What is so special about banks that their demise often triggers government intervention? In this paper we develop a simple model where, even ignoring interconnectedness issues, the failure of a bank causes a larger welfare loss than the failure of other institutions. The reason is that agents in need of liquidity tend to concentrate their holdings in banks. Thus, a shock to banks disproportionately affects the agents who need liquidity the most, reducing aggregate demand and the level of economic activity. In the context of our model, the optimal fiscal response to such a shock is to help people, not banks, and the size of this response should be larger if a bank, rather than a similarly-sized nonfinancial firm, fails.