Social media sentiment, investor herding and informational efficiency

Abstract We investigate the impact of social media sentiment on the informational efficiency of financial markets. In particular, we examine the relationship between sentiment extracted from Twitter posts and two widely used efficiency measures: return autocorrelation and variance ratio. Our results show that more positive sentiment is associated with increased return autocorrelation and variance ratio the following day, suggesting a decline in informational efficiency. Furthermore, we demonstrate that the effect of social media sentiment on informational efficiency is driven by the herding behavior among traders, with more positive sentiment leading to more pronounced herding activity. These findings suggest that elevated social media sentiment contributes to a reduction in the quality of the information environment, ultimately leading to informationally inefficient equity prices.