Herd Behaviour in Efficient Financial Markets with Sequential Trades

We describe conditions on signal distributions that are necessary and sufficient for informational herding in a stylized model of sequential specialist security trading. Curiously, there can be persistent herding even with signals that satisfy the Monotone Likelihood Ratio Property. Price paths are strongly biased in the direction of the herd but prices are also very sensitive to movements against the herd. Price movements thus become more pronounced through herding. Numerical simulations indicate that the probability of herding and the level of noise trading are inversely related. Our results contrast the existing literature which found that herding with monotonic signals is impossible, and that herding is rarely accompanied by price movements. The paper thus allows a new perspective on herding in financial markets with efficient prices. We identify that the major ingredient needed for herding is that some agents find their information confusing. JEL Classification: C70, D80, D83, D84, G12, G14.