Supervisors on Supervision

“If we had a common evidentiary basis, life would be easier. 

We don’t — at least not yet. We’ll keep looking for possibilities. I don’t know if one is possible, but I’m open-minded and would welcome it.”

“Culture is clearly important not just for conduct but for the safety and soundness of banks. 

I think Andrew Bailey said a while back that pretty much all bank failures have a behavioral driver somewhere. So the idea we can ignore this is hard to sustain.

We’ve seen failures — Credit Suisse being the most obvious — driven primarily, I’d say, by persistent cultural problems. Credit Suisse’s issues went back decades — immorality, turning a blind eye as long as someone made money; semi-independent businesses with no challenge if profitable. That’s cultural memory.

On the other hand, I struggle with how supervisors get their arms around it. At worst, it becomes vague — ’he-said/she-said’ — with supervisors giving opinions that management don’t understand or disagree with.

The challenge isn’t importance [of culture]; it’s whether supervisors can actually make a difference — and whether, given limited time, this is the best use of effort versus things easier to measure and change. 

To avoid that counsel of despair, you have to believe you can assess culture with some objectivity and then respond rationally — either encouraging or requiring change, or, recognizing the culture for what it is, and compensating with additional financial resilience.”

“Central bankers do see themselves as working toward their understanding of the public good. Our mandate comes from statutory objectives — safety and soundness. 

In a capitalist society, profit motivation is fine; but in banking, market failures mean that without regulation banks over-leverage and take too much risk — we’ve seen it. 

Politicians ultimately decide the public good — but they tend to underweight very bad outcomes when those have faded from memory. Good societies create institutions to ensure those risks aren’t forgotten; an independent central bank regulating is, in my view, good design.”

“The problematic behaviors in a supervisor differ from those in a bank. Still, techniques to identify behaviors that could cause an institution to fail or miss its mandate — yes, why should we not consider that ourselves?”

“This isn’t like capital where we want a consistent international metric of adequacy.

There isn’t a single culture in a bank. There are multiple sub-cultures — especially in large groups with acquisitions: legacy bits, trading vs. retail, geographies, disciplines. (In insurance, the actuarial side has a different culture from sales.) 

So ‘culture’ is slippery. The most useful thing supervisors can do is get specific about the behaviors they think are a problem and what’s driving them. Be concrete: a hubristic senior team that thinks it can do no wrong; a bullying CEO who won’t listen; or something pervasive, like being too driven by revenue, profit, and bonus. And then ask: how do you establish credibility with the board or senior management about your view?

I see two routes. 

One: produce something that looks like hard evidence — data, staff surveys — but ‘hard evidence’ is hard to come by here. 

Two: bring in experts with credentials people will believe — psychologists, third parties. 

Those feel like the right candidates.”

“I find neither ‘governance’ nor ‘culture’ helpful terms in this context. Lazy supervisors often point to problems with governance or culture without being sufficiently clear and specific about what they mean.

The origins of firm culture, defined as patterns of behavior, is clearly complex. In reality, large global banks have multiple sub-cultures within an overriding enterprise culture. Some patterns of behavior stem from current leaders within the firm: for example, a dominant CEO that overrides all debate within management and the board. But other patterns of behavior might be long-standing and engrained norms and habits amongst employees, or they might be specific to particular markets or geographies.

Boards and management clearly have a strong influence over culture, but they cannot control it fully. They do need to state a desired culture; to ensure that processes within the firm support this culture — e.g. tone from the top, remuneration, promotion, discipline — and to have effective ways of assessing actual culture so as to address gaps between the two.”

“Supervisors can seek to work through boards and management to change culture. But this will take a long time. In the short- to medium-term, supervisors should focus on diagnosing problems well, and then increasing the financial resilience of firms with obvious problems through higher capital requirements.

Enforcement has a place, where firms deliberately break regulations or persistently fail to address threats to safety and soundness effectively. But enforcement actions alone will not change a firm’s culture. That can only happen through consistent action from the board and senior management, and even that might not work.

Credit Suisse shows how difficult changing culture is. Since the 1970s, the firm had consistent patterns of behavior in which excessive risk-taking and sometimes morality was ignored, provided that a business was apparently profitable. It also had a siloed culture in which businesses focused on, and were rewarded for, their own revenue and did not seem to care about what was happening elsewhere in the group.”

“Very few significant prudential or conduct failures do not have some root cause in patterns of behavior.”

“From the perspective of a prudential supervisor, ‘risk culture’ (i.e., attitudes towards and behaviours around risk management) is the most common concern. 

But it is striking to what extent problematic behaviours differ across firms and some cases are not specifically about risk management. For example, behavioural problems stemming from a dominant CEO or an apparently successful firm suffering from hubris go wider. 

In general, I think prudential supervisors should be concerned about any patterns of behavior that could threaten the safety and soundness of the firm. 

And there is a very wide range of possibilities — you could argue that each case is unique. This is why it is unhelpful for supervisors to label a risk as ‘poor risk culture’ without being very clear about the specific attitudes and behaviours.”

“At worst, it becomes superficial. And supervisors might lack confidence — numbers like capital and liquidity are easier. Making these judgments — especially under strong pushback, with the risk you’re wrong — means going out on a limb. That’s hard. 

It also feels more personal for bank management to be told they’ve got behavioral problems. It’s easier when something goes wrong and you can track back to root cause. But when things are going well — which is often when behavioral issues become biggest — it’s very hard for supervisors to challenge. You see teams believing their own hype, ‘we can do no wrong,’ translating into more aggressive risk appetite. 

Ideally, supervisors would intervene early; in reality, that’s hard. That’s part of the problem — supervisors doubt their own judgments: who are we to make them? 

That’s the problem with forward-looking, judgment-based supervision. Beyond prodding bank executives to be more self-aware, it can be hard for supervisors to act in the absence of a crystalized risk.

So you either persuade the top of the bank to act — which requires compelling evidence — or you increase capital — which also requires compelling evidence, especially now, with stronger political pushback steered by industry. It’s very difficult to justify higher capital based on anything less than hard evidence — certainly not just ‘culture judgment.’ 

It may be that you have to wait for a problem to crystallize before you can act. It’s not how it’s supposed to work, and you hope for a small crystallization rather than a terminal one. A small heart attack, maybe.”

“Behaviors don’t fit cookie-cutter archetypes. Each case is somewhat unique — some things rhyme, but situations differ. With hindsight it’s easier; at the time, it’s hard to disentangle. 

Would psychologists have diagnosed more objectively? Maybe. Different ways have been tried: psychologists/experts; surveys; painstaking observation; training supervisors to be alert; some regulation — not of ‘what culture should be,’ but requiring firms to think about it and letting us check. 

The big one underused, to me, is data. Data’s probably useful for some cultural issues — but not all. We need to work out what can be done with data.”

“The right tool depends on the problem. If it’s the board, maybe you need the ‘skilled person.’ [a third-party expert consultant] 

If the first line doesn’t respect the second line, maybe you can use data. 

Silos are common — perhaps that was a part of Credit Suisse’s problem — a lack of empowerment or respect for the risk function, insufficient authority in the second line.

Maybe you can measure that — how people communicate, the language they use, and so on.”

“One important aspect is how the firm behaves towards supervisors. Recording and doing regular stocktakes of not just what people from the firm say in meetings but also how they behave is important.”

“Individual accountability regimes such as the UK’s PRA Senior Managers’ regime are helpful, because they make clear which individuals bear responsibility for taking action to address risks to safety and soundness. Remuneration also needs to be linked to these accountabilities. These measures are necessary but not sufficient.”

One approach is to set a regulatory expectation that firms define a desired culture consistent with regulatory objectives, assess actual culture effectively — this is key — and then take steps to close the gap between actual and desired culture. 

The New York Fed has convened supervisors on this for a while, and many have tried things. Maybe they’ve had better results than I’ve seen — but it doesn’t feel ‘cracked.’

We see the announcements, posters on walls, focus groups, surveys with ‘right’ answers. We see remuneration and promotion changes. But are behaviors actually changing? Everything we read says they should — yet behaviors tend to come back.”

“Success in culture supervision would involve:

1. Identifying firms with cultures that pose risks to safety and soundness;
2. having effective supervisory techniques to diagnose the specific problem behaviours accurately;
3. persuading senior management that the firm has a problem and overseeing an effective program by the firm to tackle it; while, in the meantime,
4. requiring additional financial resources against the higher risk. 

All of these elements are very hard to achieve. An open question for me is whether the cost/benefit of putting more resources into this aspect of supervision is worth it in a world of limited supervisory resources and limited political capital.”

 

“Ultimately, there are only two solutions to a cultural problem. One, more financial resources — acknowledge the bank’s riskier and push it further from failure through capital requirements. Two, get the top of the bank to become part of the solution — so they fix themselves. 

If supervisors want to change the culture, persuading the board of the firm that it has a problem is vital, as change is impossible without full buy-in from senior management. A supervisor can’t improve culture if the people running the bank aren’t fully bought in. This requires compelling evidence. Subjective supervisory assessments are unlikely to be sufficient. Possible sources of evidence are:
 

1. Opinions of experts (internal and external). A big question for culture supervision is whether all regulatory agencies should have teams of organizational psychologists (either in-house or external) in the same way as we have experts in credit and market risk. I am open minded but have yet to see strong evidence of their effectiveness;
2. Crystallized risk from actual incidents with root cause analysis leading back to culture; and
3. Data-driven analysis. This is perhaps the most interesting new area. Staff survey results are well established. But other data might include, for example, network analysis of electronic communication within the firm.”

“On culture regulation, I don’t think regulators can tell firms what their culture should be. Good regulation can be very simple: firms should define a desired culture consistent with regulatory objectives, assess their actual culture effectively and have a program to close any gap. OSFI’s recent guideline is the best. And it is very short. 

I am not sure there needs to be international standards. But I do think it is helpful for supervisors to share ideas and experience on metrics to assess culture.
For example, what has been their experience with assessment techniques such as:

1. deep dive reviews by organizational psychologists;
2. the use of staff surveys;
3. systematic recording of observations on culture by supervisors; and
4. the use of Big Data techniques.

It is likely that different techniques are suitable for different cultural problems: for example, if the problem is a passive board or dominant CEO, supervisors might use a different technique to a problem where the first line has little respect for the second line.”

“If we hope to achieve such global collaboration, there are barriers must be overcome — e.g., coordination challenges, the lack of a common taxonomy, limited ability to share data, etc. Supervisors need to be clear that this is not a ‘woke’ agenda to change the world for the better. Rather it is hard-nosed supervision designed to achieve regulatory objectives. 

Moving away from terms such as culture and talking instead about patterns of behavior may be helpful. Supervisors also need to be as specific and evidence-based as possible about the problem behaviors about which they are concerned, and the link to safety and soundness. 

I don’t think there is a single way to skin this cat. It would be better for supervisors to continue experimenting with different approaches, sharing knowledge and learning from each other.”