Media Sentiment and Institutional Ownership

Abstract

This paper examines whether institutional investors invest less into overvalued firms due to their informational advantage. Considering high media sentiment as a source of overvaluation, we show that institutional investors tend to hold fewer stocks of firms that receive high media sentiment. The size of the overvaluation is empirically shown as a channel through which media sentiment influences the ownership structure in firms. We conclude this paper by showing institutional investors do not view high media sentiment as a chance to withdraw their money from firms with poor corporate governance structure that the institutional investors cannot exert much influence on. Instead, institutional investors tend not to invest in these firms from the start.

Keywords:

• Institutional ownership
• Media sentiment
• Informational advantage
• Governance

Acknowledgments

We want to especially thank Huynh Hai Ngan for being our research assistant, and Professor Byoung-Hyoun Hwang for sharing his data on founder family CEOs for S&P1500 companies. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Declarations of interest

No potential conflict of interest was reported by the authors.

Notes

1 For T. Rowe Price Group, they reduced their Tesla holding to 505,283 shares from 3.2 million on March 31, 2018. Fidelity sold 3 million shares in the same quarter.

2 RavenPack provides the ranking of the influence and trustworthiness of each news provider. This ranking varies from one to ten, where rank one corresponds to the highest value (i.e., the most trusted source). We only use media sources that have ranks equals to one.

3 We utilize year-end data for institutional ownership due to the fact that not all institutions report their ownership quarterly. Some institutions can adopt an exception to which allows them to delay the disclosure of their holding up to one year to ensure the confidentiality of their portfolios (i.e., confidential treatment). However, our main results remain qualitatively unchanged if we utilize quarterly data. Results are available upon request.

4 https://wrds-www.wharton.upenn.edu/pages/support/applications/institutional-ownership-research/institutional-ownership-concentration-and-breadth-ratios/#references

5 We utilize annualized data for media sentiment due to the fact that the majority of news converges in the last quarter before the financial year end. Thus, if we measure media sentiment on a quarterly basis, there will be a large number of missing data, causing discontinuity in our panel dataset. Our main results remain qualitatively unchanged if we utilize quarterly data. Results are available upon request. To ensure that our results are not driven by the different combinations of good news and bad news, we also rerun our regressions using alternative measures of HS based on the number of positive news and negative news articles separately. The results are similar to our main findings in the sense that we find more positive news is associated with higher IO whereas more negative news is associated with lower IO. Results are available upon request.

6 The unreported test statistics support the validity of the instrument. For example, we perform Hansen J statistics for over-identifying restrictions and find that the equation is exactly identified. The Kleibergen-Paap Rk LM statistic allows us to reject the under-identification hypothesis. Furthermore, the Cragg-Donald Wald F statistic for the weak instrument test allows us to reject the weak IV hypothesis. Test results are available upon request.

7 Thomson-Reuters classifies institutions into five types: (1) bank trust departments; (2) insurance companies; (3) investment companies and their managers; (4) independent investment advisers; and (5) others (pension funds, endowments, etc.).

8 Special thanks to Prof. Hwang for sharing the list.