Social learning in stock markets: A lattice model

Abstract

This paper builds an artificial stock market consisting of the agents with explicit behavioral factors, by introducing a core factor, namely, “sentiment contagion”, which is a kind of “social learning”, and discusses the relation between sentiment contagion and volatility and complexity emerging from return series. In particular, the paper discusses how the emergence of critical phenomenon from micro-level interactions of agents is related to the self-enforcement of imitation propensity. The simulation results show that, the order state (market cluster) and volatility increase with the increasing of sensitivity of investors to global news, propensity to sentiment contagion and accuracy of explaining news. When the coordination reaches a critical point, a phase transition happens and asset bubble bursts with a subsequent crash.