Agent-Based Analysis of Systemic Risk in Japanese Financial Institutions under Ransomware Attacks
Hidenori KATO, Yasuyuki Tahara, Akihiko Ohsuga, Yuichi Sei · IEICE Transactions on Information and Systems · 2025
This study analyzes the propagation of systemic risk through the interbank network when a financial institution is subjected to a ransomware attack, using an agent-based simulation approach. Although based on the conventional May model, the study introduces a balance sheet adjustment algorithm that dynamically corrects imbalances between assets and liabilities, allowing for a more realistic representation of financial system behavior. The simulation considers three scenarios: (1) a megabank is attacked by ransomware, (2) a major regional bank or regional bank is attacked, and (3) a scenario comparing ransomware attacks conducted using the May model. Each scenario is examined at three levels of net worth ratio: 8%, 10%, and 12%. The results reveal that in the case of a megabank, a higher net worth ratio effectively reduces the occurrence of cascading failures. In contrast, when a regional bank was attacked, no secondary failures were observed, and the stability of the general network was maintained.