Impact of real-time public sentiment on herding behavior in Taiwan's stock market: Insights across investor types and industries

With the advancement of information technology and the rapid growth of social media, stock forums have become increasingly popular platforms for investors to discuss and share information. Online interactions reflect not only investor sentiment but also their bullish and bearish stances. This study advances the literature by utilizing real-time, high-frequency sentiment and stance data extracted from online posts, combined with intraday transaction records for Taiwanese stocks from April 1, 2020 to March 31, 2021. We construct high-frequency indicators to dynamically capture investors' prevailing sentiment and stance and incorporate these measures into regression models to examine their relationship with herding behavior. This approach allows for a more nuanced analysis across different investor types and industry sectors, providing new insights into behavioral patterns in stock markets. The empirical results indicate that both stance and sentiment significantly affect herding behavior. Moreover, the influence of public sentiment is the highest in the electronics industry, especially among investment trusts and individual investors. These findings underscore the heterogeneous effects of public sentiment on herding behavior, highlighting variations across investor types and industry classifications.