A new dimension of bank complexity: rescue agreements and default contamination

We introduce a new type of complexity, arising from the rescue arrangements among members of a banking group. Rescue is different between parent-subsidiary, branch and ring-fenced organizations, and we argue that the complexity it generates is highest in the first, mild in the second and null in the third. Banks have incentives to choose a particular, more or less complex, organizational structure and to lever it up consistently. We measure the risk associated to each form of rescue complexity with the expected discounted loss it generates, since this form of complexity affects default risk. We calibrate to US BHCs. We show that the parent-subsidiary has the highest complexity risk, as soon as we recognize the heterogeneity in leverage of BHCs. Regulatory interventions that aim at ring fencing do reduce complexity and its risk, but are likely to reduce value to stakeholders.