Loss Sequencing in Banking Networks: Threatened Banks As Strategic Dominoes

Abstract

We demonstrate in a stylized banking network that a single large loss has the potential to leave markedly different impacts on the financial system than does a sequence of moderate losses of the same cumulative magnitude. Loss sequencing matters because banks make strategic bailout decisions based on their myopic assessment of losses, yet these decisions are highly consequential to subsequent decisions and eventual losses at other banks in the network. In particular, the network mechanism enables banks to choose to bail out their creditors after every moderate loss incurred in a sequence, while walking away from the creditors should they experience a single large loss. Government policy can force threatened banks to liquidate or sell themselves or, at the opposite pole, can bail out some such banks or overlook their threatened status. The former policy would concentrate a string of losses into a single large event; the latter could prevent a massive single loss at the expense of multiple subsequent smaller losses. As this analysis shows, either policy could prove optimal depending on identifiable circumstances. These findings have important implications for ongoing policy debates that emanated from the 2008 meltdown.

Contagion in economic networks: a data-driven machine learning approach

“…Our findings corroborate those of Tran, Vuong and Zeckhauser (2016). In a stylized banking network, the authors show that sequential losses entail a smaller total loss to the system than a single larger loss of the same cumulative magnitude.…”