Dynamic modeling of systemic risk in financial networks
OA Version
Citation
Abstract
Modern financial networks are complicated structures that can contain multiple types of nodes and connections between those nodes. Banks, governments and even individual people weave into an intricate network of debt, risk correlations and many other
forms of interconnectedness. We explore multiple types of financial network models with a focus on understanding the dynamics and causes of cascading failures in such systems. In particular, we apply real-world data from multiple sources to these models to better understand real-world financial networks. We use the results of the Federal
Reserve "Banking Organization Systemic Risk Report" (FR Y-15), which surveys the largest US banks on their level of interconnectedness, to find relationships between various measures of network connectivity and systemic risk in
the US financial sector. This network model is then stress-tested under a number of scenarios to determine systemic risks inherent in the various network structures. We also
use detailed historical balance sheet data from the Venezuelan banking system to build a bipartite network model and find relationships between the changing network structure over time and the response of the system to various shocks. We find that the relationship between interconnectedness and systemic risk is highly dependent on the system and model but that
it is always a significant one. These models are useful tools that add value to regulators in creating new measurements of systemic risk in financial networks. These models could be used as macroprudential tools for monitoring the health of the entire banking
system as a whole rather than only of individual banks.