Miranda Love (Oct 24)

Can the use of AI overcome the challenges Banks face in meeting regulatory requirements such as those related to capital and liquidity risk management?

Today, many regulated banks (Banks) face heavy scrutiny when it comes to meeting the increasing regulatory demands. This is not a revelation. Banks have been bombarded with expanding requirements since 2008. The problem is many Banks have to consume many resources to operationalise the processes needed to comply with these requirements. That’s in addition to completing the somewhat onerous task of translating, interpreting, and understanding the requirements. And many Banks operate in multiple jurisdictions that may have very different perspectives or parameters for very similar principles (e.g. capital or liquidity risk management).

Let’s take an example. For instance, the requirement for regulated Banks to maintain sufficient capital, on an ongoing basis, and to ensure it minimizes harm to the market itself, and to its clients. In some jurisdictions this is quantified through a seemingly arduous exercise called (an) ICAAP or ICARA. Over the last decade or so, the landscape of what qualified as a “good” ICAAP or ICARA has changed. It started out as a numerical guesstimate based on loss history largely run by the Finance function, and has moved to a Risk-led scenario-based approach often including statistical modelling which requires senior management input and is deemed quite subjective. Alongside each of these nuanced stages came the need for sufficient tools, subject matter experts, enhanced controls and financial support. And in the end, it consumed a lot of resources. In recent years, thankfully the discussions have also led to a “so what” discussion. Regulators are asking firms “now what”?  Firms who undertake these processes ultimately have quantified a number of remote but plausible scenarios, and ensure they have sufficient capital on reserve, but is that sufficient? The regulators certainly don’t want this to be just a “tick box” exercise, they want to see “use cases” and see the capital [risk management] framework embedded into a firm’s BAU. However, after going through aforementioned processes, and ensuring the requirements are met, templates are submitted, many banks have exhausted all resources, and in some cases it’s only a brief pause before the process kick starts again. It’s a vicious cycle.

So how can banks make complying with (capital risk) regulatory requirements more efficient and meaningful process? Some would venture to throw more resources at the problem, in the form of people, or consultants. But perhaps this is an opportunity to create some economies of scale and use AI. Build a tool that could look at your inherent risks, loss history, financials; take into consideration actual or hypothetical scenarios; and help senior management derive a more accurate result. Could AI be the solution that allows senior management to focus more of their time and energy on governing, and improving their control framework to manage these risks, rather than what seems to be, spending a lot of time quantifying them?

The answer seems simple, but I went to the source. Yes, I asked AI.

I literally asked, “Can AI tools help regulated banks quantify their capital need?” As I waited in anticipation, I had to giggle at the irony of the situation. And truth be told, the AI system paused a bit longer than usual. But this is its response [ I redacted quite a bit]:

Yes, AI tools can significantly help regulated banks in quantifying their capital needs. These tools enhance the precision and efficiency of risk management and capital planning by leveraging advanced analytics and machine learning models:

1. AI models can improve the accuracy of risk assessments by analysing vast datasets and detecting patterns that may not be evident through traditional methods, like simulating various market conditions

2. AI can use predictive analytics for future capital modelling

3. AI can help banks navigate complex regulatory frameworks, and automate calculation thereby reducing human error

4. AI can automate reporting

5. AI can optimize a bank’s asset portfolio by assessing the risk-return profiles of different assets and ensuring that capital is allocated efficiently

But of course, ChatGPT is going to say this. Much like humans, AI wants to be profitable too. So, I did a little bit of “independent” research and found further supportive statements.

PwC has already gone to market with its “Project Artificial Intelligence Reporting” which aims to digitalise reporting to reduc[e] costs and mitigate[e] key reporting challenges, so resources can be re-allocated to value-adding activities[1]

Sciendo published a study in 2020, speaking to “Prospects of Artificial Intelligence and Machine Learning Application in Banking Risk Management”[2], citing that AI (or Machine Learning) demonstrated successful results when reviewing credit risk management. Whether that be for comparative stress test results; risk-weighed asset calculations; simulations as well as recovery planning. It also cited that this study was limited to Credit Risk, but the parameters could be applied to other risks.

While this is interesting, and hopeful, I do think in using AI, firms should not lose sight of the much needed, value-added activities, such as governance, and control effectiveness. I am at least intrigued. I am definitely watching this space.

[1] Artificial Intelligence for reporting (pwc.com)

[2] jcbtp-2021-0023 (sciendo.com)

Miranda Love has been a member of the BTRM faculty since 2022. She worked at State Street in Boston, Dublin and London, for over 15.5 years holding leadership positions in Internal Audit, Asset Management Risk, Treasury and Credit. She specialises in Capital Management and ESG. She also served as a Board Director at one of State Street’s UK parent holding companies, with subsidiaries in Ireland and the Caymans.

Miranda has now left State Street to pursue a Masters at the University of East London specialising in Applied Positive Psychology. With an aim to begin Executive Coaching. She also seeks to rejoin the financial sector in the near future to dive into the world of wealth management.

Miranda is a BTRM Alumnus with Distinction. She obtained a Bachelor’s degree in Business Administration from University of Southern California, and a Graduate Diploma in Accounting from Suffolk University.