II. The Merits of Greenlighting
As discussed in Part I, the enforcement process in combined-function agencies begins with an investigation conducted by agency staff members who suspect a private-sector target with wrongdoing. If the investigators conclude that the law and facts support the issuance of a complaint, they report their conclusion to agency prosecutors. In the agencies I studied, other than the NLRB, the prosecutors lack power to issue complaints on their own. Instead, they must request the agency heads to greenlight the case by authorizing the issuance of a complaint, initiating federal court action, or making a criminal referral to DOJ. In most cases, the staff and the target have agreed to a settlement before the greenlighting stage, but agency heads must approve the settlement.
Part II discusses the greenlighting function. It situates greenlighting as an accountability mechanism and as “internal administrative law” (Part A). It then discusses the arguments for and against the practice (Parts B and C). It then concludes that the benefits of greenlighting outweigh the costs (Part D).
A. Internal Administrative Law
This Article discusses charging decisions by federal combined-function administrative agencies and the mechanisms by which prosecutorial discretion in such agencies is checked. As a practical matter, no external Judicial, Legislative, or Executive Branch accountability mechanisms constrain agency enforcement discretion. Unlike most state and local criminal prosecutors, agency prosecutors and agency heads are not subject to checks through either periodic or recall elections.
The judiciary cannot review charging decisions because they are not “final order[s].” Moreover, charging decisions are typically unreviewable because they are committed to agency discretion. In any case, most administrative enforcement cases settle instead of being adjudicated to a final agency decision, leaving nothing for courts to review. For cases litigated to a final decision by the agency heads, relatively few are judicially reviewed since the review process is so slow and costly and various deference doctrines make reversal unlikely. If the case is judicially reviewed, the court considers the merits of the decision, not the preliminary decision to charge. Even a successful judicial assault on an agency enforcement decision often produces only a remand that allows the agency to reconsider the case and come to the same conclusion. Nor does Congress or the President exert any meaningful control over enforcement discretion in individual cases; indeed, it would be improper for those bodies to interfere in a pending adjudicatory process.
The charging process is largely concealed from the public and targets. An agency can close the meeting at which it considers whether to take enforcement action (in fact, at most agencies, these decisions are made by notational voting, not in meetings). Staff memoranda recommending enforcement are exempt from Freedom of Information Act (FOIA) disclosure.
Because no external checking mechanisms exist, agencies should generate internal checks on the prosecutorial process. Such checking practices are referred to as “internal administrative law,” meaning procedures created by the agency itself that the agency is not legally required to provide. Such procedures might limit the discretion of agency heads or staff, allow monitoring by superiors of staff discretionary decisions, or provide protections for regulated parties and regulatory beneficiaries.
B. The Case for Greenlighting
The greenlighting function has important structural advantages as an accountability mechanism. For a number of reasons, the agency heads, rather than the staff, should make the call when the agency’s enforcement caseload is small enough to allow them to do so.
As pointed out earlier, from the point of view of the target of administrative enforcement, the charging decision is often the most important procedural event in the entire regulatory process. If the heads believe a case brought to them by prosecutors is weak or ill-advised, or does not align with agency-head priorities, the case should be stopped before it goes any further. In such cases, the target should not be forced to agree to a settlement, and the agency should not embark on the costly adjudication process.
However, the argument for greenlighting goes well beyond protecting private interests. An agency must adopt prosecutorial priorities since its resources will never be adequate to prosecute every case that might involve a regulatory violation. It is appropriate for agency heads, rather than staff, to make the call on the allocation of limited enforcement resources. Even more significant, combined-function agencies use adjudication for policymaking, and choosing which cases to prosecute is an essential element of the policymaking process. This Section explores the arguments in favor of greenlighting in greater detail.
1. Principles of Public Administration
Accepted principles of public administration suggest that agency heads must supervise staff decisionmakers in hierarchical government agencies that exercise delegated power. Supervision ensures that the staff implements the priorities of agency heads and that agency norms and practices are respected. Agency-head supervision is particularly vital when it concerns functions like charging decisions that are unconstrained by external judicial, executive, legislative, or political checks. Of course, supervision can take many forms, but greenlighting is one effective method of accomplishing it.
The agency-head approval process serves other goals identified by public administration scholars. William Simon identifies a set of practices described as “post-bureaucratic organization.” These practices depart from the traditional choice between professionalism (that allows professionals such as prosecutors to exercise unconstrained discretion) and strict hierarchical bureaucratic control. Post-bureaucratic administration entails transparency and a requirement of justification for staff discretionary decisions, representation of stakeholders, and multi-disciplinary group decisionmaking for evaluating discretionary decisions.
Simon observes that post-bureaucratic administrative innovation has had little impact on the criminal prosecution process. However, greenlighting exemplifies post-bureaucratic administration. Greenlighting requires professional prosecutors to explain and justify their charging decisions and makes these decisions transparent to agency heads. The notice-and-comment process that precedes greenlighting exposes prosecutors to the opinions of stakeholders such as targets or victims. Greenlighting entails a multi-disciplinary collegial discussion among the agency heads. Greenlighting encourages agencies to adopt prosecution guidelines. It enables the heads to monitor prosecution decisions and to adjust their priorities when there is a change in the markets the agency regulates or when the priorities of the agency heads change.
There is another way that greenlighting serves important principles of public administration. As practiced in most agencies, greenlighting requires staff prosecutors to produce a confidential and candid memorandum to the agency heads. These memos explain the factual, legal, and policy rationales for charging a target or settling the dispute. They analyze strategic concerns such as evidentiary weak points or political implications. SEC prosecutors file “Action Memos;” FTC prosecutors file “Memoranda Recommending Complaint;” and FERC prosecutors produce the “Enforcement Staff Report and Recommendation.” The process of preparing such memoranda is likely to improve the staff’s decisionmaking process. Staff members realize they will have repeated interactions with the agency members and are anxious to preserve their credibility by writing thorough and balanced memoranda.
Still another advantage of greenlighting is that in multi-member agencies, the greenlighting function compels agency heads to make a collective and collegial deliberative decision about prosecution priorities. Multi-member agency heads must be politically balanced and are likely to have different skill sets and backgrounds. As a result, the collective agency-head decision about enforcement priorities and charging decisions may be better than leaving that decision to prosecutors (as in the criminal law process) or to the General Counsel (as in the case of the NLRB). This assumes, of course, that the agency’s caseload permits collective decisions about prosecution (as it probably does not in the case of the NLRB). This pluralistic decisionmaking process is one of the advantages of multi-headed agencies, and it applies to enforcement decisionmaking and other administrative functions such as rulemaking or final adjudicatory decisions.
2. Internal Separation of Powers
The greenlighting function is an example of “administrative separation of powers” as described by Jon Michaels. Michaels described three distinct and competitive interests at play in agency decisionmaking. The three are the agency heads, the agency staff, and outsiders to the agency (including both regulated parties and beneficiaries of the regulatory scheme). These three interests engage in constant rivalrous competition in the course of a variety of administrative functions. Michaels points out the parallels between this internal separation of powers and the traditional system of external separation of powers and checks and balances between the Legislative, Executive, and Judicial Branches of government.
Applying Michaels’ analysis to charging decisions, the agency’s investigators and prosecutors take the initiative to uncover violations and prioritize them. They make the initial decision whether to prosecute, settle, or abandon the cases. The staff is checked by the agency heads who must greenlight the staff’s decision. In practice, the heads almost always support the staff’s determinations, but the need for agency-head approval constrains prosecutorial decisions. Regulatory targets can also substantially influence the notice-and-comment system employed in many agencies, which is discussed in Part III of this Article. Thus, the three interests identified by Michaels engage in a competitive struggle within the adjudicatory process, with the same positive effects as traditional inter-branch checks and balances. Because, as discussed above, external separation of powers has almost no influence over charging decisions, internal separation of powers seems to be an attractive alternative.
3. The Principal–Agent Problem
Prosecutorial decisions or settlements can represent a principal–agent problem if the preferences and priorities of staff prosecutors fail to align with those of the agency heads. Misalignment of prosecutorial priorities between agency and staff can produce either over- or under-enforcement.
Agency prosecutors sometimes want to over-enforce by pushing the envelope to pursue targets they perceive as wrongdoers, even if the evidentiary basis for doing so is questionable or the legal theory is not well supported by existing precedent. The staff may pursue a “crackdown” by allocating resources to a particular class of cases the heads do not support. Prosecutors may be overcharging to force the target to settle. Agency attorneys may have their future careers in mind rather than the public interest. The staff may wish to use resources unwisely by concentrating on trivial cases that are easier to win or to run up the numbers. Thus, the requirement that the agency heads greenlight a charging decision helps to rectify possible misalignment of priorities between staff and agency heads. Such misalignment issues have sometimes surfaced in NLRB ULP enforcement, where the General Counsel, rather than the agency heads, controls charging decisions. In an interesting recent development, the FTC heads voted (3–2) to sue to block Meta’s acquisition of Within, rejecting the staff’s recommendation not to attack the acquisition. This vote reflects misalignment between agency heads and staff that would produce under-enforcement and was rectified by a positive greenlighting determination.
In addition to correcting the misalignment of priorities, greenlighting creates what is sometimes called a “sentinel effect.” The sentinel effect means that people make different decisions when those decisions are subject to check than when they are not. The sentinel effect exists because the staff is well acquainted with the enforcement preferences of the chair and the other agency heads. The heads may have made these preferences clear in discussions with prosecutors, or the prosecutors may discern these preferences from their experience with past greenlighting events. Staff prosecutors might not advance proposed complaints if they think the agency heads might reject or narrow them or even that there might be a contentious discussion and a split vote at the commission level. I frequently asked interview subjects a counterfactual question: If greenlighting did not exist, and the staff was free to choose prosecution targets, would the pattern of prosecutions look different than it does now? Most interviewees answered affirmatively. They believed the staff would have been more aggressive in choosing prosecution targets if their charging decisions were not subject to greenlighting.
Agency-head approval obviously is less effective as a check on under-enforcement (as opposed to over-enforcement) since the heads may not see cases on which the staff has passed. In particularly important cases, a staff recommendation not to charge may enter the greenlighting process and enable the agency heads to reject the staff’s decision by increasing the level of enforcement, as occurred in the FTC’s decision to sue Meta discussed above. In less important cases, informal communications between the chair and the general counsel often address the under-enforcement problem.
C. The Case Against Agency-Head Greenlighting
1. Confirmation Bias
I. The Confirmation Bias Problem
A number of observers of the administrative enforcement process are troubled by the problem of confirmation bias because the greenlighting process requires agency heads to discharge both prosecutorial and judicial functions in the same case. As a result, the heads may be unable to render an unbiased final adjudicatory decision when a case they greenlighted returns to them for the final agency decision. The famous 1941 Report of the Attorney General’s Committee on Administrative Procedure identified the problem of confirmation bias and offered a suggestion to partially remedy the problem.
“Institutional bias” exists independently of confirmation bias, and both can operate when agency heads make the final decision in an enforcement case. Institutional bias means that agency heads are likely to believe in strong enforcement of the regulatory regime for which they are responsible. They probably wish to support the hard work of their prosecutorial and investigative staff by validating those lower-level decisions. Institutional bias is inevitably present in combined-function agency enforcement proceedings, but confirmation bias adds an additional concerning element.
Confirmation bias might manifest itself in different ways. For example, the agency heads may be reluctant to overturn an ALJ’s decision against the target because doing so would suggest the heads were wrong to have greenlighted the complaint in the first place and thus, wasted agency resources. The converse problem also exists—commissioners who voted against greenlighting a complaint may be reluctant to uphold an ALJ’s decision against the target because doing so might suggest their earlier vote was wrong.
There is another way greenlighting could produce confirmation bias. Agency heads might rely on ex parte information, opinions, and anecdotes communicated to them by the prosecutorial staff in meetings that occurred before and during the greenlighting process. This information might predispose them to decide the case against the private party. Yet some of this material likely will not appear in the record of the hearing conducted by the ALJ. That record should form the exclusive basis on which the heads make the final adjudicatory decision. This form of confirmation bias becomes more severe as the number and intensity of contacts between prosecutorial staff and agency heads increase. In FERC, for example, such communications apparently occur frequently over a course of years during the investigatory process.
The drafters of the APA were concerned by the problems of confirmation bias, and they instituted a system of internal separation of functions to protect against it. Under § 554(d) of the APA, agency staff members who played a significant adversarial role in a case as prosecutors, investigators, or advocates are prohibited from serving as adjudicatory decisionmakers or as off-the-record advisers to adjudicatory decisionmakers in the same case. Congress had two rationales for imposing internal separation of functions. First, an adversarial staff member’s prosecutorial or investigative work may have infused a “will to win” that distorts the adversary’s ability to serve as an impartial decisionmaker or adviser. Second, the adversary may have been exposed to information about the facts of the case or other information about the target and its behavior that do not find their way into the adjudicatory record. As discussed above, these are the two ways that greenlighting might also create confirmation bias at the agency-head level.
However, for reasons to be discussed, the APA excludes agency heads from internal separation of functions. The APA allows the heads to take part in both the prosecutorial and adjudicatory phases of the case. This statutory exception creates a risk that agency heads may be subject to confirmation bias.
II. Reasons to Believe that Confirmation Bias Is Not a Serious Problem
In theory, the differences in agency-head decisionmakers’ prosecutorial and adjudicative roles should greatly reduce the risk that confirmation bias affects their decisions. For example, the burden of proof at the two stages is different. The greenlighting decision is based on probable cause to believe that a violation of law has occurred. An adjudicatory decision against the target must be supported by a preponderance of the evidence in the record—a far more demanding decisional standard.
There are important differences in the cognitive processes employed in carrying out the prosecutorial and adjudicatory tasks. When they vote to greenlight a complaint, the agency members know they are relying on a one-sided ex parte staff presentation of the evidence in favor of charging. Often, the staff has accepted the credibility of outsider witnesses for purposes of deciding to charge the target. The final adjudicatory decision, on the other hand, occurs after a trial-type adversarial hearing that will test the credibility of witnesses. The ALJ produces a reasoned opinion based on the evidence presented during the hearing. The agency heads are limited by the exclusive record rule to consider only the evidence introduced at the hearing. Whether the differences in the decisionmaking process at the two stages make any practical difference, however, is disputed. In any event, many targets and their attorneys are not impressed by the differences between the prosecutorial and adjudicatory stages. They complain that their prosecutors have turned around to act as their judges.
The confirmation bias problem arises rather infrequently. Most greenlighted cases settle—very few of them make it all the way to a final decision by the agency heads. And of those cases that traverse the entire process, confirmation bias is seldom a problem because of the high turnover rate of agency heads. The heads who are called upon to make the final decision are usually not the same people who greenlighted the case years before. But those agency heads who remain in their job for more than a couple of years are likely to see cases a second time.
III. Interviewees’ Views on Confirmation Bias
In my interviews, most former agency heads said they did not believe they were personally affected by confirmation bias, but they were aware of the issue. Indeed, some said they barely remembered the meetings at which they greenlighted the complaints. They pointed to the differences between the greenlighting and adjudication decisions that are discussed above, such as the exclusive record and differences in burden of proof. They observe that a failure to take account of evidence and arguments developed during the ALJ hearing would invite disaster on judicial review. Needless to say, however, such interview data is not very reliable. Nobody likes to admit they might have been biased. Moreover, some former agency heads stated that they found the situation uncomfortable, and others acknowledged it created an appearance of bias even though they believed they were not personally subject to confirmation bias.
Some former agency staff members believe confirmation bias is a serious problem. Other former staff members who are now in the defense bar disagree; they do not see confirmation bias as a problem with which they are concerned. Again, data from staff interviews is mostly meaningless on this issue; there is no way a former staff person could know whether the agency heads were biased in deciding cases that the staff members had been involved in prosecuting.
IV. Empirical Research on Confirmation Bias
Attempted empirical research on the existence of confirmation bias based on win rates is inconclusive. This is hardly surprising given the elusive character of psychological phenomena such as confirmation bias. Clearly, there are many other variables that predict win rates, most of them more important than confirmation bias.
Most of the empirical work on this issue concerns the FTC. The most comprehensive study of the issue covered all FTC agency-head administrative decisions between 1977 and 2016 (a total of 145 cases). It suggests that confirmation bias is not a major problem if it exists at all. When the same Commission majority both authorized the complaint and decided the case, the FTC dismissed 33% of the cases. When a different majority voted out the complaint and made the final decision, the FTC dismissed only 27% of the cases.
Several other studies of FTC decisionmaking point in the opposite direction, but they are based on a shorter time period and consider only limited portions of the FTC’s work. One study of FTC merger decisions between 1950 and 2011 indicated that confirmation bias does exist. When three, four, or five commissioners who made the final decision also participated in the charging decision, the FTC enjoyed a greater win rate than in situations where zero, one, or two commissioners participated in both decisions.
Other studies simply infer bias from the fact that the agency wins most of the cases decided at the Commission level. However, this analytical method is suspect given that the FTC is a law enforcement agency with discretion to select only strong cases to prosecute and where most of the cases settle. Agency decisionmakers are naturally subject to institutional bias; it is not surprising they would uphold most or all of the ALJ decisions that come to them, whether or not confirmation bias also exists.
2. Efficiency
A second disadvantage of the greenlighting system is based on efficiency concerns. The greenlighting process can be quite time-consuming for agency heads, especially if they are conscientious about reading the complete files. Another efficiency concern is that the need for greenlighting at the agency-head level can prolong settlement negotiations or increase the time between the prosecutorial decision to charge and the commencement of an agency hearing. Thus, greenlighting may contribute to the problem of administrative delay.
At the SEC, for example, action memos may run fifty to seventy-five pages in length and perhaps five to ten of them are circulated each week. In addition, the Wells responses or white papers are often lengthy, and some commissioners read them in full. Greenlighting is considered at formal SEC meetings, which are sometimes contentious. Thus, a substantial portion of the time of SEC commissioners is devoted to enforcement matters.
Agencies that handle the greenlighting function through notational voting spend less time in meetings, but the members must still read the lengthy files (although no doubt many of them have time only to read the executive summaries). At FERC, the agency heads make greenlighting decisions in several stages. Each stage generates lengthy memos, but voting is generally done through a notational process rather than in-person meetings. The same is true at the FCC where enforcement matters are seldom taken up during commission meetings.
SEC interviewees estimated that perhaps 40% or more of agency heads’ time is taken up in enforcement matters, including, but not limited to, charging decisions and settlement approvals. At FERC, estimates were much lower, perhaps closer to 10%, because relatively few enforcement cases make it that far. One former FTC commissioner estimated spending half of work time on enforcement issues but thought the time investment was well worth it.
Obviously, agency heads pay a substantial opportunity cost to achieve this level of involvement in enforcement issues. This is time the heads could devote to other important responsibilities such as rulemaking, ratemaking, liaisons with other agencies, study of the problems faced by the industry, development of policy, or consideration of adjudicatory records at the time of final decision.
D. Weighing Costs and Benefits
Weighing the cases for and against greenlighting, my conclusion is that the greenlighting function is valuable and should be preserved in agencies where the caseload permits it. Greenlighting produces substantial advantages in terms of public administration norms, such as the need for accountability of staff, improved supervision of staff, the sentinel effect, and the benefits of internal separation of powers. Greenlighting also mitigates the principal–agent problem. I believe these benefits outweigh the problems of confirmation bias and inefficiency, but this Article will suggest further checks and balances in Part IV that could alleviate both concerns.
III. Legality of Greenlighting
This Part discusses the arguments that greenlighting violates due process or the APA.
A. Due Process
Andrew Vollmer argues that the SEC denies a target due process when the agency heads greenlight a complaint and later issue the final agency decision. I disagree with Vollmer’s analysis. The Supreme Court has consistently rejected constitutional attacks that arise out of the structure of combined-function administrative agencies.
The leading case on this issue is Withrow v. Larkin. In Withrow, the Supreme Court assumed that the heads of a state medical licensing agency had personally investigated a physician’s conduct, authorized the filing of a criminal complaint against the physician, and then adjudicated a revocation of his license. The Court unanimously rejected the physician’s due process claim. It held that agency heads could exercise the functions of both investigation and adjudication in the same case, absent particularized facts indicating that the heads had prejudged the case. The Court pointed out that there is no incompatibility between the agency filing a complaint based on probable cause and rendering a subsequent adjudicatory decision in favor of the target when the evidence fails to establish a statutory violation. The Withrow Court was obviously concerned that a contrary decision would cast doubt on the practices of countless federal, state, and local licensing agencies.
The Withrow decision is supported by the principle of necessity—if the agency heads were disqualified by their involvement in prosecution, there would be no way to adjudicate the case and thus, no way to revoke the doctor’s license. Withrow is consistent with several earlier Supreme Court decisions that held agency adjudicatory decisionmakers are not biased simply by reason of their involvement in earlier agency proceedings.
Under Withrow, it seems clear that due process is not violated when agency heads greenlight a charging decision and later make the final adjudicatory decision in the same case, absent some further evidence they had prejudged the issues. Since Withrow, federal courts have consistently rejected arguments that agency heads who exercised overlapping functions could not fairly adjudicate a case.
Despite this authority, Vollmer argues that due process is violated when agency heads greenlight and later adjudicate the same case. His analysis is based on Williams v. Pennsylvania. Williams was a death penalty case in which Ronald Castille had previously served as a prosecutor. Castille later became a justice on the Pennsylvania Supreme Court and voted to uphold the death penalty in the same case he helped prosecute. The Supreme Court held that this combination of functions violated due process.
It seems plain that Williams is distinguishable from the administrative greenlighting issue. It is shocking and inexplicable that a justice on a state supreme court would not disqualify himself in a case he had prosecuted. It is a gross breach of judicial ethics for a judge to decide a case in which the judge served as counsel in an earlier phase of the case, let alone a death penalty case. The Williams scenario is likely a situation that will never recur. It easily fits into the Withrow exception for particularized facts that reveal prejudgment. In contrast, there is no breach of judicial ethics nor any particularized facts indicating prejudgment when an agency head greenlights a prosecution, then decides the case. Such an action is routine, generally accepted, and has occurred in countless cases.
By concentrating only on the SEC, Vollmer fails to deal with the systemic effect of holding that greenlighting plus adjudication is a due process violation. Such a decision would have an enormously disruptive effect on the state and federal administrative process because greenlighting is so common, particularly in licensing agencies. The Supreme Court is reluctant to decree due process principles that would have widespread effect of this kind. Such a decision would also have the effect of holding unconstitutional the agency-head exception in § 554(d) of the APA, a step the Supreme Court would be reluctant to take. The Williams decision does not offend the principle of necessity because the Pennsylvania Supreme Court could have decided the case if Castille had recused himself. In contrast, a decision preventing all agency heads who had greenlighted a prosecution from deciding the case would frequently immobilize the agency for lack of a quorum, therefore making it impossible to render a final decision. Thus, Vollmer’s argument that greenlighting at the SEC or elsewhere violates due process is not persuasive.
Vollmer argues that to solve the due process problem, an agency member who voted to greenlight a case should be disqualified from voting on the final decision. Whether that proposal should be adopted as a matter of policy is discussed below.
B. The Administrative Procedure Act
Section 554(d) of the APA imposes a separation of functions requirement that prevents a staff member involved in an investigation of a case from serving as an adjudicator in the same case (or a substantially related one) or as an adviser to the adjudicator. However, § 554(d) does not apply “(C) to the agency or a member or members of the body comprising the agency.”
This agency-head exception was inserted because Congress felt that application of separation of functions to the agency head would damage the agency’s ability to conduct law enforcement. As a result, according to a number of cases, the APA allows agency heads to engage in a prosecution function such as greenlighting, then participate in the agency’s final adjudicatory decision. The Withrow decision contains a dictum confirming that the APA permits agency heads to engage in investigation and prosecution in the same case they adjudicate.
IV. Accountability Mechanisms and Greenlighting
Part II of this paper discussed the fundamental rationales for the greenlighting function. Agency-head review of charging decisions serves as an accountability mechanism that promotes public administration values and helps to correct principal–agent misalignments. Nevertheless, Part II acknowledged concerns about greenlighting because of possible confirmation bias and efficiency issues. This Part discusses accountability mechanisms that might be employed in connection with the greenlighting process and that might alleviate these concerns. There are, of course, structural changes that would address these concerns, such as eliminating combined-function agencies, but such options are beyond the scope of this Article.
A. Mechanisms that Allow Targets’ Input into Charging Decisions
As discussed in Part I, the SEC, FTC, FERC, and NLRB employ pre-charging notice-and-comment procedures. These formalized procedures invite targets to submit memoranda designed to dissuade the staff from charging them and supplement the informal interchange between the target’s attorneys and agency enforcement staff that routinely occurs during the investigation process. In general, target lawyers value these pre-charging procedures as opportunities to dissuade the staff from seeking a Commission greenlight, to improve their position in settlement negotiations, and to influence the agency heads not to greenlight the complaint.
In particular, the SEC’s Wells notice, and informal notice-white paper procedures are well established. SEC commissioners told me they and their advisory staff take Wells responses or white papers seriously. These submissions sometimes persuade the SEC heads that issuance of a complaint is contrary to their policy priorities, that the case is weak, or that the complaint should be narrowed. Defense lawyers appreciate the Wells and white paper process and often use it to extract more information from the staff than might otherwise be disclosed and to improve their prospects for settlement. My interviewees expressed unanimous support for the Wells or white paper procedures.
FERC provides two distinct opportunities for the target to influence the charging decision. The target can respond both to the staff’s “preliminary findings memorandum” and to its Rule 1b.19 memorandum. Both agency heads and private lawyers favor the pre-charging notice-and-comment procedures, however, several believe that the Rule 1b.19 procedure is redundant since it duplicates the preliminary findings memorandum and response.
The regional offices of the NLRB that investigate ULP charges offer targets the ability to file a position paper in response to the charge. Targets also have the opportunity to meet with regional staff and, in cases where there is an appeal against the charge or refusal to charge, with regional directors and the General Counsel if the latter agrees to the meeting.
The FCC does not provide a formalized notice-and-comment system prior to issuance of an NAL. The target can file a detailed response after the staff issues the NAL. The response is intended primarily for the benefit of the commissioners who will be called on to greenlight the complaint and, later, will adjudicate it.
In my opinion, the notice-and-comment procedure employed by the SEC, FTC, FERC, and NLRB during the pre-charging phase of enforcement is useful and should be considered best practice. The formalized ability to comment contributes to a sense by private parties that they are being treated fairly and helps them decide whether to settle. The process facilitates reasoned decisionmaking by the enforcement staff and helps the agency heads produce an informed greenlighting decision. It furthers what Michaels called administrative separation of powers. I would not, however, recommend agencies provide two separate opportunities of this kind, as occurs at FERC. A double notice-and-comment procedure seems redundant and increases costs for both targets and agencies without corresponding benefit.
B. Ability of Target to Communicate with Individual Agency Members
Several agencies permit targets to communicate with the agency heads prior to their greenlighting decision. After the FTC staff recommends issuance of a complaint, the target is entitled to meet separately with each of the five commissioners to attempt to dissuade them from greenlighting the complaint. Commission staff are usually present at these meetings. Because no complaint has been issued, the APA’s ban on outsider ex parte communications to agency decisionmakers is not applicable. A similar practice of meetings between targets and commissioners exists at the FCC but is employed less often than at the FTC.
Several FTC interviewees support the FTC’s ex parte meeting procedure. Former FTC commissioners found these meetings enlightening, since they are otherwise exposed mostly to the staff’s arguments before greenlighting the complaint. The meetings tend to offset criticisms that the commissioners are out of touch and removed from practicalities. Some private lawyers, including former FTC staff, think the meetings can be a useful vehicle to persuade a commissioner that the case is weaker than the staff says it is. At times, the commissioners can broker settlements. Other private lawyers refer to the FTC meetings as “last rites” and think they are a costly waste of time. My interviewees were also skeptical about the value of meetings between targets and FCC commissioners, which they regarded as useless.
Staff members and former heads at other agencies were unenthusiastic about the FTC’s and FCC’s one-on-one practice. I agree with their criticisms. These meetings seem wasteful of the precious time of both the agency heads and staff members. The process is costly for clients who must pay their lawyers to engage in numerous separate meetings, even though it is unlikely the meetings will have much practical impact. The meetings can worsen confirmation bias because the heads learn still more about the case at the pre-complaint stage, including material that might never become part of the record.
FERC permits targets to submit written, but not oral, communications to the agency heads during the investigation process. This approach, a holdover from prior practice existing before the FERC acquired civil penalty authority, is less time-consuming and costly than the ex parte in-person meetings conducted by the FTC and FCC. Some FERC practitioners send letters to the commissioners frequently; others never do. Whether such communications are useful to targets or to FERC is debatable. Some interviewees thought the letters might prompt the recipients to communicate with the staff to better understand the issues. Others thought the practice was counter-productive because such communications might prejudice the staff against a target that attempted to go over their heads.
C. Separation of Functions During Investigation
William Scherman and his co-authors proposed changes to FERC’s ex parte communication and separation of functions rules. Under the existing FERC rules, as in most agencies, ex parte communications to decisionmakers or their advisers either by outsiders or by adversarial staff members, such as prosecutors and investigators, are prohibited after FERC makes a charging decision by issuing an OSC. However, such communications can and do occur before the agency decides to charge.
Under Scherman’s proposal, these prohibitions would apply at an earlier stage of the proceeding, perhaps when the Rule 1b.19 notice issues (meaning staff has decided to recommend charging the target). His article expressed concern about the fairness of allowing the staff unfettered access to the Commission during the investigatory and greenlighting phases of the case while the ability of targets to communicate with agency heads is limited to written submissions. Under Scherman’s proposal, meetings between the staff and agency heads concerning greenlighting would be on the record rather than ex parte and the target could participate in such meetings. In a subsequent article, members of the FERC staff and outside lawyers strongly criticized Scherman’s proposal.
Most interviewees opposed Scherman’s proposal, whether at FERC or at other agencies. The interviewees believe the staff needs to conduct a candid and robust discussion with agency members about whether to greenlight a case. If the target’s representatives were present or if the discussion was on the record, staff could not level with the agency heads about the weaknesses in the case, the political or policy issues it creates, or the terms on which it might be settled. The need for candid discussion about charging is the reason for the Sunshine Act exemption of meetings devoted to initiation of litigation.
D. Delegation to Enforcement Staff in Routine Cases
The 1941 Report of the Attorney General’s Committee on Administrative Procedure (the Report) expressed concern with the problem of confirmation bias resulting from greenlighting. To alleviate the problem, the Report suggested that agencies delegate to staff the decision to issue a complaint in cases that raise only applications of well-established legal principles. Such cases might present difficulties of proof but would otherwise be routine. Delegation of the charging decision in routine cases would not inhibit the agency’s use of adjudication for policymaking. However, in cases raising important policy issues or those that involve extension of existing precedents or new departures, the agency heads should be responsible for making the charging decision.
Such delegations are in effect at several agencies. The FCC staff has power to charge civil penalties below a certain amount ($100,000 for common carriers, $25,000 for others), so that the commissioners need not consider the majority of penalty cases. At FERC, penalties arising from reliability violations that are assessed by an industry self-regulatory process can be processed without agency-head involvement. In the NLRB, over 90% of complaints processed by regional offices involve routine, well-settled applications of law, and are filed without any involvement of the General Counsel or the General Counsel’s staff, even though, in theory, the General Counsel is responsible for all NLRB prosecution decisions.
Delegation to staff of complaint issuance in routine cases is a good idea, especially in agencies with substantial enforcement caseloads. Delegation would reduce the number of cases in which confirmation bias is a concern because the agency heads would never see them before the final adjudicatory stage. Delegation should also be efficiency-enhancing by reducing the amount of time the heads need to spend on enforcement decisions. Thus, agency heads could adopt procedural rules setting forth classes of cases the staff could initiate on its own. Of course, such rules are possible only if allowed by statute because some statutes require agency-head approval of every complaint.
Nevertheless, most of the interviewees opposed delegation to the staff of complaint issuance. They thought it would be difficult to identify precisely which cases are routine or unimportant. At FERC, relatively few enforcement cases are litigated rather than settled; the remaining cases tend to involve policy questions. Even if the case turns on evidentiary issues rather than disputed legal questions, these evidentiary issues may be controversial and of fundamental importance, especially in competition and securities cases. Even routine cases involve the expenditure of resources and can create precedents that may have important effects on the regulated industry. Former commission members think that complaints in routine cases should be approved by politically responsible agency heads in light of the importance of the cases to the particular parties and the sentinel effect. Private lawyers want the commission-level bite at the apple, even if the case seems routine.
E. Disqualification of Agency Heads Who Participated in Charging Decisions
Agency members who voted to greenlight a case could be disqualified from voting on the final adjudicatory decision. Andrew Vollmer, who was a former SEC staff member, has strongly advocated this proposal.
One practical problem with Vollmer’s proposal is that such disqualifications might render the agency unable to muster a quorum to vote on the final decision, causing the decision in the case to be suspended indefinitely. This would not frequently occur, given the rapid turnover of agency heads, but it would occasionally happen, especially during presidential transitions when the confirmation process causes substantial delays in filling vacancies. Virtually everyone I interviewed opposed this proposal, including many who now serve in the defense bar.
More fundamentally, Vollmer’s proposal would force agency heads (at least those who have not decided to leave the agency in the near future) to make a difficult choice. Should they disqualify themselves from greenlighting a case to preserve the ability to vote on the final decision, or should they retain the greenlighting function and give up their vote on the final adjudicatory decision? Some former agency heads who answered this question said they would opt-out of the charging decision because of the importance of being able to make policy through the adjudicatory decision. Others said they would opt-out of the final decision because the charging decisions are so important and so much more numerous than cases that survive all the way to the end of the adjudicatory process.
As argued above, participation of agency heads in the charging decision is valuable as a check on prosecutors and as an element of policymaking. It would be unfortunate if commissioners opted out of that function. And it would be equally unfortunate if some were disqualified from participating in the final decision process. That process involves collegial effort and compromise of diverse policy perspectives and often entails establishing agency policy for the future. In my view, these structural concerns are more important than preventing confirmation bias.
F. Removal of Agency Heads from Greenlighting: The NLRB Model
Since passage of the Taft-Hartley Act in 1947, NLRB members lack the power to make charging decisions in ULP cases. Instead, the General Counsel makes the charging decisions and is politically accountable for them. Much has been written about the NLRB’s separation of prosecutorial and adjudicatory functions. Several articles have recommended that other combined-function enforcement agencies follow the NLRB model.
Although NLRB members play no prosecutorial role in most ULP cases, the separation of prosecution and adjudication is incomplete. The members decide whether to approve the General Counsel’s recommendation that the Board seek a temporary injunction in ULP cases (so-called “10(j) cases”). Thus, under the NLRB model, the agency heads greenlight the particularly sensitive temporary injunction cases but not the more routine and far more numerous ULP complaints in which no injunction is sought.
NLRB regional offices file between 800 and 1,200 ULP complaints each year. This heavy caseload would make it practically impossible for the NLRB agency heads to be meaningfully involved in charging decisions. The agency heads see between ten and one hundred 10(j) cases each year, which is a more manageable task. Limiting the Board’s greenlighting function to 10(j) cases makes sense because 10(j) cases are more significant than routine ULP cases. The Board seeks an injunction when the conduct being enjoined may inflict serious injury that could not be remedied by a later adjudicatory decision. For example, the Board might seek a temporary injunction against employer violations that interrupt a union-organizing campaign. An injunction is appropriate in such cases because, otherwise, the adjudicatory decision would occur long after the momentum behind the organizing campaign dissipated.
Confirmation bias remains a potential issue in 10(j) cases because Board members exercise both prosecution and adjudicatory functions in those cases. My interview subjects doubted that the problem was serious because the General Counsel’s written request for Board approval of the injunction accepts the credibility of the complainant and does not include much of the factual and evidentiary material that the prosecutors have assembled. When a 10(j) case comes to the Board after an ALJ decision, the record looks completely different than it did at the complaint stage because it contains the respondent’s evidence and the ALJ’s credibility determinations.
The separation of prosecution and adjudication at the NLRB in ULP cases can create principal-–agent problems when the views of the General Counsel and the Board members misalign. These principal-–agent problems arise most frequently when presidential administrations change and the 3–2 political balance on the board switches, while the General Counsel holds over. The General Counsel may refuse to issue complaints in cases that the heads would have prosecuted. Alternatively, the General Counsel may issue complaints that the heads would not have authorized. Since most cases settle (at least 90%), the Board never has an opportunity to pass on the policy issues raised in settled cases.
One example of this sort of conflict arises out of the General Counsel’s valuable advice-giving function. If the General Counsel disagrees with Board-made law and hopes to change it, the General Counsel can advise charging parties to file particular types of charges and regional offices to issue complaints in those cases. Of course, the Board makes the final call and can reject the General Counsel’s initiative. Charging parties may participate in ALJ hearings and can introduce witnesses and arguments supporting their view, perhaps disagreeing with the General Counsel’s approach. On the other hand, the Board will be unable to change existing law if the General Counsel disagrees and declines to charge cases raising the issue.
Another area in which the General Counsel and the agency heads might come into conflict arises in judicial enforcement. Unlike other independent agencies, the Board’s attorneys handle litigation for the Board through the Court of Appeals. Conflict might arise when the general counsel is called on to enforce a Board decision in court with which the General Counsel disagrees (as could occur after a change in presidential administration, where the General Counsel is a holdover). Such problems have occurred in the past but not in recent years.
An additional problem with the independent General Counsel is that it creates a duplicate power center within the agency. The General Counsel is a de facto agency head whose practical power may well exceed those of the five NLRB members. Particularly during periods of budget stringency and uncertainty—as have occurred in recent years—the Board and the General Counsel have disagreed about management and budgetary issues, such as how to both make the necessary budget cuts and allocate limited resources. A recent example of General Counsel-Board conflict resulted from differences of opinion about replacement of the Board’s outmoded IT system. In addition, the General Counsel makes staff hiring decisions (except for the Board members’ personal staffs). General Counsel hiring decisions have given rise to conflict with Board members.
In my interviews, I found little enthusiasm for the NLRB model in other federal combined-function agencies. Most interviewees favored having the agency heads make charging decisions, both in the interest of constraining prosecutors and articulating policy. They were concerned by the problem of the general counsel being out of sync with the agency heads and the creation of a competing power center. They feared that an independent general counsel might increase partisanship. A minority of interviewees were open to the idea.
G. Enforcement Guidelines
One way to reduce administrative prosecutorial discretion is by adopting guidance documents that establish enforcement priorities and criteria. As conditions in the regulated industry change, or as new agency heads with different priorities are appointed, the guidelines can and should be updated. Such guidelines provide readily available guidance for both staff and agency heads and help to assure more consistent charging decisions. A concern with making such guidelines publicly available, as they must be under FOIA, is that they can undermine deterrence by informing the regulated industry of what cases are unlikely to be prosecuted.
Nevertheless, federal agencies have found it feasible to establish prosecution guidelines that at least suggest the factors that prosecutors and investigators should consider. The FCC adopted guidelines for upward and downward adjustment of forfeiture penalties. The NLRB Division of Advice furnishes detailed guidance to regional offices about enforcement criteria and policies in ULP cases. In addition, the general counsel adopted detailed case-handling instructions (publicly available) to regional offices about every aspect of processing ULP, representation, and compliance proceedings. A useful FERC guideline lists the factors that staff should consider in deciding whether to open an investigation.
H. Peer Review of Prosecution Decisions
Another approach to limiting and checking administrative prosecutorial discretion is to institute a system of peer review of charging decisions. Peer review is common in post-bureaucratic public administration, such as the “mortality-morbidity” reviews of adverse events that occur in hospitals. The idea is that a team of staff prosecutors and investigators would conduct a periodic review of a sample of prior decisions by the staff to charge or not to charge. The objective of such peer review is to enhance the learning of staff decisionmakers about the prosecutorial decisionmaking process and to achieve more consistency for future decisions. The team would ascertain whether prior charging decisions led to successful and cost-effective outcomes and whether these decisions complied with the agency’s prosecutorial guidelines and procedural requirements.
I. Structural Solutions to the Confirmation Bias Problem
There are a number of possible structural changes to the organization of combined-function agencies that would remove the possibility of confirmation bias. Except for considering the NLRB model that stripped agency heads of greenlighting power, I have not explored these options. The other options are beyond the scope of this Article and most of them do not seem politically feasible.
For example, Congress might require all enforcement adjudication be situated in federal court rather than being conducted through internal agency adjudication, or that a target would have the right to remove an administrative enforcement case to federal court (as occurs in the case of FERC), or that the agency must bring a de novo federal court action to collect a civil penalty (as in the case of the FCC).
Another set of options, often referred to as external separation of functions, calls for creation of an adjudicatory tribunal, which is common in other former British colonies, either for specific agencies (as in Canada) or for all enforcement agencies (as in Australia and the United Kingdom). Under the tribunal model, an enforcement agency engages in rulemaking, investigation, and prosecution, but a separate agency makes the resulting adjudicatory decision. The United States employs tribunals in worker safety, mining safety, and federal taxation cases. Many states situate adjudication in separate tribunals in their unemployment compensation and workers’ compensation systems.
Still another approach is delegation of the internal appeal function to an appellate review board, such as the Environmental Appeals Board, or to a judicial officer, which occurs in the Department of Agriculture. The delegation could cover certain classes of cases that are likely to present only factual issues, or it could cover all enforcement cases. The agency heads might retain discretionary review power over decisions of the intermediate review board or judicial officer in cases presenting important policy issues. Delegations of final decisional authority are quite common in the administrative state, and might be attractive for agencies with substantial caseloads or serious backlogs at the agency-head level. Delegation of the power to make the final adjudicatory decision would promote efficient use of the limited time of the agency heads and reduce delays in making final decisions. Such delegation would also limit the number of cases subject to potential confirmation bias.
Conclusion
Combined-function enforcement agencies should engage in greenlighting if their caseload permits them to do so. Greenlighting means that the agency head, including all the heads of a multiple-member agency, is responsible for approval of charging decisions made by the staff. The greenlighting process assures that the choice of enforcement targets aligns with the priorities of the agency head and is a wise allocation of scarce enforcement resources. Greenlighting is a powerful accountability mechanism to control the exercise of prosecutorial discretion—a problem that pervades the world of criminal and administrative prosecution.
Agencies engaged in greenlighting should require the staff to engage in a structured written notice-and-comment process, whereby targets can attempt to persuade the staff not to charge them. In addition, when the staff seeks agency-member approval of a charging decision, it should generate a detailed memorandum. This document should set forth the facts uncovered by the investigation and applicable legal analysis to assist the members in the greenlighting process.
Because greenlighting may present problems of confirmation bias and inefficiency, agencies should consider whether their enforcement docket includes classes of cases that are sufficiently routine that the charging decision can be delegated to the staff or that the final adjudicatory decision could be delegated to a judicial officer or a review board. Agencies should also consider adopting guidelines that set forth priorities for exercising its prosecutorial discretion and instituting a peer-review process at the staff level. With these refinements, combined-function agencies should continue to employ the greenlighting process when it is practicable to do so.
