From Regulation to Interconnection: Mapping the Evolution of the U.S. Interbank Market

Abstract The U.S. interbank market, a critical component of financial stability, has undergone significant transformations over the past four decades, driven by regulatory reforms, technological advancements, and economic shocks. This paper investigates the evolution of the interbank network structure among mid-size and large U.S. banks from 1984 to 2024. We construct quarterly interbank networks based on the cosine similarity of interbank exposure profiles derived from Call Reports and analyze their topological properties, including density, centrality, community structure, and modularity. A key focus is mapping major legislative and regulatory changes (e.g., the Riegle-Neal Act, GLBA, and Dodd-Frank Act) to observed shifts in network topology and interconnectedness. Furthermore, we employ temporal graph neural networks (TGNNs), such as EvolveGCN or TGN, to model the dynamic network and detect anomalies that may signal heightened systemic risk or financial stress. Our findings reveal a trend towards increased modularity and community size in the post-GFC period and demonstrate a discernible impact of regulatory interventions on network architecture. The TGNN-based anomaly scores correlate with historical periods of financial stress and bank failures, highlighting their potential as an early warning tool. This research provides novel insights into the long-term dynamics of interbank networks, the influence of regulation, and the application of advanced machine learning for financial surveillance.