Writing

This paper identifies how financial contagion spreads through physical and virtual spaces. Using an unbalanced panel of publicly traded U.S. bank holding companies from 2000 to 2023, we find that geographic proximity to failing institutions is positively associated with systemic risk across specifications. Social proximity exhibits no systematic association with systemic risk during normal times, as mixed signals wash out in aggregate. During acute stress episodes, however, social ties become a meaningful channel through which sentiment and panic propagate across the banking sector, consistent with Wake-up Call Theory and herding documented in the broader literature. These findings suggest that the channels through which systemic risk propagates extend beyond the balance sheet, and that monitoring the spatial and social architecture of banking networks may offer regulators early warning signals that traditional indicators miss.

This paper examines whether bank branching decisions respond to conditions in socially connected locations, and whether these responses change under economic stress. Using county-level FDIC branch data and Facebook's Social Connectedness Index, I find evidence of state-dependent peer effects. During normal times, expansion in socially connected counties predicts local contraction, consistent with strategic substitution and geographic reallocation. During COVID-19, this relationship reverses: peer expansion predicts local expansion instead. I instrument peer branching using a socially weighted shift-share exposure to address reflection and correlated shocks. The results suggest that banks use peer signals differently depending on the reliability of local information, offering a novel propagation mechanism for spatially correlated branching decisions, with suggestive evidence of downstream effects on local credit activity.