Network Debt

Given a financial network of liabilities, we consider the following question: which agent’s potential default would implicate the most number of agents in the network structure? Sim-ply comparing each agent’s total debt completely ignores the structure of the network. For propagation effects of a potential default, owing to a creditor who does not owe anybody and owing to a creditor who is heavily in debt himself clearly do not have the same implications even when the amount owed is the same. The liability network structure must be taken into account in order to capture downstream effects. We propose a general notion which addresses this issue quantitatively. Our notion yields a ranking of all agents in the network in terms of severity of default implications. Potential regulatory applications include stress testing for banks