Working Paper

Reconstructing Economics: Agent Based Models and Complexity

Download paper

In 1803 Louis Poinsot, a French physicist, wrote a book of great success, Elements de Statique, which was destined to have practical and social influences unimaginable to the same author.

All this is due to the work of Leon Walras who took as a reference the system of simultaneous and interdependent equations of Poinsot with the introduction of the auctioneer, a device that allows to reduce the economic agents to atoms, devoid of any phenomenon of learning or strategic behavior. We were, around 1870, in the midst of classical mechanics and reductionism. The physics of interacting microelements was still to come, and economics found it trapped in the “equilibrated” view of Walras. And from there, the union of axiomatism and non-falsifiability led to the Lakatosian degeneration of the paradigm of mainstream economic theory.

The increasingly evident crisis of the dominant paradigm is manifested also through the “support” offered by other disciplines, from biology to chemistry, from neurology to physics. The contribution made to economic research from econophysics consists especially in an approach that makes extensive use of the experimental methodology and that operates on the data, often of very high frequency, related to the real markets, deriving empirical regularities (not laws)1 and phenomenological models. Statistical physics provides useful tools to analyze systems composed of many heterogeneous agents (atoms) that interact in obedience to microscopic law. Unlike economic agents, however, atoms are not able to adopt either learning or strategic behavior, both of which are derived from interaction.