Modelling Social Learning in an Agent–Based Baseline Macro Model

We analyse the dynamics resulting from social learning in a simple general equilibrium (GE) model, whose structure is similar to the one underlying macroeconomic models of the New Macroeconomic Synthesis. The economy is composed by households and firms who exchange labour and consumption goods in the corresponding markets with potential rationing. Market interactions are modelled through an Agent-Based Model (ABM), in which heterogeneous agents operate under bounded rationality and social learning. We discuss the robustness of the symmetric optimal equilibrium of this GE model, by testing the ability of agents to coordinate on it through this learning scheme in the absence of substantive rationality and perfect information. Our results indicate that one needs to "discipline" social learning in order to obtain realistic dynamics out of this simple model. However, full coordination of agents on this equilibrium remains hard to obtain, and social welfare loss remains. Our results are quite robust and do not depend on our specific modelling assumptions