Does Perception Matter in Asset Pricing? Modeling Volatility Jumps and Returns Using Twitter-Based Sentiment Indices

Abstract

In this paper, I argue that we can use consumer and investor perceptions to forecast short-term fluctuations in asset prices. Using tweets scraped from Twitter between 2009 and 2019, I perform textual analysis to construct daily sentiment indices. While other scholars have relied on third-party companies to complete this task, doing so limits our potential understanding of sentiments' effects on asset pricing. The sentiment indices I constructed are numerical, not dichotomous, scores, which allows to control for sentiment strength. Results indicate that sentiments can forecast daily stock returns and volatility jumps.

Keywords: Behavioral Finance, Volatility Jumps, Twitter, Social Media, Stock Returns

JEL Classification: G40, G41, G14, G17