Quantifying the importance of different contagion channels as sources of systemic risk

A framework that allows computing contagion effects from both direct exposure contagion and overlapping portfolios is presented. The effects of the latter are broken down into loss correlation, effects from fire sales and mark-to-market accounting. The impact can be quantified for any single contagion channel as well as when multiple channels are jointly active. The model can be used to compute contagion losses that are consistent with a given macroeconomic scenario and thus provides a macroprudential extension for microprudential stress tests. Empirical results for a real-world banking system suggest that contagion effects stemming from direct exposures have the highest loss contribution.