Online social interactions and financial markets

This dissertation examines the relationship between online social interactions and financial market outcomes by analyzing user discussions from Reddit. It consists of three empirical studies that investigate how sentiment dissemination, belief disagreement, and thematic attention shifts emerge in online communities and influence key market variables. The first study shows that a small number of highly active users drive sentiment dissemination within Reddit's stock market discussions. It also finds that network-level sentiment has short-term predictive power over stock returns. The second study focuses on belief disagreement and compares its effects across the stock and Bitcoin markets. Results indicate that disagreement is strongly associated with increases in volatility, trading volume, and downside risk in the stock market, but has a weaker and more limited effect in the Bitcoin market. The third study analyzes investor attention by classifying Reddit discussions into stock-level, market-wide, and inattentive themes. It finds that shifts to stock-level attention reduce return comovement, while shifts to market-wide or inattentive themes increase it. Collectively, the findings show that online social interactions influence how investors form beliefs, process information, and affect markets.