Abstract:

A profound comprehension of the microcosmic mechanisms underpinning enigmatic asset pricing phenomena， such as the pronounced co-movement of stock returns， holds paramount significance for the effective management of risk in the stock market and the optimization of investment portfolios. In the context of the Internet， social media has become the most important opinion field for individual investors， which has improved the information environment for investors， and also led to an increasingly complex risk contagion pattern in the stock market. On the one hand， individual investors’ anticipated returns on stocks can be significantly influenced by the diffusion of information through social media platforms， thereby potentially contributing to the observed excess co-movement in stock returns. On the other hand， stock price fluctuations can attract online attention， especially in the aftermath of a stock market crash， often accompanied by heightened stock price co-movements.From the perspective of interaction effect， this study explores for the first time the dynamic lead-lag and interaction relationships between social media information diffusion and stock price comovement through a comprehensive methodology encompassing panel vector autoregression （PVAR） modeling， and the thermal optimal path （TOP） approach. Based on the herding effect and excessive attention theories， the intermediate processes of the relationships are investigated through the stepwise regression method. Finally， robustness checks are implemented by modifying the parameters of the TOP model， examining causality between information diffusion and stock price comovement， and changing the variable metrics.Results show that the change in information diffusion among stock bars leads that the corresponding stock price correlation before 2020， and this relationship reverses when the market has a high level of fluctuation after 2020. Secondly， there is a significant two-way effect between information diffusion and stock price comovement， and the effects have a non-linear characteristic of “negative→positive convergence”. Third， the trading behavior of individual investors is an important channel through which information diffusion affects stock price comovement. Fourth， the increase in stock price correlation can attract the attention of individual investors， leading to frequent postings and replies in the corresponding stock bars， and increasing the degree of information diffusion.Consequently， the findings hold practical significance for market participants and regulators. Investors should be cautious about the information on social media platforms and stock price fluctuations， and improve their ability to recognize effective information and respond to risks， so as to avoid making wrong trading decisions due to the diffusion of noisy information and temporary fluctuations in stock prices. For the regulators， they should strengthen the monitoring of abnormal proliferation of stock information on social media platforms and excessive linkage of stock prices， so as to improve the timeliness of risk disposal and prevent systemic financial risks.The study represents the inaugural endeavor to scrutinize the dynamic interplay between the diffusion of information through social media and the occurrence of stock price comovement. In doing so， it supplements existing research focused on elucidating the impact of information diffusion engendered by the word-of-mouth effect on stock price dynamics. The research also aligns with prior work exploring the role of investor behaviors in shaping the dynamics of stock return co-movement.

Key words: stock bar, information diffusion, stock price comovement, interaction effect, mediating effect