Shapley value-based allocation of system-level risk in the Hungarian banking system

The aim of the paper is to estimate the allocation of system-level risk and capital buffers due to interconnectedness (due to the correlation structure) of Hungarian Other Systemically Important Institutions using a market information-based method. The approach of Tarashev et al. (2015) is followed, with the Shapley value used to allocate the risk measure (VaR and ES) calculated from the aggregate loss distribution of the Hungarian banking system. The methodology applied is employed to investigate the role of size, probability of default and correlation structure on systemic risk. Our results confirm the important role of size in systemic risk. At the 99.9% confidence level, the Shapley value-based share of system-level risk for large banks exceeds the weights calculated based on asset value, while for smaller banks it is below the weights calculated based on asset value. As the confidence level is reduced, a significant shift in the allocation is observed due to the increasing importance of probabilities of default. The allocated share of banks with low probability of default diminishes, while the share of institutions with a higher probability of default increases. Furthermore, it can be observed that for institutions with lower asset values and high comovement with the common factor, systemic risk due to interconnectedness explains a significant share of the allocated system-level risk. Conversely, for banks with higher asset values only approximately 10% of the estimated capital buffer is allocated due to systemic risk from interconnectedness, suggesting an important role for other factors (e.g. size, probability of default and foreign ownership).