Media Sentiments on Stakeholders and Daily Abnormal Returns during COVID-19 Pandemic: Early Evidence from the US

Abstract

How does the stock market react to the media coverage about a firm’s efforts in addressing non-financial stakeholder claims during COVID-19 pandemic? Focusing on daily data in the first quarter of 2020, I have examined the relationship between daily abnormal returns (AR) of US listed companies and independent media sentiment on a firm’s issues related to customers, employees, community, and the environment. Merging Compustat security daily data and TruValue Labs daily sentiment data, I constructed a panel data of 313,049 firm-day observations, representing 5,133 listed firms in the US. I first measured sentiment for each stakeholder’s issues as four dummies: no news, negative sentiment, neutral sentiment, and positive sentiment. Random-effect panel regressions report that, compared to no news or neutral sentiment, negative and positive sentiment on customer issues may result in an additional daily AR of 13.207 p.p. (p<0.001) and 14.316 p.p. (p<0.001) respectively. On employee issues, negative sentiment would result in a lower daily AR by 6.394 p.p. (p<0.05), compared to no news, neutral or positive sentiment – the three of which appear to have no statistically significantly different effects on daily AR. On community issues, whether there is news or whether the news was in a negative, neutral, or positive sentiment would not have statistically significantly different effects on AR. On environment issues, positive and negative sentiment would cause a lower daily AR by 8.724 p.p. (p<0.01) and 7.539 p.p. (p<0.01) respectively, compared to no news or neutral sentiment. Finally, I measured sentiment as a continuous value between -1 (most negative) and 1 (most positive), and found that only employee sentiment has statistically significant linear effects on AR. Daily AR would increase by 7.325 p.p. (p<0.1) and 9.277 p.p. (p<0.05) respectively if the sentiment on employee issues increased by 1, under the assumptions of “no news is missing information” and “no news is neutral news”.