A Starling Insights Deeper Dive Report

Supervisors on Supervision

— Chapter One Preamble —

The success or failure of financial sector supervision does not depend on the care and prescience with which rules are formulated. No rule set can be complete, and the future is always radically uncertain. The success or failure of a regulatory model depends on the culture which emerges — both in regulated firms and in regulatory agencies.

Culture is not sentiment or garnish. Organisational culture is not what we celebrate on away-days and at karaoke parties. And organisational culture is certainly not defined by the self-congratulatory public relations statements which, it seems, almost all firms now distribute to new clients and new employees. But culture is, admittedly, about what many term ‘soft stuff’, rather than the ‘hard’ factors typically parameterised in the mathematical models that abound in finance. 

Culture defines and describes the attitudes and expectations through which employees relate to each other and to their clients. Anyone who does not understand how the nature of such attitudes and expectations is fundamental to the delivery of financial services understands nothing about the delivery of financial services. Culture is not a peripheral concern. It is central to both institutional trustworthiness and to supervisory effectiveness. And culture must therefore be treated with the kind of evidentiary discipline and intellectual seriousness that this report seeks to advance.

What We Talk About When We Talk About Culture

Economics has tended to treat firms as ‘black boxes’ or, at best, as contractual assemblages — nexuses of transactions governed by incentives and compliance. But this is not how successful businesses operate. Nor is it how supervisory bodies function when they are at their best. 

A central thesis of my work is that the transactional account of the firm is not just inadequate, but actively misleading. The ubiquitous models in finance that are based on this transactional fiction fail to explain why some firms thrive while others falter, why governance structures that look identical on paper produce radically different outcomes in practice, and why capital adequacy alone cannot guarantee institutional survival. 

The modern firm is not a machine, a list of equations, a bundle of incentives, or a set of contracts. The modern firm is a community — of people, capabilities, habits, and beliefs — structured by history and animated by shared understandings.

A footballer is in possession. The ball is in his hands or at his feet. Does he try to score, or pass to a possibly better placed player? The successful graduate of Economics 101 will explain that the appropriate action depends on contracts and incentives.

Do the coach’s pre-game instructions require him to pass, in a situation which the experienced coach had been able to describe in general though not specific terms?¹

The well-educated footballer will reflect on how his compensation package is structured. What proportion is based on his own scores, and what proportion is calculated by reference to the aggregate scores of the team? An alpha student may recognise that this dilemma has the character of a ‘repeated game’. Will the fortunate recipient of a pass have an opportunity to reciprocate the benefit? The analysis of repeated games allows for the possibility of multiple equilibria. There may be teams characterised by passing games, and others in dynamic equilibria better described as shooting games. And the best outcome for the individual in a shooting game may be to ‘cheat’ on a team whose other members have converged on passing game norms.

These analogistic models are not as ridiculous as they may seem to many practitioners. It is helpful, in this and in other areas of economics, to understand real world problems by focusing on simplified versions of them. The ‘Prisoners’ Dilemma’ is not a description of police interrogation tactics, but an insightful illustration of how individually rational actions may yield less than optimal outcomes. 

The mistake is to believe that these models describe the ‘real’ world, in the way many physical models do, and that they lead directly to recommendations for action, as many physical models do. No one (I hope!) really believes that the well-educated footballer is applying these classroom models mentally in the split second in which he is allowed to make an action decision. Not even — or perhaps still less — the players in the teams of elite colleges. 

And here I speak from long experience at Oxford University. The best footballer I ever taught went on to a successful career in financial services. Many years later, he told me that the most useful thing he had learnt from me was how to do discounted cash flow calculations. I am pretty confident he has found application in finance, rather than the blues match against Cambridge. He learned football in the dressing room, not the classroom. And, indeed, I sometimes wonder how much ‘operational’ finance he learned in the classroom at all…

The decision of the footballer in possession will not be explicitly calculating, but will depend principally on the historic pattern of behaviour within the club. 

Some teams will function as teams and will operate within an equilibrium in which a passing game is the normal response. Others, playing a shooting game, will remain and operate as a collection of talented individuals who play as such. The commercial and financial analogues of these kinds of behaviour are obvious enough.

The choice between passing and shooting is a rather explicit form of organisational routine. Many such routines are difficult to identify, or to write down. Rather, they shade into the realm of ‘organisational knowledge’ — the knowledge that people both implicitly and explicitly acquire when they join an organisation with a strong culture. 

Routines, Knowledge, and the Tacit Architecture of Action

Neither ‘organisational routines’, nor ‘organisational knowledge’, are codified in process manuals. They are not reducible to KPIs. Rather, they emerge from repetition, reciprocation, and context — through historic patterns of behavior so embedded within an organisation’s steady practice that even its members would struggle to articulate them as expected norms of action.

Organisational routines are the patterned ways in which groups behave — not because of formal instruction, but because of a shared sense of “how things are done here.” They are not static. They evolve. But they evolve slowly, through cumulative experience, reinforced by repetition and the expectation of future interaction.

Organisational knowledge, by contrast, is what makes an institution more than the sum of its parts. It is not just that individuals within a firm are skilled. It is that their capabilities interact — informally but reliably — in ways that produce intelligence and problem-solving capabilities greater than what any individual could marshal alone.

Organisational routines are the patterned ways in which groups behave.

The understandings captured in organisational routines and organisational knowledge is tacit. It is relational, not individual. It cannot be imposed or transferred wholesale. You cannot acquire it by purchasing a ‘best practice.’ You cannot install it via consultancy. You cannot copy it by poaching talent from a higher-performing peer. You can only build it — through time, trust, and situated experience. Such is the nature of what we term ‘organisational culture.’

Again, this is true of the cultures that enliven firms and supervisory agencies alike.     

What makes a supervisor effective is rarely written down. It lies in the interpretive reflexes of experienced examiners, the judgment shared between colleagues who have seen a failure unfold before, the memory of lessons learned that subtly shapes how new risks are viewed. These things cannot be manufactured on demand. 

Why Culture Matters to Supervision

To date, supervisory engagement with “culture” has largely fallen into two camps.

The first concerns “risk culture” — the extent to which employees understand and comply with risk-taking norms as established and imposed from above. This is the domain of rules, systems, and internal controls. It is legible. It is process-bound. And, often, it devolves into a box-ticking exercise masquerading as governance.

The second focuses on “culture risk” — a kind of non-financial residue, attributed to tone, morale, or employee engagement. This is seen as the province of HR departments, ethics trainers, or public relations teams. It is perhaps important, but it is peripheral to operations and the performance outcomes they produce. The ‘risk’ is that the wisdom of the CEO, or Board, is frustrated by the contrary norms for behaviour that prevail among  individuals within the organisation.

Neither approach to the topic of culture is sufficient. The “risk culture” frame treats culture as an implementation detail. The “culture risk” frame treats it as a reputational vulnerability. Both fail tograsp that culture is generative. It does not just influence responses to risk. It shapes whether risks materialize at all.

Organisational knowledge is what makes an institution more than the sum of its parts.

The risk governance failures at Credit Suisse — like those of Wells Fargo, SVB, and others — were not about liquidity alone or even principally. They were failures of narrative coherence, of reflexive challenge, of situational awareness. In short: they were failures of institutional culture. 

And that these risk governance failures among firms were missed — until deposit flight was unmissable — should in turn be seen as a supervisory failure. Just as cultural fragility within firms precipitated these failures, so too did cultural inertia within supervisory bodies.

We have grown used to post-mortem analyses. But we cannot navigate the future by appeal to the rear-view mirror. Instead, we must build supervisory methods that are anticipatory — capable of assessing institutional coherence and behavioral reliability upstream, before the damage is done.

Culture Is Not Imposed — It Emerges

There is a wide gap between the language of management literature and the lived reality of institutional culture. We often hear CEOs speak of “driving culture change” as though it were a policy lever. But culture is not constructed, it is ‘emergent.’ It arises from what is repeated, reinforced, and rewarded in informal daily interactions between employees — not what is declared formally by the CEO, or proclaimed by HR and PR departments.

The sociologist Michael Polanyi distinguished between “knowing that” and “knowing how.” Culture resides in the latter. It is revealed in informal practices, in who speaks up and when, in how conflict is handled and by whom. It is coded into jokes, routines, and rituals. It is not a corporate value statement, and it is not established by repeatedly chanting some pious ‘tone-from-the-top.’ 

Moreover, it is deeply path-dependent. The culture of a firm — or a supervisory agency — is not generic. It is a product of its history, of its social composition, and of the relationships between its people.

This is why attempts to replicate the culture of a high-performing organisation by transplanting individuals or copying slogans so often fail: the tacit structure of behaviour cannot be lifted out of one context and simply dropped into another.

It is also why cultural interventions must be approached with humility. Cultures that took decades to build can be dismantled in months — sometimes by accident, often by design. And while it is difficult to build a healthy culture, it is frighteningly easy to destroy one. Supervisors must keep this asymmetry in mind. When we intervene in culture, we must know that we are working not with inert materials, but with living systems — fragile, complex, and full of feedback loops.

Supervisory Culture Is Not Exempt

The temptation is to believe that supervisory authority confers cultural immunity. It does not.

Supervisory culture — like that of any institution — can drift into performativity, avoidance, and self-congratulation. It can prioritize the visible over the important, the formal over the real. And because supervisors do not face market pressure in the same way as firms do, their cultural inertia can often be more entrenched.

The events of 2023 — when major institutions failed under the watch of multiple regulators — should be seen not just as operational failures, but as cultural breakdowns. The signals were visible. The risk was escalating. But the systems for escalation were muted. The instincts to act were dulled. And the internal trust necessary to question prevailing views was missing.

Supervisors must therefore examine their own routines, norms, and memory structures with the same scrutiny they apply to firms. We must ask: how is discretion exercised? What kinds of behavior are incentivized, not just by the applause of senior management and the end-year bonus but by the day-to-day approbation of colleagues? Where does curiosity live in our processes? And how does organizational knowledge — tacit, historical, and contextual — get preserved?

From Doctrine to Grammar

The challenge before us is not to define a global doctrine of culture supervision. It is to develop what I would call a ‘grammar’ — a shared vocabulary, an interoperable syntax, a method of description that permits comparability without requiring conformity.

That grammar must be grounded in evidence. And it must be sensitive to the fact that culture, while not measurable in the same way as liquidity or solvency, is no less determinant of institutional resilience and therefore demands the same level of analytical rigor that we bring to questions of financial risk and capital adequacy, if not necessarily the same quantitative kind.

Supervisors must examine their own routines, norms, and memory structures with the same scrutiny they apply to firms.

To build this grammar, we must learn from disciplines long ignored in prudential supervision — anthropology, psychology, sociology, the study of complex systems and of ecological cascades. We must also invest in new capabilities: behavioural diagnostics, qualitative analytics, and narrative assessments. These are not ‘soft’ tools. They are the instruments of forward-looking governance.

And supervisors must begin with themselves. We ask whether firms have the right culture. But we rarely ask whether we’ve trained supervisors to recognize what good culture looks like — or what it requires to thrive. That omission is no longer tenable.

Toward Coherence, Not Uniformity

Cultures cannot be harmonized by fiat. Supervisory philosophies differ. Legal frameworks diverge. Political tolerance for ambiguity varies. But the goal is not uniformity; it is coherence.

If we can develop a common evidentiary basis, shared diagnostic tools, and a mutual commitment to early intervention, then we can begin to supervise culture for what it is — not a reputational variable, but a critical system condition.

That is the path this report lays before us. It will not be easy. But the alternative is clear: a future in which supervisory authority is undermined not by scandal, but by irrelevance — unable to act when the signals emerge, because we have not learned to read them.

Sir John Kay is an economist whose career has spanned the academic world, business and finance, and public affairs. He has held chairs at the London Business School, the University of Oxford and the London School of Economics, and is a Fellow of St John’s College, Oxford, where he began his academic career in 1970. He is a Fellow of the British Academy and of the Royal Society of Edinburgh.

He is the author of many books, including The Truth about Markets (2003), The Long and the Short of It (2009, new revised edition 2016), Obliquity (2010) and Other People’s Money (2015). Radical Uncertainty, jointly written with Mervyn King, was published in March 2020. Greed is Dead, co-authored with Paul Collier, was published in July 2020. His latest book The Corporation in the 21st Century was published in August 2024.