Herding and heterogeneity with feedback in an agent-based market model

Abstract

This paper presents an agent-based model to simulate the price formation process that exhibits certain empirically-observed features found in typical stock markets. The model involves a randomly evolving network of agents whose actions are determined by herd behavior and heterogeneity with feedback. Results from simulations show that it reproduces the necessary features of power-law distributed price fluctuations with exponential tails as well as clustering of market volatility, both commonly observed in stock markets. The large price fluctuations can be attributed to herding effects while volatility clustering is due to the feedback between the agents' heterogeneous subjective opinions and market activity.