Dynamics of Phase Transitions in Expectations for Financial Markets: An Agent-Based, Multicomponent Model

Abstract

We elaborate on behavioral foundations for agent expectations in financial markets and propose a multicomponent model. Components in bounded rational processing and imitative processing that can underlie frequently cited “herding” in financial markets are proposed in the agent-based model. We define a transition function for the dominance of a component in the distance between market price and price metrics based on fundamental value. The multicomponent model is then implemented in a network model of interacting agents. Our numerical exercises with the proposed model demonstrate the extensively cited property of phase transitions between ordered and disordered states that are commonly observed to presage critical points in financial markets is demonstrated. Additional numerical exercises examine the strength of neighbor connectivity in the component that generates “herding” as it relates to cycle in financial markets. Finally, we consider a methodology to operationalize the strength of neighbor connectivity as a policy variable for managing levels of this variable toward efficient markets.