Precautionary Saving and Un-anchored Expectations

This paper revisits monetary policy in a heterogeneous agent New Keynesian model where agents use an adaptive learning strategy named recursive least square learning in order to form their expectations. Due to the households' finite heterogeneity triggered by idiosyncratic unemployment risk, the model is subject to micro-founded heterogeneous expectations that are not necessarily anchored to the rational expectation path. Households experience different histories which has non-trivial consequences on their individual learning processes. In this model, supply shocks generate precautionary saving and possible long-lasting deflationary traps associated with excess saving. Dovish policies focused on closing the output gap dampen those effects which is in contradiction with previously established representative agent results under learning. Nonetheless, only price level targeting appears to resolve most of the problem by better anchoring long run expectations of future utility flows