Abstract:

Banks' obligations to each other involve counterparty risks. In case of a failure, the losses of counterparties of the failing bank are exacerbated by additional bankruptcy costs. By also taking the contagion risk into account, banks' potential losses incentivize them to rescue each other whenever rescues are less costly than absorbing the losses. Endogenously arising rescues reverse the standard intuitions from the financial contagion literature: A system-wide contagion risk does not necessarily imply financial instability and, surprisingly, leads to greater stability in certain networks where banks more than undo the contagious failures and take actions against any potential failure. In a framework where capital transfers between banks are more efficient than government bailouts, I characterize welfare-maximizing networks and show that they are connected through i) intermediate levels of interbank liabilities per bank, and ii) no clustering of interbank exposures among any subset of banks. Consequently, financial stability is determined by the potential bankruptcy losses internalized by banks and the loss absorption capacity of the system (i.e., banks' aggregate capital). The results provide additional insights into the historical debate on bank rescues and help us better understand the implications of current interbank regulations. The findings also offer plausible explanations for the selective rescues in the 2007-2009 period.