A Network Theory of Financial Cascades

Abstract:     
International financial markets instability, as exhibited by persistent periodic crises, is a concern for all market participants. This paper presents an alternative theory of financial crises using recent developments in network theory. Simulation is used to apply the theory to short-term currency crises. The simulated model, despite being parsimonious, is capable of generating non-Gaussian complex outcomes, including cascades. Crisis dynamics are shown to be predominately generated by the pattern of inter-agent linkages and agent characteristics.

We conclude use of agent-based complex network models will improve understanding of the dynamics of market crises and should form part of financial dynamic analysis. While periodic crises may be inevitable, we argue market participants may be able to use knowledge of topology to reduce losses from future crises.

Number of Pages in PDF File: 33

Keywords: Information cascades, foreign exchange dynamics, complex networks, heterogeneous agents, simulation, financial risk

JEL Classification: C15, F31, G12

working papers series