Early Warning and Systemic Risk in Core Global Banking: Balance Sheet Financial Network and Market Price-Based Methods

Abstract

We analyse systemic risk in the global banking system using market price-based methods and the asset-liability network approach. For the latter we use the BIS consolidated data for exposures of 18 national banking systems to the banking sector debt of the same countries relative to their respective equity capital over the period 2005Q4- 2014Q4. The network based Systemic Risk Index (SRI) extends the spectral eigen-pair method of Markose (2012) and Markose et al. (2012), which treats network failure as a dynamical system stability problem. We also provide a comprehensive application in a cross-border setting of well known market price based SRIs, viz. MES (Marginal Expected Shortfall), ∆CoVaR, (Delta Conditional Value at Risk) and SRISK. In view of the large UK and European cross-border exposures to the US mortgage backed securities which led to widespread bank capital losses during the 2007 Global Financial Crisis (GFC), it is important to see if the SRIs can give early warning unlike paradoxical risk measures which were very low in the run up to the GFC when leverage was peaking. The network based eigen-pair method simultaneously gives early warning of instability of the global banking system in terms of tipping points identified by regulatory capital thresholds and also the centrality measures for both systemically important and vulnerable banking systems. Market price-based SRIs tend to be contemporaneous with the crisis and they are found to covary with standard risk management measures, such as VaR and betas.