Social interaction, herd behaviour and the formation of agent expectations

Survey data on agent expectations appear to experience inertia, remaining relatively stable for protracted periods punctuated with the occasional structural shift initiated by exogenous changes. The data is also characterised with an underlying level of volatility which varies over time. This paper examines if social interaction and herd behaviour, based on the social learning literature, can explain the characteristics of this dynamic process. The social learning takes place in a network with small world characteristics. Moving from an ordered to a small world network dramatically increases the level of volatility and it quickly reaches a higher state (at which point increasing the randomness of the network has little effect). Assuming that all social networks have small world characteristics then there is an inherent level of volatility in expectations formation. Increasing the influence of experts, by increasing the number of connections from these agents, also increases volatility. This may explain the variability in volatility over time. Finally, it is found that under certain network structures, where the number of connections between agents is increased, herd behaviour leads to information cascades potentially leading to the formation of speculative bubbles