Modeling shock propagation and resilience in financial temporal networks

Area of Science:

• Network Science
• Econometrics
• Financial Modeling

Background:

• Modeling shock propagation in temporal networks is vital for understanding systemic risk.
• Existing research often overlooks node-centric shocks, focusing instead on link disruptions.
• Node connectivity and stability are frequently impacted by real-world shocks.

Purpose of the Study:

• To develop a novel framework for analyzing the impact of node shocks on network metrics.
• To analytically compute the Impulse Response Function (IRF) for node-initiated disturbances.
• To apply this methodology to real-world financial network data.

Main Methods:

• Utilized a vector autoregressive (VAR) framework built upon the configuration model.
• Developed a nonlinear VAR model where IRF depends on network state and shock size.
• Combined maximum likelihood estimation and Kalman filtering for parameter dynamics and IRF computation.

Main Results:

• Successfully computed the Impulse Response Function (IRF) for node shocks in a temporal network.
• Demonstrated a nonlinear relationship between shock size and network metric response.
• The IRF is shown to be state-dependent, varying with the network's condition at the time of the shock.

Conclusions:

• The proposed VAR framework offers a powerful tool for analyzing node-centric shocks in temporal networks.
• The novel econometric method provides accurate estimation of latent parameter dynamics and IRFs.
• Application to interbank deposit markets highlights the model's utility in assessing financial systemic risk.