In this paper, we consider how the intensity and channels of the relation between social networks and bank loyalty vary according to the state of the economy. We analyze bank exit over the period 2005–2012 for over 300,000 retail clients of a commercial bank that experienced a bank run in 2008 due to a shock in solvency risk. The unique and rich data we constructed in close collaboration with the bank enables us to distinguish different sorts of family networks from neighborhood networks, while controlling for a wide range of client-level and branch-level characteristics and events. Using a proportional hazards model, we show the importance of family networks. In times of financial distress, family networks become even more important and retail clients take weaker, less direct social relationships into account.