Modelling realistic economic agents

Paolo Pellizzari and Friederike Wall are working on creating the basis for a unifying model of information-based decision-making by boundedly rational economic agents within the scope of a new research project. By means of this project, the research team aims to contribute to the continued advancement of agent-based computational economics (ACE). In doing so, the two researchers are simultaneously reinforcing the long-standing collaboration within the framework of the strategic cooperation between Ca’ Foscari University of Venice and the University of Klagenfurt.

Traditional schools of economic thought argue that economic actors act in an ideal manner consistent with perfect rationality. In contrast, agent-based computational economics (ACE) builds on the famous concept of bounded rationality.allowing to capture less “gifted” and thus more realistic actors. Among other things, these actors are characterised by limited memory and information-processing capacities. At the same time, gathering and processing information usually incurs costs. However, given bounded rationality, these costs are not reflected in mathematical optimisation calculations. To address this, agent-based modelling and simulation aims to provide more realistic insights into economic processes.

However, modelling and simulation involving non-perfect agents is significantly more complex. “…there is one way to be rational but many ways to depart from rationality”, Friederike Wall (full professor at the Department of Management Control and Strategic Management in Klagenfurt) cites the US-American scholar Robert Axtell. While remarkably high predictive accuracy has already been achieved for so-called “zero-intelligence agents” with respect to financial markets, it is difficult to extend the intelligence of an agent in the corresponding agent-based models in such a way that decisions are based on the level of sophistication common to humans. Criticism of agent-based simulation centres on the potential arbitrariness of models involving boundedly rational actors.

In a joint research project, Friederike Wall and Paolo Pellizzari (full professor of Mathematics for Economics and Social Sciences at Ca’ Foscari University in Venice) aim now to lay the foundations for a unifying model of those decisions made by boundedly rational economic agents that relate directly to the (limited) information itself. The focus will be on the use (or non-use) of so-called “noisy” information or the costly acquisition (or non-acquisition) of additional information.

This project builds on several years of collaboration between Friederike Wall and Paolo Pellizzari. Further reciprocal research visits to Venice and Klagenfurt are scheduled to follow in the spring and early summer of 2023.