This paper presents a methodology to calculate the Systemic Risk Ranking of financial institutions in the European banking sector using publicly available information. The pro- posed model makes use of the network structure of financial institutions by including the stock return series of all listed banks in the financial system. Furthermore, a wide set of common risk factors (macroeconomic risk factors, sovereign risk, financial risk and housing price risk) is included to allow these factors to affect the banks. The model uses Bayesian Model Averaging (BMA) of Locally Weighted Regression models (LOESS), i.e. BMA-LOESS. The network structure of the financial sector is analysed by computing measures of network centrality (degree, closeness and betweenness) and it is shown that this information can be used to provide measures of the systemic importance of institutions. Using data from 2005 (2nd quarter) to 2013 (3rd quarter), this paper provides further insight into the time-varying importance of risk factors and it is shown that the model produces superior conditional out-of-sample forecasts (i.e. projections) than a classical linear Bayesian multi-factor model.
bank stock returns Bayesian model averaging financial networks locally weighted regression systemic risk