Volatility clustering and herding agents: does it matter what they observe?

Authored by Ryuichi Yamamoto

Date Published: 2011-05

DOI: 10.1007/s11403-010-0075-5

Sponsors: No sponsors listed

Platforms: No platforms listed

Model Documentation: Other Narrative Mathematical description

Model Code URLs: Model code not found

Abstract

Recent agent-based models have demonstrated that agents' herding behavior causes volatility clustering in stock markets. We examine economies where agents herd on others, yet they have limited sets of information on other agents to imitate. In particular, we conduct experiments on economies with agents with different levels of information sharing where agents can imitate: (1) the strategies of others but with an error, (2) the strategies of only a fraction of agents, or (3) the strategies of others, but update their parameters only by a proportion. In each experiment we change the likelihood that agents make errors to copy the strategy of others, the fraction of agents to herd, or the proportion of the parameter that agents update, in order to examine the effect of the different degrees of information sharing on volatility clustering. We show that volatility clustering tends to disappear when agents have limited information on the strategies of others, and agents need to imitate the strategy details of others in order to generate the clustered volatility.

Tags

Volatility clustering Agent-Based Learning herding