The experience of financial crises in recent decades has reinforced the critical need for economic policy to anticipate and reduce the impact of unexpected, extreme, events. In this paper, we develop a simple framework based on latent regime shifts to analyze why investors may undertake overly risky investment strategies and how policymakers may attempt to constrain such behavior. This framework explains, in particular, why banks, investors, and policymakers may decide not to hedge against extreme events, even when those events are exogenously determined and have well understood probabilities and consequences. We also examine cases in which the extreme events are endogenously created by the investment strategies themselves. Our main finding is that the private costs of the sustained suboptimal investment may be bounded and small. Thus, investors may knowingly ignore or exacerbate the likelihood of extreme events, especially if they face costs to learning the structure of the financial environment in which the events are created. These results obtain both in the theoretical model and upon calibration to the US economic experience, including the recent financial crisis. Our findings provide a strong motivation for policymakers to ensure full disclosure and dissemination of information regarding probabilities and consequences of extreme outcomes.