By Peter Farley, Capital Markets Strategist, Finastra
Artificial intelligence (AI) in the banking sphere has gained significant interest in recent months. Combined with the noise around regtech (regulatory technology), both are starting to gain real traction and are actually beginning to be put to practical use. It might even be the turning point for the evolution of wider information-technology (IT) transformation for which the industry has been crying out to deliver sustainable, lower operating costs and smarter businesses.
Turning point on compliance costs
The tsunami of new regulations since the financial crisis has been the single biggest headache for banks to deal with. Already stressed IT systems have come under pressure, as have already bloated cost bases through the huge volumes of new people required to implement and monitor ongoing compliance. It was therefore only a matter of time before banks adopted a more strategic approach to tackle legacy technology, rising costs and regulatory compliance. Some have used the term regtech as the umbrella for these activities.
Estimates suggest that tier one banks are spending well in excess of $1 billion a year on compliance-related costs, or some $270 billion a year for the industry as a whole. This can account for more than 10 percent of most banks’ operating costs, given that staffing for these commitments has doubled in the past five years. In some institutions, compliance-staff levels now match front-office staff numbers one-to-one. It is not sustainable, and something has to give.
The catalyst for turning this around is the evolution of AI and machine learning (ML), on the back of much more affordable and capable data storage and management and far superior intelligent analytics. AI has been around for decades, but it is only now that automation and intelligent-learning capabilities are being truly harnessed.
The robots are here.
The glue that binds AI and ML together and creates the fusion for it to deliver tangible capabilities is natural language processing (NLP). Together, these create solutions that can “read” new regulatory documents and perform a range of tasks, including extracting metadata, identifying reference entities and “understanding” the intent or purpose of the documents.
For regulatory compliance, this means that banks can use NLP to extract metadata, which helps to understand what the regulation is about by identifying the relevant financial products and matching them with the related regulatory topics and consequent business processes. By combining this information, it becomes possible to determine whether the regulation is relevant, what parts of the organisation are likely to be affected, and who needs to review it.
Taking that a stage further, AI can then understand what requirements or expectations are included in the new rules (banks are faced with hundreds of these every month) and what they must (or must not) do about them. At the moment every new rule has to be read and interpreted by a human being, which is time-consuming and prone to errors. NLP can help identify the requirements that are contained within a document and, using the entities and metadata, determine to whom they apply and to what products, topics and processes they relate. On an ongoing basis, machines can then continue to monitor compliance.
Wider benefits
These capabilities are not just confined to paperwork. AI can be used to analyse complex trading relationships and strategies, as well as patterns and communications between banks, exchanges and wider market participants. It is also capable of monitoring internal conduct and communication to clients using quantitative metrics such as supervisory input—which will be even more critical under new transparency obligations.
As AI relies on computer-based modelling, scenario analysis and forecasting, it can also help banks in stress testing and risk management. When dealing with financial regulations, it can simplify the complexity given the multitude of different jurisdictions, products, institutional differences and enforcement mechanisms. And when algorithms that drive AI are supported by human consultants, the systems can “learn” faster, delivering a wider synergy of capabilities.
More practically, banks can deploy AI to analyse huge volumes of conversations from phone recordings, chats and emails using voice- and text-analysis algorithms to determine unusual employee behaviour. Further analyses of these conversations can help identify potential market manipulation and collusion activities. It can help banks prevent not only accidental trading errors (or “fat finger” trades) but also deliberate ones—avoiding damage to that all-important reputational risk.
Elsewhere, potential money-laundering transactions can be better scrutinised. The application of deep-learning techniques and better analytics on the transactions, with more sophisticated business rules, can also significantly reduce the volume of activities flagged for investigation.
AI in practice
With much more demanding regulatory requirements that include delivering meaningful insight and analysis in close to real-time, along with the advances in technology and AI, banks have no choice but to switch more tasks to machines. This will help reduce costs, increase efficiency and boost compliance.
Wall Street is already embracing AI. Much of the trading from forex (foreign exchange) through equities is now done by algorithms, while high-frequency trading (HFT) is driven by the combination of machines and the speed high-performance computing can deliver. Algo-driven machines are making decisions in nano seconds, far outpacing the ability of human trade.
This is spreading down and across the capital-markets food chain with the deployment of robo-advisers, as well as introducing a wider range of self-service analytics and trading tools. These all help drive up revenues through competitive differentiation, but also make valuable contributions to driving down costs.
The AI future
Many believe the industry’s adoption of AI will bring challenges as there are many unanswered questions that banks and regulators need to consider. These will come from the rapidly changing business landscape that will be heavily impacted by the introduction and implementation of AI. Banks will need to react quickly and begin to anticipate the impact of future changes, as well as ensure their AI-driven systems remain user-friendly, transparent and comprehensive.
For the time being, banks are showing a healthy appetite for AI and ML and the benefits they offer. The appeal is not surprising given the surge in costs and consequent slump in margins in recent years, which has seen many banks’ return on equity fall significantly below their cost of capital. AI is not a panacea, but—if employed within a wider IT-transformation strategy that embraces more collaborative practices and with the deployment of cloud services to take IT investment off balance sheets—it could herald the beginning of a turnaround in investment-banking fortunes.