Herding during Market Upturns and Downturns: International Evidence

This paper applies two methodologies of herding to a number of stock markets in Africa, Asia, Europe, and America, and examines how the returns behave with regard to movements in the MSCI world index. We separate the upturns and downturns in detecting the herd behavior. The results show that the herding behavior is asymmetric to the market turns. We demonstrate that herding is significantly higher during market upturns, which contradicts the earlier findings. The results of the ARCH model for herding behavior further support the asymmetric view, and the results of EGARCH(1, 1) model for herding behavior indicate that the new measure is more accurate in detecting the herding bias. Further, the Granger causality test shows that the new herding model generates both market returns and volatility. Finally, we find that the new measure implies that herding depends on four components: a constant term which means that herding exists whatever the market conditions are; a second component indicating that herding in period t depends on the herding behavior in period t – 1; a third parameter indicating that herding depends on the asymmetric reaction to downturns and upturns; and a fourth component meaning that herding in period t is dependent on the amplitude of the shock of the previous period.