Riccardo Faini CEIS Seminars - Antonio Cabrales
Network Formation and Contagion in the Presence of Heterogeneous Risks
Venerdì 27 Marzo 2020 h. 12:00-13:30
Room A - 1st Floor – Building B
Facolta' di Economia
Universita' degli Studi di Roma 'Tor Vergata'
Via Columbia 2, Roma
Antonio Cabrales (University College of London)
This paper explores the problem of contagion in financial networks. In previous work (with Piero Gottardi and Fernando Vega-Redondo), we had investigated optimal financial network structures in the presence of shocks. That work did not explore in sufficient depth the question of network formation. The new work focuses on inefficiencies that arise in the process of decentralized network creation, because of contracting externalities that arise because of transmission of shocks. We model a financial network in which some firms are good risks and others are bad risks. The good risks generate insurance/hedging opportunities with one another, and as a result, good firms enjoy direct and indirect risks with one another. Bad risks on the other hand, benefit from having a connection with a good firm, which can bail them out and avoid bankruptcy. However, they are a cost to both direct and indirect connections, as they can create a chain of financial shocks and defaults. The key assumption we make is that contracting is bilateral, so that a bad risk can compensate a good risk for the direct harm inflicted, but indirect connections do not get a compensation.
We first show that the optimal networks in this context are often of a form in which the good risks from large connected components. Those components have a few bad risks attached, but not many and often none, since the costs are borne by all the good risks of the component. The equilibrium networks are very different. They have typically many more bad risks attached, because of non-internalized negative externality. They also look like the core-periphery structures we often see in the real world. At the core, we have the good risks, linked together. In the periphery, we have the bad risks. Another consequence of the negative externality is that components are also smaller than the efficient ones. The reason is that a small size limits the negative costs of contagion from the bad risks.