Banking on judgement: a pragmatic approach to supervision

17th September 2026 

Key takeaways 

•   In the current debates on prudential regulation and supervision there is a call to simplify regulation and focus more supervision on the fundamental points for financial stability. This implies a greater use of supervisory judgement, but for some, this requires a global approach to externally “discipline” supervisory discretion.

•   As the key role of supervisors judgement is to be able to effectively review the judgement of the bankers on risk, it is essential for its effectiveness to preserve a robust discretion.

•   To better ensure the legitimacy of this discretion, the supervisors themselves develop ex ante internal guardrails for ensuring the consistency and adequacy of judgement, and that they increase significantly their transparency on both the use of these guardrails and on the monitoring of the results.

1. The role of supervisory judgement

This Supervisory Forum began with a very interesting conversation on the use of judgement in supervision¹, an issue that deserves further thoughts in the context of or the current debates on prudential banking regulation and supervision. On both sides of the Atlantic we have indeed seen intense discussions - linked to the finalization of the Basel III framework - about the need to ensure that an undue regulatory/supervisory burden does not hamper the capacity of banks to service the economy. 

Supervisors have underlined, also in this Forum², that there is no factual evidence that regulatory capital requirements hamper the capacity of the banks to lend, and that broad discretion is a condition for effective supervision. They do however share³ the willingness to improve its efficiency: we thus have on the table different proposals for simplification of the regulatory framework and for better focusing supervisory practices on the points that are fundamental for financial stability. On both accounts supervisory judgement would have to play an enhanced role.

Eliminating excessively detailed rules to simplify regulation entails indeed a greater role for supervisory judgement⁴ to interpret how the more principles/outcome oriented rules apply to the specific cases. In addition, the focusing of supervisory practices also implies an increased role of judgement about how to prioritize the tasks.

On the other hand, we know from the “rules versus discretion” discussions since the days of the original Henry Simmons/Chicago School debate of the 1930s about monetary policy that more discretion increases the risk for arbitrariness. These risks can come from undue external influence, from industry («regulatory capture») or from government («short-sighted political bias»). But also internally, the judgement of supervisors could be subject to inconsistencies or mandate-overstepping tendencies. It is therefore not surprising to see industry calls to have a global approach to externally “discipline” supervisory judgement and discretion. We argue however that given the particular role of supervisory judgement is to review, from the external point of view of the general interest, the internal judgement of the bank on uncertain futures, it is necessary to be careful to preserve the margins of discretion that are necessary for both these judgements to be effective.

Indeed, the banking supervisor supervises an activity, banking, which in itself is essentially based on judgement, that is on conclusions that are not deducted from known firm premises but rather built on previous practical experience - hopefully by people with a good track record at it. This is obvious for all traditional banking up to the mastering of probability modelling in the twentieth century, as there were no ways other than lessons learned from experience to have an educated judgement on the likelihood of a debtor not paying its debts or of a depositor’s stampede. Later, the development of probabilistic modelling offered opportunities for a much more sophisticated approach to risk, and these models were indeed recognized in the late 20th century regulation. But the great financial crisis led to the painful realisation that irreducible uncertainty remains in banking, and that ignoring it was a recipe for disaster. An important lesson of the crisis was the importance of having sophisticated risk measurement tools controlled by banking judgement, not replacing it. Having sufficiently diverse, reactive and challenging risk management practices and prompt governance responses has been confirmed as a key factor for banking resilience. Supervisors should thus use their own judgement to promote more, not less use of qualitative professional judgement by the bankers, but a more prudent and robust one.

To do this requires a very specific balancing act, involving not only technical knowledge and experience but also behavioural and cultural norms. However, we think there is a general agreement on two principles: the supervisor cannot be a part of the internal process of decision of the bank and must have the powers needed to effectively object to unsound judgement. This second point will lead to our main conclusion: that external review should not render too fragile this capacity to object.

 i) Non-interference 

As supervisors’ mission is to review the judgements of bankers, the first point - that they cannot be a part of the internal process by which the responsible bodies of a bank take decisions - is only the application of the impossibility for a reviewer to take part in the reviewed activity. It should apply to any form of participation, from the weakest (providing advice) to the strongest (co-decision or a fortiori ex ante dictating them).

This elementary and undisputed principle, applicable also for instance to external auditors, has however in supervision some delicate borderline aspects, linked to the nature of the relationship. The supervisor has indeed also a duty of transparency which includes explaining how a regulatory/ supervisory principle applies to a specific situation. When discussing whether this or that measure envisaged by the bank will comply or not with a requirement, typically a supervisor might refer to a positive example of compliance or issue a negative warning about some proposals of a bank. This early interaction is in fact a good practice that should be fostered, not hindered. So, the supervisor should always take extreme care in making clear that it targets prudential outcomes, but that the choice of the means is the exclusive responsibility of the bank’s judgement.

ii) The power to object

The second point relates to the crucial issue of what supervisors can or cannot do. Historically the basic powers of the supervisors have been 1)investigation and 2)withdrawal of the authorization to exercise banking. Experience has abundantly shown that this life-or-death power of withdrawal is not suitable for tasks of early intervention, and so supervisors have now much more ample powers of authorization or direction, and also on going and point-in-time unfettered access to information.

These wide powers of action, embedded in the first of the Basel core principles for banking supervision, are a commonly accepted standard, but what remains under discussion is how these powers should be used in relation to the banker’s own use of judgement. Our stance is that they are not meant to replace it, but only to correct the points where a deficiency has been assessed in relation to the supervisory objective of safe and sound risk management and governance. It is thus important to insist in their ultimate and negative nature.

Ultimate, because as the objective of supervision is to induce the banks into action, its best outcome is to make the banks understand the reasons of the supervisory findings and requests for remediation, what is usually called “moral suasion”. This ensures that the banks themselves embed the adequate level of prudence in their own practices. Use of binding powers is just the ultima ratio⁵ when explanations do not work.

Negative, because at the end of the day, a supervisor can just oppose what the bank is doing or intends to do but cannot do it itself. It can limit the authorization/capacity of the bank to act, but not direct positively its business. In the extreme cases of crisis/near default it can replace the direction of the bank, but even then it does not exercise itself the management of the bank⁶.

The use of such binding powers should be accompanied by due process obligations: explaining the content of the measure, stating its reasons and providing an adequate right to be heard before a binding decision is made. These obligations are in fact just the formalized “ultimate” version of the basic supervisory practice of explaining and discussing. The binding nature of these powers implies also that their use should be challengeable. But to what extent, this is the main point under discussion. 

2. Challenging supervisory judgement

The capacity to challenge supervisory decisions is in itself a welcomed part of the Rule of Law. But differences of view between the industry and supervisors arise on the desirable extent of this challenge. Industry has expressed its concerns that in judgemental areas the supervisory discretion should be more disciplined, to ensure predictability and avoid arbitrariness. The supervisory judgement should be effectively scrutable and challengeable, on its evidential base but also on the clarity of its legal base and the proportionality of its application. It is sometimes argued that supervisors’ judgement on banks should be subject to the same rules and scrutiny than those applied to the banks by supervisors, to ensure that the guardians are well guarded, a kind of a modern instance of the old worry from the Roman author Juvenal⁷. But this parallel does not take sufficiently into account the difference between the two levels of judgement and thus risks jeopardizing the effectiveness of supervision. 

Our main objection is that in fields where the concrete balancing of facts and potential risks is difficult we cannot afford the luxury of a fragile, lengthy and cumbersome decision-making process scrutiny. It is of course absolutely necessary to check that a supervisor is not manifestly misusing its powers. But on the substance of the assessment of whether some situations and practices are not prudent enough from the point of view of preventing external damages, if a higher instance than the supervisors can replace completely its assessment, then this higher instance becomes the real supervisor, the one taking the ultimately binding decisions. And that “second-level supervisor” by construction will intervene with more procedural delays and less concrete experience of the field than the primary supervisor. Hardly a recipe for efficient supervision.

This is why, within different legal traditions (US, UK, EU Law) we have seen the higher courts in their review of prudential supervisory decisions focus on the respect of the due procedures and on the absence of any manifest error of appreciation, but leaving a large room for the discretion of the supervisors’ judgement on the substance of the risk issues. Moving away from this remarkable convergence would be extremely unwise and contrary to the push for greater supervisory efficiency. But this does not mean that we should not strive to enhance the quality of supervisory judgement.

3. Enhancing supervisory judgement

To preserve the legitimacy of a large amount of supervisory discretion, supervisors should be seen as trying themselves to enhance the quality of their judgement and to allow its stakeholders to meaningfully assess it. This could be done by combining internal guardrails in the internal organization of the supervisors, with external guardrails that will focus on the quality of the procedures and organizational structures and the effective ex post delivery of the objectives.

On the internal guardrails of the supervisor the parallelism with the situation of the banks is in fact more pertinent than for the legal review, when adapted to its specificities. To manage the risks involved in any kind of judgement, it is necessary to frame and discuss ex ante how to deal with them – that is identifying the risks, specifying ex ante the criteria for accepting or limiting them, and defining how to review them ex post. The clear difference with a bank risk appetite framework is that risks are not balanced versus income, but versus the delivery of the public policy objectives of supervision. Pure quantitative limitations are thus not such a relevant tool, while an essential role is played by the strategic operational orientations of the supervisor. Within this background, however, we do not find any obvious reason not to apply approaches equivalent to those foreseen by international banking standards, with first, second and third internal lines of review. These reviews should ensure ex ante the consistency of the use of judgement with the mandate and strategy of the supervisors.

On substance, as a supervisor may overrule the judgement of the bank, it is very important to clearly relate such overruling to a deficiency towards the standards of additional prudence. This is why we deem important to have as much as possible ex ante transparency on the process that can lead to this, what are the expectations of the supervisor, its standards – which most commonly are best practices gathered from practical experience - and the motivation that guides its action. This effort to provide ex ante transparency can only facilitate the ex post explanation of individual decisions.

These general transparency approach should be complemented by accountability on the delivery of the objectives to the policy makers, which is within the democratic accountability framework where the supervisors explain their policies and conduct against the objectives they have been assigned. In these areas in which judgement plays such a considerable and specific role, it is particularly important to allow public understanding of outcomes that are not necessarily mechanistic. To that purpose it would be useful to make a deeper use of what I propose to name “mixed intervenors”.

With this term of ‘intervenor’⁸ we mean reviewers of the supervisors which cannot supersede the judgement of the supervisor on individual cases – they are not shadow judges – but are able nevertheless to meaningfully examine its track record and express ex post a judgement on the supervisory arrangements. There are in fact already quite a number of this kind of reviewers, starting with the International Monetary Fund examination of supervision in the framework of its Financial Stability Assessment Program, but also some Auditors of Public Institutions, from time to time parliamentary enquiries, or experts reports commissioned by the public authorities, typically after a crisis. There are also reports commissioned or performed internally by the supervisors themselves. As these reports serve different purposes and are made by different institutions with very different resources, they are of different value and depth, and essentially uncoordinated. So, the point is certainly not to encourage the proliferation of more uncoordinated reports, but to reflect on how there could be a better use of them by ensuring their coordination and relevance.

While a thorough examination of this question would exceed the limits of this contribution, I would like to conclude by highlighting some points particularly relevant for our discussion of supervisory judgement.

First, as sound judgement comes from experience, this review of the way supervisors apply judgement on banks should be carried by people having relevant experiences⁹, ideally diverse ones. This means concrete supervisory experience (from fellow or former supervisors), but also from the other kind of reviews already mentioned, public auditors or international experts, and depending on the field the review, of relevant external experts.

Second, the outcome of these reviews should be made public in a meaningful way, so that not only supervised entities or other reviewers can take them into account, but the public at large, especially for issues where democratic accountability is so important.

And last but not least, the most practical way to get this result is that supervisors themselves take this kind of initiatives. This will give them additional pragmatic instruments to improve their strategies and organization and be a clear proof that they are not trying to be defensive and escape public scrutiny.

Summing up, in the present context, it is essential to avoid creating legal fragilities for the use of supervisory judgement and thus inhibiting the challenge of the banker’s judgement. To enhance the use of supervisory judgement, supervisors should be more proactive in their use of internal framing mechanisms. They should also rely more, and in more transparent ways, on internal and external reviews centred of how the use of judgement contributed to the delivery of their strategic objectives. 

It is indeed our firm belief that it is not with more defiance but with more explicit framing and transparent assessments of their practice of judgement that supervisors will increase their effectiveness.

Endnotes

1. Cf. “Supervisory Judgement – A fresh perspective?” 26th June 2026by Sam Woods, in conversation with Sir Andrew Likierman.
2. Claudia Buch, “Banks and competitiveness: promoting competition, protecting resilience, 27th February 2026”.
3. There has been a number of speeches from supervisors and also private sector initiatives around these issues. See in this blog “Moving the Dial: Fostering a culture of Supervisory Effectiveness, 9th October 2025 “by Andrea Enria and recently the speech by Pablo Hernandez de Cos, 22th May 2026 “the Quest for supervisory effectiveness” at www.bis.org.
4. Unless we adopt some very drastic rules simplifying the framework, like the Mervyn King proposal of a “Pawnbroker for all seasons” put forward in 2016. Unsurprisingly though, ten years after, this approach has not attracted many supporters: industry and governments want to have more banking activity, not less.
5. Ultimate of course does not mean late, the pace to reach this ultimate step is itself a matter for supervisor’s judgement and should be flexible for urgencies.
6. This principle is applicable to a supervisory function, a resolution authority may have direct powers to act.
7. Addressed in the Nobel lecture of Leonid Hurwicz, December 8, 2007, “But who will guard the guardians?” available online https://www.nobelprize.org/uploads/2018/06/hurwicz_lecture.pdf. Hurwicz examines the question from a very general point of view of economics and games theory. The present reflections are very freely inspired by his three level response to the implicit Juvenal’s pessimism that it is not possible to guard the guardians as this will lead to an infinite regression (the guardians of the guardians will need to also have guardians and so forth). 
8. Borrowed from Hurwicz, with a different sense however, to designate the reviewers of the guardians that are not themselves second level guardians, to avoid the infinite regression and efficiency issues mentioned in footnote 2.
9. But, it goes without saying, no conflict of interest. 

To comment, please go to the discussion forum