This thesis investigates the effect of social media on asset prices. The three chapters in the thesis each target one aspect of the social media effect.
Chapter 1 and 2 look at social media and post-earnings-announcement drift in response to companies’ quarterly earnings announcements. In Chapter 1, I attempt to build a theoretical model to estimate the price response that is caused by investors’ attention through utilizing Bayesian learning. Using data from quarterly earnings, Twitter and StockTwits data (17 quarters from the fourth quarter of 2010 to the fourth quarter of 2014), I utilize Twitter volume and a residual methodology to generate an attention proxy that is orthogonal to the growth of Twitter accounts. In Chapter 2, I demonstrate how the new attention brought about by social media after the earnings announcements, positively affects the cumulative abnormal returns, resulting in magnitudes that are larger than the earnings surprise effects. Finally, I find that even companies reporting bad news can still have positive immediate cumulative abnormal returns if they attract enough attention from investors after an earnings announcement. Chapter 3 examines social media effects from a practitioner’s point of view. I develop a method for measuring profits for a pairs trading strategy which has previously been used by institutional and hedge fund investors. Building on Engelberg et al. (2009), who provides possible explanations for pairs trading profits, I identify social media as another driver of pairs trading profits. The work builds on a large body of literature that investigates the economic drivers of stock return co-movement.