Ranking consistency of systemic risk measures: a simulation-based analysis in a banking network model

In a banking network model, I analyse the ranking consistency of common systemic risk measures (SRMs). In contrast to previous studies, this model-based analysis offers the advantage that the sensitivity of the ranking consistency with respect to bank and network characteristics can easily be checked. The employed network model accounts, among others, for bank insolvencies as well as illiquidities, stochastic dependencies of non-bank loans as well as of liquidity buffer assets across various banks, bank rating-dependent volumes of deposits and interbank liabilities, and the funding liquidity reducing effect of fire sales of other banks. Within the assumed banking network model, I find that, in general, the ranking consistency (measured by the rank correlation) of various SRMs is rather low. A further finding is that the ranking consistency can significantly vary in statistical terms, for example for an increasing correlation between the returns of the liquidity buffer assets across banks, an increasing volatility of these assets or an increasing default rate in the non-bank loan portfolios. However, forecasting which effect a specific change in parameters, bank behavior or network characteristics has on the ranking consistency of SRMs seems to be difficult because the sign of the effect can be different for different pairs of SRMs. Furthermore, the economic significance of these changes on the overall ranking consistency as measured by Kendall’s coefficient of concordance in general is rather low.