The effect of return jumps on herd behavior

Return jumps increase in frequency and are considered to reflect the arrival of non-trivial information. We thus question the impact of return jumps on herd behavior in global equity markets. New herding detection models that incorporate return jumps overcome multicollinearity and sample-splitting problems found in prior studies. While the traditional model does not detect herd behavior in most cases, our augmented model incorporating return jumps detects more cases of herd behavior. We find the strongest effect on jump and negative return dates, supporting existing evidence of asymmetric herd behavior. In general, incorporating return jump dummy variables underlines the existence of herd behavior and an information cascade-argument helps explain this phenomenon well. Our results are robust to altering the identification of return jumps and the herd behavior model.