External Assurance and Sustainability Reporting:
A Signaling or Legitimacy Mechanism for ESG Performance?
Name: Frank van Wiefferen Student number: 6070884
Thesis supervisor: dr. G. Georgakopoulos Date: June 25, 2018
Word count: 12985
MSc Accountancy & Control, specialization Control
Statement of Originality
This document is written by student Frank van Wiefferen, who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Abstract
Over the last two decades, sustainability reporting and assurance of such reports has become an increasingly relevant topic in both the business and the academic world. This study aims to determine if firms voluntarily adopt sustainability reporting and assurance to signal positive ESG performance to the public or to disguise weak performance and protect the legitimacy of the firm towards society. In addition, this study examines if a firms’ involvement in ESG controversies changes this behavior. In order to determine this, a global sample consisting of the 250 largest firms by revenues is examined between 2012 and 2016. The effects of ESG performance and a firms’ involvement in ESG controversies on a firms’ decisions regarding sustainability reporting, assurance, assurance provider and level of assurance is examined. The findings show a positive relation between ESG performance and publishing a sustainability report, in addition to the adoption of external assurance on these reports. Also, this study finds a negative relation between involvement in ESG controversies and publishing a sustainability report, indicating that sustainability reporting can both be used as a mechanism to protect the firms’ legitimacy and to signal superior sustainability information to society. This research is the first research to take both the effect of ESG performance and ESG controversies on a firms’ sustainability reporting choices into consideration, creating a starting point to further research the effects of those factors on corporate behavior regarding sustainability reporting.
Keywords: Sustainability reporting, Sustainability reporting assurance, ESG performance, ESG controversies
Acknowledgements
I would like to express my gratitude towards dr. Georgios Georgakopoulos of the University of Amsterdam for providing me with insights that helped me with my research and for the supervision of my thesis. Furthermore, I would like to thank Bianca of GRI for providing me with the necessary data to conduct this research. Finally, I would like to express my appreciation towards my family and friends for listening to my thesis experiences and sharing their views on my research.
Table of Contents
1 Introduction... 6
2 Literature and development of hypotheses ... 9
2.1 Sustainability reporting & assurance ... 9
2.2 Levels of assurance...11
2.3 Providers of assurance ...13
2.4 Development of hypotheses ...14
3 Data and methodology...19
3.1 Sample selection ...19
3.2 Explanation of model...21
3.3 Measurement of variables ...21
4 Findings and discussion...24
4.1 Descriptive statistics ...24
4.2 Empirical results and discussion ...27
4.3 Robustness checks ...32
5 Conclusion ...37
References ...39
Appendix A: Summary overview of key studies used ...44
1 Introduction
Over the last two decades, the attention of investors, customers and other stakeholders towards issues on sustainable development within companies has increased (Lai, Melloni & Stacchezzini, 2016; Kim, Park & Wier, 2012). Through a survey among 320 institutional investors around the world, EY (2017) found that 92% of the respondents agree that sustainability related topics have real and quantifiable impacts over the long term, which they take into account in their investment decision-making. In addition, these respondents think that CEOs should lay out long-term board-reviewed strategies regarding sustainability related themes on a yearly basis. To serve the increased attention towards sustainable development within companies, sustainability reports that focus on sustainability performance have been functioning as an important communication tool between organizations and stakeholders (Junior, Best & Cotter, 2014). Within these sustainability reports, sustainability indicators are linked to all the socially responsible activities of a firm and therefore often referred to as Environmental, Social and Governance (ESG) performance indicators (GRI, 2013; EY, 2017).
The interest in sustainability reporting and the publication of reports also increased the demand for accuracy of these reports (GRI, 2013). In KPMG’s survey on Corporate Responsibility Reporting it is stated that the amount of firm’s that seek external assurance for their sustainability report is growing steadily since 2005. In 2017, 67% of the G250 firms1 have a certain level of assurance of their sustainability information. Also, the GRI Standards, provided by the Global Reporting Initiative2, remain the most widely adopted sustainability reporting framework. The increased application of assurance practices and usage of reporting guidelines makes sustainability reporting assurance a relevant topic for the business world.
The increased attention for sustainability and assurance also caused a growth in literature studies on this topic (Hahn & Kühnen, 2013). Previous research on the determinants of adopting external assurance (e.g. Kolk & Perego, 2015; Michelon, Pilonato & Ricceri, 2015) focused on identifying factors that can influence a firm’s decision to assure their sustainability report. Where some studies explain the adoption of sustainability reporting and assurance as a mechanism to signal superior ESG performance to the public (Lai et al., 2016;
1 The G250 refers to the world’s 250 largest companies by revenue based on the Fortune 500 ranking
of 2016 (KPMG, 2017).
2 The Global Reporting Initiative (GRI) promotes the usage of sustainability reporting and aims to
Jain, 2016; Casey & Grenier, 2015), other papers claim it is used to disguise weak performance and for protecting legitimacy towards shareholders (Clarkson, Li, Richardson & Vasvari, 2008; Michelon et al., 2015). In addition, specific characteristics as the assurance provider and the level of assurance are also examined (Cuadrado-Ballesteros, Martinez-Ferrero & Garcia-Sanchez, 2017). Following up on this, Junior et al. (2014) suggest to further examine the possible motivations of sustainability reporting and assurance.
Compared to assurance of financial reports, assurance of sustainability reports is a relatively new practice and it evolves quickly (Kolk & Perego, 2010). Many firms seek assurance to increase the credibility and transparency of sustainability information (Kolk & Perego, 2010). In essence, it is a mechanism to positively influence the users’ perception of the reliability of the reported ESG performance (KPMG, 2017). The perception of users can also be challenged by media attention (Gillet-Monjarret, 2015). In a study on 120 French companies, Gillet-Monjarret concludes that the media can influence the use of sustainability assurance, because firms lose their legitimacy towards the public when the media is reporting socially controversial activities of the firm. Nevertheless, other studies (Jain, 2016; Casey & Grenier, 2015) claim that mechanisms to increase the reliability of information are avoided when the firm is in disrepute and showing weak ESG performance. Therefore, further research about the motivations of a firm to adopt external assurance on their sustainability report is needed.
This paper aims to determine if a firms’ decisions regarding sustainability reporting and assurance can be explained as a mechanism to signal superior ESG performance or to disguise weak performance and protect legitimacy. In addition, this study examines if a firm’s involvement in controversial social activities affect sustainability reporting decisions. To determine this, the influence of ESG performance and a firms’ involvement in ESG controversies on a firms’ decisions regarding sustainability reporting, assurance, assurance provider and level of assurance is tested. This empirical study is conducted on a sample consisting of firms from the Fortune 250 largest companies by revenue from 2012 to 2016. The focus in this research is on both stand-alone sustainability reports and integrated reports. An integrated report is an annual report that contains both a financial and non-financial (or sustainability) section. When referring to sustainability reports, both type of reports are taken in to account.
The findings of this study show that ESG performance is positively related to issuing a sustainability report and adopting external assurance, which indicates that firms are more
sustainability report and assurance as a mechanism to protect the legitimacy. Nevertheless, the findings also show that the sustainability report does function as a mechanism to protect a firms’ legitimacy when a firm was involved in ESG controversies.
This study makes several literary contributions. First, this is the first study to test the effect of both ESG performance and ESG controversies on sustainability reporting choices from a signaling and legitimacy perspective. Second, the findings are relevant for stakeholders, investors and other users of sustainability reports interested in the effect of media coverage and sustainability performance on a firms’ sustainability reporting choices. On firm-level, these findings can be used to assess when sustainability reporting and assurance will be perceived as a signaling or legitimizing mechanism by society. Finally, this study contributes to a better understanding of using sustainability reporting and assurance and thus to the understanding of corporate behavior.
The paper proceeds as follows: The literature background that supports the development of the hypotheses; the description of the sample and research methods; a discussion of the descriptive statistics and the results of testing the hypotheses, followed by the conclusions and implications for further studies.
2 Literature and development of hypotheses
The aim of the literature review is to describe and compare current theories and ideas in the field of sustainability reporting and assurance. First, section 2.1 describes the current literature on sustainability reporting and specifically the external assurance of such reports. Second, section 2.2 provides an overview of the different levels of assurance. After that, section 2.3 delves into the providers of assurance. Finally, section 2.4 revolves around the development of hypotheses regarding the effect of ESG performance and controversies on sustainability reporting assurance.
2.1 Sustainability reporting & assurance
In recent years, there has been an increase in the number of firms that voluntary report information on sustainability related topics (KPMG, 2017). Firms need to disclose this type of information as a response to the needs of stakeholders concerned with social and environmental performance. In addition, investors rely on this type of nonfinancial data to make investment decisions (Kolk & Perego, 2010). This increased interest can be explained by the growing impact of problems such as climate change, poverty and human rights violations (Kolk & Van Tulder, 2010). According to Kolk & Van Tulder, organizations worldwide are called upon to play a positive and leading role to contribute towards a more sustainable development. As a response, firms issue a sustainability report that provides information on the most critical aspects of the organization’s economic, social, environmental and governance impacts and the relation of those impacts with its performance (GRI, 2013). To stimulate the publication of sustainability reports, independent organizations like the GRI produce sustainability reporting guidelines in order to help organizations prepare such reports. The common framework provided by the GRI has been an important driver in improving the quality of sustainability reports (GRI, 2013). The growth of sustainability initiatives and the disclosure of those activities have also increased the demand for assurance of sustainability reports (Alon & Vidovic, 2015).
The International Auditing and Assurance Standards Board (IAASB) provides the following definition of assurance: “An engagement in which a practitioner aims to obtain sufficient appropriate evidence in order to express a conclusion designed to enhance the degree of confidence of the intended users other than the responsible party about the outcome of the measurement or evaluation of an underlying subject matter against criteria” (IAASB, 2013, p. 7). Furthermore, the IAASB proposed the ISAE3000 (International Standard on
Assurance Engagement). This standard, which originated from the accounting profession, also organizes sustainability assurance and positions it as a mechanism to enhance reporting reliability. Its objective is to serve as a general reference framework for accountants that carry out financial and non-financial auditing engagements (IAASB, 2013). However, the low amount of specifications that focus on sustainability reporting is considered to be a limitation of the standard (Manetti & Becatti, 2009). The Assurance Standard AA1000AS, launched in 2003 by AccountAbility, is another widely accepted framework for assurance services (Gürtürk & Hahn, 2016). While this framework has a clear sustainability focus, it is mostly used by smaller consulting firms that provide assurance services, because big accountancy firms tend to apply the ISAE3000 due to its professional background and network effects.
There have been multiple studies that investigate the determinants of the adoption of external assurance of sustainability reports (e.g. Kolk & Perego, 2010; Alon & Vidovic, 2015; Gillet-Monjarret, 2015). Still, most of the available research that investigates external assurance is focused on the assurance of financial reports, which is mandatory. Nevertheless, the increasing demand for voluntary sustainability assurance has also caused an increase in the amount of studies on this topic. However, there is a lack of research on how the judgements of auditors differ when assuring sustainability rather than financial information (Moroney & Trotman, 2016). In 2016, Moroney and Trotman suggested that sustainability assurance teams, which consist of both auditors and subject-matter experts (e.g., environmental scientists, geologists), will judge information differently compared to auditors with a financial background. Furthermore, the high amount of competitiveness in the market and diversity within sustainability information, next to the lack of well develop criteria and analytic rigor from double-entry bookkeeping, make sustainability reporting engagements a different field to study (Cohen & Simnet, 2015).
The evolvement of sustainability reporting has increased the importance of transparency when providing valuable information to stakeholders (Junior et al., 2014). Therefore, firms seek third-party assurance to ensure transparency, but also to increase the reliability of this information (KPMG, 2017). Furthermore, firms use assurance as a tool to increase the credibility of their reported sustainability performance (Casey & Grenier, 2015). In their study, Casey and Grenier point out that assurance makes sustainability information more usable for internal decision-making purposes regarding the strategy, risk management and management control of a firm. From an external perspective, their findings show that
stakeholders are more likely to question the credibility of strong performance.
Contradictive to positive statements about external assurance, Gürtürk and Hahn (2016) did not find any evidence that the current assurance practice can improve the transparency of sustainability information and add credibility to the sustainability report. Casey and Grenier (2015) add to this that possible influence of management can create a bias towards the relevance and independence of the provided assurance, because it is common that the assured firm itself determines under which conditions the assurance takes place, instead of the stakeholders of the organization. Also, due to the lack of regulation on sustainability assurance providers, there are currently no severe consequences for providers of the assurance service when inaccurate, incomplete or misleading reporting isn’t identified. This makes it possible for the management of a firm to use assurance as a tool to make misleading or unsubstantiated claims about sustainability performance look more reliable, which is commonly referred to as “greenwashing” in the literature (Cho & Pattern, 2007). Furthermore, the value of external assurance for internal use is vague, and there is a limited potential to use it for decision-making and organizational change (Gürtürk & Hahn, 2016).
Nevertheless, Cheng, Green and Ko (2015) demonstrated by using an experiment that the perception of strategic relevance of reported sustainability performance increases investors’ willingness to invest in a firm, and external assurance stimulates that perception. Alon and Vidovic (2015) add to this that disclosed sustainability performance is positively associated with the sustainability reputation of the firm, although their findings indicate that external assurance does not increase this reputation from an external stakeholder’s point of view. The absence of clear and universal policy objectives with regards to sustainability reporting and the role of assurance makes it difficult for users to judge such reports (Sonnerfeldt, 2012).
Besides providing a mechanism to report sustainability information, the GRI Sustainability Reporting Guidelines also aim to standardize the sustainability reporting process (GRI, 2015). The most recent G4 guidelines make disclosure of topics related to environmental, social and governance impact less diversified. Likewise, a universal assurance standard will also decrease the diversity in the current assurance process (Sonnerfeldt, 2012).
2.2 Levels of assurance
In theory, it is possible to provide an almost unlimited number of levels of assurance for various types of engagements (Mock, Strohm & Swartz, 2007; Hasan et al., 2005). However,
it is not considered practical to provide a high number of different assurance levels. Therefore, the IAASB recognizes two different levels of assurance: reasonable and limited assurance (IAASB, 2013). Reasonable, or high levels of assurance provide conclusions regarding the content of the report in a positive form. Positive statements include phrases like ‘fairly stated in all material aspects’ and ‘are free from material misstatements’, which are indicators of a high level of assurance (Mock et al, 2007). In contrast, such conclusions can also be stated in a more negative form. Therefore, statements as ‘nothing has come to our attention’ are indicators that the level of assurance provided was limited, or moderate, and of a lower quality. To sum up, a higher level of assurance indicates that the procedures during the assurance process have been executed more rigorous (GRI, 2013).
In 2009, Manetti and Becatti determined that there are some innovative elements that are not always addressed by the ISAE 3000, but could lead to standards’ improvement if taken into consideration by the IAASB. One of these elements concerns that different parts of a report might require different levels of assurance, because it requires deeper or less deep procedures of verification. As a result of cost constraints and other feasibility issues, firms can nowadays also decide to have a reasonable level of assurance for some parts of the report and limited for others (GRI, 2013; IAASB, 2013).
It has proven to be difficult for users to distinguish the difference between limited and reasonable assurance (Hasan et al, 2005). Since a sustainability report consists of both financial and nonfinancial information, it can be challenging to assess the quality of the assurance process. The experiment by Hodge, Subramaniam and Stewart (2009), shows that the perceived confidence of users of a sustainability report increases when the assurance level was reasonable. However, in their experiment it also appeared that users may not be able to fully distinguish the two assurance options.
Furthermore, Hasan et al. (2005) state that firms consider the costs and benefits when deciding to choose for limited or reasonable assurance. The time and effort it takes to collect and prepare sufficient evidence to apply for reasonable assurance will lead to higher costs compared to a limited level of assurance. Therefore, firms first need to determine if the increased credibility of the report, from a user’s perspective, is worth the extra costs of increasing the level of assurance (Cuadrado-Ballesteros et al., 2017). Cuadrado-Ballesteros et al. also found that a reasonable/high level of assurance increases analysts’ forecast accuracy. This in turn could be beneficial with regards to attracting investors.
2.3 Providers of assurance
There are currently no norms or regulations that guide the contract between a firm and an assurance provider (Fernandez-Feijoo, Romero, & Ruiz, 2016). In order to contract the provision of any assurance service, the client requests to hire a specific auditor, and this request has to be accepted by the auditing firm. Although a broad range of possible professions could conduct a sustainability assurance service, there is not much knowledge of the best practice for providing high-quality assurance (Cohen & Simnett, 2015). Cohen & Simnett state that, in this competitive and voluntary market, firms will only demand certain assurance services when the expected benefits exceed the costs, and when there are differences between the value-proposition of assurance providers from within or outside the accounting profession.
Providers of external assurance services are generally divided in three groups; Accountancy, engineering & sustainability services firms (GRI, 2013). In literature, the division is commonly made between accountancy firms, which have acquired expertise regarding financial and nonfinancial auditing over the years, and consultancy firms, which provide more specialized knowledge on certain sustainability related topics (Gillet-Monjarret & Riviere-Giordano, 2017).
The usage of accountants instead of consultants for assurance services seems to increase confidence (Hodge et al., 2009) and credibility in sustainability reporting (Pflugrath, Roebuck & Simnett, 2011). The experiment by Hodge et al. showed that the perceived reliability of the sustainability report was greater when the assurance was provided by a accountancy firm. In addition, it also causes investors to find the assurance statement more credible (Casey & Grenier, 2015). Financial analysts view that professional accountants provide a higher level of independent and expert assurance compared to sustainability consultancy firms (Pflugrath et al., 2011). According to Pflugrath et al., this could be due the reputation of accountancy firms in the auditing field, their well-developed international standards and ethics, and their usage of quality control mechanisms during the assurance process. These elements help accountants to ensure that they have the required auditing expertise, but also the necessary specific subject matter expertise.
However, Gürtürk and Hahn (2016) state that non-accountants on average apply a wider diversity of assurance methods and achieve a slightly higher quality score in this regard. Perego (2009) adds to this that the quality of the provided recommendations and opinions are higher when the assurance is done by a consultancy firm. Still, neither type of
assurance provider considers the issues of stakeholders in their assurance procedures or recommendations (Gürtürk & Hahn, 2016).
Currently, it seems that the market only values sustainability assurance done by professional accountants, because it takes time for practitioners of new and developing assurance services to convince users of their added value (Peters & Romi, 2015). This indicates that firms will most likely chose an accountancy firm to provide the assurance service. However, Cheng et al. (2015) state that the rapidly developing market of sustainability assurance will keep it necessary to question if the current assurance provider’s profession is appropriate to fulfill its task (For an overview of the key papers used for this literature overview, see appendix A).
2.4 Development of hypotheses
It seems that both individual and institutional investors are becoming more interested in those activities of a firm that go beyond maximizing profits and shareholder wealth (Gillan, Hartzell, Koch & Starks, 2010). In addition to the impact on sustainability reporting and assurance of such reports, this increased interest also led to three extensively examined indicators of nonfinancial performance; Environmental awareness, social responsibility and corporate governance. As a result, ESG has become a popular abbreviation when referring to a firms’ activities in these specific areas.
Alon and Vidovic (2015) found that firms with superior sustainability performance are more likely to seek assurance of their sustainability disclosures. The disclosure of sustainability efforts is a signal that firms can use to convey their performance. The signaling theory is often used to describe behavior between two parties (individuals and organizations) that have access to different information (Connelly, Certo, Ireland & Reutzel, 2011). One party, the sender, must choose whether and how to communicate (or signal) information, and the other party, the receiver, must choose how to interpret the signal. The signaling theory implies that sustainability reporting serves as a signal to the stakeholders about how the firm is performing with the intention to convince readers of its social responsibility (Jain, 2016). Jain (2016) found evidence that firms signal this information during good times, but such signals are not carried through during times of crisis. Since publication of a sustainability report is still on voluntary basis, it is possible for a firm to signal ESG performance in such a way. However, the growing interest and expectations from society regarding sustainability development increase the pressure on firms to report sustainability information on a yearly
basis (Kolk & Van Tulder, 2010).
Casey and Grenier (2015) state that firms with high sustainability performance are more likely to choose external assurance, because stakeholders are more critical about the credibility of high performance. The above argumentation assumes that a distinction could be made between high- and low performing firms. High-performing firms will benefit the most from signaling their true performance (Connelly et al., 2011), which makes it possible to differentiate themselves from low-performing firms. However, this benefit won’t apply to low-performing firms, because reporting low performance is not considered to be beneficial. This implicates that users will know that firms who signal their sustainability performance on a voluntary basis, will be the high-performing firms.
In contrast, Hummel and Schlick (2016) state that not only high-performing, but also low-performing firms could possibly benefit from sustainability reporting. In their study, they found that low-performing firms use sustainability disclosure as a legitimation tactic to influence the public perception of their true sustainability performance. The legitimacy theory considers that an organization’s existence will be threatened if the public perceives that the organization has broken its social contract with society (Deegan, 2002). Deegan states that when society feels that the organization is not behaving in an accepted, legitimate manner, then society will revoke the ‘contract’ that disables the organization to continue its operations.
The legitimacy theory predicts a negative association between sustainability performance and the level of disclosures (Clarkson et al., 2008). Clarkson et al. suggest that social disclosure is a function of external pressures facing a company. Firms that show poor performance face more external pressures and their legitimacy towards shareholders is being threatened. You could argue that firms that voluntary seek external assurance do this to change their shareholders’ perspective regarding their low performance and to regain legitimacy. This interpretation is strengthened by Michelon et al. (2015), who concluded that assurance of a sustainability report could be seen as a symbolic practice used by firms to influence stakeholder’s perceptions of corporate commitment to sustainable reporting.
In light of “greenwashing”, a firms’ decision to assure its sustainability report could also be a way of making bad ESG performance look better (Casey & Grenier, 2015). When taken into consideration that sustainability reporting and assurance of such reports is a way to increase users’ perception of the reliability of the disclosed information, benefits could also arise for low-performing firms. In contrast, Lai et al. (2016) found that reporting on
gaining strategic legitimation.
As a starting point, this study will initially take on the assumptions of the signaling theory. Therefore, the first two hypotheses of this study are:
H1: A firm’s ESG performance is positively related to sustainability reporting.
H2: A firm’s ESG performance is positively related to assurance of a sustainability report.
Once a firm has chosen to seek external assurance, there are several other decisions that need to be made. According to Kim et al. (2012), firms that experience high ESG performance are less likely to manipulate their reported activities. In this way, such firms provide more transparent and reliable information towards investors. The experiment from Cheng et al. (2015) found that the benefits that may accrue to a firm from assuring their ESG disclosures are higher if they can clearly communicate the link between the company’s strategy and its ESG performance towards investors. When both findings are taken into consideration, it could indicate that firms with high reported ESG performance, and thus more capable of communicating this link, would benefit more from external assurance. Since firms outweigh possible benefits and costs when choosing the level and provider of assurance (Hasan et al., 2005; Cohen & Simnett, 2015), the high ESG performers will likely chose a reasonable/high level of assurance. In addition, Casey and Grenier (2015) also found a positive association between choosing an accounting provider and high sustainability performance. These arguments lead to the following hypotheses:
H3a: A firm’s ESG performance is positively related to a reasonable/high level of assurance of a sustainability report.
H3b: A firm’s ESG performance is positively related to assurance of a sustainability report by a accountancy firm.
In addition to ESG performance, this study also takes possible effects of ESG controversies on sustainability reporting assurance into consideration. ESG controversies are corporate environmental, social and governance news stories such as undesired social behavior or product-harm scandals that place a firm under increased attention of the media, which results
in more attention from investors (Cai, Jo & Pan, 2012; Aouadi & Marsat, 2016). Controversies are therefore more focused on specific events captured by society (mostly through media coverage), whereas performance captures a firms’ overall achievements.
As previously mentioned, Jain (2016) concluded that firms signal sustainability information during good times, but avoid doing this during bad times. This indicates that firms are less likely to report sustainability information when they were involved in ESG controversies. Also, firms genially don’t report more negative information than required (Casey & Grenier, 2015). Next, Casey and Grenier state that positive information needs assurance to increase the users’ perception of its credibility, but negative information is already perceived as being credible. Under the signaling theory, this would indicate that ESG controversies, which is negative information, will already be perceived as being credible information in the firms’ sustainability report. Therefore, the argument to seek external assurance in order to gain additional reliability would not be valid when a firm has been negatively exposed in the media.
Contradictive towards the signaling theory, Gillet-Monjarret (2015) found that the media, which reflects the expectations of the public, is able to influence the use of sustainability assurance practices. The results of her study indicate that firms are more likely to seek external assurance when they are losing legitimacy through negative media coverage. In addition, the level of media exposure of a firm positively influences the use of assurance practices. In this way, the media has indirect power over companies and their legitimacy (Deegan, 2002). Consequently, ESG controversies through media coverage could lead to firm’s using sustainability assurance as a mechanism to manage social issues, with the goal of obtaining or maintaining legitimacy (Perego & Kolk, 2012).
To maintain consistency in the direction of the hypotheses, the assumed effect of ESG controversies will also initially be explained through arguments of the signaling theory. This leads to the following hypotheses:
H4: A firm’s involvement in ESG controversies is negatively related to sustainability reporting.
H5: A firm’s involvement in ESG controversies is negatively related to assurance of a sustainability report.
Based on current insights, the effect of ESG controversies on the level and provider of assurance has not been studied so far. Casey and Grenier (2015) found that firms experiencing sustainability concerns, and thus use assurance for impression management purposes, are more likely to use non-accounting providers for assurance. Casey and Grenier suggest that firms avoid such providers to minimize the chance of falsely reported sustainability information being identified. This and previously stated argumentation indicates that firm’s involved in ESG controversies are less likely to choose the type of assurance that is perceived to be of the highest quality. Therefore, in line with signaling theory, the last two hypotheses of this study are as follows (See Appendix A for a summary overview of the key studies and their outcomes that are used for the literature review and development of hypotheses):
H6a: A firm’s involvement in ESG controversies is negatively related to a reasonable/high level of assurance of a sustainability report.
H6b: A firm’s involvement in ESG controversies is negatively related to assurance of a sustainability report by a accountancy firm.
3 Data and methodology
This quantitative archival research will focus on the differences in firms’ choices regarding sustainability reporting and assurance, depending on their ESG performance and involvement in ESG controversies. First, further details regarding the data and methodology are explained. Section 3.1 describes the way that the final sample for this study is determined and distributed. Next, section 3.2 explains the model that is used to test the hypotheses of this paper. Lastly, the measurement of variables is discussed in section 3.3.
3.1 Sample selection
To test the hypotheses, an initial sample consisting of the G250 largest companies by revenue, based on the Fortune 500 ranking of 2016, is used. This sample has been selected because large global companies are typically leaders in sustainability reporting and their behavior often predicts trends that are adopted more widely (KPMG, 2017).
Data regarding sustainability reporting, assurance, level and provider for the initial sample is retrieved from the Sustainability Disclosure Database of the Global Reporting Initiative (http://database.globalreporting.org). This database is accessible online, but by contacting the GRI a more structured, analyzable overview is retrieved. In addition, company websites were visited when the database didn’t include the required data. When the data is not available at both the database and company website, the assumption is made that the firm did not publish any sustainability report. With the usage of an unique ISIN code for firms in the initial sample, data concerning the ESG performance and involvement in ESG controversies is retrieved from Thomson Reuters, which is accessible through Datastream. The initial sample consists of 1250 observations for 250 firms from 2012 to 2016. This period is chosen because the GRI database did not keep track of data regarding the level and provider of assurance for firms before 2012. Also, 2017 is excluded because a large amount of data was not yet available.
Table 1 Panel A shows how the final sample is selected. For 257 observations it was not possible to retrieve data about the ESG performance and controversies. As for missing data regarding sustainability reporting, firms that did not publish any form of sustainability report were still usable as an observation. This has led to a final sample consisting of 202 unique firms and 993 observations.
Table 1: Sample selection and distribution
Panel A: Sample selection
Number of observations
Initial sample: 2012-2016 Fortune Global top 250 companies 1250
Less: missing values for ESG Scores -257
Final Sample 993
Panel B: Year
Publication Observations Sustainability Assured Reports/Observations Assured/reports
year reports reports (%) (%)
2012 196 145 62 73.98% 42.76% 2013 199 168 76 84.42% 45.24% 2014 200 174 85 87.00% 48.85% 2015 201 178 86 88.56% 48.31% 2016 197 176 91 89.34% 51.70% Total 993 841 400 84.69% 47.56%
Panel C: Sensitive industries vs. Others
Industry [1] Observations Sustainability Assured Reports/Observations Assured/reports
reports reports (%) (%) Mining 5 5 5 100.00% 100.00% Oil Exploration 35 34 17 97.14% 50.00% Chemical products 50 50 29 100.00% 58.00% Petroleum refining 69 59 38 85.51% 64.41% Metals 15 14 8 93.33% 57.14% Utilities 48 36 19 75.00% 52.78% Others 771 643 284 83.40% 44.17% Total 993 841 400 84.69% 47.56%
[1] Paper (SIC code 26) is excluded, because firms in this industry are not part of the sample.
Table 1 Panel B and C show the distribution of the firms in the sample by year and industry. In addition, the ratios of sustainability reports versus observations and assured reports versus sustainability reports are provided for each of the provided distributions. The second and third column of Panel B shows that the amount of sustainability reports and assurance has steadily increased over the years. This is consistent with recent developments in society (KPMG, 2017) and also with prior research (Peters & Romi, 2015).
Panel C shows how the behavior of firms in environmentally sensitive industries (See section 3.3 for further elaboration on these industries) regarding sustainability reporting and assurance compares to other industries. The fifth and sixth column show that both the ratio of sustainability reports versus observations and assured reports versus sustainability reports are
higher for almost all sensitive industries compared to the others. To the best of my knowledge, there is no previous research stating similar ratios for sensitive industries, which makes it difficult to compare the distribution to other studies. However, expectations are that the relatively high amount of published sustainability reports and assurance in sensitive industries is partly due to construction of the sample and should be taken into consideration when interpreting the results.
3.2 Explanation of model
A logit regression model, similar to the study of Casey and Grenier (2015), is used in order to test the hypotheses. This statistical model is commonly taken to apply to a binary dependent variable. The hypotheses are tested by using the following regression formula (see Appendix B for a definition of the variables used in the model):
SUS_REPit/ASRit/ASR_PRVit/ASR_LVLit = α + β1*ESG_SCOREit + β2*ESG_COMBit +
β3*SIZEit + β4*ROAit + β5*LEVERAGEit + β6*AGEit + β7*SenIndi + ε 3
Due to the nature of the data used in this research, the assumption that the underlying population data for all variables is normally distributed is violated4. Sheskin (2003) concludes that non-normally distributed variables (excessive skewness or kurtosis) or substantial outliers of variables can distort statistical tests. Although normality is not necessary for a logit regression model, it does provide a more stable solution and should therefore be taken into consideration as a limitation of this study. In addition, by using the chi-square test for homogeneity possible heterogeneity in the sample was found. To account for this, the logit regression model is undertaken with an added robust check.
3.3 Measurement of variables
The dependent variables used in the model are all binary, because the values are the result of a simple yes/no question. To make this measurable, dummy variables are created that take on a value of 1 if the answer was ‘yes’ and a value of 0 if the answer was ‘no’. First, the
3 Variables are measured for firm i in year t
4 Winsorizing of the data, in addition to removing the outliers (below 1st and above 99th percentile) of
dependent variable SUS_REP takes on a value of 1 if a firm has issued any form of a sustainability report, and 0 if not. Second, the dependent variable ASR takes on a value of 1 if the sustainability report of a firm has any form of external assurance, and 0 if not. Third, the dependent variable ASR_PRV takes on a value of 1 if the sustainability report of a firm is audited by a accountancy firm, and 0 if assurance was provided by another type of company. Final, the last dependent variable ASR_LVL takes on a value of 1 if a sustainability report of a firm received reasonable or a combination of limited and reasonable assurance, and 0 if
only limited assurance was provided.
To measure the independent variables of this study, the ESG performance and controversies, the ESG scores of Thomson Reuters are used. Thomson Reuters reviews company-reported data across 10 main themes5, categorized into an environmental, social and governance score. Together, these scores are combined into one overall ESG Score (Thomson Reuters, 2018). This score, weighted on a scale of 0 to 100, is used to measure the ESG_SCORE variable. In addition, a company’s involvement in scandals across any of the 10 main themes is also taken in to consideration. These scandals are captured from global media sources and reflected in a ESG Controversies Score. To take these controversies in to consideration, Thomson Reuters introduced the ESG Combined Score. The main objective of this score is to discount the ESG Score based on negative media stories. Therefore, if a company was involved in ESG controversies, the ESG Combined Score is calculated as the weighted average of the ESG Score and ESG Controversies Score per fiscal period, with recent controversies reflected in the latest complete period. When a company was not involved in ESG controversies, the ESG Combined Score is equal to the ESG Score. Involvement in ESG controversies (ESG_COMB) is therefore measured by using a firms’ ESG Combined Score.
Based on previous research, several control variables on firm- and industry-level are included in the model. The first control variable will be the size of the firm (SIZE), because there is reason to assume that a company’s size influences the disclosure of ESG performance (Lang & Lundholm, 1993). Lang and Lundholm argue that larger firms face more external pressure to issue sustainability reports and to disclose positive information. This could influence the outcome of their ESG score. The size of a company is measured by the natural
5 The ESG Score is measured through 10 different categories. Resource use, emissions and innovation
are indicators of environmental performance. Workforce, human rights, community and product responsibility are used to assess the social performance. Management, shareholders and CSR strategy
logarithm of the company’s assets. Next, the profitability of the firm will be measured by the return on assets (ROA). Also, the leverage (LEVERAGE) of the firm, which is measured by the ratio of total debt to total assets, is included. Finally, the last control variable on firm-level will be the age (AGE) of the firm, which is measured as the natural logarithm of the observed year minus the founding year of the company.
On industry-level, a control variable similar to studies of Peters and Romi (2015) and Cho and Pattern (2007) is used to measure sensitive industries (SenInd). Firms in environmentally sensitive industries experience greater political and social pressures related to sustainability issues (Peters & Romi, 2015). These firms have an increased need to manage risks and to provide credible and reliable sustainability information. Therefore, firms in sensitive industries are more likely to seek external assurance. To capture this, a dummy variable is created that takes on the value of 1 if the industry of the firm is considered environmentally sensitive and 0 otherwise. Similar to the study of Cho and Pattern (2007), this is determined based on the two-digit SIC industry-code: 13 (Oil exploration), 26 (Paper), 28 (Chemical and allied products), 29 (Petroleum refining), 33 (Metals) or 49 (Utilities). If a firm is part of any of these industries, it is considered to experience greater external pressure about the credibility of sustainability information and thus has a greater need to get assurance.
4 Findings and discussion
4.1 Descriptive statistics
In Table 3, the descriptive statistics and Spearman correlation of the selected variables are presented. Panel A shows that, from the total of 993 observations, for 841 observations a sustainability report was issued, and for 400 of these reports any level of assurance was provided. 75.3% (301 observations) of these reports received assurance from an accountancy firm, and 24.7% (99 observations) from any other company. Of all assured reports, only 11.5% (46 observations) received a reasonable, or combination of reasonable and limited level of assurance, versus 88.5% (354 observations) of reports with only limited assurance. Both the distribution of the level and provider of assurance are in line with previous studies (Cuadrado-Ballesteros et al., 2017; Fernandez-Feijoo et al., 2016).
The third column of Panel A shows that ESG_SCORE has a mean value of 70.315 and a median of 73.540. In contrast, the research of Gospodinov (2015) shows a mean of 62.779 and a median of 69.500 for the ESG_SCORE of 204 North American companies in the Oil & Exploration industry between 2009 and 2014. Chollet and Sandwidi’s (2018) study shows a mean of 53.68 and a median of 53.89 for the ESG_SCORE of a worldwide sample consisting of 23.194 observations between 2003 and 2012. The sample and period used in these studies clearly differ from this research, although the fact that large global companies are typically leaders in sustainability reporting (KPMG, 2017) could explain the higher mean and median values of Panel A, since the ESG score is based on company-reported data (Thomson Reuters, 2018). For ESG_COMB, Panel A shows a mean value of 46.917 and a median of 43.400, which indicates that firms in the sample were involved in ESG controversies. To the best of my knowledge, a similar score that measures ESG controversies like ESG_COMB has not been used in previous studies. The fifth column of Panel A shows that the standard deviation of both ESG_SCORE and ESG_COMB is close to 15. This indicates that, for a score on a scale of 0 to 100, the data points are just slightly spread.
Panel B of Table 3 compares the mean and median of the variables for assured and non-assured reports. It can be observed that ESG_SCORE is significantly higher (p < 0.0001) for assured reports compared to non-assured reports (Mean=75.550 and median=76.480 for assured reports versus mean=70.130 and median=72.940 for non-assured reports). This is also the case for ESG_COMB, although the difference is slightly lower (Mean=48.550 and median=44.500 for assured reports versus mean=46.822 and median=43.370 for non-assured
reports) and less significant (p < 0.05).
To sum up, it can be concluded that firms with an assured sustainability report show higher ESG performance and are slightly less involved in ESG controversies than firms without assurance. Furthermore, it can be observed from Panel B that firms with an assured report are larger, have lower return on assets, higher leverage and consist of slightly more companies from a sensitive industry.
A comparison of the descriptive statistics of the variables for firms that issued a sustainability report and firms without such a report is given in Panel C. Observing the data shows that ESG_SCORE is significantly a lot higher (p < 0.0001) for the first mentioned group of firms compared to the second (Mean=72.708 and median=75.030 for firms with a sustainability report versus mean=57.073 and median=59.500 for firms without). As for ESG_COMB, the difference is again slightly lower (Mean=47.773 and median=43.990 for firms with a sustainability report versus mean=42.179 and median=37.750 for firms without), but also highly significant (p < 0.0001). Although the different sample size of both groups (841 versus 152) is quite large, the difference in mean and median is remarkable. Shortly, it can be concluded that firms with a sustainability report show higher ESG performance and are less involved in ESG controversies than firms that did not issue such a report. In addition, Panel B shows that firms with a sustainability report are larger, have higher return on assets and higher leverage. Also, it seems that firms that issued a sustainability report are founded earlier than firms without a report. The statistics stated in Panel B and C of Table 3 are in line with expectations of H1, H2, H4 and H5. Naturally, the mean and median values are mostly used as summary statistics, so findings of the main regression model could differ.
Panel D of Table 3 shows the Spearman correlation between the dependent variables, variables of interest and control variables. This measure is chosen over the Pearson coefficients, because (as previously stated) based on the characteristics of the variables in the sample, the assumption that the data is normally distributed is violated. This makes the Spearman’s rank correlation the most appropriate measure to determine the dependencies between the variables (Lehman, 2005). Observation of the table reveals that sustainability reporting has a significant, positive association with ESG performance, also when the score is discounted for ESG controversies. These results provide support for H1 and H4: Firms with high ESG performance are more likely to issue a sustainability report. A similar association for sustainability reporting is found with the size, leverage and age of a firm.
correlation with sensitive industries is slightly significant for assured reports. In contrast, the variables of interest are both not significantly correlated with the level and provider of assurance. Therefore, support for hypotheses H3a, H3b, H6a and H6b is not found.
The variables ESG_SCORE and ESG_COMB are significantly correlated. This indicates that there is multicollinearity between the variables of interest of this study. This relationship between the independent variables might cause difficulties when assessing the reliability of the studied relations in the logit regression analysis.
Table 3: Descriptive statistics of selected variables Panel A: Full sample (N = 993)
N Mean Median Std. Dev. 25th Perc. 75th Perc.
SUS_REP 993 0.847 1.000 0.360 1.000 1.000 ASR 841 0.476 0.000 0.500 0.000 1.000 ASR_PRV 400 0.753 1.000 0.432 1.000 1.000 ASR_LVL 400 0.115 0.000 0.319 0.000 0.000 ESG_SCORE 993 70.315 73.540 15.028 63.530 81.390 ESG_COMB 993 46.917 43.400 14.887 38.060 50.360 SIZE 993 8.624 8.269 1.035 7.819 9.295 ROA 993 0.036 0.028 0.043 0.009 0.058 LEVERAGE 993 0.244 0.225 0.141 0.148 0.325 AGE 993 1.750 1.813 0.362 1.491 2.053 SenInd 993 0.198 0.000 0.399 0.000 0.000
Panel B: Assured vs. non-assured Reports
Assured Reports Non-Assured Reports Difference test N Mean Median N Mean Median T (p-value)
ESG_SCORE 400 75.550 76.480 441 70.130 72.940 0.000 ESG_COMB 400 48.822 44.500 441 46.822 43.370 0.043 SIZE 400 8.823 8.558 441 8.605 8.311 0.002 ROA 400 0.032 0.024 441 0.040 0.032 0.010 LEVERAGE 400 0.263 0.247 441 0.233 0.220 0.001 AGE 400 1.793 1.872 441 1.776 1.820 0.486 SenInd 400 0.233 0.000 441 0.181 0.000 0.063
Panel C: Firms with Sustainability Report vs. firms without
Firms with Sustainability Report
Firms without Sustainability Report
Difference test N Mean Median N Mean Median T (p-value)
ESG_SCORE 841 72.708 75.030 152 57.073 59.500 0.000 ESG_COMB 841 47.773 43.990 152 42.179 37.750 0.000 SIZE 841 8.709 8.405 152 8.155 7.814 0.000 ROA 841 0.036 0.030 152 0.033 0.027 0.000 LEVERAGE 841 0.247 0.231 152 0.227 0.188 0.000 AGE 841 1.784 1.839 152 1.562 1.544 0.000 SenInd 841 0.206 0.000 152 0.158 0.000 0.172
4.2 Empirical results and discussion
The outcome of the statistical analyses of a firms’ sustainability reporting choices, as explained in the logit regression model, is presented in Table 4. The dependent variables SUS_REP, ASR, ASR_PRV and ASR_LVL are regressed on the ESG score variables and control variables. To determine the goodness of fit of a logit regression model, the Pseudo R2
is commonly used. This measure is an estimate of the probability likelihood of the predicted outcome of the model. Next, the adjusted R2 is calculated to penalize the Pseudo R2 for
including too many predictor variables in the model. The analysis shows that the probability likelihood of the model is 23.3% (Adj. R2=0.233) for choosing sustainability reporting, 5.8%
(Adj. R2=0.058) for assurance choice, 2.8% (Adj. R2=0.028) for choosing the provider, and
3.4% (Adj. R2=0.034) for the chosen level of assurance.
The first variable of interest, ESG_SCORE, shows a positive relation (z = 8.44; p < 0.01) with issuing a sustainability report. This gives support for H1, indicating that higher ESG performance positively affects a firms’ decision to issues a sustainability report. This is in line with previous research (Lai et al., 2016; Jain, 2016; Casey & Grenier, 2015) and signaling theory, but contradictive to other studies (Clarkson et al., 2008; Michelon et al., 2015) explaining sustainability reporting as a mechanism of regaining legitimacy. Next, a positive relation (z = 6.25; p < 0.01) between ESG performance and seeking external assurance is also found, which is in line with H2 and can also be explained by the signaling theory. However, the model shows a negative relation between ESG performance and choosing a accountancy firm with a reasonable level of assurance, although these relations are not significant. Nevertheless, no support for H3a and H3b is found. In contrast, Casey and Grenier (2015) did find a positive relation between high sustainability performance and choosing a accountancy firm as assurance provider.
Panel D: Spearman correlation between SUS_REP, ASR, ASR_PRV, ASR_LVL, ESG_SCORE, ESG_COMB and control variables
SUS_REP ASR ASR_PRV ASR_LVL ESG_SCORE ESG_COMB SIZE ROA LEVERAGE AGE SenInd
[1] [2] [3] [3] SUS_REP 1 ASR - 1 ASR_PRV - - 1 ASR_LVL - - - 1 ESG_SCORE 0.329 *** 0.182 *** 0.041 -0.035 1 ESG_COMB 0.187 *** 0.091 *** 0.037 -0.076 0.409 *** 1 SIZE 0.225 *** 0.119 *** -0.134 *** -0.036 -0.183 *** 0.017 1 ROA 0.017 -0.089 *** -0.116 ** 0.041 0.072 -0.005 -0.258 *** 1 LEVERAGE 0.103 *** 0.121 *** 0.070 -0.001 -0.117 ** -0.008 -0.054 -0.085 * 1 AGE 0.226 *** 0.027 0.049 -0.096 * 0.032 -0.164 *** -0.148 *** 0.080 -0.012 1 SenInd 0.043 0.063 * 0.028 0.079 0.061 0.089 * -0.059 0.053 0.003 -0.103 ** 1
*, **, *** indicate if correlation is statistically significant at the α = 0.10, 0.05, and 0.01 levels, respectively.
[1] Refers to 993 observations [2] Refers to 841 observations [3] Refers to 400 observations
Also, signaling theory predicts that firms with high ESG performance want the highest level of assurance, although such a relation is not found. This can be explained by the fact that firms outweigh possible costs and benefits when choosing the level of assurance (Hasan et al., 2005), and that currently, these additional benefits are not lucrative enough for large global firms.
For the next variable of interest, ESG_COMB, a negative relation is found (z = -1.67; p < 0.10) with issuing a sustainability report. This indicates that a firm’s involvement in ESG controversies, which will lower the ESG_COMB score, also increases the likelihood that the firm will issue a sustainability report, which is opposite to the assumption under H4. This relation is in line with the findings of Gillet-Monjarret (2015), who found that media coverage can influence a firms’ sustainability reporting decisions, and that higher media exposure increases this influence. Since the sample of this study is covering large global firms, it can be expected that these firms are under a high level of media exposure. Therefore, such firms will likely use sustainability reporting as a mechanism to regain legitimacy when it is affected by controversies reported in the media. In contrast, a significant relation between ESG_COMB and assurance (ASR) is not found. You could argue that, when a firm is involved in ESG controversies, sustainability assurance is also a tool to regain legitimacy (Gillet-Monjarret, 2015). However, since negative information is already perceived as being credible (Casey & Grenier, 2015), the increased credibility from assurance won’t be as effective as is the case with positive information. Still, no significant relation for H5 is found. As for a firm’s decision regarding provider (ASR_PRV) and level (ASR_LVL) of assurance, no significant relation is found with a firm’s involvement in ESG controversies either. Therefore, the model found no support for H6a and H6b. To the best of my knowledge, no previous studies focused on the relation between provider and level of assurance and a firms’ involvement in ESG controversies.
The regression of the dependent variables on the ESG scores shows somewhat contradicting results. It seems that firms are more likely to signal good performance and use assurance as a mechanism to improve the reliability of such information, but also issue a sustainability report during controversial times. These findings are somewhat consistent with the findings of Hummel and Schlick (2016). They state that firms disclose sustainability information to reveal their superior performance to the market and distinguish themselves from poor sustainability performers, which is also in line with the findings of Connelly et al. (2011). Next, such reported performance needs assurance, because stakeholders are more
sustainability performance is still being reported to contribute to a sustainable company image (Hummel & Schlick, 2016). Naturally, this image needs to be restored when a firm is involved in ESG controversies. It is likely that this information is disclosed without assurance, because information that lacks reliability and comparability makes it easier to disguise the firm’s poor sustainability performance. To sum up, it seems that sustainability reporting can be predicted by both signaling and legitimizing behavior of firms.
The relationship between the other control variables and the dependent variables are mixed compared to previous studies. As for SIZE, it seems that larger firms are more likely to issue a sustainability report (z = 6.24; p < 0.01) and seek assurance (z = 3.51; p < 0.01), but also choose a non-accountant provider (z = -2.98; p < 0.01). Research by Casey and Grenier (2015) found similar results. Casey and Grenier state that the decision to seek assurance from non-accountants is because accounting firms show ineffective marketing of their assurance services towards large firms. The return on assets (ROA) of a firm is positively associated with issuing (z = 2.06; p < 0.05) and negatively associated with assuring (z = -2.45; p < 0.05) a sustainability report, but an increase in ROA is also associated with choosing a non-account provider (z = -2.83; p < 0.01). Furthermore, a positive relation between higher leverage and adopting external assurance can be observed from the model (z = 3.2; p < 0.01). This is inconsistent with findings of Casey and Grenier (2015), who find that highly leveraged firms are less likely to obtain assurance due to financial constraints and bank monitoring. As for the age of a firm, older firms are more likely to issue a sustainability report (z = 3.77; p < 0.01), but less likely to obtain a reasonable level of assurance (z = -2.71; p < 0.01). Finally, findings show that firms from a sensitive industries (SenInd) are more likely to seek external assurance (z = 1.99; p < 0.05). This could be due to the previously mentioned construction of my sample, but such firms likely face more pressure from the media to seek external assurance due to their environmentally unfriendly activities. However, Peter and Romi (2015) did not find such a relation.
Table 4: Logit regression of main logistic model Results from the logit regression of main model
SUS_REP (N= 993) ASR (N=841) ASR_PRV (N=400) ASR_LVL (N=400)
Coefficient Z-score Coefficient Z-score Coefficient Z-score Coefficient Z-score
ESG_SCORE 0.065 *** 8.44 0.042 *** 6.25 -0.008 -0.77 -0.016 -0.94 ESG_COMB -0.014 * -1.67 -0.0030 -0.52 0.007 0.79 -0.023 -1.51 SIZE 0.845 *** 6.24 0.269 *** 3.51 -0.383*** -2.98 -0.205 -1.36 ROA 4.848 ** 2.06 -4.076 ** -2.45 -8.916*** -2.83 4.814 1.35 LEVERAGE 1.049 1.26 1.764 *** 3.2 0.232 0.23 -0.238 -0.19 AGE 1.138 *** 3.77 -0.0190 -0.09 0.303 0.92 -1.190*** -2.71 SenInd 0.141 0.5 0.343 ** 1.99 0.187 0.62 0.558 1.59 Constant -11.465 -8.44 -5.728 -5.88 4.508 2.55 3.848 1.8 Pseudo R2 0.239 0.066 0.045 0.051 Adj. R2 0.233 0.058 0.028 0.034 Wald Chi 135.92*** 64.94*** 16.12** 16.07**
Logit regression with Sustainability Reporting, External Assurance, Assurance Provider and Level of Assurance as dependent variable. *, **, *** indicate if Z-score is significantly different from zero at the α = 0.10, 0.05, and 0.01 levels, respectively, for two-tailed tests.
4.3 Robustness checks
In other studies, the usage of two different variables to distinguish strong and weak sustainability performance is common (Dhaliwal, Li, Tsang & Yang, 2011; Casey & Grenier, 2015). These scores are often based on the ratings of independent bureaus like Thomson Reuters. According to Hummel and Schlick (2016), the rating process is often not fully transparent, which makes it hard to assess the reliability of the data. Therefore, they developed their own and more refined measure by hand-collecting company data regarding sustainability performance. Next, they transformed all the data into a continuous [0, 1] scale by assigning ‘0’ to the worst and ‘1’ to the best performance indicator values. As a robustness check of the findings in this study, Panel A of Table 5 presents the outcome of a similar approach. The independent variable ESG_SCORE is rescaled into a dummy variable that equals 1 if the ESG score of the observed firm is in the top 50% of all scores (High ESG performers), and 0 if the score is part of the 50% lowest ESG scores (Low ESG performers). The outcome shows a positive relation between high ESG performers and issuing a sustainability report (z = 6.59; p < 0.01). In addition, high ESG performance is also associated with seeking external assurance (z = 4.81; p < 0.01). These results are similar to the main regression model of this study and also support H1 and H2. Additionally, as in the main model, no significant relation between the ESG score and a firms’ choices regarding level and provider of assurance is found.
The results of a second robustness check are presented in Panel B and C of Table 5. Studies regarding firms’ sustainability reporting are often done solely for US firms (Casey & Grenier, 2015) or European firms (Hummel & Schlick, 2016) due to different country- and continent-specific factors, like law regulations and stakeholder behavior. To assess if such factors affect the outcome of this study, the sample is divided in three groups; North American (which solely consists of American companies), European and Asian firms. Next, the dependent variables SUS_REP and ASR are regressed on the ESG Scores for each of the newly constructed samples. Regression for level and provider of assurance choice was not possible, because it would cause the model to be insignificant due to the smaller sample size. Panel B of Table 5 shows a positive relation between the ESG score and issuing a sustainability report for North American (z = 6.4; p < 0.01) , European (z = 6.18; p < 0.01) and Asian (z = 3.68; p < 0.01) firms. This is consistent with the findings and supports H1. Panel C shows a positive relation between the ESG score and obtaining external assurance for European (z = 5.96; p < 0.01) and Asian (z = 3.35; p < 0.01) firms, which also supports H2.
However, no significant relation is found for North American firms. Casey and Grenier (2015), who’s study solely focused on U.S. firms, also found that there is a low U.S. demand for sustainability reporting assurance. They explain this by stating that in the U.S., intense regulatory oversight is serving as a substitute form of enhancing the credibility of sustainability reports. Peters and Romi (2015) add to this that internal audit is often used as a substitute for external assurance of report reliability, because of the fear of litigation in the U.S. This makes it challenging for assurance practitioners to provide external assurance on the U.S. market. Hence, no support for H2 is found when solely taking U.S. firms in to consideration.
