The relationship of aggregate herd behavior and retail investor attention : a multinomial logistic regression

This paper investigates the relationship between aggregate market herding and investor attention in seven selected equity markets over the period of 2004-2019. The multinomial logistic regression employed in this study provides a more direct, comprehensive, and straightforward test than other methodologies found in existing studies, as it demonstrates a true occurrence of herding. Investor attention is positively (negatively) related to (anti-)herd behavior, in which information obtained from internet searches is associated with rational and unintentional herding. Thus, internet search improves stock market efficiency. Nonetheless, the role of investor attention is weaker during negative market returns and global financial crisis periods because investors are less attentive to their psychological discomfort explained by the Ostrich effect. The results call for policymakers to gear to the digital economy.