Multi-source information fusion and deep-learning-based characteristics measurement for exploring the effects of peer engagement on stock price synchronicity

Abstract

By combining financial information from the financial market with social textual information from social media, we apply a two-level information fusion approach to examine the effects of peer engagement on social media on stock price synchronicity and compare the effects between epidemic and non-epidemic contexts. On the first level, single pieces of information are fused at the firm-year level and deep learning models are used to measure the characteristics of peer engagement – informativeness, diversity, information diffusion degree, and expert proportion – which are constructed grounded in the theory of the wisdom of crowds (WoC). On the second level, all measurements at the firm-year level are fused into a full sample to conduct regression analysis. The experimental results show that peer engagement reduces stock price synchronicity. This suggests that high synchronicity could be mitigated through effective guidance from peer engagement activities. We also find that during epidemics, synchronicity is much higher, and group diversity and experts have stronger effects in lowering synchronicity, while the effects of informativeness and information diffusion are hampered. This has implications for combatting the adverse effects of epidemic outbreaks on financial markets.