Ana Badour, Kirsten Thompson, Carole J. Piovesan and Brianne Paulin
On November 1, 2017, the Financial Stability Board (the “FSB”) published its report on the market developments and financial stability implications of artificial intelligence (“AI”) and machine learning in financial services. The FSB noted that the use of AI and machine learning in financial services is rapidly growing and that the application of such technologies to financial services are evolving.
“Use Cases” of AI and Machine Learning in the Financial Sector
The FSB identified current and potential types of use cases of AI and machine learning in financial services, including: “(i) customer-focused uses, (ii) operations-focused uses, (iii) uses for trading and portfolio management in financial markets, and (iv) uses by financial institutions for Regulatory Technology (“RegTech”) or by public authorities for supervision (“SupTech”).”
The FSB found that “financial institutions and vendors are using AI and machine learning methods to assess credit quality, to price and market insurance contracts, and to automate client interaction.” Specifically in the insurance industry, machine learning is being used to analyze big data, improve profitability, and to increase the efficiency of claims and pricing processes. The investment to global InsurTech totaled $1.7 billion in 2016.
Such application of AI and machine learning can increase market stability as financial institutions have a greater ability to analyze big data to enhance their knowledge of trading patterns and to better anticipate trades. The FSB warned, however, that due to the lack of data on how the market would react to an increase use in AI and machine learning by market participants, a market shock could occur. In fact, market participants could be enticed to apply such technologies if their competitors, in applying AI and machine learning to customer-focused uses, are increasing profits and outperforming them. This increased use by market participants could cause a market shock and bring instability to the market.
Operations-Focused Uses and Trading and Portfolio Management
Trading firms would be able to better assess market impacts and shifts in market behaviour, increasing market stability. An example of such use is ‘trading robots’ than can react to market changes. The ‘trading robots’ can perform and assess market impact of certain trades, which allows trading firms to collect more information which, in turn, allows these firms to modify their trading strategies. The FSB also identified back-testing as an area of growth for the use of AI and machine learning. Back-testing is important for banks in their assessment of risk models. AI would provide a greater understanding of shifts in market behaviour and the FSB stated that this could potentially reduce the underestimating of risks in such instances.
Uses of AI and Machine Learning by Financial Institutions
The FSB found that AI and machine learning is used by financial institutions for regulatory purposes and by authorities for supervision purposes. The RegTech market is expected to reach $6.45 billion by 2020. Several regulators around the globe are using AI and machine learning to facilitate regulatory compliance, such as applying AI and machine learning to the Know-Your-Customer process. In terms of SupTech, the report noted the implementation of AI and machine learning in various supervision functions by authorities, such as monetary policy assessments. A 2015 survey of central banks’ use of AI and machine learning, cited by the FSB, found that central banks anticipated using big data reported by third parties for economic forecasting and for other financial stability purposes.
Implications of AI and Machine Learning on Market Stability
The FSB warned that, though AI and machine learning would benefit market stability by reducing costs, increasing efficiency and increasing profitability for financial institutions, financial institutions must implement governance structures and maintain auditability to ensure that potential effects beyond the institutions’ balance sheets are understood. Governance structures include ‘training’ to ensure that users understand the technologies and applications of AI and machine learning, promoting algorithmic transparency and accountability to ensure decisions made by the algorithm, such as the credit score assigned to a particular customer, can be understood and explained.
Without sound governance structures, the application of AI and machine learning could increase the risk to financial institutions. The report noted that “beyond the staff operating these applications, key functions such as risk management and internal audit and the administrative management and supervisory body should be fit for controlling and managing the use of applications.”
The benefits of using AI and machine learning systems for consumers and investors could translate into lower costs of services and greater access to financial services. AI and machine learning could allow financial institutions to assess big data to tailor financial services to specific customers and investors. The FSB noted that proper governance structures must be in place to protect the privacy and data of both consumers and investors.
The FSB also raised concerns over the small number of third party providers of data in the financial system. Bank vulnerability could grow if the financial institutions rely on the same small number of third-party providers, using similar data and algorithms. On dependency, the FSB noted that “third-party dependencies and interconnections could have systemic effects if such a large firm were to face a major disruption or insolvency.” If financial institutions are unable to use big data from new sources, dependencies on previous data could develop, potentially leading to market shocks and bringing instability in the financial system.
This same concern was recently echoed by the Bank of Canada in its November 2017 Financial System Review, in which it said:
“As financial services rely increasingly on information technology, there are growing operational risks from third-party service providers. Since providing services such as cloud computing, big data analytics and artificial intelligence requires a critical mass of users to remain cost-effective, global markets could become dominated by a few large technology firms. Higher industry concentration would raise systemic risks from operational disruptions and cyber attacks. Investments by service providers to avoid disruptions have benefits beyond the individual firm and can be considered a public good.”
Legal and Ethical Issues
The FSB also provided an analysis on certain legal issues that arise in the use of AI and machine learning with big data, specifically in the context of data protection and data ownership rights. The FSB highlighted the efforts in several jurisdictions to adopt guidelines for the protection of data ownership and privacy. Some jurisdictions are also assessing whether consumers should have the ability to understand certain techniques used in the application of AI and machine learning to credit systems. Other issues that arise in the use of AI and machine learning with big data include anti-discrimination laws and equal opportunity laws. The FSB noted that the use of AI and machine learning could lead to discriminatory practices and results, even without the inclusion of gender or racial information. Finally, liability issues could also arise, such as determining whether experts who rely on algorithms could be liable for their decisions.
The FSB noted that it will continue monitoring the uses of AI and machine learning in the financial markets, especially as the application of such technologies to the financial sector is growing.
To view all formatting for this article (eg, tables, footnotes), please access the original here.