In March 2020, safe asset markets experienced surprising and unprecedented price crashes. We explain how strategic investor behavior can create such market fragility in a model with investors valuing safety, investors valuing liquidity, and constrained dealers. While safety investors and liquidity investors can form a symbiotic relationship with offsetting trades during times of stress, strategic interactions among liquidity investors harbor the potential for self-fulfilling fragility. When the market is fragile, standard flight-to-safety can have a destabilizing effect and trigger a “dash-for-cash” by liquidity investors. Well-designed policy interventions can reduce market fragility ex ante and restore orderly functioning ex post.

We model an anxious agent as one who is more risk averse with respect to imminent risks than with respect to distant risks. Based on a utility function that captures individual subjects’ behavior in experiments, we provide a tractable theory relaxing the restriction of constant risk aversion across horizons and show that it generates rich implications. We first apply the model to insurance markets and explain the high premia for short-horizon insurance. Then, we show that costly delegated portfolio management, investment advice, and withdrawal fees emerge as endogenous features and strategies to cope with dynamic inconsistency in intratemporal risk-return trade-offs.

We estimate a structural model of resource allocation on work hours of Federal Reserve bank supervisors to disentangle how supervisory technology, preferences, and resource constraints impact bank outcomes. We find a significant effect of supervision on bank risk and large technological scale economies with respect to bank size. Consistent with macroprudential objectives, revealed supervisory preferences disproportionately weight larger banks, especially post‐2008 when a resource reallocation to larger banks increased risk on average across all banks. Shadow cost estimates show tight resources around the financial crisis and counterfactuals indicate that binding constraints have large effects on the distribution of bank outcomes.

We analyze the effects and interactions of monetary policy tools that differ in terms of their timing and their targeting. In a model with heterogeneous agents, more productive agents endogenously expose themselves to higher interim liquidity risk by borrowing and investing more. Two inefficiencies impair the transmission of monetary policy: an investment- and a hoarding inefficiency. Heterogeneous agents respond disparately to ex-ante, conventional and ex-post, unconventional monetary policy. However, we show that the two policies are equivalent due to the endogeneity of hoarding. In contrast, targeted interventions such as discount-window lending can alleviate both inefficiencies at the same time.