cases, however, like Medicare’s recent billion-dollar settlement with over 1,900 hospitals,244 have generated significant interest in Congress. As these examples illustrate, aggregation procedures may offer agencies another way to efficiently and consistently expand access to agency tribunals, while improving the caliber of representation and information provided to them.

Nevertheless, adjudicators and staff highlighted concerns with legitimacy raised by aggregate proceedings. The model for administrative adjudication typically imagines individualized hearings where claimants enjoy their own day in court before a neutral decision maker. EEOC AJs, for example, noted that the inability of parties to opt out of class actions seeking damages was an additional source of “pressure” for adjudicators to make appropriate decisions and narrowly define the class.245 Some hospitals and medical suppliers reported that they resisted OMHA’s statistical sampling program out of a fear that a single adjudicator’s view about the medical necessity of a small sampling of claims would be extrapolated to thousands of others.246 Even omnibus proceedings raise interesting questions about the legitimacy of using an adjudication process to settle complex scientific questions. Some plaintiffs in the OAP were anxious about commencing cases together; members of the public health community “f[ound] it unsettling that the safety of vaccines must be put on trial before three ‘special masters’ in a vaccine court.”247

Each of these systems have responded to these concerns by diversifying decision-making bodies, assuring adequate representation, and increasing opportunities for individual participation and control in the aggregate proceeding. Special masters in the Vaccine Program, for example, relied on a panel of three adjudicators in the OAP to allay concerns about bias. Similarly, as OMHA expands its statistical sampling initiative, it will also consider permitting multiple adjudicators to hear different samples of claims.248 For example, instead of a single ALJ hearing a sample of 100 cases, ten ALJs might each hear ten cases from the sample.249 Spreading the sample among more than one randomly selected ALJ will help alleviate the concern that the entire universe of claims will be decided by an ALJ that the party hopes to either avoid or obtain.

Finally, the EEOC relies on many rules adopted from the Federal Rules of Civil Procedure to increase legitimacy and participation. Among other procedures, adjudicators screen class counsel to ensure they adequately represent class members; hold “fairness hearings” where class members can voice their concerns with any proposed resolution or settlement; and, in a departure from the federal rules, require mini-trials to test individual claims and defenses remaining in adjudications involving damages.250

Agencies adopting aggregation mechanisms must ensure that aggregate proceedings are transparent and legitimate. To do so, they should adopt rules to ensure diverse, independent decision making, police potential conflicts within groups, provide opportunities for parties to be heard or in some cases opt out from aggregate proceedings when they believe their interests would be served best by proceeding on their own. Such rules may do more than just improve fairness. They may enhance other forms of agency policymaking, such as rulemaking and enforcement. We take up these issues in more detail in the next Part.

v. The Forms and Limits of Agency Adjudication

Notwithstanding the challenges class actions and other complex procedures pose, the EEOC, NVICP, and OMHA demonstrate the potential of aggregation to improve agency adjudication in a variety of ways. But our study also yields broader lessons about policymaking, public enforcement, and adjudication itself. We discuss each in turn.

A. Aggregation Complements Rulemaking

First, our study demonstrates how group litigation techniques offer agencies another form of decision making. Many scholars argue that Congress or agency policymakers can resolve large groups of claims more competently and openly through prospective rulemaking. However, our case studies illustrate that rulemaking does not necessarily eliminate the common issues of law and fact that must be repeatedly resolved in case-by-case adjudication.

Scholars have long contended that Congress or administrative policymakers promote efficiency by adopting prospective rules in a legislative process.251 For example, the question of whether vaccines cause autism is a general fact about the world that could certainly have formed the basis for legislation. If vaccines are dangerous and should be regulated differently, or should or should not be mandatory, that is something Congress could address through legislation. In fact, Congress has created specialized courts and compensation programs to compensate employees suffering from radiation exposure, occupational disorders, and black lung disease.252

Even inside many agencies, policymakers can resolve common questions by other means, most prominently through rulemaking. Many, if not most, agencies make prospective rules to “resolve certain issues of general applicability.”253 Consider the SSA, which is “probably the largest adjudicative agency in the western world.”254 Applicants for Social Security disability benefits must establish not only that they are unable to do their previous work but also that they are unable, considering their age, education, and work experience, to engage in any other kind of available gainful employment.255 For many years, ALJs reviewing disputed disability claims relied on the testimony of vocational experts to determine the types and number of jobs in the national economy that could be performed by the claimant, which led to criticisms of “inconsistent treatment of similarly situated claimants.”256 Consequently, the Secretary of HHS created medical-vocational guidelines to reduce ALJ reliance upon the testimony of vocational experts.257 The Supreme Court upheld the guidelines against a legal challenge, explaining that “the agency may rely on its rulemaking authority to determine issues that do not require case-by-case consideration.”258

Rulemaking can thus uniformly and definitively resolve common issues of law or fact that arise consistently in adjudications, relieving adjudicators of the burden of repeatedly addressing the same issues in individual cases.259 However, our case studies illustrate that shifting the resolution of cases from Article III courts to administrative tribunals with rulemaking power will not necessarily eliminate the need to aggregate common issues of law and fact that repeatedly arise in case-by-case adjudication, nor the need for parties to harness expertise and skilled counsel to represent them in complex cases.

First, rulemaking has not proved to be an effective tool for resolving all common issues of law or fact in agency adjudications. The law generally disfavors retroactive rulemaking.260 Therefore, it is less effective for addressing administrative backlogs or high volumes of filed claims such as those faced by OMHA or the NVICP.

Second, just as legislation leaves gaps for agencies to fill with rules, rules leave gaps that agency adjudicators must fill. As the Supreme Court has observed, “problems may arise in a case which the administrative agency could not reasonably foresee, problems which must be solved despite the absence of a relevant rule.”261 To return to the example of the SSA’s medical-vocational guidelines, those guidelines do not address claimants with mental or psychiatric conditions.262 Similarly, the NVICP was confronted with claims that were not anticipated by the Vaccine Injury Table, but nevertheless had to be resolved. And the EEOC, which has no power to issue substantive regulations interpreting Title VII, is frequently confronted with new issues raising discrete civil rights claims by federal employees.263

Third, administrative program beneficiaries impacted most significantly by agency adjudications often have the least access to the rulemaking process.264 While rulemaking is often a “top-down” proceeding, initiated and managed by the agency’s political leaders and influenced by organized interests with significant resources,265 aggregation can provide a “bottom-up” remedy, in which the individuals most impacted by adjudications “play a role in crafting discrete, retrospective forms of relief.”266 Federal employees bring to light previously unnoticed civil rights violations, persons injured by vaccines provide evidence on whether a particular vaccine causes a particular type of injury, and medical service providers highlight common problems in reimbursement.

Some may worry that aggregate decisions give administrative judges policymaking power beyond their authority or capacity, or even allow agencies to make an end-run around costly rulemaking procedures.267 As we have argued elsewhere, however, these concerns overlook the fact that most agency heads have final say over the rules adopted in adjudicatory proceedings.268 Moreover, the Supreme Court has long given agencies substantial discretion to choose the best procedural format for decisions that affect large groups of people.269 To that end, at least one commentator recommends agencies use adjudication when rulemaking proves infeasible or impractical.270 Finally, the aggregate proceedings we reviewed often provided more procedural safeguards than informal rulemaking, which only requires notice and an opportunity to comment.271 There is no reason to fear the loss of the minimal procedural requirements of informal rulemaking so long as class members and other interested parties can be heard in aggregate proceedings.

In sum, an agency with rulemaking power may still find useful the tool of aggregation in certain circumstances. Even agency adjudicators may need flexibility, in the trenches, to aggregate “all the way down.”272 In particular, agencies may prefer aggregate adjudication to rulemaking when the relief sought (1) is retroactive, (2) responds to backlogs of already filed claims, and (3) involves discrete problems, and when parties’ concerns may not be easily heard or represented by sophisticated representatives or counsel.

B. Aggregation as an Enforcement Tool for Agencies

Second, our study found that agencies have generally avoided using aggregation to accomplish what the Supreme Court once called “the policy at the very core of the class action”—deterring misconduct by enabling claims where the damages in each individual case are too small to justify the costs of litigation.273

OMHA and the NVICP generally aggregated cases in order to resolve existing claims, not to enable more claims. To be sure, the NVICP’s omnibus proceedings and OMHA’s statistical sampling initiative make it easier to recover small dollar claims.274 Nevertheless, the NVICP developed its omnibus proceedings in response to an influx of arthritis and autism cases; OMHA developed its pilot programs to address an existential crisis created by its mounting caseload, and was careful to roll them out slowly in order to avoid creating new backlogs. Only the EEOC explicitly uses class action to encourage suits to enforce federal anti-discrimination policies.

More importantly, in all three case studies aggregation is being used to group claims against the federal government. Agencies also adjudicate disputes between private parties. But with the exception of the NLRB275 and the Department of Education,276 we have not seen agencies use aggregation in the adjudication of disputes between non-governmental parties. In fact, both the CFTC and the FCC explicitly rejected the use of agency class actions in such contexts.277

Agencies’ reluctance to use class actions in this way seems to reflect a broader concern about giving private parties control over how the law is enforced.278 Commentators both praise and criticize aggregate litigation, particularly class actions, for enabling private attorneys to sue for mass harm much like “private Attorney Generals [sic].”279 Several scholars have pointed out agencies’ concern that class actions or other forms of aggregate litigation will encourage private attorneys to bring cases that would upset an agency’s own carefully calibrated enforcement regime.280 As noted above, the FCC rejected a proposal to hear class actions in its own adjudications out of a fear that it would “needlessly divert” the resources of its lone ALJ to adjudicating extremely “fact-intensive” and “complex” cases.281 The CFTC similarly rejected the use of class actions in the adjudication of broker-dealer disputes due to fears of burdening its adjudicators.282 And even after the Department of Education adopted an opt-out class action proceeding to resolve common claims by thousands of student borrowers, it emphasized that only designated Department officials could commence the class proceeding. In so doing, the Department of Education hoped to avoid creating a “‘cottage industry’ of opportunistic attorneys attempting to capitalize on victimized students and unleash a torrent of frivolous lawsuits.”283 In some ways, the Department’s procedure resembles rules adopted in many European countries, which similarly give public agencies near-exclusive control over class actions in order to avoid recreating America’s infamous litigation culture.284

Agencies might not just fear that aggregation will spur more frivolous litigation; they may also worry about the good cases. First, some argue that zealous private attorneys general may bring legitimate cases that agencies would not pursue in the exercise of their prosecutorial discretion.285 An agency might dispute “the notion that all laws warrant enforcement to the letter in all instances,”286 particularly when doing so leads to unnecessarily harsh punishment or undermines important regulatory goals. Second, in some cases, private attorneys general may advance innovative legal theories that conflict with the way the agency interprets the law.287 Third, private attorneys may waste resources through duplicative enforcement.288 Finally, there is the “who the heck are you” critique289—the perceived illegitimacy of allowing a few plaintiffs, or their attorneys, to usurp the traditional role of public enforcers in their own proceedings. As Richard Nagareda once observed, “The question here is: if the function of the class action today is indeed to operate in parallel with public regulation, then can that function achieve fruition without supplanting the institutional boundaries on regulatory power?”290

But such fears may ignore important benefits of aggregation, while exaggerating their costs, particularly in the context of agency adjudication. First, enabling group litigation may help resolve cases more efficiently in the long run. As the Vaccine Court experience illustrated, aggregation can attract the support of skilled counsel and medical experts in early stages of the litigation, helping resolve both large and small claims.291 Second, aggregation may save cash-strapped government enforcers money by encouraging private parties to police misconduct when they bring claims on their own. Agencies have long supported class actions in federal court—a private complement to otherwise overburdened government actors unable to respond to fraudulent investment schemes,292 unconscionable consumer contracts,293 and predatory for-profit colleges.294 Ensuring aggregate adjudication in agencies may become even more important as judges threaten to shut down federal lawsuits under doctrines of Article III standing, primary jurisdiction, or other theories that do not apply to federal agencies.295

Moreover, in many cases, aggregate agency adjudication is less threatening to agency control over enforcement than private class actions in federal court. Agencies enjoy substantial authority to regulate private attorneys and control litigation costs. Unlike in federal court, the agency’s political appointees control an agency’s final interpretations of law.296 The head of an agency may interpret the law without regard to the decision below and may even overturn the ALJ’s findings of fact under certain circumstances.297 Thus, agencies may have more power to stop plaintiffs from advancing unwanted legal theories or duplicative cases in their own proceedings.

An agency’s control over how the law is interpreted cannot, of course, prevent plaintiffs from bringing what are in fact meritorious claims, even when agency enforcers judge them as unnecessary and possibly even counter-productive.298 However, this has not been a concern of the EEOC, which cedes significant control over enforcement to the federal employees who prosecute class actions before EEOC adjudicators. Moreover, both the NVICP and OMHA have been able to adjudicate even low-dollar claims against the federal government without jeopardizing the public treasury.

Finally, unlike courts, agencies can tweak procedural rules to better accomplish their specific policy objectives without running afoul of judicial constraints like the Rules Enabling Act. Thus, when deciding whether to aggregate, administrative judges could consider the consistency of aggregation with the agency’s enforcement priorities. The adjudicator could hear from the agency’s enforcement arm on the question—if the agency was not already a party to the proceeding—or the agency could craft rules requiring the enforcement office’s assent to any formal aggregation. If the adjudicator or enforcement officials determined that aggregation was not in the interest of the agency’s enforcement goals, the members of the proposed class would have to proceed on a case-by-case basis.

The Department of Education has proposed something like this for individuals seeking debt relief from student loans under federal law. The Secretary of Education, through a recently created Student Aid Enforcement Unit, will decide whether to initiate aggregated proceedings for groups of borrowers based on the existence of common facts or claims. The Secretary’s Notice of Proposed Rulemaking states that in addition to commonality, the Secretary will consider such enforcement concerns as “the promotion of compliance by the school or other title IV, [Higher Education Act] program participants.”299 The Department’s proposal demonstrates how agencies can craft their procedures to avoid ceding control over their enforcement priorities to private attorneys general.300

Scholars have long examined the rise of private attorneys general and the relationship between public and private enforcement in federal and state courts.301 Our study illustrates that the use of private attorneys general in agency adjudications bears further examination due to agencies’ distinct institutional capacities. Federal agencies have only begun to explore the forms and limits of aggregation in their adjudicatory proceedings.

C. Protecting the Legitimacy of Adjudication Through Active Management and Bargaining

Both of the preceding concerns may reflect an even more fundamental anxiety with aggregate agency adjudication: is it legitimate? Prominent legal theorists have long argued that adjudication cannot legitimately address sprawling, interconnected claims among large groups of people.302 Similarly, the Supreme Court has barred federal courts from aggregating mass tort claims because they “defy” judicial resolution—calling on Congress to establish a “nationwide administrative claims processing regime [which] would provide the most secure, fair, and efficient means of compensating victims.”303 And yet, our study of similar “administrative claims processing regimes” supports a different conclusion—that far from pushing the limits of adjudication, aggregate procedures form an integral part of the adjudication process.

Professor Lon Fuller long ago famously defined the “‘essence’ of adjudication” as the right of affected parties to participate in the proceeding by “presenting proofs and reasoned arguments” to the decision maker.304 He suggested that adjudication was not well suited for what he described as “polycentric” problems involving large groups of people, where any one decision could have countless, unforeseeable consequences on others.305 Such disputes, in his words, were like “a spider web,” in which “[a] pull on one strand will distribute tensions after a complicated pattern throughout the web as a whole.”306 For example, when the U.S. government regulated prices and wages during World War II,307 no adjudication could have “take[n] into account the complex repercussions that may [have] result[ed] from any change in prices or wages.”308 A rise in the price of aluminum could have ripple effects on the price of steel, plastic, wood, or other materials. Large, interdependent cases were better handled outside the courts, through private bargaining or by elected officials.

Scholars and judges have diverged from Fuller’s analysis by focusing on the many ways in which judges actively manage and oversee fluid forms of relief, like structural reform efforts.309 However, there is no denying that Fuller’s framework has deeply influenced the way we think about when courts, Congress, or private parties may legitimately resolve disputes.

The same thinking often animates criticism of class actions in federal court. In Amchem Products, Inc. v. Windsor,310 for example, the Supreme Court was troubled by the impact of a proposed settlement on parties who had not yet filed claims, had distinct interests, and did not have their own representatives. According to the Court, settlement payouts to people already suffering from asbestos-related injuries “tugs against the interest” of those exposed plaintiffs who had not yet manifested any injury.311 Accordingly, the Court found that the global compromise lacked “structural assurance of fair and adequate representation for the diverse groups and individuals affected.”312 One can almost hear the echo of the “pull” on Fuller’s spider web in the “tugs” of Justice Ginsburg’s opinion.

This framework for defining the appropriate roles of courts and legislative bodies also took hold inside administrative agencies.313 Notably, Fuller himself saved his most significant criticism for adjudication inside the administrative state. “[H]owever inappropriate” it may be, said Fuller, “[i]f we survey the whole field of adjudication and ask ourselves where the solution of polycentric problems by adjudication has most often been attempted, the answer is: in the field of administrative law.”314

Reflecting these concerns, the APA provides distinct sets of rules and procedures for “adjudication” and agencies’ broader policymaking powers using rulemaking and enforcement.315 Adjudicatory decisions are rendered after a hearing on the record conducted by neutral adjudicators insulated from agency policymakers, enforcement officers, and even the President. Policymakers often use rulemaking, by contrast, to address polycentric problems; they may hold meetings and solicit comments from the general public, informally talk to parties, officials, and others without notifying other interested parties, and form distinct views about policies and problems.

But this narrow definition of adjudication may perversely threaten the legitimacy of adjudicators who, in case-by-case adjudication, lack tools to resolve critical backlogs of similar claims consistently, efficiently, and accurately. Our study illustrates that even as cases move from the judiciary to administrative agencies, adjudicators may still need to engage in the kind of bargaining and active case management that Fuller viewed as inconsistent with adjudication. Without the ability to consolidate and aggregate cases, rely on steering committees, sub-class interest groups, and turn to statistical consultants, adjudicators cannot efficiently hear and consistently resolve large groups of cases within already aggregated systems. Far from being inconsistent with adjudication, tools that allow judges to actively organize and manage cases have proven to be an essential part of an adjudicative process that must rely on “presenting proofs and reasoned arguments.”316

Consider the NVICP’s no-fault alternative to federal court for vaccine injuries. Congress created the very administrative process that the Supreme Court in Amchem endorsed for resolving vexing scientific questions more efficiently, consistently, and openly than courts. But when confronted with a large influx of claims involving the same vaccine and the same injuries, the NVICP special masters turned to the very same tools used by courts in mass-tort cases.317 The special masters created an ad hoc system to pool claims before the same adjudicator and form steering committees of claimants’ counsel, who then coordinated to offer the best expert testimony they could in support of their clients’ claims. Thus, even after Congress consolidated vaccine cases before a specialized tribunal, the tribunal still could not avoid using aggregation to resolve its caseload.

OMHA is coming to the same realization in the context of Medicare appeals. It too is a specialized tribunal with unique expertise to resolve complicated medical disputes in its jurisdiction. Yet OMHA now faces an “existential crisis,” forcing it to turn to aggregation to handle a deluge of appeals regarding similar types of claims by the same parties. Bipartisan support currently exists in Congress to expand funding for OMHA’s statistical sampling program.318

Specialized administrative courts, including the Vaccine Court, have recently come under scrutiny for failing to deliver the promised expeditious and rationalized compensation decisions.319 Specialization, expertise, and informal procedures may not be enough for administrative agencies and other non-Article III courts to address these concerns. Advocates may underestimate the expertise of Article III judges and overestimate the expeditiousness and informality of agency procedures.320 Our study contributes to this debate by suggesting that just like Article III courts, when confronted with large numbers of similar cases, agencies may need to turn to aggregation to resolve similar claims consistently, rationally, and legitimately.


Moving cases to administrative agencies does not eliminate the risks inherent in individual adjudication of large groups of similar claims: long backlogs, inconsistent results, and obstacles to justice for those without access to legal and technical expertise. But agencies have shown that they can respond to such problems by using their existing authority to aggregate cases themselves—with proper attention to avoiding diseconomies of scale and ensuring the legitimacy and accuracy of their decisions.

More broadly, aggregate agency adjudication raises questions about the way we think about the nature of adjudication. Rather than building formal walls between policymaking and adjudication to make adjudication legitimate—which we have done in both class action law and within the administrative state—some judicial proceedings require integrating rulemaking and other managerial tools to ensure the legitimacy of adjudication itself. The central question raised by such cases turns not on any abstract concept of adjudication or policymaking, but instead, how to best adapt procedure to “fairly insure[] the protection of the interests” at stake.321

To view the Appendix, please see the PDF version of this Article.