Heterogeneity, learning and monetary policy

The goal of this project is to analyze key issues related to monetary
economics while relaxing the assumption of representative rational agents
(RRA) in economic models. The models will instead feature heterogeneity
and learning in expectation formation. The motivation for this line of
research is threefold. First, there is increasing evidence showing the failure
of traditional approaches to explain key features in economic and financial
fluctuations. Second, the presence of heterogeneity and large interactions
between agents, markets and macroeconomic aggregates has been widely
acknowledged. Third, monetary authorities have extensively used the socalled
unconventional monetary policy (UMP), raising many questions about
the pros and cons of such policy. This project comes with multiple
challenges, be it theoretical, as there are infinite ways to model non-RRA
behaviours, or empirical, as usual estimation techniques are not suited to
introduce heterogeneity or learning.
Three concrete and complementary projects, subject to future adjustments
over the course of the PhD have been identified. First, we will empirically
revisit the debate of “rules vs. discretion” in monetary economics, in the light
of the recent shift to UMP. To that end, we will pay a particular attention to
feature learning in agents’ behaviour in the estimation process. Next, we will
concentrate on potential side effects of UMP. Drawing on HAM models with
chartists and fundamentalists, we will more specifically investigate the
impact of UMP in explaining the proportion of the different types of agents in
the market. As a final step, we aim to estimate a structural model with
heterogeneous learning agents. Such a model could be of interest for
monetary authorities to explain fluctuations of inflation for instance. For this
project, the main challenge will be to find the right trade-off between
precision and tractability of the estimation method.