Abstract:We introduce a new method to gauge firm-level mood using Twitter engagements on non-financial tweets. By counting likes on such tweets, we assess positive sentiment, impacting market reactions to earnings surprises. This sentiment measure doesn't foresee earnings surprises and its market impact is short-lived, indicating it reflects mood more than informational content. However, it underscores mood's significance in stock investment and the utility of big data for novel financial market analysis tools.
What problem does this paper attempt to address?
The main problem that this paper attempts to solve is to explore how firm - specific sentiment (measured by non - financial tweet interactions on Twitter) affects investors' market reactions to the company's earnings announcements (EAs). Specifically, the author introduces a new method to evaluate firm - level sentiment using the number of likes on non - financial tweets on Twitter and studies the impact of this sentiment on market reactions, especially on market reactions to earnings surprises.
### Main research questions:
1. **Relationship between sentiment and market reaction**: Research whether firm - specific sentiment will affect investors' market reactions to earnings announcements.
2. **Nature of sentiment**: Distinguish between mood and opinion and verify whether sentiment mainly reflects investors' immediate sentiment rather than expectations of future earnings.
3. **Short - term impact of sentiment**: Analyze whether the impact of sentiment on market reactions is short - lived and whether this impact is independent of information and attention channels.
### Research background:
- **Behavioral finance theory**: Asset values may deviate from their intrinsic values due to cognitive biases, such as the impact of external factors such as weather or sports events on market sentiment.
- **Application of social media data**: Social media platforms provide a new way to measure sentiment, but existing research mostly focuses on market - level sentiment and pays less attention to firm - specific sentiment.
- **Sentiment and market reaction**: Sentiment can affect stock returns through different channels, such as attention, information, and sentiment itself.
### Research methods:
- **Data sources**: Collected tweet and interaction data (likes, retweets, replies) of 2,213 companies from Twitter data from 2015 to 2019.
- **Sentiment indicators**: Measure firm - specific sentiment through standardized like numbers (StandLikesi,t) and "ratio - liked" (Ratioedi,t) indicators.
- **Control variables**: Use variables such as cumulative abnormal return (CAR), analyst forecast error (Surprisei,t), and institutional ownership ratio (InstOwni,t) to control other possible influencing factors.
### Research findings:
- **Sentiment and market reaction**: The standardized like number (StandLikesi,t) predicts a more positive cumulative abnormal return (CAR), indicating that positive sentiment is associated with a more positive market reaction.
- **Impact of replies**: The impact of the standardized reply number (Repliesi,t) on future CAR depends on the direction of earnings surprises, supporting that replies convey ambiguous sentiment.
- **Negative sentiment**: The "ratio - liked" (Ratioedi,t) predicts a more negative cumulative abnormal return (CAR), indicating that negative sentiment is associated with a more negative market reaction.
- **Short - term impact of sentiment**: The impact of sentiment on market reactions is short - lived, lasting about 2 - 3 weeks, further supporting that sentiment reflects immediate sentiment rather than persistent information - driven sentiment.
### Contributions:
1. **New method for sentiment measurement**: Introduced a firm - specific sentiment measurement method based on Twitter likes and "ratio - liked", which is different from traditional financial discussions.
2. **Relationship between sentiment and market reaction**: Systematically excluded the influence of opinion, information, and attention, and verified the direct impact of sentiment on market reactions.
3. **Impact of digital communication**: Provided insights into non - financial tweets for corporate social media management, which is helpful for enterprises in brand management and crisis response applications.
Through these studies, the author not only enriches the theoretical basis of behavioral finance but also provides practical tools and strategies for corporate and social media management.