The nonlinear risk amplification of overlapping structures in multiplex financial networks

To investigate the origins of systemic risk in multiplex financial networks, this paper addresses a central question: Do overlapping structures—where financial institutions form same-direction linkages with the same counterparties across different markets—generate risk amplification effects? By developing a theoretical model and conducting empirical tests using data from 32 listed Chinese banks, we confirm that overlapping structures are a key driver of risk amplification. This effect exhibits an asymmetry of being “easily activated but hard to disappear”, meaning it can be triggered by minor shocks yet only subsides when the system is near a complete collapse. Our empirical analysis reveals that assessments based on single-layer networks could underestimate systemic risk by a factor of up to 1.72, highlighting the significant limitations of traditional regulatory frameworks. Furthermore, we decipher the non-monotonic structural dependence of this amplification. Contrary to simple linear expectations, risk amplification shows an inverted U-shaped relationship with both connectivity and leverage. Finally, we use a leverage multiplier as a calibration device to locate the empirical interbank system on the overlap-induced amplification curve, illustrating the procedure with data from the Chinese interbank market.