MRN-based connectedness: A nonlinear approach for capturing systemic risk dynamics in financial systems

Measuring connectedness among financial institutions is critical for monitoring systemic risk, understanding its formation and transmission, identifying key institutions, and formulating effective regulatory policies. Traditional methods, often based on parametric models, typically represent financial relationships using linear correlations or rely on idealized nonlinear mappings, limiting their ability to capture the inherent nonlinear dynamics and complex interdependencies in financial systems. To address this limitation, this study constructs connectedness indicators using multiplex recurrence networks (MRNs). The MRN-based approach embeds time series into phase space to capture their temporal structures and leverages mutual information to quantify nonlinear dependencies among institutions. Additionally, it requires minimal preprocessing, avoids strong assumptions, and reduces reliance on precise parameter estimation. Simulation experiments demonstrate that the MRN-based approach effectively captures changes in tail dependencies across multidimensional returns, closely reflecting systemic risk dynamics. Empirical analyses of China’s publicly listed banks further illustrate its ability to track the evolution of systemic risk, identify systemically important banks, and highlight the increasing role of state-owned banks in economic adjustments. These results suggest that the MRN-based method offers advantages over VAR-based approaches, providing a more nuanced and timely reflection of systemic risk. By emphasizing the nonlinear characteristics of financial variables, this study complements prudential regulatory tools and enhances the understanding of systemic risk evolution in complex financial systems.