Description:
This paper investigates the predictability of stock market returns conditional on herd behaviour states (intense/adverse) using a fixed effects model to capture cross-sectional and time variability covering the European region. We show that herd behaviour negatively forecasts stock returns on average. From December 1992 to December 2020, the mean returns following an intense herding state are 0.26% lower per month over a six-month holding period than following an adverse herding state. Our results are robust to using risk-adjusted returns and a continuous herding variable. Intense herding emerges during periods of lower returns and higher volatility than adverse herding