A bank panic is an expectation-driven redemption event that results in a self-fulfilling prophecy of losses on demand deposits. From the standpoint of theory in the tradition of Diamond and Dybvig (1983) and Green and Lin (2003), it is surprisingly difficult to generate bank panic equilibria if one allows for a plausible degree of contractual flexibility. A common assumption employed in the standard banking model is that returns are linear in the scale of investment. Instead, we assume the existence of a fixed investment cost, so that a higher risk-adjusted rate of return is available only if investment exceeds a minimum scale requirement. With this simple and empirically-plausible modification to the standard model, we find that bank panic equilibria emerge easily and naturally, even under highly flexible contractual arrangements. While bank panics can be eliminated through an appropriate policy, it is not always desirable to do so. We use our model to examine a number of issues, including the likely effectiveness of recent financial market regulations. Our model also lends some support for the claim that low-interest rate policy induces a “reach-for-yield” phenomenon resulting in a more panic-prone financial system.