Empirical Studies related to Corporate Disclosure on Social Media: A Content Analysis

This paper portrays the literature on voluntary corporate disclosure via social media networks over the course of two decades (2002-2022) through a content analysis framework. Focusing on papers in English, Turkish and Arabic, it reviews a total of 65 studies from 15 countries, based on number of posts, type of information disclosure, time span and social media platforms used. Additionally, it provides a detailed break-down of the reported effect between voluntary disclosures on social media and a total of 37 different variables used to proxy for market, accounting, corporate governance and firm-specific (size, age and risk) measures. Drawing from an exhaustive analysis encompassing more than 51 thousand companies and nearly 70 million social media posts, the prevailing body of knowledge pertaining to corporate disclosures via social media predominantly hinges upon empirical evidence based on the United States and the United Kingdom. We feel that more country comparison studies and multi-lingual samples as well as meta-analyses could significantly improve our understanding of this field. © 2024 by the authors. Licensee


Introduction
Timely dissemination of corporate information stands as a crucial cornerstone within a robust and transparent financial framework.Corporate transparency not only bolsters financial markets but also eliminates barriers that disrupt the information flow to stakeholders.This, in turn, mitigates information asymmetry, fosters equity among investors, and diminishes agency costs (Mohamed and Basuony, 2015).Additionally, improved disclosure and more transparency are shown to correlate with decreased instances of insider trading and reduced uncertainty especially with regard to investment decisions (Klapper and Love, 2004).
The internet has transformed corporate reporting, empowering firms to integrate basic annual reports with supplementary information presented in various formats (Jones and Xiao, 2004).Internet Financial Reporting (IFR) is the standard today for quoted firms across the world, by providing not just a full set of financials including footnotes and auditors' report but also financial highlights in a userfriendly presentation (Bozcuk, 2012).Disseminating corporate information online offers companies the opportunity for significant cost savings regarding information production and distribution, while also facilitating outreach to a broader spectrum of stakeholders at a comparatively reduced cost (Debreceny et al., 2002).Additionally, it represents a potent avenue for delivering tailored information to targeted stakeholders as part of a legitimacy strategy (Campbell, 2003).Furthermore, the rapid proliferation of social media usage has presented companies with a fresh platform for disseminating financial, non-financial and strategic corporate information.
The purpose of this study is to provide an account and a review of the literature on voluntary corporate disclosure through social media over the course of two decades, using a content analysis framework.It aims to paint a picture regarding our understanding of using social media as a funnel for voluntary corporate disclosures as well as the associated effects reported for a number of different accounting, market and corporate governance variables.

Theoretical framework
Timely dissemination of corporate information stands as a crucial cornerstone within a robust and transparent financial framework.Prior studies have conceptualized the corporate information dissemination process in terms of 2 main categories: statutory (compulsory) and voluntary (discretionary) disclosures.Statutory disclosures are those that the company is required to make either by law or in accordance with the stock market quotation requirements.Voluntary disclosures, on the other hand, are those that complement the statutory disclosures., at least in theory.In practice though, this just means that the company has an unregulated outlet for sharing corporate information with stakeholders, should it wish to do so.This brings with it a multitude of considerations in terms of risk and ethics, especially due to the potential for misinformation.It is this 'unregulated' nature of voluntary disclosures that makes it an interesting and challenging topic for academic interest.The determinants, consequences and market response to voluntary disclosures are just a few of the many areas explored in prior literature.The research papers that are included in the sample of this content study all attempt to describe the landscape of voluntary disclosures that are specifically made through social media channels.
These research papers have used several different theories that assist in our understanding of corporate disclosure and the factors influencing it.As such, the agency theory and the information asymmetry theory are two of the most predominant frameworks (Healy and Palepu, 2001).According to these theories, firms disseminate information not only to mitigate agency costs but also to alleviate information asymmetries, during the process of adhering to the requirements of stock exchanges, and meeting the expectations of investors and analysts (Fuertes-Callén et al., 2014).Conversely, signaling theory posits that companies that are motivated to divulge more information are usually those that are profitable (Fuertes-Callén et al., 2014).While those with weaker financial performance tend to adopt disclosure strategies that offer no additional information on top of the basic legal requirements (Debreceny and Rahman, 2005).Moreover, large corporations lean towards disclosing more information in an attempt to demonstrate a favorable reply to societal pressures (Cormier et al., 2010;Boubaker et al., 2012).
It is important to note that the underlying paradigm differences that characterize each of the above theories lead to a different set of incentives and costs related to the decision to disclose or not to disclose (Cotter et al., 2011).However, Cotter et al. (2011) also explain that, regardless of the choice of theory, the voluntary information disclosed is predominantly 'good' news, while the disclosure of 'bad' news is usually either delayed or completely withheld.

Methodology
We start the sample selection procedure by reviewing all articles on Web of Science, Science Direct, Research Gate, EBSCO Host, Google Scholar, Dergipark, and Jasj, using the keywords "social media AND corporate/ voluntary/ information disclosure" in three languages; English, Turkish, and Arabic.Following Basuony et al. (2018), the scope of this study is limited to the concept of 'enterprise-generated content (EGC)'.Hence, we concentrate only on disclosures made by the firms themselves rather than those made about the firm by investors, stakeholders or various other investment platforms since the firm has no direct control over the latter.
Table 1 presents the procedure used for sample selection.Eliminating studies that are solely qualitative (anecdotal evidence or case studies), unsuitable for empirical analysis (e.g.does not involve hypothesis testing) or written in a language outside the scope of this paper (Persian) leaves us with a sample of 65 studies.After a detailed review, each study was coded into a worksheet providing the following information: language of the study, country the sample is drawn from, the number of firms included in the sample, the number of posts analyzed, the type of information disclosure analyzed ( a break-down based on financial, non-financial, intellectual capital and strategic information), type of social media platform used (Twitter, Facebook, Instagram, LinkedIn, Youtube, Wikipedia, Google+, Slideshare, and Flickr), the time period analyzed, the specific variables (37 different variables classified into accounting, market, corporate governance and control groups) that were used in each study and the reported positive or negative effects with regard to social media disclosures.To ensure coding reliability, all coding was performed by one author and independently checked for accuracy by the other author.

Findings
Table 2 provides the breakdown for the 65 studies that are included in our content analysis, based on total number of studies and total number of firms by each country.Looking at   To provide a different perspective, Table 3, provides the country and study breakdown based on the total number of posts analyzed by the studies in our sample.Out of the 65 studies in our sample, 23 studies formulate their research question based on a sample made up of the number of posts, videos or tweets they analyze, rather than a number of firms.These 23 studies analyze a total of nearly 117 million posts in order to understand and question the interaction between firm performance and social media usage in a country-specific approach.
Interestingly, in Table 3, a total of 60% of the posts analyzed are from the UK, followed by 25% from the US and 14% from India.These 3 countries together account for 98.6% of all the posts that are analyzed.The total number of posts from the remaining countries, Spain, Turkey, Poland, and China account for a mere 1.4%.
It is fair to say that, based on the number of firms (Table 2) and on the number of posts analyzed (Table 3), what we know so far about voluntary corporate disclosures through social media is mainly dominated by country specific evidence from the US, the UK, China and India.Evidence pertaining to all other countries make up only 5% based on the number of firms and 1.4% based on the number of posts analyzed.It is important to keep this point in mind especially when making inferences and generalizations regarding this research area.Table 4 shows the distribution of studies based on the type of information disclosed.This classification uses the definition based on Cotter et al. (2011).In exploring the relationship between social media disclosures and firm performance, nearly half of the studies (46%) focus on financial information only.While, 14% of the studies focus on non-financial information only, another 14% of the studies focus at both financial and non-financial information.There is only one study that focuses on a combination of disclosures related to financial, non-financial and strategic information.Meanwhile 23% of the studies approach the research question without referring to a specific type of disclosure.5 provides the distribution of studies by the social media platforms analyzed.Twitter is by far the most popular platform in the literature.Remarkably, 56% of the studies base their analysis on social media disclosures through Twitter alone.All the other platforms pale in comparison to Twitter, even when the studies focus on more than one platform.The percentage of studies that focus on only Facebook and Twitter and Facebook together are 9% and 11%, respectively.In an effort to capture the multitude of dissemination channels available, 13% of the studies focus on a combination of more than 3 social media platforms.Table 6 shows the distribution of studies by the time period analyzed.Majority of the studies (55%) concentrate on one year or less.This is to be expected, given the sheer size of the content that needs to be analyzed, as previously presented in Table 3.The percentage of studies that look at a period of 2 to 4 years and a period of 5 years or more is 20% and 19%, respectively.The remaining 6% are mostly case studies.Table 7 summarizes the types of variables used in the literature into categories and reports the findings of studies in each category, in terms of their relationship with social media disclosures.To measure the interaction between disclosure and several different variables, most studies use a selection of variables from each of the categories which have been previously identified in the related literature to proxy for market, corporate governance, accounting and control environment.Across all these four categories, the number of studies reporting a positive effect outnumber those reporting a negative effect.
There is a striking difference especially for market variables, followed closely by corporate governance variables.Out of the 69 studies that look at the interaction between market variables and social media disclosures, 78% report a positive effect, while only 22% report a negative effect.Similarly, out of the 54 studies that look at the interaction between corporate governance variables and social media disclosures, 70% report a positive relationship and 30% report a negative relationship.For accounting and control variables, although the number of studies reporting a positive effect still outweigh those that report a negative effect, the difference is not as pronounced as that reported for market and corporate governance variables.
Overall, since most studies look at more than one type of variable, the total number of studies in Table 7 amounts to 195, despite our sample size of 65 studies.Out of those 195 studies, 135 (69%) report a positive effect and 60 (31%) report a negative effect between social media disclosures and a range of market, corporate governance, accounting, and control variables.The next table provides a detailed breakdown of each of these categories of variables.
Table 8 reports the breakdown of the 4 categories into 37 individual variables and their relationship with social media disclosures.It provides a detailed account of the picture displayed in Table 7, sorted by total number of studies from largest to smallest by category and also within each category.This breakdown provides valuable insight which is usually lost by categorizing and aggregating the reported findings.
The first 2 columns in Table 8 provide the number of studies that report a positive effect and a negative effect, respectively, between each individual variable and social media disclosures.The third column provides a total of the studies reporting a positive and negative effect, for each individual variable.The last 2 columns provide the percentages within each category of variables in order to give some idea as to the relative importance of each variable within the specific category.
Overall, for 5 variables, there is an equal number of studies reporting a positive and a negative effect between the respective variables and social media disclosures.These are Tobin's Q, Return on Assets, Earnings, Firm Age and Industry.For 6 variables, the percentage of studies reporting a negative effect is higher than those reporting a positive effect.These variables are, Board Independence and Duality, Ownership Concentration, Leverage, Debt-to-Equity, Earnings Before Interest and Tax (EBIT) and Enterprise Value to Sales (EVS) ratio.For all other variables listed in Table 8, (a total of 26 variables), the percentage of studies reporting a positive effect is higher than those reporting a negative effect.
Firm size is by far the most researched variable.There are a total of 15 studies reporting a positive effect and 7 studies reporting a negative effect, totaling 22 studies over two decades.Stock Returns are the second most researched variable with 13 studies reporting a positive and just one study reporting a negative effect.Stock Price and Stock Volatility, rank a close third and fourth with a total of 12 and 11 studies, respectively, for both of which more than 60% of the studies report a positive effect.
Looking at Table 8 in terms of extreme values, there are a total of 11 variables for which no negative effect has been reported so far.They are Market Capitalization, Stock Liquidity, Stock turnover, Free Float-to-Market, Board Diversity, Audit Committtee Quality, Audit Firm (Big4), Credit Risk, Revenue Growth, Dividend Payout, and External Operations (%).Conversely there are only 2 variables for which no positive effect has been reported so far.They are EBIT and EVS.It is important to also consider the number of studies in each of these cases, to put these findings into perspective.For EBIT and EVS, there is just one study looking into each of these variables over the course of two decades of literature.So, the 100% negative effect found between these two variables and social media disclosures should be interpreted with care.
Although the number of studies is not as small as the 100% negative group for most variables in this category, the 100% positive effect should also be interpreted with care.For example, except for Free Float-to-Market, Audit Committee Quality, Audit Firm (Big4), Credit Risk and External Operations (%), there are 2 or more studies looking into each variable in the 100% positive effect group.In this group, the most remarkable finding is for Market Capitalization and Board Diversity for which all 10 and all 5 studies, respectively, have reported a positive effect.

Discussion and Conclusion
It is fair to say that, based on more than 51 thousand firms and nearly 70 million posts analyzed in 65 studies over the course of two decades, what we know about corporate disclosures through social media is primarily dominated by US and UK evidence.
Out of the 65 studies we conducted a content analysis on, there is just one study addressing strategic information disclosures such as corporate strategy, acquisitions and divestments, research and development, and future information.Within the scope of our study, it is hard to evaluate whether this is due to the number of strategic disclosures being scarce or that research attention just has not focused on this area yet.We suspect that the former rather than the latter is true, understandably so, since strategic information usually entails price-sensitive information (Holland, 1996).
Twitter is by far the most popular social media platform among the studies in our sample.It is debatable whether this is driven by Twitter being the companies' preferred venue for social media disclosures or just by the advanced analytical tools it offers researchers which other platforms do not.
Based on what we know so far, the percentage of studies reporting a positive effect between corporate disclosures and a number of market, corporate governance, accounting and control variables is considerably and consistently higher than those reporting a negative effect.However, differences in methodology, as well as proxy variables and statistical tests used by the studies in our sample makes it hard for us to infer generalizable conclusions based on a specific variable.It is important to note that, the variable based interactions reported in the literature should be interpreted with caution due to the highly country-specific nature of the evidence provided so far.The fact that all other countries except the US, the UK, India and China account for a mere 5% based on the number of firms and 1.4% based on the number of posts analyzed (from Tables 2 and 3), skews the findings reported above and impairs the generalizability of these inferences.
Hence, in terms of future direction in the literature, studies based on multi-lingual samples preferably in countries other than the US and the UK and especially meta-analyses would significantly improve our understanding of interactions and consequences of voluntary corporate disclosures through social media.
Appendix: Studies included in the analysis (in chronological order)

Table 1 :
The elimination procedure for sample selection Table2, it is fair to say that US studies dominate the literature by far.More than half (54%) of the literature on social media and corporate performance is based on US studies.This picture becomes even more pronounced when we look at the total number of firms included in the samples of each of those 65 studies There are a total of 51,192 firms analyzed in the social media and corporate performance literature, out of which 36,879 (72%) is based on US firms.The next highest representation is 22% for Chinese firms.The number of firms analyzed for all other countries is almost negligibly small in percentage terms.

Table 2 :
The number of studies and number of firms by Country GCC: The Gulf Cooperation Council (Bahrain, Kuwait, Oman, Qatar, Saudi Arabia, United Arab Emirates(UAE))

Table 3 :
The Country and Study breakdown of the total number of Posts analyzed by the studies in the sample

Table 4 :
The distribution of studies by types of Voluntary Information Disclosure

Table 5 :
The distribution of studies by Social Media Platforms

Table 6 :
The distribution of studies by time period

Table 7 :
The distribution of studies by category of variables used and its relationship with disclosure

Table 8 :
Distribution of studies by type of variable used and its relationship with disclosure