3% rules the market: herding behavior of a group of investors, asset market volatility, and return to the group in an agent-based model

摘要

This study presents an agent-based model of capital markets by adopting simple trading rules for bounded rational agents who maintain different expectations regarding a tipping point at which price starts to change its direction from rising (falling) to falling (rising). The effect of herding behavior on the volatility of stock market prices and the rate of return to the herding group are investigated by dividing agents into one or more groups. Herding behavior by a group of agents leads to high market volatility and high return for the agents in the group. Maximum rate of return is reached when the group size is approximately 3% of the total number of agents. This finding is consistent with the actual degree of herding behavior in markets found by empirical studies. However, the rates of return decrease when the group size exceeds 3%, and the premium of the herding group tends to disappear when the group size reaches a certain level (20%) compared with that of non-herding groups. Reducing the number of groups (or increasing the average size of the herding groups) leads to high price volatility.