Does regulatory bank oversight impact economic activity? A local projections approach

Existing research generally finds that the magnitude of the effect of supervisory rating shocks on real economic activity is small and short-lived. Theoretical work on this topic is also limited. This paper addresses both issues. First, we develop a model that combines a signal extraction problem with a costly adjustment of a policy instrument to demonstrate that bank supervisory changes can be decomposed into a “systematic” component (driven by fundamental banking and macroeconomic shocks) and a “non-systematic” component (driven by shifts in uncertainty). Second, we use the local projections approach to estimate impulse responses from a vector autoregression (VAR) model. We show that the effect of supervisory stringency shocks is larger than the one estimated with the standard Cholesky structural VAR approach. We find that the effects are asymmetric: bank downgrades lead to a pronounced decline in real activity, while upgrades do not result in its increase. The linear framework averages out these effects, overstating the impact of upgrades, and understating that of downgrades. Furthermore, we document the presence of nonlinear effects for the downgrade shocks, as their impact increases disproportionately with its size. Such effects are not observed for upgrade shocks. In effect, this demonstrates that, similar to monetary policy, macroprudential policy may be "pushing on a string"; this is, it may be effective at curbing real activity but not at stimulating it. Finally, we demonstrate that our results are robust to the inclusion of a variety of controls, including other macroeconomic variables, banking variables, and the real-time forecast of future macroeconomic activity.