Quantifying culture and its implications for bank riskiness

Joel Suss, David Bholat, Alex Gillespie and Tom Reader

‘Bad cultures’ at banks are often blamed for scandals and crises, from the global financial crisis to the mis-selling of payment protection insurance (PPI) in the UK. Yet surprisingly little research has tested this claim. This is because quantifying culture is difficult to do. Our working paper gives it a go. Leveraging unique access to data available to regulators, we diagnose the cultural health of the UK banking sector. We find that banks with organisational cultures two standard deviations below the sector average are associated with a 50% increased risk of failure.

Observing culture

Culture is defined as a set of shared values that shape how a group of people perceive and act. Cultures develop in different group settings, including in workplaces at different scales; for example, within teams, and across an organisation.

While organisational culture is easy to define, it’s difficult to observe empirically. As the Governor, Andrew Bailey, once noted ‘we cannot go into a firm and say show us your culture.’

That’s because organisational culture is both pervasive and elusive. It’s pervasive in that it isn’t manifest in a single practice, but an assembly of them. As with a person, the habits that define an organisation are many and complex.

Yet organisational culture is also elusive. While employee surveys and interviews are the typical approach used to understand organisational culture, a growing body of evidence suggests that these sources are unreliable.

There are a number of reasons for this. First, surveys and interviews lend themselves to impression management, ie deliberate attempts by those under scrutiny to create a favourable perception that may be at odds with reality. Second, and less intentionally, employees embedded in particular organisations tend to take for granted their cultural context as ‘normal’ when, in fact, it may be anomalous and specific to that firm. This can make employees unreliable witnesses even if they are committed to revealing the truth about their firm’s culture. Indeed, because culture shapes how people respond to surveys, it is not clear whether responses are perceptions or products of culture.

Alternatively, regulators might take a cue from ethnographers and undertake a period of participant observation to see how an organisation works and arrive at an independent assessment. But the trouble here is that staff may modify their behaviour once aware they are being studied. This is also very costly from a time and resource perspective.

In sum, since organisational culture is pervasive, we need to measure it with a battery of indicators rather than one. And because it’s elusive – and especially slippery under scrutiny – it is best to measure it unobtrusively. Our paper does exactly this.

Quantifying culture

Using multiple, unobtrusive indicators, we measure the organisational culture of individual banks and the banking sector as a whole. In total, we compile 20 unobtrusive indicators of culture (UICs) from six different sources. Each chosen indicator conforms to one of the dimensions identified by the Organisational Culture Profile, and we are able to cover five dimensions in total – detail orientation, inclusivity, customer orientation, risk orientation and integrity.

To measure how detail oriented banks are, we look at the punctuality and accuracy of the regulatory data they submit to the Bank of England. To proxy how inclusive their cultures are, we look at the spectrum of gender and ages represented at senior levels. For capturing a bank’s customer-centricity, we look at the volume, value and outcomes of complaints levelled against firms by consumers. We also judge an organisation’s risk appetite by its risk-weights relative to peers, and capital add-ons applied by supervisors. Finally, we estimate bank integrity by looking at the frequency and financial impact of internal fraud, and the number of (anonymised) whistleblowing referrals to the PRA.

We scale each indicator to range between 0 and 1 to make them comparable. We then calculate a simple average for each of the 150 banks and building societies (shorthanded banks) in our sample, providing a composite culture score. Chart 1 shows the distribution of scores by firm type. Building societies tend to have better cultures than banks. Taking the average of all these composite culture scores, we can also get a sense of the overall trends in the UK banking sector. In aggregate, the cultural health of banks has neither improved nor deteriorated. The stability we observe using UICs echo findings from the Banking Standards Board survey published in 2019.

Chart 1: Culture by bank type

Culture and bank riskiness

After constructing composite cultural scores, we explore whether these are linked to bank risk. We measure bank risk in various ways and our findings are robust to different definitions. Here we focus the discussion on the distance to default (z-score) measure, which is commonly used in the banking literature. The z-score tells us the number of standard deviations a bank’s return on assets have to fall to offset its equity. Once equity is offset, a bank is insolvent.  

We regress the z-score on our composite cultural score with quarter fixed effects and robust standard errors at the bank level. Our specifications include a range of control variables, including bank size and balance sheet characteristics. We lag the composite cultural score and the control variables to forecast default probability. Below we report results where our independent variables are lagged by eight quarters (two years), but our results are robust at different horizons.

We find a statistically significant relationship between organisational culture and bank risk. The effect is substantive – a one standard deviation improvement in culture is expected to reduce risk by 25%.

In the paper, we perform additional tests to account for endogeneity, in particular, that bank risk affects organisational culture rather than the other way around. This might be because riskier banks, might attract individuals with greater risk appetites, fostering bad cultures. Both our instrumental variable estimate and coarsened exact matching analysis give us further confidence to believe there is a causal chain running from organisational culture to bank risk. 

What our findings mean for bank supervision

In sum, a bank’s culture is a leading indicator of its risk. While this claim has often been made, it has been far less frequently substantiated. Our research provides evidence to support it.

This finding has a few implications for banking supervision. First, it suggests that supervisors should seek to measure organisational culture explicitly and use cultural indicators as inputs into microprudential models of bank risk. While culture is already monitored by supervisors qualitatively and directly, via surveys and interviews of bank staff, we suggest it might be done quantitatively and unobtrusively, in ways both more objective and less costly to obtain. Our own composite indicator is an initial attempt that could be further developed, with additional data added. It is also likely that the way we measure organisational culture needs to change over time. As per Goodhart’s Law, if firms are aware that regulators are looking at a handful of indicators to measure their culture, they may seek to manipulate them, or optimise them while allocating resources away from other, equally important practices that make for a good organisational culture.  

Second, our research underscores the importance of analysing firms holistically. Many of our indicators, such as the frequency of internal fraud or customer complaints, are traditionally thought to relate to conduct and consumer protection. However, our research shows they are directly relevant for prudential supervision.

---

Joel Suss works in the Bank’s Advanced Analytics Division, David Bholat worked in the Bank’s Advanced Analytics Division while the post was written, and Alex Gillespie and Tom Reader work in the Department of Psychological and Behavioural Science, London School of Economics.

If you want to get in touch, please email us at bankunderground@bankofengland.co.uk or leave a comment below.

Comments will only appear once approved by a moderator, and are only published where a full name is supplied. Bank Underground is a blog for Bank of England staff to share views that challenge — or support — prevailing policy orthodoxies. The views expressed here are those of the authors, and are not necessarily those of the Bank of England, or its policy committees.