Ambiguity aversion and rational herd behaviour

This article reviews the literature on herd behaviour in financial markets in the context of the sequential trading model and points out the importance of incorporating ambiguity into the framework. Although Ford et al. (2005) have applied the Choquet-expected-utility theory to analyse the relationship between ambiguity and herd behaviour, their model does not allow for the separation between ambiguity and ambiguity aversion, therefore how ambiguity and ambiguity aversion affect herd behaviour cannot be analysed by comparative statistics. This article adopts the smooth model suggested by Klibanoff et al. (2005), and applies Gollier's (2006) value function to describe decision makers' welfare under ambiguity. Using very general assumptions, we prove that if the value functions of market makers and traders are homogeneous, herd behaviour will never happen even if ambiguity exists; if some types of traders have different attitudes towards ambiguity from market makers, then herd behaviour will happen with a positive probability. Our numerical simulation suggests that herd behaviour is one of the reasons behind stock price bubbles, and the probability of herd behaviour is positively correlated with the ambiguity of the distribution of stock returns as well as the disparity between traders and market makers' attitudes towards this ambiguity. © 2010 Taylor & Francis.