Purpose: This study analyzes the structure of mutual dependence and market risk clustering in Iran's banking industry. Specifically, it identifies connectivity hubs, bridging institutions, and behavioral groups of banks, examines the time-varying co-movement of their returns, and provides actionable recommendations for network-based risk management and financial stability monitoring.



Methodology: Weekly log returns of nine listed banks over 2015–2026 (593 weeks) were computed and, after percentile winsorization, a Pearson correlation matrix was estimated. The matrix was converted into a distance metric to build a minimum spanning tree and a threshold correlation network (0.25), yielding centrality measures. Ward's hierarchical clustering grouped the banks, and 52-week rolling correlations tracked temporal dynamics.



Findings: Correlations are positive yet heterogeneous; the highest coefficient links Saderat and Tejarat (0.773), while the lowest involve Shahr. In the minimum spanning tree, Saderat (betweenness 23) is the primary gatekeeper and Tejarat the secondary hub. In the strong network, Tejarat (strength 3.95) and Saderat (3.88) hold the greatest connection intensity. Clustering yields three groups: a six-bank core, a middle layer (Eghtesad Novin and Post Bank), and a single peripheral bank (Shahr). Rolling correlations rose during COVID-19 (mean 0.438) and the regional tensions (mean 0.430) relative to the overall mean (0.367), indicating intensified co-movement and temporary weakening of diversification during crises.



Originality/Value: By jointly applying network tools to long-horizon weekly returns and covering recent shocks, the study offers a multi-layered map of Iranian banking connectedness and distinguishes behavioral patterns from connection intensity—an approach rarely adopted in prior studies and complementary to interbank-data research.