Structure and Stability in Payment Networks – A Panel Data Analysis of ARTIS Simulations

The purpose of this study is the investigation of the importance and impact of network structure, both, at the network level across days and at the node level across days and scenarios (stricken ARTIS participants) for the stability of payment systems in the face of operational shocks. The analysis is based on a large number of simulations of the Austrian large-value payment system ARTIS that quantify the contagion impact of operational shocks at participants' sites. The analysis uncovers that only few payment system participants are systemically important and that contagion displays substantial variation across time and across scenarios. A subsequent panel data investigation tries to explain the variation across time and network participants by structural differences of the payment network across time and the position of the stricken account within the network. It uncovers that (i) standard variables such as liquidity and liquidity loss can explain a substantial fraction of variation, both, across time and across scenarios, that (ii) the structure of the network itself adds very little and (iii) the position of the stricken account within the network indeed contributes somewhat to explaining the variations of contagion. Relative explanatory power is higher when the analysis focuses on contagion measured by the number of banks with unsettled payments or the value of unsettled payments than in the case of the measure based on the number of unsettled payments. In light of the fact that those structural indicators add only little – in terms of explanatory power – to the more traditional measures of the role of an individual participants in the payment system (value and volume of payments) we conclude that at this stage network indicators seem to be of limited use for stability analysis