Mood swings and insufficient information acquisition: A study of cross-sectional stock returns

This paper studies mood, as measured by the general public’s Twitter messages, and its implications for asset pricing. Using this Twitter-based measure, we find that mood swings negatively predict investors’ acquisition of earnings-related information. We argue that this contributes to the failure of classical (unconditional) pricing models. Tests on cross-sectional stock returns show that stocks that are more sensitive to mood earn higher expected excess returns than less mood-sensitive stocks. We quantify the risk premium on the mood risk factor to be around 0.56% per month and show that this survives the usual battery of asset-pricing tests. Our results are consistent with the theoretical prediction that investors mistakenly use mood as information rather than learning enough about fundamental information about assets, thereby inducing mispricing.