Agent Based Modeling and Neoclassical Economics: A Critical Perspective*

Abstract

Agent Based Modeling and Neoclassical Economics based modeling naturally generates complexity whereas neoclassical economics is incompatible in principle with complexity. The reasons that preclude complexity in neoclassical economic models also ensure that neoclassical economics cannot describe any society ever observed or that could ever be observed. The meaning of complexity has been developed, mainly by physicists, to cover unpredictable, episodic volatility and also particular network topologies. In both cases there are nodes representing the components of a system and there are links among the components that can represent interactions amongst those components. Unpredictable, episodic volatility can result from particular forms of behavior by components and the interactions amongst those components. I am not aware of any investigations into relationships between that type of complexity and network topology. The point to be made here is that the core assumptions and the methodology of conventional neoclassical economics preclude the emergence of episodic volatility and render social network topology inconsequential. When elaborated with heterogeneous agents, network topologies might have some effects on the outputs from computational neoclassical economic models – but, again, I am not aware of any systematic investigations into this possibility