Brown Bag Seminar with Professor George Evans, University of St Andrews Adaptive learning and cognitive discounting: implications for macroeconomic dynamics and policy

Abstract: Nonlinear New Keynesian models can have multiple rational expectations steady states, and it is possible for there to be more than one steady state that is locally stable under adaptive learning (AL). Gabaix has recently introduced cognitive discounting (CD) as a behavioral modification that has the potential to resolve some well-known puzzles in long-horizon models. We argue that boundedly rational agents would naturally combine AL and CD, leading to novel results. This can yield a model with two locally stable steady states: the steady state targeted by policy, and a ‘Great Depression' steady state. We explore the dynamics and policy implications of this framework. In particular, starting at the targeted steady state, an unlucky series of shocks can push the economy away from the targeted steady state and into the basin of attraction of the stagnation steady state. We also examine how policies can dislodge the economy from stagnation and return it to the target steady state.