Information contagion within small worlds and changes in kurtosis and volatility in financial prices

An agent based artificial market is developed to determine the impact of the interaction between investors on prices. It consists of sentiment investors, a single fundamental investor and a market maker. Sentiment investors live in a small world network and have limited liquidity. They trade based on their assessment of the future direction of the market. Consistent with the social learning literature, there are two types of sentiment investors; social learners and experts. Experts only consider private information while social learners also consider the views of neighbours. It is found that the interaction between the agents generate kurtosis and persistence characteristics of volatility in returns. In addition, the level of kurtosis and volatility depends on the inter-connectedness of the network as well as the number of experts and the number of connections from these experts to social learners. Cluster coefficient and characteristic path length analysis show that kurtosis and volatility are lowest within the small world region of the network. This effect is negated as the number of experts increases beyond a threshold.