Network banks exposures and risk spillovers in the Euro area

The US Subprime and European Sovereign bond crisis, sparkled a renaissance in the research fields related to contagion and Systemic Risk. Even if those concepts are almost 20 years old and largely related to currency crisis still there is not complete consensus around their definitions. For this reason we refrain from the use of these terms and refer to the phenomena we are willing to investigate as variance and covariance spillovers. We define, in line with the literature, a variance spillover as the change of the variance (Risk) of a recipient entity due to the change of the variance of another entity in the period before. This definition is signed, directional, include the time dimension and can account for feedback effects. It excludes instead a systematic shift of variances due to the contemporaneous exposure to a common factor. A Covariance Spillover is a change in the covariance (dependence) of two entities due to the change in the variance (risk) of a third entity in the previous period or to the change in the covariance among other two entities in the previous period. These definitions are clearly related to contagion and Systemic Risk but are not capturing all the evidence generated by contagion and systemic risk. In this paper in addition, the emphasis is not only in the detection of risk spillover effects but also on the investigation of their fundamental transmission channels of these risk spillovers thanks to the use of economically based distance relationship among entities.

More specifically, we propose a spatial approach to model risk spillovers using financial time varying proximities based on actual claims among entities. We show how these methods could be useful in (i) isolating influential and fragile entities and important risk channels, (ii) investigating the role of portfolio composition in risk transfers, and (iii) computing target exposures able to reduce system volatility. Our empirical application uses banks foreign exposures provided by BIS as a proxy for the euro area cross country holdings.

We find that Ireland, Greece and France are playing a central role in spreading risk in the European stock markets and this spillover effect can be traced back to a physical claim channel: banks foreign exposures.