Artificial intelligence pinpoints patterns in financial transactions with lightning swiftness, making it invaluable for banks and regulatory authorities. Professor Dirk Broeders, currently a fellow at the world’s “bank for central banks” in Basel, is studying the rise of algorithmic supervision
If you want to find exciting technological innovations that deal with big data, just look for data warehouses. However, one field with impressive data warehouses that may not spring immediately to mind is that of financial supervision. Central banks, and the authorities that supervise the conduct of financial institutions and markets, work with astronomical amounts of data. These data are now being scrutinised by data scientists. The authorities’ objective is to identify emerging risks in the financial system, to trace malicious behaviour by market participants, and to monitor the compliance of banks and other financial institutions. Across the global community of financial supervisors, algorithmic supervision – or SupTech – is on the rise.
Spotting economic shifts or potential misdeeds in the blink of an eye
One important area of application for SupTech is Anti-Money Laundering (AML) and Combating the Financing of Terrorism (CFT). Artificial intelligence, or intelligence demonstrated by machines, is able to detect patterns and anomalies in transactions and relations in the economy that cannot be spotted by humans, and it can do so with incredible swiftness. But the process doesn’t stop there. Natural Language Processing (NLP), which refers to the capacity of programs to understand human language, can be used to provide virtual assistance to consumers and businesses.
Many more fascinating SupTech applications are on the way. Central banks are experimenting with algorithms that try to forecast inflation, or the likelihood of a bank run. They do so by analysing sentiment expressed in thousands of tweets. What’s more, algorithms can be trained to predict the likelihood of fraud or mis-selling by “reading” annual reports or advertisements. The most progressive algorithms are inspired by the structure and functioning of the brain. These neural networks may, for example, detect emerging risks from the innumerable transactions that run through payment systems. Regulators are also looking to introduce machine-readable regulations: in other words, regulations issued not as legal text but as programming codes that feed directly into the IT systems of banks.
More data mean a stronger supervisory toolbox
The availability of more and especially better data, the exponential increase in computing power, and the march of scientific progress are all enriching the supervisory toolbox. The use of innovative technology will help supervisory agencies to digitise regulatory processes, which will lead to more efficient and more forward-looking assessment of risks and compliance. Supervisors often work together with the academic community to explore new solutions, and academics, in turn, are inspired by the wealth of data available.
Will algorithms replace humans in financial supervision? My answer is no…. at least not for the foreseeable future. Algorithms cannot be perfect predictors, and are known to produce “false positives”, raising alarms about behaviour that is, in fact, not liable. Moreover, they can also yield “false negatives”, when algorithms fail to spot misconduct. In that sense, algorithms are just like humans – they are not perfect. Furthermore, we have yet to create algorithms that can decide on the most appropriate regulatory intervention. A better question, therefore, is who is better at the margin in finding needles in a particular haystack: humans or algorithms?
Although they are not a universal solution to every challenge facing regulators, algorithms will continue to gain ground, and every financial supervisor should have at least some basic knowledge of them. More broadly, of course, their prominence is on the rise across society: “algos” are everywhere, and coding should have a prominent place in the curriculum at universities.
Rich research opportunities at the Swiss heart of world banking
I am currently spending three months as a Fellow at the Financial Stability Institute of the Bank for International Settlements (BIS) in Basel, Switzerland, surveying algorithmic supervision around the world. Although its name may not be widely familiar to the general public, the BIS is an organisation of great global importance, playing a central role in setting regulatory standards for banks worldwide. Hence, bank regulation is often referred to as “Basel”.
Equally importantly, the BIS is an international knowledge organisation, with over 60 nationalities working at its headquarters, and it has the ideal network for amassing and disseminating knowledge about these fields. The survey of algorithmic supervision that I am carrying out is intended to facilitate the exchange of knowledge and experience amongst supervisory agencies on a global level. Basel, and the BIS, is certainly a stimulating and fruitful place to work on such an exciting topic.
Dirk Broeders is special professor of pension finance and regulation at the School of Business and Economics of Maastricht University. He is also a senior strategy advisor in the Supervisory Policy Division of De Nederlandsche Bank.