Towards a Microfounded Agent-Based Macroeconomic Model

Abstract

The literature on agent-based macroeconomics has grown substantially over the past decade. However, a typically small set of common features impedes an instructive comparison of model results and obscures the implications of specific model assumptions. While validation and model selection techniques can contribute to consolidating the literature and thereby achieve a more concerted research agenda, I propose a different route in this paper. By transcribing the common core of modern DSGE models into an agent-based framework, a well-known and parsimonious representation of the economy is developed that can serve as a point of departure for studying a wide range of research questions. The flexibility of the agent-based approach thereby helps to address common critiques of mainstream macro, such as the complexity-limit to microfoundations, the fallacy of composition and the lack of realism associated with rational expectations. To demonstrate this flexibility, simulations focus thematically on the aggregate impact of non-financial uncertainty, which has proven to be very challenging to incorporate into a DSGE setting. Simulation results show that an increase in uncertainty leads to a new equilibrium with unemployment. A reduction of nominal wages, discretionary fiscal and monetary policy can all lead to economic recovery. However, each measure bears its own risk. Furthermore, when households learn about uncertainty from past experience, discriminatory behavior by firms on the labor market can result in a strong and detrimental impact on aggregate dynamics.