[3U1-IS-3] Agents

[3U1-IS-3-02] Stability analysis of a financial system

〇Igor Zavialov¹, Kazushi Ikeda¹ (1. Nara Institute of Science and Technology)

[[Online, Working-in-progress]]

Keywords:systemic risk, Lyapunov exponents, Value-at-Risk, graph neural network

Systemic risk is a complex system phenomenon that defines the possibility of a system collapse after a failure of one of its components. In this research, we consider a systemic risk of a financial system. Due to the complexity of a financial system, various actors, and their relations, the precise mathematical definition of systemic risk is difficult to formulate. However, we can indirectly define a financial system's stability using Lyapunov exponents (LEs). By simulating system dynamics with parameters, we analyzed different states of the system and concluded about system stability. Using Wolf algorithm, we calculated Lyapunov exponents of the system dynamics with the setting of two different topologies of the financial network. We found that ring topology is less stable than the fully-connected topology that supports the related research in the field of economics.