Network Analysis of Global Banking Systems and Detection of Suspicious Transactions

Abstract

A novel network-based approach is introduced to analyze banking systems, focusing on two main themes: identifying influential nodes within global banking networks using Bank for International Settlements data and developing an algorithm to detect suspicious transactions for anti-money laundering. Leveraging the concept of adversarial networks, we examine Bank for International Settlements data to characterize low-key leaders and highly-exposed nodes in the context of financial contagion among countries. Low-key leaders are nodes with significant influence despite lower centrality, while highly-exposed nodes represent those most vulnerable to defaults. Separately, using anonymized transaction data from Rabobank, we design an anti-money laundering algorithm based on network partitioning via the Louvain method and cycle detection, identifying unreported transaction patterns indicative of potential money laundering. The findings provide insights into system-wide vulnerabilities and propose tools to address challenges in financial stability and regulatory compliance.