by

In the world of economic modeling, assumptions about how people form expectations play a crucial role. Traditionally, many macroeconomic models rely on the assumption that all agents are “representative” and form expectations rationally. However, this approach often fails to capture the diversity and complexity of actual human behavior—especially when it comes to how we respond to economic policies and external shocks.

Our recent paper, “A Dynare Toolbox for Social Learning Expectations,” published in the Journal of Economic Dynamics & Control, aims to address this gap. Through a novel extension of the popular Dynare platform, we provide economists with a new toolbox to explore the impacts of social learning (SL) on macroeconomic outcomes. This research marks an important shift, enabling the integration of learning games into large-scale DSGE (Dynamic Stochastic General Equilibrium) models, thus introducing a richer, more realistic framework for modeling heterogeneous agents.

Why Social Learning?

Most economic models assume rational expectations—a setup where agents predict future outcomes based on a complete understanding of the model. While this simplifies analysis, it often generates patterns that conflict with observed behavior. For example, in real life, people’s beliefs about inflation, interest rates, and growth vary widely. These beliefs are not always perfectly rational, and they evolve over time as agents interact, observe, and learn from one another.

Social learning is a way to capture these interactions. Originating from behavioral economics, social learning models assume that people update their expectations based on their peers’ experiences and actions. Agents are influenced by their social networks, adopting strategies that they perceive as successful or imitating the most credible sources in their environment.

A Tribute to Jasmina Arifovic’s Work

This work builds on a long-standing collaboration with Jasmina Arifovic, whose pioneering work with learning models inspired many of the concepts in this toolbox. Her contributions laid the foundation for embedding social learning mechanisms in economic models. Jasmina was one of the first to demonstrate how agent-based models and genetic algorithms could capture the diversity in expectations and decision-making processes. Her approach showed that social learning can yield unique insights into macroeconomic dynamics that differ from those based solely on rational expectations.

To learn more about this project, you can access our open-access publication:

Additional References

  • Arifovic, J. (2000). Evolutionary dynamics of inflationary expectations. Macroeconomic Dynamics, 4(3), 373–394.
  • Arifovic, J., Evans, G. W., & Salle, I. (2013). Learning to believe in simple equilibria in a complex macroeconomy. Journal of Economic Dynamics & Control, 37(10), 2187–2206.
  • Coibion, O., & Gorodnichenko, Y. (2015). Information rigidity and the expectations formation process: A simple framework and new facts. American Economic Review, 105(8), 2644–2678.
  • Evans, G. W., & Honkapohja, S. (2001). Learning and Expectations in Macroeconomics. Princeton University Press.
  • Hommes, C. (2021). Behavioral and Experimental Macroeconomics and Policy Analysis: A Complex Systems Approach. Journal of Economic Literature, 59(1), 149–219.