Artificial Intelligence and Systemic Risk

Abstract

Artificial intelligence (AI) is rapidly changing how the financial system is operated, taking over core functions for both cost savings and operational efficiency reasons. AI will assist both risk managers and the financial authorities. However, it can destabilize the financial system, creating new tail risks and amplifying existing ones due to procyclicality, unknown-unknowns, the need for trust, and optimization against the system.

AI ethics and systemic risks in finance

“…The financial crisis of 2008 was a paramount example of a systemic event that drew the attention of ethicists in the fields of business and finance ethics [7,[9][10][11][12] and the political economy [45,46] to systemic risks. The merit of this excellent research was to, first, clearly argue the case of the moral relevance of systemic risks and, second, highlight the challenges that existing ethical approaches face, while analysing moral aspects of systemic risk imposition.…”

“…Importantly, the literature has already discussed at length whether applications of advanced technologies such as AI could contribute to systemic risks in financial markets [7,[9][10][11][12]. It has explored whether these technologies can potentially harm the financial system-and the economy as a whole-and thus jeopardise the life chances of citizens only marginally involved in investing and trading.…”

The Norms of Algorithmic Credit Scoring

“…As such, it remains to be seen whether observed improvements in predictive accuracy simply reflect the relatively benign macroeconomic environment since 2008 in which many ML models used for algorithmic credit scoring have been trained. 65 Indeed, there are ominous parallels between the use of alternative data for algorithmic credit scoring and the loosening of loan underwriting standards for high-risk borrowersincluding the notorious "Alt-A" mortgagesthat foreshadowed the 2008 global financial crisis. 66 A related concern is the impact of algorithmic credit scoring on the overall volume of household debt and the rate of credit expansion in the economyparticularly to vulnerable consumers for whom debt can quickly become unaffordable.…”

The promises and pitfalls of SupTech for corporate governance-related enforcement

“…Evidence suggests that authorities use advanced analytics such as machine learning, natural language processing, text mining and network analysis to enhance their capacities -especially with regards to detecting networks of related transactions, identifying anomalies and unusual behaviours, and drawing insights from extensive amounts of structured and unstructured data (Coelho, De Simoni and Prenio, 2019 [13]). For instance, Mexico's National Banking and Securities Commission (CNBV) has developed a prototype for a Natural Language Processing (NLP) application to detect what a suspicious Anti-Money Laundering/Combatting the Financing of Terrorism (AML/CFT) network is 'talking about', thus facilitating the detection of unusual transactions, relationships, and networks events to identify potential money laundering issues that cannot be identified by people.…”