dataset: First, a fat API pipeline receives real-time tweets from Twitter, making our information
source timely. Tweets are forward-looking because they capture expectation formation and hence
help us overcome the challenge of time lag. Third, unlike rankings computed on arbitrary guidelines
and/or companies’ own reports, Twitter is an open speech forum in that it provides everyone who
feels entitled a voice on a topic of interest can present their analysis. Fourth, from an investment
perspective, most ESG disputes and severe ESG events, such as corporate frauds, have originated
from a whistleblower inside or outside the organization. Nowadays, ESG issues are voiced and
flagged on social media, especially Twitter, which has become the global ”virtual speaker’s corner.”
Hence, Twitter detects anomalies or disruptions in a real-time format. In addition, we have taken
newspaper articles as the other sources to diversify the forums and integrate an official source in
the study, which also considers the press releases and company updates.
Hence, with the help of the dataset of tweets and news, we aim to calculate the sentiment of
each single data point. Subsequently, we will analyze the stock trends of specific companies and
form a correlation between the sentiment of the dataset with the price changes we observe in the
stock.
2 Further Relevant Literature
Firms’ performance in the area of Environmental, Social, and Governance (ESG) issues appear to
be particularly actively scrutinized by their investors and the public, as seen by the establishment
of sustainable investment funds or demand for “green finance” (Gilbert, 2019). The amount and
quality of data available to explore whether and how investors respond to information about a
company’s ESG performance has also evolved. As the empirical evidence to answer this question is
still scarce and contradictory in the academic literature, this paper seeks to fill this research gap by
extracting ESG information from publicly available news articles (Twitter and newspaper article
sources) and investigating its relationship with the stock market performance of four companies’
stocks namely, Amazon, Tesla, HSBC Bank and Goldman Sachs.
We are basing our research on the work of various authors, all of whom have disputed hypotheses,
conclusions, and opinions on the subject. According to (Friedman, 2007), a firm’s only social
responsibility is to generate lawful profits. This viewpoint would imply that any ESG activity
that is not part of a company’s core business should not be undertaken by the company and that
investors should not incorporate ESG-related information into their investment decisions other
than by withdrawing capital from companies that engage in such activities. (Brammer et al,
2006) present supporting evidence for this effect of ESG activities, finding that organizations with
greater social performance scores had lower returns than those with lower social performance scores.
Similarly, (Kr¨uger, 2015) and (Capelle-Blancard and Petit, 2019) find that positive ESG-related
information can hurt a firm’s market value in the near run. (Cheong et al, 2017) discover that most
organizations have a reactionary attitude toward ESG matters, which provides greater insight into
the likely mechanism behind such observations. These corporations participate in ESG activities
excessively only after experiencing poor market and investor sentiment in the preceding year, where
market and investor sentiment is recorded by a modified version of Baker and Wurgler’s index (Baker
and Wurgler, 2006). While this conclusion would simply have ramifications for how altruistic
a company’s motivations behind its ESG initiatives are on its own, (Goss and Roberts, 2011)
demonstrate the potential consequences of such behavior. The authors suggest that corporations’
ESG operations, which are launched in direct response to negative media and investor sentiment,
are frequently regarded as window-dressing, lowering the company’s perceived creditworthiness and
increasing its cost of financing.
2