Agent-based Multi-layer Network Simulations for Financial Systemic Risk Measurement: a Proposal for Future Developments

Abstract

The paper addresses the topic of measuring the systemic risk and of identifying Systemically Important Financial Institutions (SIFIs) with an agent- based multi- layer network simulation. The paper starts from the shortcomings of the models currently proposed in the literature and suggests directions for future researches and guidelines to realize a methodology able to accurately model the direct network contagion channel (interconnectedness of balance sheet of financial institutions, including direct losses and liquidity hoarding), also integrating the indirect contagion channel (fire sales and bank runs), in order to reach the full representation of the financial systemic risk.