Default propagation in customer-supplier networks: An agent-based model approach to the isolation of companies in default

The analysis of customer-supplier relationships is crucial to analyze both risk assessment problems and operational supply-chain management optimization. Although the underlying network properties are essential to analyze default propagation and default cascades, empirical changes of the network or realistic models for network topology evolution have not been previously taken into account. Here, lead by experimental evidences, we propose a variation of the Microscopic Markov-Chain Approach (MMCA) model to study the details of default propagation in customer-supplier networks. We modify companies’ behaviour towards default to make them act according to several heuristics that better reflect real-world behaviour, as observed on a real customer-supplier network. The so-called isolation of the defaulted is simulated through several experiments where defaulted companies are obliged to reshuffle their outgoing edges (payments) towards other companies when their suppliers are not willing to risk their goods anymore. Our results show how exposure of different sectors to default changes according to the policies that the agents undertake. Besides, we show that individual agents decisions are highly relevant to decrease the global network sensitivity to default.