Early warning of systemic risk in global banking: eigen-pair R number for financial contagion and market price-based methods

Abstract

We analyse systemic risk in the core global banking system using a new network-based spectral eigen-pair method, which treats network failure as a dynamical system stability problem. This is compared with market price-based Systemic Risk Indexes (SRIs), viz. Marginal Expected Shortfall (MES), Delta Conditional Value-at-Risk (Delta-CoVaR), and Conditional Capital Shortfall Measure of Systemic Risk (SRISK) in a cross-border setting. Unlike paradoxical market price based risk measures, which underestimate risk during periods of asset price booms, the eigen-pair method based on bilateral balance sheet data gives early-warning of instability in terms of the tipping point that is analogous to the R number in epidemic models. For this regulatory capital thresholds are used. Furthermore, network centrality measures identify systemically important and vulnerable banking systems. Market price-based SRIs are contemporaneous with the crisis and they are found to covary with risk measures like VaR and betas.