Robust optimal macroprudential policy

We consider how fears of model misspecification on the part of the planner and/or the households affect welfare gains from optimal macroprudential taxes in an economy with occasionally binding collateral constraints. In this setup, the decentralized equilibrium may differ from the social planner's equilibrium both because of the pecuniary externalities associated with the collateral constraint and because of the paternalistic imposition of the planner's beliefs when designing policy. When robust agents have doubts about the model, they create endogenous worst-case beliefs by assigning a high probability to low-utility events. The ratio of worst-case beliefs of the planner over the household's captures the degree of paternalism. We show that this novel channel could render the directions of welfare gains from a policy intervention ambiguous. However, our quantitative results suggest that doubts about the model need to be large to make a “laissez-faire regime” better than an intervention regime.