Faculty of Economics and Business
Does Assurance Better
Explain the Link Between
CSR Disclosure and Firm
Value?
E.M. Dijksma
ABSTRACT
Recent research highlights the importance of the determinant and effects of external assurance on
corporate social responsibility (CSR) disclosure. This study therefore empirically investigates that
influence of CSR assurance (CSRA) on firm value as proxied by share price. Based on legitimacy theory
and signaling theory, CSRA is expected to have a positive influence on the market value of the
respective firm, as it increases the perceived reliability. This positive relationship can be strengthened
by the quality of CSR disclosure, the higher the quality, the more valuable enhancing the reliability will
be. Moreover, specific characteristics of the assurance engagement, i.e. scope, level, and provider are
taken into account to examine their influence on market value. After examining 2063 European listed
firms, the results do not support the predictions. These findings contribute to the existing knowledge
regarding CSRA, and is theoretically relevant since a research gap has been closed. Practical
implications can be derived from the study as well since it holds some relevant insight for companies
as well as policy-makers.
2
Does Assurance Better Explain the Link Between CSR Disclosure and Firm Value?
University of Groningen
Faculty of Economics and Business
MSc Accountancy
Name: Esther Manon Dijksma
Student number: S2473305
Address: Anna Paulownastraat 17, 9725 JP, Groningen
Phone Number: +31617090017
E-mail: E.m.dijksma@student.rug.nl
Supervisor: dr. N. Hussain
Word Count: 10.140
Date: 25-06-2018
3
Table of Contents
Introduction ... 4
Theoretical Framing... 6
Legitimacy theory ... 6
Signalling theory ... 6
Hypotheses Development ... 8
Research Method ... 11
Sample ... 11
Dependent variable ... 11
Independent variables ... 12
Control variables ... 12
Statistical model ... 13
Results ... 15
Descriptive statistics ... 15
Correlation analysis ... 15
Regression analysis ... 16
Conclusion ... 19
Findings ... 19
Theoretical implications ... 20
Practical implications ... 21
Limitations and future research ... 21
4
Introduction
The last 20 years are characterized by a steadily increasing emphasis on decision-useful disclosure of corporate social responsibility (CSR) activities. Companies are shifting from solely focussing on increasing their profit, to more socially responsible decision-making. Research has shown that due to a growing global concern for CSR, companies behave more responsible in terms of their business activities regarding the community and the environment (Darus, Sawani, Mohamed Zain & Janggu, 2014). ‘Large corporations, and especially large multinational corporations, that have a significant footprint in terms of environment and sustainability, human rights protections, treatment of workers, and global transactions that are vulnerable to charges of bribery and corruption, must anticipate public pressure of information and take actions to minimize regulatory investigations and address concerns for institutional investors and other external stakeholders’ (Sethi, Martell & Demir, 2017, p.61). In order to anticipate this growing amount of pressure, companies are engaging in CSR and subsequently voluntary CSR disclosure, as the latter is proven to be an effective means to tackle contemporary pressures (Adam & Shavit, 2008; Sethi et al., 2017). Besides anticipating public pressures and minimizing regulatory investigations, CSR can lead to increased firm performance as it enhances firm reputation and leads to cost savings (Kang, Lee & Huh, 2010).
Although, as argued above, CSR is expected to have a positive influence on firm performance, there is no consensus regarding the value relevance of CSR and CSR disclosure within academic literature. This lack of consensus can be explained on the basis of two contradicting theories. On the one hand, Friedman’s theory (1962, 1970) states that the sole responsibility of an organization is to increase shareholders’ wealth. Thus engaging in CSR would be considered to be a waste of resources. On the other hand, Carroll (1979, 1999) acknowledges that CSR goes beyond shareholders’ wealth and includes legal and ethical responsibilities. Consequently, empirical evidence supports this ongoing debate regarding the value relevance of CSR disclosure. Whereas, several researchers find that CSR disclosures are enhancing firm value (Cahan, De Villiers, Naikers & Van Staden, 2015; Dhaliwal, Li, Tsang & Yang, 2011; Dhaliwal, Radhakrishnan, Tsang & Yang, 2012; Servaes & Tamayo, 2013), others find a negative relationship between CSR disclosure and financial performance (Vance, 1975; Wright & Ferris, 1997; Cordeiro & Sarkis, 1997; Campbell & Slack, 2011). Despite the lack of consensus, Tschopp and Huefner (2015, p. 565) conclude that reporting on CSR activities (CSRR) ‘seems destined to become a key part of the overall accounting reporting framework, joining external financial reporting, income tax reporting, regulatory reporting, and internal reporting.’ Hence, more research on CSR, the effects hereof and perchance mediating factors is required to enlarge the existing knowledge regarding CSR disclosure.
Perceived (un)reliability might explain the previously described differences in empirical results. Servaes and Tamayo (2013) argue that firms disclosing CSR information ‘signal’ to their stakeholders that there is above average quality. Contrarily Sethi et al. (2017) argue that CSR has led to voluminous and visually attractive reports wherein information provided is objective and substantive and therefore lacks reliability. Expenditure allocated to CSR disclosure cannot be justified if the costs outweigh the benefits. Therefore, the associated reliability of the respective CSR report should be strengthened. Voluntary assurance has the possibility to bend the arguments that CSR disclosure leads to unreliable reports, and hence has the influential power to alter the perceived reliability. Voluntary CSR assurance (CSRA) might thus be a potential mediating factor that can be used to explain the benefits of CSR disclosure. According to Pflugrath, Roebuck and Simnett (2011) the need for third-party assurance rises, as argued above, due to a lack of trust between companies and society, so an audit should give the audience more assurance regarding the information disclosed. The lack of public trust stems from amongst others, the financial crisis, and recent accounting scandals e.g. Enron and WorldCom. Academics confirm the need for enhanced reliability, concluding that the primary factor to engage in CSRA was the need for enhanced credibility (Cho, Michelon, Patten & Roberts, 2014; Park & Brorson, 2005; Simnett, Vanstraelen & Chua, 2009). Confirmatory, the results of this assurance indeed lead to enhanced credibility among the users of the CSR report (Hodge, Subramaniam, & Stewart, 2009; Pflugrath et al., 2011). Whereas multiple researchers focus on the identification of drivers behind engaging in purchasing third-party assurance and the choice of assurance
5
provider (Bagnoli and Watts, 2017; Mock, Strohm & Swartz, 2007; Mock, Rao & Srivastava, 2013; Park & Brorson, 2005; Simnett et al., 2009), few studies focus on the results hereof (Hodge et al., 2009; Pflugrath et al., 2011). Since external assurance is a costly decision, it may be assumed that benefits outweigh the costs; firms would only undertake a respective costly decision when the expected value outweighs the costs (Defeo & Falk, 1998; Simnett et al., 2009). Despite of these predictions, the research regarding the beneficial effects of CSR assurance on firm value is divided. No statistical significance has been found yet to support this relationship (Cho et al., 2014; Peters & Romi, 2015). Therefore this study aims to close the research gap of CSR disclosure on the one hand and firm value on the other, by considering the mediating effect of external assurance.
In order to explain the mediating effect of CSRA on firm value, multiple theories are included. Firstly, legitimacy theory can be used to explain the CSR assurance, since assuring is closing a legitimacy gap when a CSR report is seen as a self-promotional document (Sethi et al., 2017). Additionally, as mentioned already, CSR disclosure is an effective means to tackle the increased public pressure. Besides legitimacy theory, signalling theory can be used to explain the phenomenon of CSRA, since disclosing information regarding CSR activities signals to the market that there is a superior sustainability strategy, and assurance enhances the credibility of the information that is being disclosed. This will, in theory, lead to market participants who value firms who disclose CSR information with an assurance statement more, thus increasing firm value (Cho et al., 2014). This study is related to, but differs from, the work of Cho et al. (2014), Peters and Romi (2015), Simnett et al. (2009), Pflugrath et al. (2011) and Hodge et al. (2009) as they examined the effects of CSRA within the United States and this study focusses on European firms. Therefore, this study is contributing to the existing literature as another region will be examined, including a recent time-period of multiple years, namely 2012-2017. European is used instead of any other region, since Europe has the highest assurance rate, and additionally, CSR awareness is highest within Europe (Ackers, 2009; Zorio, García-Benau, & Sierra, 2013). Furthermore, this study does not only include the presence of external assurance, but includes quality characteristics of the assurance engagement as well. These characteristics are the assurance level, assurance scope, the type of assurance provider, and whether the provider belongs to a Big4 auditing firm. Taking multiple characteristics of the assurance engagement into account contributes to the existing literature, as the combination of these characteristics has not been examined yet.
CSR, CSRR and external assurance hereof are concepts that are exponentially growing nowadays within academic literature, as already discussed. Likewise, CSRA receives a growing amount of attention in practice. KPMG studies CSR reporting among the world’s top companies (G250) and N100 each year. The percentage of companies engaging in CSRA increased from 30% in 2005 to 67% in 2017 (KPMG, 2005; KPMG, 2017). Thus, the number of companies choosing to seek external assurance of their CSR report is increasing significantly. All in all, recent research does acknowledge that purchasing external assurance is a costly decision; and a growing amount of attention is devoted to CSRA by academic scholars as well as practitioners, however whether this decision is beneficial in regard to firm value is underexposed. As purchasing third-party assurance is costly, the value-relevant question, to both academic literature and practitioners, which will be addressed in this study is: ‘What
is the influence of purchasing external assurance on CSR reporting on firm value?’
The predictions within this study, based on legitimacy theory and signalling theory states that assurance and the quality characteristics of the assurance engagement will be positively related to firm value. Although, contrary to the predictions, no statistical evidence has been found to support this relationship confirming the results of Cho et al. (2014) and Peters and Romi (2015).
In the next section, the literature review, the relevant theories regarding the research question, will be discussed. The relevant theories that will be presented here, are legitimacy theory and signalling theory. Subsequently, the hypotheses generation will be discussed. Thereafter, the methodology section will be disclosed, as well as the data collection. Then the results will be discussed. Eventually the last section will contain the conclusion, where the main findings will be discussed, and the discussion with the limitations and suggestions for future research.
6
Theoretical Framing
In this section, legitimacy theory and signalling theory are discussed to develop insights about the relationship between CSR reporting and external assurance thereof on the one hand and firm performance on the other hand.
Legitimacy theory
Within academic literature, legitimacy theory has been used to explain CSR disclosure. Legitimacy theory is based on the relationship between organizations, the government, individuals and society. Conflicts within society are acknowledged and society, politics and economics cannot be separated as these concepts are intertwined (Deegan, 2002; An, Davey and Eggleton, 2011). How society perceives the legitimacy of a company is crucial for its survival. Organizations namely have an existence right when their practices are being perceived as legit. There is a so-called ‘social contract’ to which companies need to adhere. Legal requirements are the basis of this contract enriched with implicit conditions by non-regulatory expectations. Although the starting position is the social contract, it is not sufficient to only adhere to this contract, companies should take all precautions to assure that they live up to the norms, values and expectations of society (An et al., 2011). Managers can tackle the described legitimacy issues with multiple strategies. According to Lindblom (1994) and Dowling and Pfeffer (1975) voluntary disclosure can be used as an effective means to execute those strategies. Organizations namely have access to ‘private’ or ‘insider’ information, i.e. information that is not available to the public. Disclosing information decreases the information gap between the company and its stakeholders, which can help to address and reduce the legitimacy concerns. Hence, companies respond to societal concerns and public pressures by voluntarily disclosing information. CSR reports can be seen as a product of increased public pressure (Sethi et al., 2017). Further on, Sethi et al. (2017) stress out that when companies do not adhere to society’s expectations, it can pose a threat to the legitimacy of a company. Regulatory investigations may then be conducted which might have a harmful effect on the perceived legitimacy.
Besides explaining the need for voluntary disclosure, legitimacy theory can be used to underline the importance of CSR assurance. As Sethi et al. (2017) properly emphasize, companies who disclose CSR information and self-asses the quality hereof impose a threat to reliability. Deegan (2002) acknowledges the reliability issues as well, noticing that when managers are left free to decide what information will be disclosed and what information will be held private, the reports might be biased. When the CSR reports are being perceived by the public as biased, or unreliable, this poses threats to a company’s legitimacy and thus survival. Therefore, a strategy that can be implemented is to provide the public with transparency and verification of the information disclosed, i.e. external assurance (Sethi et al., 2017). Otherwise the CSR report can be seen as a self-promotional document, leading to a legitimacy gap. Thus, when CSR assurance is purchased, this will contribute to firm value, as the legitimacy gap is closed and the perceived legitimacy by the public will most likely be enhanced.
Signalling theory
Another possible explanation of CSR disclosure can be addressed to signalling theory. Signalling theory was first introduced by Spence (1973), and relates to the information asymmetry problem as described by agency theory. Here the underlying idea is that two parties who are bounded to asymmetric information, can reduce the information asymmetry by sending a signal. One party sends a signal, then based on this signal, the other party can adjust its behaviour. Assuming that the signal is positive, it is beneficial to both parties to lessen the information asymmetry. According to An et al. (2011, p. 575) ‘signalling theory is concerned with how to address problems arising from information asymmetry in any social setting. It suggests that information asymmetry should be reduced if the party possessing more information can send signals to other interest-related parties. A signal can be an observable action, or an observable structure, which is used to indicate the hidden characteristics (or quality) of the signaller.’ Whenever a signal is send, the sender assumes that it will be favourable to the receivers, meaning that it indicates a higher quality than its competitors for example. The more difficult the signal will be to imitate, the more effective it will be.
7
As mentioned already, managers have access to private information. This private information can contain, expected future cash flows, upcoming investments, lawsuits, risk management etc. This information is not available to the public, i.e. there is asymmetric information. Based hereon, shareholders cannot optimally distinguish the quality among multiple firms. As explained by An, et al. (2011, p. 576) ‘consequently the firm with a quality above average incurs an opportunity loss because its superior quality is not perceived by the investors, while the firm with a low quality obtains an opportunity gain. Under the circumstances, the high-quality firm has an incentive to highlight its superior quality in order to attract more investors.’ In conclusion, the assumption behind signalling theory is that companies who are in possession of higher quality products or services will signal this to their stakeholders to gain an advantage compared to its competitors. Firstly, the signal is interpreted by all relevant stakeholders leading to a revaluation of the company. Hereafter the stakeholders will make decisions that are more beneficial to the company then when the signal was not send (Whiting & Miller, 2008). Secondly, as the stakeholders make decisions that are more favourable to the company, the costs of raising capital would be decreased since a company can obtain more investment. Thus, the beneficial effects for the company is magnified. According to An, et al. (2011) CSR disclosure is one of the most effective ways to disclose information about the company. Morris (1987) states that as long as the difference in price received after the signal exceeds the costs, signalling will be an on-going process.
Servaes and Tamayo (2013) build upon signalling theory as well claiming that firms can signal better quality products by disclosing CSR information. The perceived effort that a company undertakes to acquire external assurance, i.e. the required amount of time and money spent on the assurance, ‘signals’ to consumers that the company is willing to invest in CSR, and their disclosure send a ‘signal’ that they are confident enough concerning their practices to make this private information public. Eventually, this leads to a positive relationship between CSR disclosure and firm value (Servaes & Tamayo, 2013). The results from Sethi et al. (2017) indicate that firms with higher quality CSR reports are more likely to have their reports assured. As can be concluded based on signalling theory, their incentive to signal the respective ‘high-quality’ report is higher.
8
Hypotheses Development
This sections deals with the specific predictions about the relationship between CSRA and firm performance. Moreover, the specific predictions related to the quality characteristics of the assurance engagement will be discussed. The concepts of CSRA and the specific quality indicators will also be explained in more detail.
As already described, the beginning of the 21st century was characterized by more a demanding market for CSR activities. Companies did not only need to publish their financial data, CSR disclosure was demanded by the public as well. This made room for more integrated reporting, i.e. financial and CSR activities combined. Meaning that the focus is moving from solely publishing quantitative data to more qualitative data (Zorio et al., 2013). This growing concern for CSR activities is driven by a diverse group of stakeholders, namely shareholders, consumers, societies and the government (Kang et al., 2010). In addition to an increased concern, there is another trend visible. Namely Gallup (2014) notes that public trust in firms and more specifically in large organisations is decreasing. According to Sethi et al. (2017, p. 61) ‘a company’s self-assessment of the quality of its CSR report carries a measure of public scepticism, which must be mitigated by (a) providing information that meets societal expectations, (b) meets or exceeds industry standards for similar information, and (c) contains robust measures of verification and transparency.’ When one of these conditions is absent, the CSR report bears the risk of being viewed by the company’s stakeholders as a ‘greenwashing’ product. Meaning that its sole purpose is to self-promote the company, without actually engaging in CSR activities. As a consequence, the CSR report loses its value as enhancing public trust and decreasing information asymmetry (Sethi et al., 2017). The emerging trend of CSR disclosure, in combination with decreased public trust, calls for the need of external assurance.
In order to enhance the perceived reliability of the CSR report, CSRA can play a vital role. As discussed in the introduction, voluntary adoption of assurance is not commonly studied although it is an emerging topic, only a small number of empirical studies examined the drivers behind voluntary assurance. This is due to the fact that assurance mostly relates to assurance of financial reports which is mandatory in most developed countries (Simnett et al., 2009). Chow (1982) was one of the first to examine the concept of voluntary assurance. He found that agency costs were correlated to voluntary adoption of external assurance. Blackwell, Noland and Winters (1998) argue on their turn that the need for external assurance arises when the company wants to decrease the information asymmetry with regard to institutional creditors. They conclude that assurance is an effective means to decrease this information asymmetry. This finding is supported by Carey, Simnett and Tanewski (2000) who studied the demand for assurance within family businesses. Multiple studies have focussed already on the relationship between voluntary CSRA and enhanced perceived credibility (Cho et al., 2014; Park & Brorson, 2005; Pflugrath et al., 2011; Simnett et al., 2009). Pflugrath et al. (2011) support the notion that assurance is used to increase credibility, assured information tends to be perceived as more credible compared to non-assured information. In addition, Simnett et al. (2009) find that undergoing a voluntary assurance engagement facilitates greater user confidence.
In addition Darus et al. (2014) state that it is merely timely that non-financial information is provided with third-party assurance to enhance credibility. The driver behind third-third-party assurance stems from the quest by shareholders who rely on this verification in their decision-making process. Present-day, public trust is decreasing and concerns regarding transparency, reliability and accountability of information provided by organisations are growing (Gillet, 2012). Providing stakeholders with assurance is a response by the organisation to provide them with external verification. CSRA will verify the information presented by the company and thus increase the credibility. This statement is supported by the findings of Adams and Evans (2004) voluntary assurance enhances the perceived credibility.
From the discussion above, it can be concluded that external assurance will increase the credibility of the information that is being disclosed and reduce information asymmetry. Thereafter, based on the described
9
theories, assurance will close the legitimacy gap and ‘signal’ above-average performance to the market. This is expected to have a positive influence on firm value. Therefore, this first hypothesis will lead as follows:
H1: Firm value is positively associated with external assurance of CSR.
As Cho et al. (2014) point out, the relationship between firm value and CSRA could be influenced by the actual social and environmental performance. They argue that the higher the quality of CSR disclosure, in combination with CSRA, the higher firm value. Increased credibility is one of the main determinants and effects of CSRA. Therefore, Cho et al. (2014) predict that when a company is a better performance in terms of social and environmental activities and the disclosure quality is higher, the effect that assurance has on firm value will be higher. Legitimacy theory and signalling theory predict the same positive effect of CSR disclosure on firm value. Therefore, the interaction term of CSR disclosure and assurance will be included to test the second hypothesis, which leads as follows:
H2: Greater CSR disclosure in combination with CSRA is positively associated with higher firm value.
Sethi et al (2017) argue that the current market of CSRA providers can be divided into four categories. The first category consists of regular accounting firms, including Big4 auditing firms. The Big4 auditing firms are EY, KPMG, PwC and Deloitte. The second group consists of assurance provider firms who are specialized in integrity (e.g. Bureau Veritas). The third group includes academic groups or scholars and independent NGOs. Where the last group includes the internal audit department. They classify the first two assurance providers as ‘formal’ assurance providers, since they provide the report with a formal certificate. The division in organizations providing assurance is also recognized by Darus et al. (2014). They acknowledge that there are three parties engaging in CSRA, namely, consultants, accounting firms and non-governmental organizations. Assurance provided by formal assurance providers creates greater credibility compared to the latter two groups identified by Sethi et al. (2017). Additionally, Big4 accounting firms have access to more resources compared to non-Big4 accounting firms, their capacity is greater and thus they can provide clients with more effective monitoring (DeAngelo, 1981; Simnett et al., 2009).
Researchers Pflugrath et al. (2011) find that financial analysts perceive information to be more trustworthy and credible when assurance is provided by a professional accountant compared to a non-formal accountant. Manetti and Becatti (2009) draw the same conclusion, using a professional or formal auditors to provide CSRA will decrease the credibility gap regarding CSR reporting. Zorio et al. (2013) acknowledge that there is a large difference in quality among CSRA. They also address this difference in quality to the different type of assurance providers. From this research the conclusion is that auditing firms deliver higher quality CSRA than non-auditing firms, and that among auditing firms, the quality provided is the highest when provided by a Big4. Analysis shows that CSRA is growing, more specifically the number of CSR audits performed by audit firms is increasing relatively to audits that are performed by non-formal accounting firms (KPMG, 2017).
This leads to the second and third hypotheses:
H3: Firm value is higher when CSRA is provided by a formal auditing firm compared to a non-formal auditing firm. H4: Firm value is higher when CSRA is provided by a Big4 auditing firm compared to a non-Big4 auditing firm.
The prediction made by Cho et al. (2014) that higher CSR disclosure quality accompanied by external assurance leads to higher firm value, can also be translated to the specific characteristics of the assurance engagement. The higher the quality of CSR disclosure, in combination with higher credibility in terms of assurance provider, the bigger the effect on firm value. Therefore, the following hypotheses are generated:
10
H5: Greater CSR disclosure in combination with a formal auditing firm providing the CSRA will lead to greater firm value compared to a non-formal auditing firm providing the CSRA.
H6: Greater CSR disclosure in combination with a Big4 auditing firm providing the CSRA will lead to greater firm value compared to a non-big4 auditing firm.
Providing an assurance statement does not go without challenges for assurance providers. Namely, CSR reports are focussed on qualitative data instead of quantitative data and due to the lack of rules and guidelines there is an inconsistent format. Auditors therefore rely on documents provided by the organization, which might cause a lack of independent verification (Zorio et al., 2013). Hodge et al. (2009) relate to the problems that arises from the lack of generally accepted reporting criteria for CSR as well. They argue that the lack of generally accepted reporting criteria creates problems for the provision of assurance statement. The lack of guidance does also impose restrictions on the usefulness of CSRA, since as a result significant differences might arise in scope, level and standards applied during the auditing process. The scope of the assurance engagement relates to the sections of the CSR report that are being assured, varying from GHG only to the entire CSR report. The level of the assurance engagement relates to the statement that is provided, limited assurance versus reasonable assurance. Limited assurance is stated negatively, nothing has come to the attention that the CSR report was not in accordance with respective guidelines. Whereas reasonable assurance is stated positively, that the CSR report is in accordance with the respective guidelines (Hodge et al., 2009). Highly reputable accounting firms, i.e. formal accounting firms including the Big4, are more likely to provide ‘limited assurance’ due to the difficulties associated with the assurance engagement (Ballou, Heitger, Landes & Adams, 2006).
However, the higher the amount of time and money spent on the assurance engagement, the higher the quality of respective engagement. As more resources are being allocated to the assurance engagement, the scope of the assurance engagement can be increased, as can the level of the engagement. As a result, the reliance on internal documents will be reduced and thus the amount of independent verification will be enhanced. Hence, the credibility associated with the assurance engagement is magnified. When the level and scope of the engagement are higher, firms ‘signal’ their above average quality to the market. Furthermore, the legitimacy gap can be further reduced, since the higher the level and the scope of the engagement the lower the amount of information asymmetry and the higher will market value be.
Hence, the seventh and eighth hypothesis will be stated as follows:
H7: Firm value is higher when the scope of the assurance engagement is greater. H8: Firm value is higher when the level of the assurance engagement is greater.
The assurance engagement characteristics scope and level can also be influence by the extent of CSR disclosure. Thus relating to the arguments of (2014) that higher CSR disclosure quality accompanied by external assurance leads to higher firm value. The higher the quality of CSR disclosure, in combination with higher credibility in terms of scope and level of the assurance engagement, the bigger the effect on firm value. Therefore, the following hypotheses are generated:
H9: Greater CSR disclosure in combination with a wider scope within the assurance engagement will lead to greater firm value compared to a more narrow scope within the assurance engagement.
H10: Greater CSR disclosure in combination with a higher level of the assurance engagement will lead to greater firm value compared to a lower level of the assurance engagement.
11
Research Method
Sample
In order to conduct an investigation on the relationship between CSRA and firm value, to test the hypothesis, a sample of publicly-listed European firms during the period 2012-2017 is used. This data is collected using the DataStream (worldscope and ASSET4 ESG) and GRI database. The data on CSR reports, and environmental, social and governmental information is collected from ASSET4 ESG. Moreover, financial data is collected from worldscope. Data on assurance, including level and scope of assurance and provider of assurance is collected from the sustainability database of the Global Reporting Initiative (GRI). GRI provides students with free access to this database. The time period 2012 until 2017 is used taken the availability of data into account. This is the most recent period for which all data is available.
The data on European listed firms during 2012 to 2017 that was retrieved from worldscope yielded 6456 firm-years. As the worldscope data and GRI data could not be automatically matched, they two databases are manually compared. After this procedure, a list of 3691 firm-years remain for the time period 2012-2017. Hence, the initial sample, consists of 3691 companies that issue a CSR report. Thereafter, companies with missing or non-available data on assurance are deleted as well, together with companies with missing or non-available data on financial variables i.e. the worldscope database. These companies are not relevant to take into account, as no causal relationship between the dependent and independent variables can be analysed. Within this group, 718
companies had a fiscal year-end did was not equal to the 31st of December for that given fiscal year. These
companies were dropped from the sample, as the dependent variable is collected for the data March 31st. In
total, the sample consists of more than 2063 firm-year observations for 615 publicly-listed European firms issuing CSR report. Within this sample, there are companies with a CSRA statement and companies without a CSRA statement. The resulting sample therefore consists of organisations from 21 different European countries. The sample selection that was followed after the two datasets are matched is shown in table 1.
Table 1: Sample Selection
European listed firms during 2012-2017 (in firm-years)
3690 Less:
Firm-year observations without Worldscope data available (432)
Firm-year observations without GRI data available (477)
Firm-year observations where fiscal year did not end on 31st of December (718)
Final sample 2063
This study takes European companies into account since previous studies focussed mostly on U.S. firms (Cho et al, 2014; Peters & Romi, 2015; Simnett et al., 2009; Pflugrath et al., 2011; Hodge et al., 2009). Moreover, Europe is the most active CSR-reporting region and has the highest assurance rate, so it is valuable to investigate the impact of CSRA not on U.S. firms but on firms where apparently CSR awareness is high (Ackers, 2009). In addition to this, Ackers (2009) states that Europe has the highest external assurance rate, namely 64% of all CSRA statements internationally are provided within Europe.
Dependent variable
Based on Cho et al. (2014), Peters and Romi (2015) and Ohlson (1995), the market valuation model is used to evaluate the influence of assurance on firm value. In line with the existing research model, share price is used to represent firm value. Share price is collected from worldscope at the time of t+3months, meaning the share price
12
Independent variables
In order to perform the regression analysis, multiple independent variables are included. As the research model developed by Cho et al. (2014), Peters and Romi (2015) and Ohlson (1995) will be used to examine the influence of assurance on firm value, the independent variables used in this model will be transferred to this study. The first independent variable is the book value of equity (BVE). Which is measured as the book value of equity divided by the number of shares outstanding. The book value of equity is expected to have a positive influence on the price of common shares. Thereafter, earnings per share (EPS) is included as the second independent variable. This variable is measured as the earnings of a company before extraordinary items divided by the number of outstanding shares. Similar to book value, this second independent variable is expected to have a positive relationship with the share price.
To be able to test the hypothesis related to assurance, and the specified characteristics of assurance, data is obtained from the GRI database. Following Simnett et al. (2009) assurance (ASSURANCE) is included as the third independent variable. It indicates whether the CSR report contains an independent CSRA statement or not. If the CSR report is accompanied by a CSRA statement provided by an independent party, the indicator variable equals 1, when the CSR report is not assured the value equals 0. The fourth variable relates to the provider of the CSRA statement (PROV). When the assurance statement is provided by a formal accounting firm, the indicator value equals 1. Consequently, the indicator value equals 0 when the assurance is provided by a third party but a non-formal accounting firm. The fifth variable is Big4 (BIG4), which incorporates the fact that information is perceived to be more credible when it is assured by an accounting firm included in the Big-4, than a formal accounting firm not included in the Big4. Indicator value 1 is given to CSRA statement that are provided by accountants belonging to the Big4 (meaning: EY, KPMG, PwC or Deloitte), and value 0 is given to any other CSRA provider. The sixth independent variable that is included is assurance scope (SCOPE). The GRI sustainability database differentiates within this dimension between specified sections, entire sustainability report, GHG only and not specified. Thus, scope relates to the extent of the assurance engagement, and the amount of time and resources spent. The broader the scope, the higher the credibility, i.e. a CSRA statement which covers the entire sustainability report is more reliable and credible than a statement which only covers specified sections. The indicator value for this variable will equal 1 when the assurance engagement covers the entire CSR report, and it will equal value 0 otherwise. The seventh variable is assurance level (LEVEL), this relates to the level of assurance provided. The GRI sustainability database differentiates within this dimension between reasonable/high assurance, limited/moderate assurance, a combination and not specified. The level of assurance also relates to the amount of time and resources spent on the assurance. It is expected that reasonable/high assurance has the highest impact on the dependent variable, therefore this type of assurance is given the value 1. The indicator value equals 0 when the assurance level equals the other types (limited/moderate, combination and not specified). The last independent variable that will be included is the quality of CSR disclose, i.e. the ESG score(ESG). This variable is collected the Thomas Reuters/Asset4 ESG database. Thomas Reuters ESG scores are an enhancement of CSR disclosure scores, the CSR performance of a company is rated based on publicly available information for more than 400 measures. Thus, examining the quality of CSR performance as well as CSR disclosure.
Control variables
In order to ensure that the research is free from any bias, control variables are included. In accordance with Simnett et al. (2009) size (SIZE) and leverage (LEV) are included as control variables, these variables cause higher agency costs, which is associated with demand for voluntary assurance. Due to agency costs, leverage and size will be expected to be negatively related to share price. Firm size will be measured as the natural logarithm of a firm’s total assets at the end of the year. Leverage is the ratio of total liabilities divided by total assets. In addition to this, based on Cho et al. (2014), return on assets (ROA) is included as a control variable, which is expected to be positively related to share price. Moreover, industry (IND) is included as a control variable. Companies in environmentally or socially sensitive industries can experience a greater demand/need to engage in external assurance to enhance their credibility. For example, the need for a company to assure its CSR report is greater in
13
the mining industry compared to the retail industry, there are namely more incentives to misstate such information on purpose (Pflugrath et al., 2011). Assuring their CSR report will enable them to better manage their risks and improve stakeholders’ confidence (Simnett et al., 2009). These industries are defined by Simnett et al. (2009) and include, mining, paper, chemicals, petroleum, metals, utilities, finance, insurance and real estate. Hence, industry is included as dummy variable, where companies that are operating in environmentally sensitive industries are accompanied by the value 1, and otherwise this value will equal 0. Socially and environmentally sensitive industries are considered to experience more risks within these areas. Lastly, a dummy variable representing the years (YEAR) in this sample is included following Peters and Romi (2015).
Statistical model
This research conducts an ordinary least squares method on the following statistical model to test the first hypothesis:
(1) 𝑆𝑆𝑃𝑃 = 𝛽𝛽0 + 𝛽𝛽1(𝐵𝐵𝐵𝐵𝐵𝐵) + 𝛽𝛽2(𝐵𝐵𝑃𝑃𝑆𝑆) + 𝛽𝛽3(𝐴𝐴𝑆𝑆𝑆𝑆𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐵𝐵) + 𝛽𝛽4(𝑆𝑆𝑆𝑆𝑆𝑆𝐵𝐵) + 𝛽𝛽5(𝐿𝐿𝐵𝐵𝐵𝐵) + 𝛽𝛽6(𝐴𝐴𝑅𝑅𝐴𝐴) + 𝛽𝛽7(𝑆𝑆𝐴𝐴𝐼𝐼) + 𝛽𝛽8 − 13(𝑌𝑌𝐵𝐵𝐴𝐴𝐴𝐴) + 𝜀𝜀
The second hypothesis, where the interaction term will be added, will be tested according to the following statistical model:
(2) 𝑆𝑆𝑃𝑃 = 𝛽𝛽0 + 𝛽𝛽1(𝐵𝐵𝐵𝐵𝐵𝐵) + 𝛽𝛽2(𝐵𝐵𝑃𝑃𝑆𝑆) + 𝛽𝛽3(𝐴𝐴𝑆𝑆𝑆𝑆𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐵𝐵) + 𝛽𝛽4(𝐵𝐵𝑆𝑆𝐸𝐸) + 𝛽𝛽5 (𝐴𝐴𝑆𝑆𝑆𝑆𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐵𝐵 ∗ 𝐵𝐵𝑆𝑆𝐸𝐸) + 𝛽𝛽6(𝑆𝑆𝑆𝑆𝑆𝑆𝐵𝐵) + 𝛽𝛽7(𝐿𝐿𝐵𝐵𝐵𝐵) + 𝛽𝛽8(𝐴𝐴𝑅𝑅𝐴𝐴) + 𝛽𝛽9(𝑆𝑆𝐴𝐴𝐼𝐼) + 𝛽𝛽10 − 15(𝑌𝑌𝐵𝐵𝐴𝐴𝐴𝐴) + 𝜀𝜀
The third, fourth, fifth and sixth hypotheses will be tested using the following statistical model:
(3) 𝑆𝑆𝑃𝑃 = 𝛽𝛽0 + 𝛽𝛽1(𝐵𝐵𝐵𝐵𝐵𝐵) + 𝛽𝛽2(𝐵𝐵𝑃𝑃𝑆𝑆) + 𝛽𝛽3(𝑃𝑃𝐴𝐴𝑅𝑅𝐵𝐵) + 𝛽𝛽4(𝐵𝐵𝑆𝑆𝐸𝐸4) + 𝛽𝛽5(𝑆𝑆𝐴𝐴𝑅𝑅𝑃𝑃𝐵𝐵) + 𝛽𝛽6(𝐿𝐿𝐵𝐵𝐵𝐵𝐵𝐵𝐿𝐿) + 𝛽𝛽7(𝑆𝑆𝑆𝑆𝑆𝑆𝐵𝐵) + 𝛽𝛽8(𝐿𝐿𝐵𝐵𝐵𝐵) + 𝛽𝛽9(𝐴𝐴𝑅𝑅𝐴𝐴) + 𝛽𝛽10(𝑆𝑆𝐴𝐴𝐼𝐼) + 𝛽𝛽11 − 16(𝑌𝑌𝐵𝐵𝐴𝐴𝐴𝐴) + 𝜀𝜀
Lastly, the remaining hypotheses, relating to the interaction term of CSR disclosure and assurance engagement characteristics, will be tested using the final model:
(4) 𝑆𝑆𝑃𝑃 = 𝛽𝛽0 + 𝛽𝛽1(𝐵𝐵𝐵𝐵𝐵𝐵) + 𝛽𝛽2(𝐵𝐵𝑃𝑃𝑆𝑆) + 𝛽𝛽3(𝑃𝑃𝐴𝐴𝑅𝑅𝐵𝐵) + 𝛽𝛽4(𝐵𝐵𝑆𝑆𝐸𝐸4) + 𝛽𝛽5(𝑆𝑆𝐴𝐴𝑅𝑅𝑃𝑃𝐵𝐵) + 𝛽𝛽6(𝐿𝐿𝐵𝐵𝐵𝐵𝐵𝐵𝐿𝐿) + 𝛽𝛽7(𝐵𝐵𝑆𝑆𝐸𝐸) + 𝛽𝛽8(𝑃𝑃𝐴𝐴𝑅𝑅𝐵𝐵 ∗ 𝐵𝐵𝑆𝑆𝐸𝐸) + 𝛽𝛽9(𝐵𝐵𝑆𝑆𝐸𝐸4 ∗ 𝐵𝐵𝑆𝑆𝐸𝐸) + 𝛽𝛽10(𝑆𝑆𝐴𝐴𝑅𝑅𝑃𝑃𝐵𝐵 ∗ 𝐵𝐵𝑆𝑆𝐸𝐸) + 𝛽𝛽11(𝐿𝐿𝐵𝐵𝐵𝐵𝐵𝐵𝐿𝐿 ∗ 𝐵𝐵𝑆𝑆𝐸𝐸) + 𝛽𝛽12(𝑆𝑆𝑆𝑆𝑆𝑆𝐵𝐵) + 𝛽𝛽13(𝐿𝐿𝐵𝐵𝐵𝐵) + 𝛽𝛽14(𝐴𝐴𝑅𝑅𝐴𝐴) + 𝛽𝛽15(𝑆𝑆𝐴𝐴𝐼𝐼) + 𝛽𝛽16 − 21(𝑌𝑌𝐵𝐵𝐴𝐴𝐴𝐴) + 𝜀𝜀
Provided that ßi are coefficients and ε is the error term. The various variables are further defined in table 2.
14
Table 2: Definition variables
Variable Name Measurement Retrieved from
Dependent variable
Share Price SP Share price three months after fiscal year-end Worldscope
Independent variables
Book value of equity BVE Book value of equity divided by number of
shares outstanding Worldscope
Earnings per share EPS Earnings of a company before extraordinary
items divided by the number of outstanding shares
Worldscope
Assurance ASSURANCE Value 1 for companies with accompanying CSR
assurance statement, 0 otherwise GRI database
CSR disclosure ESG Company’s ESG
performance based on publicly available data (i.e. CSR disclosure) on a scale of 0-100
Asset4 ESG
Provider PROV Value 1 when assurance is provided by formal
accounting firms, 0 otherwise GRI database
Big-4 BIG4 Value 1 when assurance is provided by a Big-4
auditing firm GRI database
Scope SCOPE Value 1 when the assurance engagement
considered the entire sustainability report, 0 otherwise
GRI database
Level LEVEL Value 1 when the level of assurance was
reasonable/high, 0 otherwise GRI database
Control variables
Firm size SIZE Natural logarithm of firm’s total assets at
year-end Worldscope
Leverage LEV Total liabilities divided by total assets at
year-end Worldscope
Return on Assets ROA Net income divided by total assets at year-end
Industry IND Value 1 when operating in an environmentally
sensitive industry, 0 otherwise GRI database
15
Results
Descriptive statistics
An overview of the descriptive statistics can be found in table 3 for the total sample. The descriptive statistics indicate that on average, 47% purchased assurance on their CSR disclosure, meaning 962 companies relative to the entire sample of 2063 companies. The average of scope and level of the acquired assurance are relatively low, 42% and 9% respectively. This might be addressed to the relative costs of CSR assurance, as it is very costly to ensure the entire sustainability report to the level of reasonable/high, firms are more likely to purchase assurance to a smaller extent. Furthermore, on average 82% purchased external assurance from an accounting firm, whereas the percentage of assurance provided by a Big4 auditing firm is 81%. Indicating that when assurance is purchased from an accounting firm, the vast majority is purchased from either KPMG, Deloitte, EY or PwC. Additionally, regarding the control variable industry, the environmentally sensitive and insensitive industries are almost equally represented within this sample, where the mean is 0,44.
Table 3: Descriptive statistics
N Minimum Maximum Mean Std. Deviation
SP 2063 0,02 5614,24 66,8852 265,48432 ASSURANCE 2063 0 1 0,47 0,499 SCOPE 962 0 1 0,42 0,396 LEVEL 962 0 1 0,09 0,200 PROV 962 0 1 0,82 0,486 BIG4 962 0 1 0,81 0,484 ESG 2063 11,37 94,86 65,7323 14,00837 BVE 2063 -14,17 1587,75 28,7939 96,04413 EPS 2063 -84,57 202,10 2,7691 12,30128 IND 2063 0 1 0,44 0,496 SIZE 2063 4,354185 9,384993 7,06414984 0,786062319 LEV 2063 0,00 150,26 22,6548 19,17281 ROA 2063 -57,36 65,96 3,8571 6,64887 Valid N (listwise) 2063
Correlation analysis
Table 4 displays the correlation coefficients between the various variables. Coherent with the expectations that share price is influenced by book value of equity and earnings per share, there is a positive and significant correlation, r=0,758, p=0,00 and r=0,850, p=0,00 respectively. Unfortunately there is no indication of H1, as contrary to the prediction, the correlation coefficient between share price and assurance is negative and insignificant (r=-0,040, p=0,115).
Generally, two independent variables can cause multicollinearity issues when the correlation coefficient exceeds 0.7 (Valentine, 1969). In this case there are multiple variables that exceed this threshold, namely BIG4 and provider (r=0,961, p=0,00), as well as the variables provider and assurance (r=0,842, p=0,00) and BIG4 and assurance (r=0,830, p=0,00). The high correlation is a result of logical reasoning. As already could be seen from the descriptive statistics, the data on the variables provider and Big4 was almost similar. When companies choose to assure their CSR report using a ‘formal’ accountant, this ‘formal’ account is almost always an auditing firm belonging to the Big4. Therefore, the high correlation was expected. Since, the two variables are highly correlated, exceeding the threshold of 0.7, and are almost identical in their measurement, only the variable provider is included in the regression to prevent multicollinearity issues.
16
Thereafter, assurance is measured as a dummy variable, subsequently, various factors related to assurance engagement are measured. Such as, the provider of the assurance. The provider of the assurance will only be included when there actually is an assurance engagement, hence the correlation was expected to be high. Despite the high correlation, it will not cause multicollinearity issues since the characteristics of the assurance engagement will be included in the regression model and the assurance variable will be excluded.
Table 4: Correlation matrix
SP BVE EPS SIZE LEV ROA ASSU ESG IND SCOPE LEVEL PROV BIG4
SP BVE ,758** EPS ,850** ,666** SIZE -,057** ,059** -0,016 LEV -,063** -,078** -,087** 0,016 ROA ,118** 0,017 ,191** -,180** 0,037 ASSURANCE -0,040 -0,033 -0,018 ,296** ,099** -,058** ESG -0,012 -0,012 0,016 ,486** 0,029 0,003 ,354** IND -,098** -0,040 -,081** ,373** 0,001 -,222** ,144** ,075** SCOPE -0,016 -0,016 0,004 ,092** ,075** -0,019 ,526** ,165** ,075** LEVEL -0,003 -0,007 0,003 ,070** ,046* -0,004 ,223** ,088** 0,036 ,204** PROV -0,031 -0,018 -0,010 ,294** ,071** -,074** ,842** ,293** ,161** ,441** ,155** BIG4 -0,030 -0,021 -0,013 ,296** ,061** -,073** ,830** ,280** ,148** ,442** ,134** ,961** *p<.05 (two-tailed), **p<.01 (two-tailed)
Regression analysis
To test the hypothesis, a regression analysis using the ordinary least squares method is performed, homoscedasticity is assumed.
Table 5 provides an overview of the results of the regression analysis. The first model, i.e. model I, only includes the control variables, that explains approximately 2,0% (adj. r²=0,020) of the variance in share price and is significant (F=5,713 and p=0,000). Consistent with academic literature, leverage (ß=-0,920, p=0.000), ROA (ß=4,107, p=0,000) and industry (ß=-61,607, p=0.000) have a statistically significant influence on the share price (Cho et al., 2014; Simnett et al., 2009). Contrastingly, no significant relationship has been found between share price and size. The year dummies are included in the regression model and do not have a statistically significant influence on share price. The year dummies do not have a statistically significant influence.
Model II present the test statistics for the regression performed to examine hypothesis 1. From the hypothesis development, in combination with legitimacy theory and signalling theory, CSRA is expected to be positively related to share price. Unfortunately, no support has been found for this prediction, as assurance is not statistically significant (ß=1,101, p=0,845). In accordance with the expectations, book value and earnings per share are positive and statistically significant. Model II explains around 79,3% of the variation in market value (adj. R²=0,794) and is statistically significant (F=660,727, p=0,000). Contrary to the prediction, industry is not statistically significant related to share price. Size, leverage and ROA are however statistically significant.
17
Model III adds the interaction term of CSRA and quality of CSR disclosure i.e. ESG score. Based on Cho et al. (2014) the interaction term is expected to be positively associated with share price. Hypothesis 2 therefore states that greater CSR disclosure leads to a higher share price with a mediating effect of CSRA. This hypothesis is not supported as the interaction term is insignificant (ß=0,568, p=0,174). Again, as expected book value and earnings per share are positively related to share price and statistically significant. Model III explains in similar amounts the variation in share price (adj. R² 0,794) and is significant (F=567,541, p=0,000).
The results of the regression related to the specific characteristics of the assurance engagement are presented in model IV. Contrary to the predictions, the various characteristics do not have a statistically significant influence on share price. Level and provider have a positive sign, as expected, however scope has a negative sign contrary to what was expected. As mentioned, none of the variables, scope (ß=-10,093, p=0,113) level (ß=3,507, p=0,748) and provider (ß=1,743, p=0,831) are statistically significant. Therefore, the third, fourth and fifth/sixth hypothesis are not supported. The explanatory power of the model decreased compared to model II and III (adj. R²=0,748). Lastly, model V adds the interaction term again to perform the regression related to the seventh, eighth, ninth, and tenth hypothesis. As can be seen, the interactions terms are insignificant. The interaction term provider*ESG (ß=0,069, p=0,921) is positive, whereas the interaction terms scope*ESG 0,145, p=0,797) and level*ESG (ß=-0,642, p=0,921) are negative contradicting the expectations. Hence, the respective hypotheses are not supported by these results. The explanatory power of the fifth regression model is equal to model IV (adj. R²=0,748). To control for multicollinearity, the highest variation inflation factor (VIF) is provided for every model. There are no indications of multicollinearity when the VIF’s stay below the threshold of 10 (Kennedy ,1992). Interactions terms are associated with high VIF’s, contrary to normal regressions do interactions terms not cause multicollinearity issues, in general. Therefore, in model III and model V the highest VIF is provided as well as the second highest VIF where interactions terms were not taken into account. As can be seen the VIF’s of the normal variables stay well below the threshold of 10.
18
Table 5: Regression results
Variables Model I Model II Model III Model IV Model V
Coeff. t-statis. Coeff. t-statis. Coeff. t-statis. Coeff. t-statis. Coeff. t-statis.
Independent variables BVE 0,968 25,670** 0,973 25,756** 0,391 4,493** 0,381 4,374** EPS 13,394 44,756** 13,367 44,648** 19,972 29,911** 19,908 29,762** ASSURANCE 1,101 0,195 -39,787 -1,387 ESG 0,197 0,709 0,640 0,945 ASS*ESG 0,568 1,359 SCOPE -10,093 -1,585 -0,087 -0,002 LEVEL 3,507 0,321 49,130 0,640 PROV 1,743 0,213 -1,932 -0,039 BIG4 SCOPE*ESG -0,145 -0,257 LEVEL*ESG -0,642 -0,606 PROV*ESG 0,069 0,100 BIG4*ESG Control variables SIZE -3,503 -0,438 -22,466 -5,881** -26,125 -6,092** -19,513 -4,421** -24,120 -4,895** LEV -0,920 -3,040** 0,277 1,974* 0,283 2,018** 0,290 1,619 0,303 1,670 ROA 4,107 4,567** -0,879 -2,072* -0,942 -2,215* -3,756 -5,966** -3,876 -6,114** IND -38,019 -2,979** -7,361 -1,252 -5,854 -0,988 0,535 0,080 1,837 0,273
Year controls Yes Yes Yes Yes Yes
Observations 2063 2063 2063 962 962 F-test 5,713** 660,727** 567,541** 204.865** 159,727** Adj. R² 0,020 0,793 0,794 0,748 0,748 Highest VIF 1,541 1,919 32,733 (1.923) 4,093 50,528(4,387) **p<0.01 (one-tailed), *p<0.05 (one-tailed)
19
Conclusion
This study examines the mediating effect of CSR assurance on the relationship between CSR disclosure and firm value. Due to the inconsistent findings in prior studies regarding the value relevance of CSR disclosure, assurance was taken into account to calculate its influence. Since CSR reporting is destined to become a key part of the modern accounting framework, empirical evidence regarding the value relevance is influential (Tschopp & Huefner, 2015). Besides the presence of assurance, certain characteristics of the assurance engagement are taken into account as well. These characteristics include the level and scope of the assurance engagement, and the provider of the assurance statement. To examine the role of CSR assurance on firm value, data is derived from European listed firms during the period 2012-2017. A total of 2063 firm-years is used in the regression analysis. The sample is diverse, as companies differ in terms of size, industry, country and firm performance and is therefore representable. To measure the dependent variable, data is retrieved from Worldscope whereas data on the independent variables is retrieved from Worldscope, Asset4 ESG as well as the GRI database. The findings indicate no support for the specific predictions that have been made in this study. Concluding that CSR assurance does not seem to have an effect on firm value.
Findings
The first hypothesis states that CSR assurance has a positive influence on firm value. This hypothesis is based on legitimacy theory as well as signalling theory. According to signalling theory, companies disclose information about their CSR practices to signal to the market that their CSR practices are of ‘high-quality’. Purchasing assurance magnifies this effect, as it signals that the respective company is willing to purchase costly assurance (Servaes & Tamayo, 2013). Companies will only engage in costly assurance when their CSR report is of high-quality. Legitimacy on the other hand states that companies interact with society based on a social contract. In order to be viable, the company should be perceived as legit. When a CSR report is seen as a self-promotional document it leads to a legitimacy gap. External assurance can close this legitimacy gap and therefore enhance firm value (Sethi et al., 2017). Contrary to the expectation, the first hypothesis is not supported by the regression results. The second hypothesis is not supported by the results as well, where the interaction term between CSR disclosure and assurance is taken into account.
A possible explanation for this lack of significance is that share price is influenced by many factors. Although assurance might have an overall impact on share price, the influence is too small to be substantial and significant. Moreover, as suggested by Cho et al. (2014), there is a lack of awareness regarding CSR and CSR assurance, which might explain the insignificant results. The expectation is that when the market is more mature, the beneficial effects will increase. In addition to this, as Friedman (1962) argued, the sole responsibility of an organization is to increase shareholders’ wealth, and allocate resources accordingly. Different perspectives on the responsibilities of an organization have led to inconclusive results regarding CSR. As purchasing external assurance is seen as a waste of resources, the insignificant results can be explained. Moreover, another explanation for the lack of significance can be addressed to the ideas from DiMaggio and Powell (1983). They argue that through the process of institutional isomorphism, firms become similar in their practices. A specific form of institutional isomorphism is mimetic isomorphism. Mimetic isomorphism relates to a situation wherein a firm follows that behaviour of other firm, regardless of the behaviour’s effectiveness. When mimetic isomorphism is translated to CSRA, it could be that firms purchase external assurance, not because CSRA has proven to be effective, but to imitate other firms within their industry (Peters & Romi, 2015).
The fourth regression model dealt with the hypotheses relating to specific assurance engagement characteristics. The results hereof will be discussed in pairs, as was the discussion of hypotheses development. The third and fourth hypotheses predicted that when assurance is provided by an accounting firm, this would be valued more by investors, and thus be reflected in firm value to a greater extent. When the assurance statement was provided by a Big4 accounting firm this effect was expected to be magnified. As could be concluded from the correlation
20
matrix, and the descriptive statistics, the variables were almost identical, therefore only the second hypothesis is tested, whereas the results apply for the third hypothesis as well. The results however do not support the third/fourth hypothesis as well as the fifth/sixth hypothesis in which the interaction variables are taken into account. Indicating that the type of assurance provider for the CSR assurance engagement does not influence firm value, and this relationship is not magnified when CSR disclosure is included.
According to Sethi et al. (2017) the majority of non-financial assurance providers, is provided by formal accounting firms, in specific Big 4. As Big4 auditing firms own are relatively large percentage of the market, it is not uncommon that a Big-4 auditing firms is auditing the financial statements as well as the non-financial information. This might pose a threat to the independence of the engagement (Sethi et al., 2017). When conflicts of interest are acknowledged by the financial market, the reliability of the assurance engagement will decrease, and thus the value thereof. In addition to this, formal accounting firms are in possession of a high level of competence, resources and human capital related to auditing financial statements, however this high level of competence is not achieved regarding non-financial data (yet) (Sethi, et al., 2017). Besides, when companies acquire assurance by the same audit firms who carries out the financial statement verification there might be conflicting interests. Namely, a firm’s financial audit constitutes a relatively large portion of the fee received by the auditing firm compared to the fee received for the CSRA. Auditing firms may provide companies with an assurance statement, although the company did not have the right to receive one. The auditing firm may value the company more as a client, because it represents a large portion of their income. In addition to this, when the auditor is auditing the financial statement of a company for multiple years already, there is a strong relationship build on trust. Hence Ballou et al. (2006) report that those auditors give high credence to the information disclosed in the CSR report. Due to the conflicts of interests and the possible incompetence of formal auditing firms, the reliability of the engagement is reduced, which might explain the insignificant results.
The last hypotheses state that the higher the scope and level of the assurance engagement, the greater the beneficial effects of assurance on firm value. In combination with high-quality CSR disclosure, the effects are expected to multiplied. As previously mentioned, these predictions were not statistically significant. Therefore, from these results it cannot be concluded that a higher level and greater scope of the assurance engagement lead to higher firm value. As already mentioned during the hypothesis generation, the lack of generally accepted reporting criteria imposes challenges for auditors (Hodge et al., 2009). It is more difficult to provide a statement based on reasonable/high reliability. In addition to this, auditors rely on documents provided by the organizations due to the lack of guidelines, leading to a decreased extent of independent verification (Zorio et al., 2013). When this causes auditors to issue only statements based on limited reliability, this reduces the expected positive impact of CSRA on firm value. This might explain the insignificant results.
Theoretical implications
This study contributes to the growing academic literature about CSR and more specifically CSR assurance (Hodge et al., 2009; Pflugrath et al., 2011). Specifically, this study aims to close an existing gap in the literature, by examining the relationship between CSR assurance and firm value for European listed firms over the period of 2012 to 2017. Where previous research mostly focused on US firms, this study takes European firms into account as research suggested that the assurance rates and CSR awareness is highest in the European region (Ackers, 2009). Additionally, this study answered the call by Cho et al. (2014) to test assurance over multiple years instead of taken one year into account. Moreover, Cohen and Simnett (2015) include recommendations for future research, stating that the effect of CSRA on changes in share price should be examined. This study addresses this question by examining the effects of CSRA on firm value in a region that has not been examined yet in this specific field. In addition, this study enriches the existing theory by not only taking assurance into account, but several characteristics of the assurance engagement as well, namely the CSRA provider and level and scope of the assurance engagement. The explanatory power of these characteristics have not been examined yet in empirical evidence. Thus, this study is closing a gap in the existing literature.
21
Practical implications
Likewise, practical implications can be derived from this study. The results of the study are relevant for companies disclosing CSR information, with regard to their assurance decision-making. The decision for assurance is explained on the basis of legitimacy and signalling theory, arguing that assurance will lead to increased firm value. As the results show this effect is not statistically significant. Therefore, companies could decide not to purchase assurance, as it is a costly decision and the effects are negligible. Additionally, the more extensive the scope of the assurance, the higher the costs. When the auditing firm provides an assurance statement relating to the entire report, the amount of time and resources spent will be higher than when only specific sections are considered. The same conclusion can be drawn relating to the level of the assurance engagement. Assurance of non-financial information is subjective and time-consuming, therefore limited/moderate assurance is more common over high/reasonable assurance. The results show that the effect of scope and level on firm value are not statistically significant. Therefore, from this results there cannot be concluded that assurance of the entire report and at a high/reasonable level is valued over the other assurance engagements. An acquiring firm can decide based on these results what the scope and level of the assurance engagement should be. Firms purchasing assurance to comply with public pressures for example, should keep this in mind, because it could be cost-saving to limit the amount of resources and time spent by the auditing firm on the engagement.
The results hold implications for policy-makers as well. Previous research namely already indicated that there are positive effects associated with voluntary CSRA. Amongst others, CSRA increases the perceived reliability (Hodge et al., 2009; Pflugrath et al., 2011) and it decreases information asymmetry and thus agency costs (Blackwell et al., 1998; Carey et al., 2000; Chow, 1982). However, as indicated by the results of this study, CSRA does not seem to have an influence on firm value. As companies can only justify the costs allocated to assurance when the benefits outweigh the costs, this decreases their incentive to engage in CSRA. Therefore, policy-makers should take mandatory assurance into consideration. Reliable CSR disclosure is demand by the public, and assurance closes the reliability gap and is therefore relevant. When the incentive for companies to engage in CSR is too low, no assurance would be purchased, and the reliability gap would not be closed. If this trend becomes visible, it should be investigated whether CSRA should and could be mandated by law.
Limitations and future research
Limitations are present in this study, and therefore the results should be interpreted with caution. Firstly, there could be a sample bias, accompanied with limited generalizability of the results. After matching the data from worldscope and the GRI database, the sample was halved. Although the number of firm-years included is still relatively high, this limits the generalizability of the results. Furthermore, the database of GRI is dependent upon companies openly disclosing their CSR assurance information. As mentioned before, after matching only half of the sampling companies were present in both datasets. It could be that there is a sampling-bias regarding companies who voluntarily disclosure information about their assurance engagement. Moreover, GRI developed their own guidelines related to assurance, therefore the sample could be biased towards companies that use this guideline to report on their CSR practices. Future research could conduct a similar study, except use hand-collected data instead, to make sure that the results are free of this bias. Thereafter, there are several control variables included in this study, size, leverage, return on assets and industry based on previous studies. However, the amount of control variables is limited, therefore a possibility arises that the problem is misspecified, as some relevant controls are missing. Although by following published articles regarding the control variables, this possibility is reduced. Lastly, during the regression the VIF’s relating to the interaction terms significantly exceeded the threshold of 10 that is set by Kennedy (1992). Although high VIF’s are common when interactions terms are used and they do not necessarily indicate multicollinearity issues, it does hamper the quality of the respective regressions.
22
In order to expand the knowledge regarding CSRA, more research is required. Kang et al. (2010) state that not engaging in CSR could reduce firm value in the long-term, as it damages reputational capital. Therefore, instead of focussing on the positive effects of engaging in CSRA, the negative effect of not engaging in CSRA could be studied. Secondly, Zhou, Simnett and Green (2017) study the effects of integrated reporting (IR) specifically, on the financial market. Their findings conclude that IR does matter to the capital market. The GRI database also provides the user with an overview of which CSR reports are integrated reports, and which reports are not. Hence, future research could use this data to see whether integrated reports in combination with assurance have a statistically significant effect on firm value. Lastly, Vormedal and Ruud (2009) discuss the influences of different regulations on CSR disclosure, namely mandatory versus voluntary. Future research should take this distinction into account when evaluating whether assurance enhances firm value or not. The implications hereof are relevant to policy-makers as well.
