We study the optimal design of stress scenarios. A principal manages the unknown risk exposures of agents by asking them to report losses under hypothetical scenarios before taking remedial actions. We apply a Kalman filter to solve the learning problem and we relate the optimal design to the risk environment, the principal’s preferences, and available interventions. Applied to banking stress tests, optimal capital requirements cover losses under an adverse scenario while targeted interventions depend on covariances among residual exposures and systematic risks. Our calibration reveals that information is particularly valuable for targeted interventions as opposed to broad capital requirements.
delegation complementarities stress test information bank regulation filtering learning stress tests information bank regulation filtering learning