Central banks around the world are vested with enormous powers, for good or evil, over the lives of the countries in which they function. All central banks enjoy authority over monetary policy. Most supervise and monitor banks and the banking system. While a country with an effectively functioning central bank may not necessarily be happy, one with a dysfunctional central bank is decidedly glum.

Central banks around the world are vested with enormous powers, for good or evil, over the lives of the countries in which they function. All central banks enjoy authority over monetary policy. Most supervise and monitor banks and the banking system. While a country with an effectively functioning central bank may not necessarily be happy, one with a dysfunctional central bank is decidedly glum.

Central banking is a high-stakes game, and it is important that the actions a central bank takes be wise, fair, and grounded in reality. Many features of institutional design can be understood as intended to further this goal. One of the more important such features is the possibility that the central bank’s actions will be reviewed, and potentially vetoed, by some other agency—usually a court, but potentially another entity, such as a finance ministry. This paper discusses the process of “external review” of central bank and bank regulatory decisions, with particular reference to the law of the United States.

Basic Principles

Why do we have external review of agency decisions at all? Why not just let the agency decide, and decree that the agency’s decision will be final and binding with no further review? In other words, why not allow the agency to function like the Vatican’s College of Cardinals, which is infallible with respect to the selection of a Pope. The Catholic Church conclusively presumes that the decisions of the College of Cardinals reflect God’s choice of a Pontiff. Why shouldn’t the legal system declare, likewise, that a central bank’s action on a matter within its jurisdiction is final and binding?

Improving Accuracy of Decisions

Part of the answer to the question of why we have external review of central bank decisions is that we want to have the best possible decisions on issues of financial policy and regulation, and we believe that in some cases external review of a central bank’s actions will improve the decision process.

But how do we know when, and to what extent, the decisions of the central bank should be subject to external review? This question may seem obvious: one might suppose that a decision can always be improved by being checked through review by an external reviewer. But it takes little reflection to realize that this is not true. For example, suppose we had a cello competition at a music conservatory, and Yo Yo Ma, the famous cellist, kindly agreed to judge the winner. Suppose also that to assure fairness in the competition, the organizers decided to subject Mr. Ma’s decision to review by the pop vocal group, the Spice Girls, who also attend the competition. The Spice Girls, let’s assume, know nothing about the cello and have never heard of Mr. Ma. It would not seem to be a particularly good idea to subject Mr. Ma’s decision to external review by the Spice Girls: they may be an excellent pop group, but their judgment about cello doesn’t rival Mr. Ma’s.

Now, it clearly would be preferable to reverse the order of judging in this contest. The Spice Girls make the initial decision about the winner of the competition, and their decision is reviewed by Yo Yo Ma. This is likely to yield a good outcome, since while the Spice Girls may pick more or less at random, Mr. Ma can substitute his own judgment and select the best cellist.

However, this scenario raises another problem. Given that Mr. Ma is a better decision maker than the Spice Girls (at least with respect to cello playing), what benefit is there to having the Spice Girls involved in the first place? We could simplify things by letting Mr. Ma make the decision without input from the Spice Girls. In this competition, he would be the musical equivalent of the College of Cardinals.

So, if we have two possible decision makers, and one has superior knowledge and expertise, the sensible course would seem to be to give the decision to the better decision maker and cut the less well qualified one out of the action altogether. It’s awkward, but we ask the Spice Girls not to come to the competition. But if we can do just as well with a single decision maker as with two, we have not yet provided a justification for when an external review process would make sense, since the essence of external review is the use of more than one decision maker.

We have been assuming that both decision makers are basing their judgments solely on the information set available to them, and that they both have access to the same information. That is, the Spice Girls have never heard of Mr. Ma, and so don’t give any weight to his decision; and Mr. Ma is very confident in his own judgment and probably horrified at the Spice Girls’ selection, so he completely discounts the credibility of their decision. Perhaps we can find a justification for external review if we relax these assumptions.

Consider the following hypothetical. The Spice Girls and Yo Yo Ma are both participating in a contest. There is a container filled with white and black beads. The contestants can’t see into the container, but they can take out ten beads, examine them, and then replace them. The object of the contest is to guess whether there are more white beads or more black beads in the container.

The Spice Girls draw first, and get nine white beads and one black bead. They decide, quite reasonably, that there are more white beads in the container. Their decision now is reviewed by Yo Yo Ma. Mr. Ma is allowed to draw his own sample of ten beads, and he gets six black beads and four white beads. He now has to decide whether to “affirm” the Spice Girls’ decision, or “reverse.”

The situation is thus as follows:

article image

Mr. Ma, based only on his own sample, would conclude that there are more black beads in the container than white beads. However, he knows that the Spice Girls have had a different draw, and if he puts the draws together he gets the following:

article image

He would therefore decide that there are more white beads than black beads, based on his knowledge of the Spice Girls’ decision, even though deciding on his own he would come out the other way. In the jargon of administrative law, he would “defer to” or “uphold” the decision of the Spice Girls. On the other hand, if the situation were reversed—if the Spice Girls had drawn six black and four white and Ma had drawn nine white and one black—he would decide, putting together the two draws, that the Spice Girls probably had made a mistake; he would “reverse” their decision.

Here, we have expanded the model to take account of the fact that the reviewer can get information, not only from his or her own investigation of the data, but also from the first decision maker’s decision. We have thus provided a model in which the process of review improves the decision, and where we can’t achieve the same degree of improvement simply by substituting the better decision maker. Both decisions add information, and the combination of the two is better than either of them alone.

It’s easy to extend this model to one where the reviewer has more information or better judgment than the initial decision maker, and where there is still a benefit to having two decisions rather than one. Suppose the Spice Girls get to draw and replace 10 beads, and Mr. Ma gets to draw and replace 20. Now, ordinarily, since Mr. Ma is drawing a larger sample of beads, his result will be more reliable than the result of the Spice Girls. Thus if Mr. Ma is the reviewer, he will ordinarily prefer his own judgment to that of the Spice Girls, as we would expect given his superior expertise. Sometimes, however, it will be rational for him to “defer,” even when his own judgment is to the contrary.

For example, suppose the parties get the following draws:

article image

Mr. Ma, based only on his own sample, would conclude that there are more black beads in the container than white beads. However, he also knows that the Spice Girls have had a different draw, and if he puts the draws together he gets the following:

article image

So even though Mr. Ma is a superior decision maker (because he has drawn more beads), we cannot dispense with the Spice Girls without incurring a cost because they can improve the result. Moreover, this model explains why we want the second in time, and not the first, making the decision. The second reviewer has access to two sets of information: (i) his or her own investigation of the underlying facts (i.e., in the model, his or her own draw out of the container); and (ii) the first agency’s investigation. Because the second reviewer has more information, his or her decision is likely to be superior even if the second reviewer is not a better decision maker in other respects.

This model gets us quite far toward explaining the presence of external review, but it does not go far enough, for two reasons. First, it doesn’t matter who goes first in this model. Either Yo Yo Ma or the Spice Girls can go first, and the result will be the same. In either case, the second decision maker adds up the total beads drawn by both parties and decides accordingly. This doesn’t seem to comport with the real world, where we see distinct differences between the character of the initial decision maker and the character of the external reviewer.

A second problem with this model is that it doesn’t explain why we need two decision makers. We can achieve exactly the same result with less hassle by simply combining them. You don’t need two separate decisions; you can have only one based on a draw of 30 beads. The result will be the same as when there are two decision makers. This is a problem for the model, because we are trying to explain why we might want to use more than one decision maker in order to improve the outcome of the process.

Let’s imagine a different scenario: a patient goes to a general practice doctor with a set of symptoms. The doctor takes a medical history, examines the patient, and orders tests. In the great majority of cases, the doctor is able to diagnose the patient and prescribe treatment. In some cases, however, the doctor may refer the patient to a specialist with a tentative diagnosis. The specialist reviews the chart, examines the patient, and decides whether to accept or reject the first doctor’s diagnosis.

This is a situation in which both decision makers are necessary for the best outcome. If we get rid of the general practitioner, we will not have a detailed initial examination conducted by a party competent to evaluate a wide range of problems. It is unlikely the specialist will be as good at conducting the initial evaluation as the general practitioner. On the other hand, we obviously can’t dispense with the specialist without losing the benefit of his or her expertise, which the general practitioner lacks.

This is also a case where the order of decision is important—we need the general practitioner to rule out possible syndromes that are outside the area of expertise of the specialist before the case is presented to the second doctor for analysis. And it is a situation where the option of simply combining the decision makers into one is not particularly palatable. We could certainly combine them by having both doctors present at all times, but doing so would be inefficient because it would waste the time of both practitioners to be present while the other is doing his or her work.

I suggest that this model gives us a better insight into why we might want to have external review of agency action, rather than simply collapsing the review process into a single procedure. We have a review process because we have different types of expertise or competence by the decision makers that cannot be combined without wasting the time of one or both.

The purpose of external review is then to conserve resources by allowing multiple decision makers with different competencies to examine the issues, and to allocate the final decision on any given matter to the last in time. This, I take it, is the basic rationale for external review of all administrative agency action, including actions by banking agencies and central banks.

Conserving on Resources

The analysis above might lead us to conclude that, far from being unnecessary, external review of agency action is always indicated. After all, we can almost always find people or agencies with different types of expertise, and nearly every decision will have multiple dimensions so that different types of expertise can improve the ultimate decision. But it would be an error to think that we should always have external review. The reason is simple: review itself is costly.

We might imagine a situation, for example, where we can continuously improve a decision by successive rounds of external review costing $1,000,000 each, but where the amount of improvement becomes smaller each time. Obviously, in this situation we can continue forever with review, and in the process can continue to hone and improve the decision. But we would not want to do so, because the costs of the review would soon outweigh whatever benefits it conferred. As a general matter, it seems that the more elaborate the external review, the greater the accuracy of the ultimate decision, but also the greater the cost of the decision process. How do we know when to stop? In principle, the answer to this question is easy. We stop when the marginal benefits of external review, in terms of improving the decision-making process, exactly equal the marginal costs of conducting the review, in terms of delay in reaching a decision and the transaction costs involved.

Another way to express the same idea is that we are trying to minimize the sum of two costs: the cost of error in the outcome and the cost of the procedure used to reach a decision.

Consider the following figure:

One can see that at point A, we have quite low error costs but high procedural costs. Point A is not optimal from the standpoint of social policy because the marginal costs of review exceed the marginal costs of error. At point C, we have very low costs of review, but we have purchased this saving with even higher error costs. Because the marginal costs of error exceed the marginal costs of review, point C is not socially optimal either.

The optimal point is point B, where the marginal costs of error equal the marginal costs of review. Note that this implies that we have to live with some error, because it is simply too expensive to eliminate all mistakes from central bank decisions, just as it is impossible to eliminate them from life.

Now in reality, we can’t know the relevant costs of error and costs of procedure, so we have to make fairly crude guesses. The following seem to be reasonable assumptions.

  1. The benefits of external review (i.e., the error costs avoided) are likely to be greater when the external reviewer has more expertise than the initial decision maker with respect to the particular matter or aspect of the decision being reviewed.

  2. The benefits of external review (i.e., the error costs avoided) are likely to be greater when there are major interests at stake than when there are only minor interests at stake.

  3. The costs of a procedure are likely to be greater as the number of levels of review increase.

  4. The costs of a procedure are likely to be greater as the complexity of review increases.

Applications to U.S. Law

Let’s now apply this basic framework to the context of central bank decisions. There are two basic axes for external review to take: (1) the stringency of review and (2) the timing of review.

Stringency of Review

The framework described above suggests that the stringency of review depends on the importance of the issue and the relative expertise of the reviewing body compared with the expertise of the central bank. In U.S. administrative law, a menu of options exists ranging from no review at all to plenary review with no deference to the initial decision maker.

No Review

In the United States at least, and in most (perhaps all) other countries, there is no judicial review of the central bank’s conduct of monetary policy. The principal reason for the lack of judicial review appears to be an enormous difference in expertise. When it comes to monetary policy, all the relevant functions are better performed by a central bank than by a court. The central bank is clearly a better fact finder when it comes to deciding on monetary policy: a court is not going to be good at gathering economic data, analyzing trends, or engaging in modeling or forecasting. Similarly, the central bank is better than a court in determining the applicable procedures to use in this fact-finding process. A court is not likely to have much information on what sorts of procedures are going to result in the most accurate determination of the facts of monetary policy formation. And a court has little ability to determine whether the proper decision rule has been applied, especially because for most central banks, the decision rule on monetary policy is quite discretionary. Not surprisingly, courts rarely, if ever, upset central banks with respect to the determination or conduct of monetary policy.

This doesn’t mean, however, that the central bank will be completely immune from external review when it comes to monetary policy. Although judicial review is not likely to occur, there is a form of external review through the political process. Directly or indirectly, politicians bring pressure to bear on central bankers when they believe that the policies being implemented are not advisable—or not in accordance with the interests of the politician’s constituents. Here, the independence of the central bank becomes a paramount concern from a legal point of view.

On the one hand, one might suppose that politicians are unlikely to be better able to assess the pros and cons of particular monetary policies than central bankers, and therefore that political review of central bank policies is entirely inappropriate. This would counsel for a central bank completely insulated from the political process. On the other hand, politicians have a certain form of expertise in matters of state. When the interests of a state vary from the interests of the central bank, a form of political review of central bank decisions on issues of monetary policy may be desirable. An example is the case of German unification. It would have been impossible, or at least much more difficult, to unify East and West Germany if the Bundesbank had rigidly insisted on price stability as the sole objective of monetary policy. Despite its formal legal independence, the Bundesbank relaxed its usual commitment to price stability in order to allow the political unification to occur. This was surely the right decision for the bank to have made. As this example illustrates, some degree of political review of central bank decisions in the area of monetary policy may be important.

Outside the monetary policy context, most (but not all) decisions by banking agencies will be subject to some sort of judicial review, although the standard of review may be quite minimal. However, even in the regulatory context, certain agency decisions will be absolutely protected from review. For example, suppose the Federal Reserve Board considers, but ultimately decides against, promulgating a regulation defining certain new activities as being “closely related to banking” and therefore permissible for nonbank subsidiaries of bank-holding companies under the Board’s Regulation Y. It is very likely that this decision will be completely immune from judicial review at the behest of a party aggrieved by the agency’s inaction. By the same token, it is fairly clear that the courts will refuse to review an agency’s decision not to take enforcement action against a party.1

Minimal Review

In other cases, a court will review decisions by a bank regulator, but the review will be deferential. The general model described above suggests that review will be minimal when the central bank has much greater expertise than the reviewer with respect to an issue. Recall the example of the guessing contest with white and black beads. If the Spice Girls as initial decision makers draw a sample of 100 beads, and Yo Yo Ma as reviewer draws a sample of only 10 beads, Yo Yo Ma is going to act “deferentially” toward the Spice Girls because only in unusual cases will Yo Yo Ma’s draw, when combined with the Spice Girls’ draw, result in a decision to reverse a decision based on the Spice Girls’ draw alone.

Perhaps the most common standard for judicial review of administrative action in the United States takes this form. Under the federal Administrative Procedure Act (APA), a reviewing court is instructed to set aside the actions of an agency only if the court concludes that the agency’s decision is “arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with the law.” The arbitrary or capricious standard applies to informal agency action in individual cases as well as informal rulemaking procedures.

A classic case is Camp v. Pitts.2 The Comptroller of the Currency denied a license to open a new national bank. The disappointed applicant petitioned for review from a federal district court that, upon examination of the administrative record, upheld the Comptroller’s decision on the ground that while the court may not have ruled this way on its own, the agency did not act arbitrarily or capriciously. The U.S. Supreme Court eventually sustained the district court’s decision, holding that the court’s job was to review the administrative record, not to conduct any sort of trial-type procedure, and that the agency’s obligation was only to provide an adequate explanation for its actions.

As a practical matter, the arbitrary or capricious standard presents an all-but-insurmountable obstacle to a party who is disappointed by an administrative decision. Because the reviewing court cannot go outside the agency’s own record, the agency has a great deal of control over the material that the court examines in determining whether the agency acted arbitrarily. Although the agency cannot falsify or doctor the record, it can exercise a good deal of control over its contents, and thus can ensure, to a considerable extent, that the reviewing court has before it sufficient facts to justify the agency’s action. Similarly, the agency is required to explain its actions, but its burden is light: it need only persuade the reviewing court that its decision was not irrational. In the hands of a good government lawyer, nearly any agency action can be explained on nonarbitrary grounds.

While the deferential standard of review under the APA might be criticized as effectively promising more than it delivers—suggesting a bona fide judicial scrutiny over agency action while administering a rubber stamp—there is a sound justification in policy for this approach. Given that the agency has far greater expertise in the subject matter and much more involvement with the specific facts, the chance that judicial scrutiny will change the result for the better is small. Thus the courts are encouraged to engage in a very light review and to reverse the agency only when obvious error or bias is displayed.

A more controversial area of deferential review arises under the Chevron doctrine. This rule instructs federal courts not to substitute their judgment for that of the agency, even with respect to pure issues of law, when the issues are within the core area of agency function.3

This doctrine is controversial, and complicated, because two types of expertise are involved. First, courts are presumably better than agencies at interpreting statutes. Judges are in the business of interpreting a wide range of statutes whereas bureaucrats usually have only a few statutes to worry about. Courts, moreover, have a superior sense of the overall framework of the law and a better understanding of how decisions in one area translate over to others. These considerations might suggest that courts should have extensive powers of review of purely legal issues.

However, a competing set of considerations counsel for deferring to an agency even on an issue of statutory interpretation. As the law has become increasingly statute-driven, judicial power over law making (conceptualized in the English-speaking world as the “common law”) has faded. Agency expertise in legal interpretation has increased. Meanwhile, the issues dealt with by administrative agencies have become more complex, so that the interpretation of the agency’s governing statutes are likely to require an understanding not only of a complex statutory scheme, but also a grounding in various technical and scientific issues. Finally, the courts have recognized that when Congress delegates key decisions to an administrative agency, it probably intends that the agency have power to interpret its governing statutes to take account of all sorts of changing circumstances, including changing political realities. These factors suggest that courts should not, in fact, exercise plenary review over agency interpretations of their own governing law.

The Chevron case confirmed the primacy of the latter view by instructing federal courts to defer to legal interpretations made by agencies acting within the scope of their statutory powers. An example is Independent Insurance Agents v. Board of Governors.4 The issue was whether the Federal Reserve Board had authority, under the Bank Holding Act, to regulate the activities of banks that were held in holding company form, or whether this was a matter relegated to the state and federal chartering agencies. The Board argued that it lacked such authority. The reviewing court found that the statutory language was ambiguous and accordingly deferred to the Board’s interpretation, even though it later acknowledged that it would also have upheld the Board had it reached the opposite conclusion.

The Chevron rule is not absolute, however. If the statute is unambiguous—if Congress has “directly spoken to the precise question at issue”—the courts will not defer to the agency’s interpretation. Nor will the courts defer if the agency’s position, while reasonable with respect to the particular statutory provision under review, cannot be reconciled with positions the agency has taken elsewhere. A sequel to the case just discussed illustrates the latter point. In Citicorp v. Board of Governors,5 the Board argued that, while it did not have power to regulate the activities of bank subsidiaries of bank-holding companies, it did have authority to regulate the activities of subsidiaries of bank subsidiaries. The appeals court rejected this generation-skipping approach, concluding that the Board’s interpretation of the statute was irrational given the arguments the Board itself had advanced (and the court had accepted) in the earlier case.

Mid-Level Review: The Substantial Evidence Test

If the agency has engaged in a formal trial-type proceeding, review is typically more demanding than under the arbitrary or capricious standard. For example, if a banking agency concludes that a depository institution or an affiliated party is engaging in unsafe or unsound practices, or is violating a law, rule, or regulation, the agency may commence a “cease and desist” proceeding against the alleged wrongdoer by serving a notice of charges.6 The hearing is to be conducted under the provisions of the APA.7 The APA gives the accused person the right to receive notice of the offenses with which he or she is charged and the penalty sought by the agency, to present oral and documentary evidence, to submit rebuttal evidence, and to conduct cross-examination—in other words, the substantial equivalent of a U.S. trial procedure. The agency’s decision must generally be based on consideration of the whole record and must be supported by “reliable, probative, and substantial evidence.”8

Judicial review of this sort of on-the-record trial proceeding is conducted under the “substantial evidence” test, under which a court will set aside the agency’s decision if not supported by substantial evidence contained in the record of the agency hearing.9 The scope of substantial evidence review is not completely clear-cut. U.S. Supreme Justice Antonin Scalia, in a banking law decision written while he was still an appeals court judge, declared that there is effectively no difference between the substantial evidence and arbitrary or capricious standards.10 Notwithstanding this opinion, the substantial evidence test, in the view of most observers, is more stringent than the arbitrary or capricious standard, representing a middle level of judicial review.

There are good reasons why judicial review of agency adjudications should be more searching than judicial review of informal agency action. First, the interests involved in agency adjudications are typically very large for the persons involved. In a cease and desist proceeding, for example, the person charged may be prevented from pursuing his or her business plans and may be subject to costly penalties. Because the potential costs of an erroneous decision to the individual are high, there is a stronger argument for meaningful judicial review. A second reason for stricter judicial review of agency adjudications is that courts traditionally oversee trial-type procedures and, presumably, have a high degree of competence in that task. Thus there is a reason to give them a substantive role in the decision process based on their expertise.

De Novo Review

In rare cases the courts may review the decisions of banking agencies under a “de novo” standard. This approach allows the courts essentially to ignore the decision of the agency and to review the matter as tabula rasa. The classic case for de novo review is that of statutory interpretation in cases where Chevron deference is not warranted. For example, a banking agency may have occasion to interpret a statute outside the banking field (for example, a state insurance law). In such a case, the reviewing court is unlikely to defer to the banking agency’s expertise. Instead, the judge will probably follow the dictates of the APA, which declares that the reviewing court shall “decide all relevant questions of law.”11

Although not ordinarily conceptualized as such, another form of de novo review occurs when a party sues a federal banking agency for violation of personal rights under a theory of contract or tort. In general, it is difficult for individuals to convince courts to take on such cases, since the government is likely to claim a defense of sovereign immunity from suit. However, the government’s sovereign immunity is not absolute, having been partially waived by Congress in the Federal Tort Claims Act (for torts) and the Tucker Act (for contracts). Where the government has consented to suit, private parties may be able to obtain a remedy.

Implicitly, judicial review of agency action in the context of a tort or contract suit against a federal agency will be de novo. The party seeking redress from the government has only the ordinary burdens of proof that all plaintiffs face in obtaining judicial relief.

An example of private litigation against banking agencies in the tort context is United States v. Gaubert.12 Owners of a failed savings and loan institution sued the (now-defunct) Federal Home Loan Bank Board, claiming that it had negligently supervised the affairs of the institution and had thereby caused its demise. The Supreme Court threw the suit out, concluding that the government’s actions were a discretionary function as to which the United States had not waived its sovereign immunity from suit under the Federal Tort Claims Act. If the Court had allowed the suit to go forward, however, it is likely that the litigation would have been conducted on what amounts to de novo judicial scrutiny of the agency’s actions.

Private parties were more successful in the contract setting in an important recent case, United States v. Winstar Corporation.13 During the darkest days of the U.S. savings and loan crisis, the Federal Savings and Loan Insurance Corporation (FSLIC), which insured deposits in thrift institutions, tried to attract new capital to the industry by means of supervisory mergers in which well-capitalized institutions took over failing ones. Part of the inducement for such mergers was the use of creative accounting practices that gave the acquiring institutions significant economic advantages. In the absence of this favorable accounting treatment, most of these mergers would not have taken place. To safeguard the deals, the acquiring institutions obtained commitments from the FSLIC to continue to allow the favorable accounting treatment. However, Congress thereafter abrogated the contracts and obligated the Office of Thrift Supervision (OTS), the new savings and loan regulator, to eliminate the favorable accounting treatment—with disastrous consequences for many institutions acquired in supervisory mergers. The Supreme Court, by a divided vote and without a majority opinion, held that the government had breached its contractual obligations and was liable for damages to the complaining parties—a decision worth billions of dollars to the affected institutions. The point for present purposes is that the Court never considered the possibility of “deferring” to the agency when it evaluated the government’s liability; the only issue was whether the government could assert certain defenses that would be unavailable to a private contracting party. Effectively, it reviewed the agency’s actions under the de novo standard.

Timing of Review

Distinct issues arise in connection with the timing of judicial review of central bank and banking agency actions.

The normal rule is that review occurs only after administrative procedures are final—a principle embodied, in U.S. law, under the rubrics of “ripeness” and “exhaustion of administrative remedies.” Ripeness and exhaustion are similar doctrines; the former refers to the condition of the issues and the latter to the stage within the hierarchy of the agency at which the issue is pending.

Here, the trade-off, in terms of public policy, is between the costs of premature review (the external reviewer may intervene without sufficient information, and before the agency has had the chance to correct matters in-house) and the harm that the party seeking review will suffer from waiting. Suppose, for example, that the party seeking review is going to be required to comply with a rule that the party deems invalid, in a situation where the party will suffer irremediable harm if it does comply. Here, the harm to the party seeking review may outweigh the benefits of waiting until the decision process within the agency is completed. In such a case, the courts may allow review to go forward even though the agency has not completed its internal decision processes.14

Ripeness and exhaustion of administrative remedies deal with situations where a private party wants the external reviewer to step in before the agency’s decision is final. In other settings, a private party may complain that the agency has moved too fast rather than too slowly. Suppose the agency believes that an emergency situation exists requiring immediate action to ward off serious harm to the public. The agency then has an incentive to take urgent action prior to an opportunity for full external review of its action. However, in acting, the agency may impose harm on private parties, who may find their interests dramatically affected before they have had the chance to obtain external review.

The general rule under U.S. law is that there must be a hearing before private property can be taken. However, this rule may be relaxed in an emergency, provided that the law offers an opportunity for prompt and fair judicial review of the agency’s action after the fact. The trade-off is between the harm to the public that may occur if the agency does not act, on the one hand, and the harm to the individual that may occur if the agency does act, on the other. The problem is exacerbated by the high chance for error on both sides. Because the situation is almost by definition an emergency, there is inevitably a fairly high probability of error simply because the agency has not had the chance for due deliberation.

The matter can be handled without excessive disruption if the agency’s action does not create a risk of irremediable harm to the private party. If the party can be made whole after the fact in the event that the agency is proved wrong, the agency can act without obtaining prior judicial authorization or review, and the status quo can be restored after the fact. But often the party against whom the action is taken cannot be made whole, or must pay a high price in order to obtain redress. Here, the problem becomes particularly poignant because of the grave risks that someone will suffer harm.

In the banking context, the law recognizes that agencies may need to act quickly without a full-scale prior hearing because of the special features of a banking institution. There is a danger that a dishonest banker may abscond with the assets or substitute bad assets for good ones. Because the assets of a banking institution are principally financial claims, the banker can engage in this behavior without being publicly observed, and can do so very rapidly. Thus the banking regulators need to be able to intervene quickly to prevent losses to depositors and to the deposit insurance funds.

Even more important, the law recognizes that banking depends on public confidence. The courts and banking agencies perceive a danger that the public will lose confidence in a bank if it gets wind of trouble and may run the institution. No bank has sufficient assets on hand to survive a run. Even worse, if one bank fails after a run, the public may lose confidence in the banking system as a whole, sparking a generalized banking panic.

Because of these concerns, the banking laws allow regulators to take a fairly wide range of actions without a prior judicial hearing. Most important, the agencies may close a bank without notice and hearing. In the United States—at least in the days when banks were failing—the regulators often would shut a bank on Friday at the close of business without notice or prior hearing, cause it to be sold over the weekend, and allow it to open under new ownership the following Monday. These emergency closures often had the tenor of a police raid, with bank regulators fanning out around the bank’s home office and its branches with synchronized watches and walkie-talkies. The purpose of all this haste was to minimize disruption in financial transactions and avoid loss of public confidence in the banking system.

In addition to closing banks, bank regulators enjoy a fairly wide range of other powers that can be used against private parties on an emergency basis without prior notice and hearing. For example, the agency may issue a temporary cease and desist order against a depository institution or affiliated party without obtaining prior judicial approval. It is then incumbent on the affected party to seek a judicial order setting aside the agency’s order.15 Similarly, the agencies may, without notice or hearing, issue an order suspending an officer, director, or affiliated party of a depository institution from participating in its affairs.16 Again, the affected party has a right to obtain prompt judicial review of the agency’s action.

Despite these assurances of prompt review after the fact, it may be doubted that the complaining party will be made whole, even if the agency’s decision is erroneous. Often, the actions that the agency has taken in the meantime—most notably, the sale of a bank to a third party—are difficult or impossible to unwind. Although it may seem unfair to impose such losses on innocent persons, the risk of this type of harm is, perhaps, one of the burdens that people take on when they elect to work in the banking industry.


This paper has covered some, but by no means all, of the issues that arise in connection with procedures for external review of actions by central banks in their capacities as custodians of monetary policy and supervisors of banks and the banking system. In general, the goal of external review is to improve the quality of decisions, consistent with keeping the costs of such review within reasonable bounds. Technically, the object is to design a system of external review such that the marginal benefits to be obtained from the process exactly equal the marginal costs. This kind of technical precision is impossible in real life, but one can quite readily discern an implicit cost-benefit calculus in the various and complex procedures and standards we observe in the real world. This paper has focused on the United States, and it is quite possible that procedures for external review under other systems will differ in substance and detail. However, it is equally likely that at least some efforts to balance the costs and benefits of external review occur regardless of the jurisdiction in which the procedure occurs.



It is difficult to comment on the presentations made this afternoon because they were so thorough, but I would like to add some further thoughts.

Judicial Review: Deference to Agency Interpretations

With respect to the comments on the external review of decisions by central banks, I thought I would try to expand on the points raised regarding judicial deference. As mentioned, the seminal ruling on deference to U.S. administrative agency decisions is the U.S. Supreme Court Chevron decision.1 In Chevron, the Supreme Court established a two-part test to determine if an agency is entitled to deference. First, has the legislature spoken to the precise issue? More particularly, does the legislative language directly address the question at hand? If so, then that ends the matter because agencies are obliged to operate within the boundaries set by Congress.

If a court concludes that the legislature has not addressed the precise issue and there is a gap in the statute—whether due to silence on the point or ambiguity—then the agency is permitted to fill the gap. The agency can fill the gap, however, only so long as the agency’s interpretation is “reasonable” and is based on a “permissible construction” of the statute. An agency’s interpretation is to be given controlling weight unless it is “arbitrary, capricious or contrary to the statute.” Under these circumstances, a court is not permitted to impose its own construction of the language; instead, the court is to uphold the agency’s interpretation.

In my view this concept of deference is a logical result and is good public policy. A legislative body does not have the expertise to foresee and address every possible issue—particularly in an area as complex and technical as bank regulation and financial markets. Agencies have the technical knowledge and expertise in these areas to address problems and develop policy as issues arise. More important, financial markets and financial institutions are continually changing and evolving. Banking agencies need to be free to interpret the law in light of new information.

The concept of deference also needs to permit an agency the flexibility to change its views over time and to alter its policy in light of significant events and developments. The U.S. courts have recognized such shifts in agency policy and still granted deference. The greatest deference is given where an agency has a longstanding interpretation of a statute—particularly where the interpretation is contemporaneous with the enactment of the law. However, although courts may view shifts in policy or interpretation with some skepticism, deference is granted where an agency shows the shift is due to an application of its expertise and the agency provides a reasoned deliberation for its shift.

Beyond the good policy reasons for deference, the U.S. courts also have recognized political reasons for deference. The U.S. federal judiciary is not elected and federal judges hold lifetime positions, removable only for cause. As a consequence, the courts have recognized that where the elected legislature has left gaps in the law, the executive branch agencies rather than the judiciary should fill the gaps. The agencies, and not the judiciary, are at least accountable to the public through the president.

The concept of deference does have limits. First, the courts generally will not give deference to an agency interpretation of its jurisdiction. In these situations the issue is less one of expertise. An agency may have a built-in bias and is likely to err on the side of expanding its authority rather than limiting it.

A second, and more important, limit is set when the legislature has specifically addressed an issue in the statute; then the agency must follow the language even if in the agency’s expertise this causes anomalous results or inequities. The seminal decision in this area was the decision in the Dimension case.2 There the U.S. Supreme Court stated the Federal Reserve Board could not use its rule making authority to correct what the Federal Reserve Board believed to be a flaw in a statute where the statutory language was clear. If the result of applying the statute was wrong, then it was up to the Federal Reserve Board to go to Congress to change the legislative language.

Although I am a believer in the appropriateness of deference, I am obviously biased since I work for an administrative agency. The concept of deference has been criticized by many as too simple and too limited. Deciding deference by merely referencing the specificity of a statute could ignore other factors that should influence a court’s ultimate decision. Others believe deference is an abdication by the courts of their responsibility. In the final analysis, however, I believe it represents good public policy and ensures the most reasonable approach to resolving difficult policy issues.

Liability and Immunity of Central Banks

With respect to the comments on the immunity of central banks and the Foreign Sovereign Immunities Act, I thought I would try to supplement them by summarizing U.S. law regarding sovereign immunity as it pertains to the Federal Reserve Board—as well as other federal agencies.

U.S. law is based on English common law and the concept of sovereign immunity applies in the United States. Therefore, the United States and its agencies, including the Federal Reserve Board, are immune from liability, except where that immunity has been waived. The federal government has waived its immunity from certain types of actions in tort or for personal injury through passage of the Federal Tort Claims Act.3 Under that statute, damage actions may be brought against the United States “for money damages, injury, or loss of property or personal injury or death caused by the negligent or wrongful act or omission of a Government employee while acting within the scope of [his] employment….” The statute sets forth a number of exceptions to this waiver of sovereign immunity. Probably most important from the standpoint of the federal banking regulators are claims based upon the performance of a discretionary function. The remedies provided by the statute are exclusive and if a claim does not fall precisely within the terms of the statute’s limited waiver, then the waiver is not permitted and pursuant to the statute the court hearing the case is deprived of subject matter jurisdiction.

In essence, the Federal Tort Claims Act waives sovereign immunity of the United States to claims for negligence to the same extent that a private person would be liable in identical circumstances. For example, if a government employee drives an automobile on a mission connected with her official duties and negligently causes personal injury, the United States may be liable for accidents caused by that driver in the scope of her duties. But the discretionary function exception prohibits judicial second-guessing of legislative and administrative decisions grounded on social, economic, and political policy. The exclusion protects government actions and decisions based on public policy.

Probably the best example of discretionary function in a banking context is the U.S. Supreme Court decision in United States v. Gaubert.4 In the Gaubert case, a federal banking agency was alleged to have assumed day-to-day control of a financial institution. The plaintiff claimed that the agency’s actions involved running the institution and were outside the scope of the discretionary function exception. The Court held, however, that even informal regulatory actions which take place at the “operational” rather than policy level are protected as long as policy-based discretion is exercised and no statute or regulation prohibits the use of that discretion. In particular, the Court said:

A discretionary act is one that involves choices or judgment; there is nothing in that description that refers exclusively to policymaking or planning functions. Day-to-day management of banking affairs, like the management of other businesses, regularly requires judgment as to which of a range of permissible courses is the wisest. Discretionary conduct is not confined to the policy or planning level.5

Similarly, in Franklin Savings v. United States,6 a failed financial institution and its holding company sued the United States alleging that federal regulators had negligently supervised the institution’s officers and directors, as well as the institution’s day-to-day operations, when it was placed into conservatorship. The court held the discretionary function exception to the Federal Tort Claims Act barred any cause of action for money damages alleged to stem from the agency’s actions. The court stated:

The objectives of governmental agencies overseeing [financial institutions] are often in direct conflict with the officers and directors of financial institutions. While private officials maintain a loyalty and fiduciary duty to shareholders, federal regulators owe their allegiance to depositors and the general public. If private financial institutions could sue regulatory agencies for negligently performing discretionary functions, the ability of those agencies to act in the public’s best interest would be compromised. Sanctioning such suits also would put courts in the difficult, if not impossible, position of judging the propriety of the policymaking acts of [the Executive] branch.7

The court in Franklin Savings held the agency’s actions were related directly to public policy considerations regarding oversight of the financial institution industry; therefore, the discretionary function test was met and the government was insulated from liability.

Besides tort liability, the United States has waived sovereign immunity in contract as well. Some agencies, such as the Federal Deposit Insurance Corporation, have specific “sue and be sued” clauses, which permit them to be sued for money damages in contract. Other federal agencies without such authority may be sued in contract by suit against the United States under a law known as the Tucker Act.

Finally, in some narrow areas the United States legislature has determined for public policy reasons to provide direct remedies for agency actions and has created limited liability. For example, the Privacy Act protects certain information held on individuals by Federal agencies.8 The Act provides for damages from an agency if it “intentionally or willfully” discloses personal information protected by the statute. Similarly, the Equal Access to Justice Act provides that a prevailing party in an action with a Federal agency may be entitled to the recovery of attorney fees and other expenses unless the agency can demonstrate its position in the action was “substantially justified.”9


Looking at the issue of judicial deference to agency decisions and the exclusion from liability for discretionary acts reveals a general tenet of U.S. law: to avoid the review or judicial replacement of agency actions in areas of general policymaking. It provides a fine balance between the court’s review of the proper application of law while avoiding the courts involving themselves in policymaking.


See Heckler v. Cheney, 470 U.S. 821 (1985).


411 U.S. 138 (1973).


See Chevron v. Natural Resources Defense Council, Inc., 467 U.S. 837 (1984).


890 F.2d 1275 (2d Cir. 1989), cert. denied, 498 U.S. 810 (1990).


936 F.2d 66 (2d Cir. 1991).


See 12 U.S.C. § 1818(b) (1994).


5 U.S.C. §§ 554–57(1994).


5 U.S.C. § 556(d) (1994).


5 U.S.C. § 706(2)(E) (1994).


Association of Data Processing Service Organizations, Inc. v. Board of Governors, 745 F.2d 677 (D.C. Cir. 1984).


5 U.S.C. § 706(1994).


499 U.S. 315 (1991).


518 U.S. 839(1996).


See Abbott Laboratories v. Gardner, 387 U.S. 158 (1967).


See 12 U.S.C. § 1818(c) (1994).


See 12 U.S.C. § 1818(e)(3) (1994).


Chevron U.S.A., Inc. v. Natural Resources Defense Council, Inc., 467 U.S. 837 (1984).


Board of Governors of the Federal Reserve System v. Dimension Financial Corp., 474 U.S 361 (1986).


28 U.S.C §§ 1346(b), 2671 et seq. (1994).


499 U.S. 315 (1991).


Id. at 325.


970 F. Supp. 855 (D. Kansas 1997).


Id. at 866–867.


5 U.S.C. § 552a (1994).


5 U.S.C. § 504 (1994).