Geometry of Diffusion: structure of connectivity and systemic risk in financial networks.

Abstract

Ever since the 2008 global financial crises, both scholars and practitioners have argued that the interconnectivity of financial and banking networks is a key determinant of global systemic risk. Employing a simple diffusion model, we study the effect of short network cycles on the diffusion of idiosyncratic shocks on financial networks, an unexplored but important topic of research. We show the presence of trade-offs in the conditional effect of short cycles, and find preliminary theoretical evidence suggesting that cycles might at the same time stimulate and hinder diffusion. We subsequently undertake Monte Carlo analysis using core- periphery networks to analyze which direction prevails and under which conditions. We show that network cycles play a quantitatively important but complex role in the diffusion of financial shocks. In opposition to the notion of connectivity associated to short cycles, but in line with the trade-off, we detect non-monotonicities and significant interaction effects between triangles and squares that motivate further theoretical analysis.