Date & Time: June 12th, 10:30 - 11:30
Location: Seminar Room M 3170-73 at Dept. of Automatic Control, LTH 
Author: Ossian Relander and Oscar Clarke Nilsson
Title: Systemic Risk and Default Contagion in Financial Networks: Identifying Systemically Important Banks
Supervisor: Emma Tegling
Examiner: Giacomo Como

Abstract:
This thesis uses real data from the Bank of International Settlements to create a financial network for the interbank market using five different methods for network reconstruction. The goal is to analyze how defaults propagate to assess the importance of banks. By applying network theory, communicability theory, and the DebtRank algorithm, we aim to identify which banks are the most vulnerable and which propagate the largest losses to the system. We also investigate how DebtRank scores correlate with traditional centrality measures and measures derived from communicability. Our results will be compared to the Basel Committee's annual assessment of global systemically important banks.

Our findings show small differences between the network reconstruction methods. The most noticeable difference is that the minimal density method produces more resilient networks when capital buffers are small. The small-world method, on the other hand, results in networks with slightly higher losses, especially when capital buffers are in the middle range. Additionally, our results indicate that JP Morgan is the most systemically important bank in most scenarios, matching the Basel Committee's conclusions. However, we believe our methods overestimate the importance of some of the largest Chinese banks. We also show that the DebtRank centrality measure has the highest correlation with PageRank centrality and considerably lower correlation with other network dynamics measures. Finally, we conclude that impact diffusion correlates strongly with DebtRank diffusion, but impact susceptibility shows weak correlation with DebtRank vulnerability.