Supervisors on Supervision

“Organizational culture is that intangible thing that aligns each employee's attitudes and behaviors to corporate values. 

We often say that employees are guided by policies, processes, rules within an organization, that is true but it's not the whole truth. More than abiding by policies and rules, employees often choose to do what is acceptable in the eyes of their bosses and their colleagues. 

They do this when interpreting rules and in situations where there are no prescribed rules, and this can happen since rules cannot cover each and every situation. They will ask, ‘what will my boss, what will my colleagues expect or think about this thing I'm doing? What can I do to get me recognized, appreciated, even promoted?’ That is organizational culture at work.”

“A sound organisational culture strengthens alignment of attitudes and behaviours within an organisation to positive corporate values. It ensures consistency and quality in how financial institutions execute their policies and processes, and how they make decisions on a daily basis at all levels within the organisation. Examining organisational culture has thus been an increasing focus for a number of financial regulators, to minimise misconduct and internal control failures at financial institutions.”

“Sound organizational culture will align employee behaviors to positive corporate values. Be it to treat customers fairly or manage risk prudently, it strengthens the consistency and quality of how financial institutions execute their policies and processes, and how they make decisions on a daily basis at all levels within a financial institution. So this helps to minimize misconduct and internal control failures even when the supervisor is not watching. And that's why it's so important for us.”

“Fostering sound organizational culture has also increased in complexity. Under the current remote hybrid work arrangements, which some organizations are going to extend indefinitely, it is harder for managers to model the desired codes of ethical behavior, corporate values, codes of conduct. It is also challenging to onboard new employees into organizational culture when they spend limited face time with their peers and with their supervisors. 

The long-term adverse impact of remote working on an organization's culture, team dynamic, creativity and overall ability to innovate bears close watching. So firms and supervisors must stay keen, be aware of some of these challenges as they try to strengthen organizational culture.”

“Financial sector supervisors are in the business of strengthening safety and soundness and fair treatment of customers in our regulated institutions. Increasingly, it is recognized that fostering strong organizational culture within regulated institutions can help us to do that well.”

“We've been intensifying our efforts, and we have a three-pronged approach. 

The first prong approach in supervising organizational culture is what we call promote and cultivate. So MAS promotes awareness and cultivates commitment within financial institutions to build strong organizational culture. This first step recognizes that building strong organizational culture is something that financial institutions need to own. 

MAS engages in regular dialogues with financial institutions, their boards and senior management, to discuss the benefits of good culture to get the mind share. And we also lean in to listen to their operational challenges, and then we try to bring the industry together to facilitate sharing of good practices so that good solutions to common problems can be tackled very much more quickly. 

The second prong is to monitor and assess. Having secured commitment to build sound organizational culture, MAS as supervisor must monitor progress and nurture improvements. We assess both the hardware, things like frameworks, policies, procedures, they still have a role to play, as well as the software, such as tone from the top, leadership attitudes. 

We evaluate whether a financial institution's organizational culture incentivizes ethical behaviors and responsible risk taking. We also look out for potential red flags, such as for example, perhaps this empowered risk control function. The supervisory techniques that are involved are rather different from what our inspectors are ordinarily accustomed to. 

For example, great reliance is placed on in-depth conversations and interviews with bank employees on topics such as their perceptions of front office culture, tone from the top in terms of organizational values, and their perceptions of desired employee behavior hence aside from iteratively improving our methodology of assessing organizational culture, we are also deepening our supervisor's capabilities of performing these assessments. 

And the third and final prong is to enforce and deter. We can take a variety of actions ranging from just issuing a warning to closing a financial institution, to civil penalties or referring it to our attorneys generals chambers for criminal prosecution. Of course the penalties will have to be commensurate with the severity and nature of the misconduct and be sufficiently tough, adequate to achieve effective deterrents. 

While the three prongs are mutually reinforcing and important, our predominant focus now is on the first two prongs, because cultivating and monitoring industry norms of desired behavior are more preemptive towards minimizing the likelihood of serious lapses in the industry.”

“We cannot undertake this work … without the help of technology. Indeed, we are fully aware of the problems about people telling you what they think you want to hear. And so to solve this problem, it’s really about going down to the masses, taking a very elapsed sample and then lobbing off at the extremes. And in order to do this, you need technology to help you.”

“Recognising the immense potential that technology offers, MAS coined the term “SupTech” in 2017, and has since gone on to expand our capabilities to harness technology for more effective supervision, including in the area of culture and conduct supervision. 

Some examples are as follows:

• Dashboarding and Visualisation — At the most basic level, MAS monitors a broad range of data and metrics, including complaints received by MAS and financial institutions, misconduct cases, financial disputes, and product revenues, among others. Dashboarding and visualisation tools allow us to quickly identify trends, outliers, as well as specific issues of concern from these voluminous datasets.
• Natural Language Processing — To analyse the misconduct reports that MAS receives from financial institutions each year, MAS uses a combination of Natural Language Processing (NLP) techniques such as topic modelling, sentiment analysis, and regular expressions to tease out various misconduct modus operandi, and monitor these for spikes and trends of concern.
• Automatic Speech Recognition — MAS is also exploring the use of Automatic Speech Recognition technology to automatically transcribe the interviews and conversations we have with financial institutions’ staff, then running these through NLP and machine learning models for topic classification and other analyses. This will significantly improve the efficiency of our inspection work that places heavy reliance on interviews with staff of financial institutions and help us to identify more quickly, common themes and patterns that may be indicative of cultural issues.”

“One of the technologies that we've been trying to experiment with is automatic speech recognition. MAS is exploring the use of this technology to automatically transcribe the many, many interviews and conversations that we have with financial institution staff. 

This has the potential to significantly improve the efficiency of our inspection work, that places such heavy, and it's really inordinately heavy reliance on interviews. And it will help us identify very much more quickly, common themes and patterns that may be indicative of cultural issues.

[Another] example would be the use of natural language processing. To analyze misconduct reports submitted by financial institutions, MAS uses a combination of natural language processing techniques such as topic modeling, sentiment analysis and regular expressions to tease out various misconduct modus operandi, and to monitor these for spikes and trends of concern.”

“We've had some early success in developing a simple, multifactor logistic regression model to score the likelihood of a representative, and by representative I mean someone who sells and provides advice on investment products. The model scores the likelihood of a representative committing misconduct over a specified time period.

The model draws on supervisors input, on predictive factors such as working experience and misconduct history of a representative, it also has tested various other factors that MAS thought was relevant. And the scores are now used by supervisors to identify higher risk transactions for samples for scrutiny. 

Another example would be the use of augmented intelligence systems. For market manipulation cases, MAS often appoint industry experts to obtain a specialist assessment of the trading behavior of the suspects. However, it can take some time before the experts come back with a complete analysis and provide an opinion. 

So we developed a tool in-house in 2018, using an augmented intelligence system, and it performs automated trade analysis and assesses the likelihood of certain types of market manipulation and the likelihood that, that has occurred.”

“A common criticism of conduct supervision is that it tends to be reactive, as disciplinary actions are taken ex-post, when the misconduct has already been committed. We hope to make conduct supervision more pre-emptive, by using machine learning technology for analysis of larger and multiple datasets.”

“In recent years, regulators have been turning to newer techniques for culture and conduct supervision. Advancements in data science and technologies such as artificial intelligence are fundamentally changing the type of data that supervisors can efficiently collect and analyse, opening up possibilities for fresh data-driven insights on the culture of financial institutions.”

“Today, MAS has well-established methodologies and capabilities to assess traditional risk areas such as financial, operational and money laundering risks of our regulated entities. In contrast, our supervisory approach for the less tangible areas such as tone-from-the-top, leadership, attitudes and behavior are more nascent and still evolving. 

In the next phase of work, we are looking to further strengthen our approach to culture and conduct supervision in the following ways:

• Develop a framework to assess financial institutions’ organizational culture in a structured and consistent manner. The framework and supporting methodology will leverage behavioral science and organizational psychology concepts and techniques, accompanied by an intervention framework to guide supervisory actions in addressing cultural risks in a financial institution.
• Deepen expertise in the area of culture supervision. Our supervisors, while attuned to analyzing financial data, are extending their capabilities to collect and analyze non financial and qualitative data relevant to the culture of financial institutions. This requires learning and application of new knowledge areas — for example, an understanding of how heuristics and biases could lead to excessive optimism and risk taking when making decisions. MAS has set up a behavioral science unit to help supervisors better understand how culture and conduct issues affect financial institutions.
• Monitor emerging culture and conduct risks. We are monitoring risks arising from financial institutions’ digitalization efforts which have accelerated in the current COVID-19 environment. One risk to watch out for is the potential for financial institutions to embed subconscious nudges within their digital platforms to unduly pressure customers into taking up products and services. For example, time limits to complete transactions could increase the risk of hasty financial decisions. We do not think this is prevalent now, but nonetheless bears close monitoring, amongst other potential risk areas.”

“Our on-site inspections … typically involve in-depth conversations with a large number of employees across all levels of seniority and across the three lines of defence. Qualitative data from these conversations are analysed to distil insights on organisational values and shared norms, as well as areas of potential concern. The work is resource-intensive, but it brings a level of understanding that we would otherwise not have gained. 

Common themes and patterns emerging from these conversations are shared with senior management for reflection and action — a common one being how organisational values, even though regularly communicated by senior management, are not translating into desired behaviours on a day-to-day basis.”

“I really think culture and conduct will continue to be an area of supervisory focus for regulators. We're really in a period of significant risk and uncertainty. Sound organizational culture, will be all the more important and needed in times like these to help financial institutions make sound day-to-day decisions, manage their risk well and build brand reputation, and confidence and trust as they navigate the adaptations and changes needed in such an environment. 

I will be the first to say, we do not have all the answers we never had, but all the more, that means we have to work closely together to identify common pitfalls, challenges and share solutions as quickly as possible.”

“If you look at the way that we supervise in culture and conduct, it's way more intrusive than what we would do for any other area of risk… When we try to broach the topic of how do you measure good conduct? How do you foster good conduct and good organizational culture within your organization? Firms very often tell us, ‘How do you define culture? How do you measure culture?’ And we found ourselves stuck in that level of conversation.

So by taking the bull by the horns, going in using insights from behavioral science and techniques from psychology, we hope to demonstrate to financial institutions that it is possible to define this, to measure this and to help make improvements in this area. 

Over time, the hope is that supervisors will increasingly shift the burden of measurement and monitoring to financial institutions. And we, over time, would pick up that supervisory oversight as we ordinarily do.”

“MAS closely partners with the industry to elevate culture and conduct standards in our financial institutions. The Association of Banks in Singapore Culture and Conduct Steering Group (ABS CCSG) and the Insurance Culture and Conduct Steering Committee (ICCSC), both established in 2019 and chaired by banks and insurers respectively, are two industry working groups at the heart of such partnerships.”

“While we have progressed in our culture and conduct supervision, much more work lies before us for all involved. The regulatory community can benefit from working closely together to share good ideas and sharpen our approaches.

The industry must also own this journey because rules and regulations can only go so far in influencing behaviour, and it is the day-to-day decisions at the financial institution that determine the final outcomes in risk-taking and how consumers are treated. The ultimate goal is clear — a financial sector where sound organisational culture drives prudent risk management and excellent customer service. Strengthening this will remain a priority in our supervisory agenda.”