Quality of Sustainability Disclosure and Information Asymmetry Master’s thesis MSc A&C Accountancy 2017/2018 University of Groningen

(1)

Quality of Sustainability Disclosure

and Information Asymmetry

Master’s thesis

MSc A&C Accountancy 2017/2018

University of Groningen

(2)

Page 2

Personal details

Name Jeroen W. Brink

Student number s2713004

Email j.w.brink.2@student.rug.nl University University of Groningen Faculty Economics and Business

Master Accountancy & Controlling, track Accountancy Supervisor RuG Dr. N. (Nazim) Hussain, PhD

Date 25/06/18

(3)

Page 3

Abstract

This research investigates the relationship between the quality of sustainability disclosure and the level of information asymmetry. Two interacting variables were tested on this relation, namely the presence of an external sustainability auditor and the presence of a CSR sustainability committee. Robust results were found for the relation between the quality of sustainability disclosure and information asymmetry. Indicating that higher (lower) quality sustainability disclosure results in a lower (higher) level of information asymmetry. Second, an additional analysis found a negative association between the presence of a CSR sustainability committee and information asymmetry. The most important implication of this research for practice is that organisations should increase the quality of sustainability disclosure and implement a CSR sustainability committee to reduce the level of information asymmetry.

Keywords: accounting, assurance, sustainability reporting, information symmetry, quality, auditor, CSR sustainability committee & auditing.

(4)

Page 4

I. Introduction

pril 2010, at BP, the Deepwater Horizon oil spill indicated the shortcomings of their sustainability disclosure; April 2016, H&M is accused of misleading their customers with all its talk of sustainability; September 2016, Exxon Mobil is accused of misleading the public on climate change; October 2017, Amazon is accused of being opaque when discussing its environmental practices (Lewis, 2011; Bain, 2016; Barret & Philips, 2016; Fingas, 2017). These are only a few examples of organisations that disclosed sustainability in a misleading way. Together with the growing interest of society in sustainability disclosure (Arjaliès & Mundy, 2013), these developments indicate the importance of high quality sustainability disclosure.

Past research studied the effect of Corporate Social Responsibility (herafter CSR) reporting on information asymmetry. Cormier et al., 2011; Leuz and Verrecchia, 2000; Healy et al., 1999; Dhaliwal et al., 2012 all showed that the issuance of sustainability disclosure lead to significantly lower analyst forecast errors, however they especially focused on the presence or quantity of sustainability disclosure, which does not imply that the quality of sustainability disclosure is sufficient. Therefore, this research takes a more in-depth approach, by looking at the quality of sustainability disclosure and not only looking at whether CSR reports are issued.

The present study is concerned with the impact of the quality of sustainability disclosure on information asymmetry, and aims to give organisations advice on whether they should attract an external sustainability auditor and implement a CSR sustainability committee to enhance the quality of sustainability disclosure, which would lead to reduced information asymmetry. The present study found a negative association between the quality of sustainability disclosure and information asymmetry. Furthermore an additional analysis showed that the presence of a CSR sustainability committee reduces information asymmetry.

In most of the countries, organisations are not required to report on CSR (Cheng & Courtenay, 2006). This is remarkable because CSR, and more specifically sustainability disclosure, provides valuable information for investors (Cohen et al., 2011) that could increase the accuracy of disclosed information (Glaum et al., 2013). According to the Global Reporting Initiative (GRI), an independent international organisation that is a pioneer in the field of sustainability disclosure, sustainability reports are a way of both internally and externally communicating sustainability performance and are part of CSR reporting (GRI, 2017). Organisations could benefit from incorporating high quality voluntary disclosure

A

(6)

Page 6 through, for example, lower cost of capital, caused by reduced information asymmetry (Lundholm & Van Winkle, 2006; Dhaliwal et al., 2011; Healy & Palepu, 2001). So organisations could create value by disclosing voluntary information.

With regard to voluntary disclosure, information asymmetry is more relevant in comparison to required financial reporting, because of the possibility to provide selective information. Sustainability disclosure can be seen, in this sense, as an example of managerial discretion. If sustainability disclosure is applied neutrally, it can provide stakeholders with relevant information. However, organisations might also use the discretion opportunistically, by being selective with regard to the information disclosed. So, organisations should align their information disclosure tendencies with the interests of their stakeholders (Cheng & Courtenay, 2006) and follow a stakeholder-oriented approach. However, Li and Yang (2016) stated: “A firm’s voluntary disclosure decision is an equilibrium outcome of its underlying incentives and disincentives for disclosure” (p. 935). The decision to disclose certain information voluntarily can be seen as a strategic choice for which a decision needs to be made about the quantity and quality of voluntary disclosure (Li, 2010). A possible strategical choice could be to disclose more on either good or bad news. According to Verrecchia (1983), organisations will only disclose voluntary information if there is good news, because organisations are only willing to incur the costs of voluntary disclosure if the benefits outweigh the costs. Contrarily, however, Lang (2013) concludes that organisations often disclose more bad news than good news in voluntary disclosure in order to decrease analyst’s expectations to create lower earnings forecasts that are easier to achieve. Voluntary disclosure could thus be influenced by the strategical decisions made by the organisation, either to increase or decrease the perceived performance of an organisation.

Information asymmetry provides organisations with incentives to voluntary disclose CSR activities, but it does not create sufficient incentives to report in an honest way (Bagnoli & Watts, 2017). More available information for stakeholders should lower information asymmetry (Lambert et al., 2012). However, more information does not imply that the quality of the information is sufficient, because according to Van Caneghem & Van Campenhout (2012) both quantity and quality should be sufficient. Brown and Hillegeist (2007) for example found that there is a negative association between the quality of disclosure and the level of information asymmetry. So, high (low) quality disclosure will lead to a lower (higher) level of information asymmetry. Therefore, it is expected that sustainability disclosure will lead to a higher level of information asymmetry.

(7)

Page 7 By attracting an external sustainability auditor, the risk of organisations presenting selective or unjust information towards their stakeholders could be limited (Sun et al., 2017). Therefore, this research will investigate whether the presence of an external sustainability auditor could have an interacting effect on the effect of the quality of sustainability disclosure on information asymmetry.

Furthermore, this research will investigate the interacting effect of the presence of a CSR sustainability committee on the effect of the quality of sustainability disclosure on information asymmetry. A CSR sustainability committee has a critical coordinating function, accounting for the CSR strategy and performance of an organisation (Shah et al., 2014), so they could ensure the quality of the disclosed sustainability information. Second, the presence of a CSR sustainability committee shows the commitment of an organisation towards sustainability (Ricart, Rodríguez & Sánchez, 2005); higher commitment could indicate organisations trying to meet informational needs of their stakeholders with regard to sustainability. Therefore, it is expected that the presence of a CSR sustainability committee will have an interacting effect on the relation between quality of sustainability disclosure and information asymmetry. In sum, this research aims to answer the following question:

Does using an external sustainability auditor and a CSR sustainability committee moderate the association between sustainability disclosure quality and information asymmetry?

For organisations, this research will be highly relevant, because their focus is increasingly shifting from financial disclosure towards voluntary disclosure (Kolk, 2003). Research conducted by KPMG (2013) concluded that almost all of the world’s top 250 largest companies disclosed on CSR. Reporting on CSR is becoming the norm and at least 62% of the companies issue a CSR report. However, according to Adams and Evans (2004) this did not increase the accuracy, credibility and trustworthiness of the information.

Nowadays, there is an increasing amount of research in the field of CSR (Fifka, 2013). According to the literature review of Fifka (2013), research is especially done in highly developed western industrialized countries. The present research, however, will be conducted globally. More importantly, the main contribution of this research is that it relies on modern measures. The quality of sustainability disclosure will be measured by using the new ESG scoring methodology developed by ESG ASSET4 that was released in 2017. This ESG score takes controversies into account, which can serve as an indicator for the quality of sustainability disclosure. Another advantage of ESG ASSET4 is that the data is updated for previous years and recalculated regularly. To the researcher’s knowledge, this study is the

(8)

Page 8 first study that uses this new methodology developed by Thomson Reuters’ in 2017 to determine the quality of sustainability reporting.

In the following section, the theoretical background is discussed. Thereafter, suitable literature is reviewed, and the hypotheses are developed. In Chapter 3, the research method that was used is explained. After the presentation of the empirical results in Chapter 5, these outcomes are discussed, and finally the conclusions and limitations of this research are presented.

(9)

Page 9

II. Theoretical Background and Literature Review

In this chapter the theories and literature that form the basis of this research are discussed. First, the theory behind voluntary sustainability disclosure is discussed and thereafter two theoretically grounded approaches with regard to sustainability disclosure quality are discussed. This chapter ends with the theoretical background and hypotheses concerning the presence of an external sustainability auditor and a CSR sustainability committee.

Voluntary Sustainability Disclosure

Voluntary sustainability disclosure is concerned with meeting the informational needs of stakeholders with regard to sustainability, with the goal of reducing information asymmetry and agency conflicts between the managers and stakeholder (Healy and Palepu, 2001).

The articles of Dye (1985), Verrecchia (1983) and Lang and Lundholm (1993) form the basis for the voluntary disclosure theory. These articles discussed disclosure as a way of communicating sustainability information towards stakeholders. Dye (1985) suggests that if disclosure of information reveals proprietary information, it should not be disclosed. Verrecchia (1983) concludes that organisations will only disclose voluntary information if there is good news, because there are costs concerned with disclosing voluntary information. Verrecchia (1983) also concludes that only organisations with a large amount of good news are willing to incur these costs. According to Lang and Lundholm (1993), organisations with a performance above a certain threshold disclose on sustainability whereas organisations performing below a certain threshold do not disclose their sustainability performance. The reason for this is that organisations with high performance on sustainability want to show off that they are performing better on sustainability, whereas organisations performing below a certain threshold do not want to show their shortages (Clarkson et al., 2008). Poor performance on sustainability could thus serve as an incentive for disclosing sustainability information in a selective manner or to not disclose at all.

On the topic of voluntary disclosure, and more specifically sustainability disclosure, there is a growing amount of research (Fifka, 2013). It is important to note that there are multiple definitions of sustainability. The most commonly used definition of sustainability is developed by the World Commission on Environment and Development (WCED) (Sneddon, Howarth & Norgaard, 2006). According to the WCED, sustainability is concerned with meeting “[…] the needs of the present without compromising the ability of future generations

(10)

Page 10 to meet their own needs” (p. 54).

There are also multiple concepts with regard to sustainability disclosure (Van Marrewijk, 2003). Elkington coined the Triple Bottom Line (1997). The Triple Bottom Line is a commonly used concept that distinguishes between the economic, social and environmental dimensions, a concept which is also known as the People, Planet and Profit concept. According to Elkington (1997), organisations use sustainability reporting as a public relation vehicle. This research will partly follow the Triple Bottom Line approach to determine the quality of sustainability disclosure and will look at the social, governance and environmental disclosure scores of organisations. These aspects of sustainability are in line with the main topics of the widely accepted GRI guidelines. The economic dimension of the Triple Bottom Line is omitted and replaced by the governance score, because this research is especially focused on voluntary disclosure. The economic dimension is characterised by a higher degree of mandatory disclosure, therefore the economic dimension is less relevant for this research.

Quality of Sustainability Disclosure and Information Asymmetry

In general, organizations that disclose on sustainability could follow two theoretically grounded approaches, which could potentially explain the difference in the level of information asymmetry. The first one is concerned with organisations trying to disclose information that is requested by their stakeholders/community (An et al., 2011), which is in line with the legitimacy theory and the stakeholder theory by Freeman (1984). This will be further referred to as the positive view. The positive view is concerned with what ideally should be the aim of sustainability disclosure, from the perspective of the stakeholders.

The second approach that organisations could use is disclosing information that is especially beneficial for themselves (An et al., 2011). This approach will be further referred to as the negative view. The quality of sustainability disclosure will be lower for stakeholders under the negative view, because according to this view, organisations have the incentive to selectively disclose sustainability information. The negative view will be discussed more detailed, because this research is especially concerned with the point of view that organisations use sustainability disclosure for their own benefit. First, the agency and signalling theory are discussed because these theories provide the theoretical grounding for this view.

(11)

Page 11 The agency theory by Jensen and Meckling (1976) is a theory that is often used in the field of reporting (An et al., 2011). The agency theory is concerned with the principal-agent relationship, which is accompanied by misalignment of goals and desires. Jensen and Meckling (1976) defined the principle-agent relationships as follows: “[…] a contract under which one or more persons (the principal) engage another person (the agent) to perform some service on their behalf which involves delegating some decision making authority to the agent” (p.308). Based on the agency theory, organisations can reduce market friction by increasing corporate disclosure and especially by increasing sustainability disclosure (Healy & Palepu, 2001).

Several other researchers showed that sustainability disclosure reduces information asymmetry, by increasing information accuracy, which will lead to reduced agency costs (Leuz & Verrecchia, 2000; Cormier et al., 2011; Healy et al., 1999). Following the negative view, it is expected that lower (higher) quality sustainability disclosure will lead to a higher (lower) level of information asymmetry.

Another theory in line with the negative view is the signalling theory. The signalling theory (Spence, 1973) is also concerned with information asymmetry between two or more parties. According to this theory, information asymmetry between the organisation and their stakeholders will be lower if an organisation has an information advantage whereby it can signal information to other interested parties (An et al., 2011). An organisation that has an information advantage over, for example, its stakeholders is expected to signal favourable information in order to indicate the higher quality of the organisation and its disclosure (An et al., 2011). According to the signalling theory, organisations try to reduce the possibility of an opportunity loss, which could occur when a higher quality organisation is not perceived as such (An et al., 2011). Therefore, organisations will signal their high quality. According to research conducted by Xiao et al. (2004) and Watson et al. (2002), voluntary disclosure is the most effective way of increasing the perceived quality of an organisation. So following the signalling theory, organisations could use sustainability disclosure to signal their high quality, which could lead to selective disclosure in favour of the organisation, causing an increase in information asymmetry.

One of the main reasons for the present research is the perspective that organisations only disclose information that is in their own interest, which leads to selective disclose of sustainability information that is not beneficial for organisations themselves (Lang, 2013; Verrecchia, 1983; Bagnoli & Watts, 2017). It is expected that this selectiveness leads to a higher level of information asymmetry, because of a lower quality of sustainability disclosure.

(12)

Page 12 The relation between the quality of general disclosure and information asymmetry has already been supported by previous studies (Cormier et al., 2011; Leuz and Verrecchia, 2000; Healy et al., 1999; Dhaliwal et al., 2012). All studies concluded that if an organisation discloses on sustainability, there will be lower information asymmetry. However, this research will not only look at the presence of disclosure, but will also look at the quality of sustainability disclosure. According to Brown and Hillegeist (2007), there is a negative association between the quality of disclosure and the level of information asymmetry. So high (low) quality disclosure should lead to a lower (higher) level of information asymmetry. It is currently unknown whether a higher quality of sustainability disclosure will lead to lower levels of information asymmetry. Based on the research of Brown and Hillegeist (2007) it is expected that a lower quality of sustainability disclosure will lead to higher information asymmetry.

Determining the quality of sustainability disclosure is difficult. Information disclosed on sustainability is organisation-specific; there are standards like the GRI guidelines that are universally applicable. However, these standards also include topics that are irrelevant for certain types of organisation in certain types of industries. Therefore, this research distinguishes between certain industry types and will use the ESG score as indicators for the quality of the sustainability disclosure.

Furthermore, this research will test the components of the quality of sustainability disclosure (environmental, social and governance disclosure) separately on information asymmetry. This will be done to determine whether one of the components is stronger or weaker related to information asymmetry. Through this methodology, this research could provide a more in-depth understanding of the effect of these components on the amount of information asymmetry. This research follows the Triple Bottom Line coined by Elkington (1997), which consists of the economic, social and environmental dimensions. The framework by Elkington is economically orientated. Therefore, this research partly follows the Triple Bottom Line by replacing the economic dimension by the Corporate Governance dimension, which is in line with the ESG component scores.

The following hypotheses are proposed:

Hypothesis 1: The quality of sustainability disclosure is negatively associated with

information asymmetry.

Hypothesis 1a: The quality of environmental disclosure is negatively associated with

(13)

Page 13 Hypothesis 1b: The quality of social disclosure is negatively associated with

Hypothesis 1c: The quality of governance disclosure is negatively associated with

External Sustainability Auditor

The problem of voluntary disclosure of sustainability information is the aim of the management when making certain voluntary disclosure decisions. Organisations could make sustainability disclosure decisions in their own favour and users of information disclosed on sustainability may perceive these decisions as strategical, according to Coram et al. (2009). This problem makes this research relevant for organisations to determine whether organisations should attract an external sustainability auditor. Previous research concluded that external assurance reduces the amount of information asymmetry (Healy and Palepu, 2001; Pittman and Fortin, 2004).

There are multiple types of assurance providers; some examples are audit firms, management systems and several types of consultants (Deegan et al., 2006). This research will focus specifically on auditors as assurance providers, because according to Perego (2009) audit firms have greater experience and have a competitive advantage compared to other assurance providers. Furthermore, auditors are more objective and are more independent compared to assurance providers that are not auditors (Knechel et al., 2006). Therefore, it is expected that external assurance provided by auditors will be most effective in reducing information asymmetry. Furthermore, by attracting an external sustainability auditor, the stakeholders of an organisation will less likely perceive disclosure on sustainability as a set of strategical decisions in favour of the organisation. Therefore, this research will investigate whether the presence of an external sustainability auditor will moderate the effect of the quality of sustainability disclosure on information asymmetry, and will lead to lower information asymmetry. The following hypothesis is proposed:

Hypothesis 2: The presence of an external sustainability auditor enhances the

negative association between quality of sustainability disclosure and information asymmetry.

(14)

Page 14

Internal CSR Sustainability Committee

The main issue of voluntary disclosure of sustainability information is the aim of the management when making certain voluntary disclosure decisions. According to Coram et al. (2009) users may perceive these voluntary disclosure decisions as strategical decisions. This problem makes this research relevant for organisations to determine whether they should implement a CSR sustainability committee. By implementing a CSR sustainability committee, the users of the disclosed information will less likely perceive disclosure on sustainability as a set of strategical decisions. According to Shah et al. (2014) an independent CSR committee serves a critical coordinating function for organisations and enables organisations to transform their CSR strategy into a competitive advantage. Currently, there is no research on whether the implementation of an internal sustainability committee could influence the effect of the quality of sustainability disclosure on information asymmetry.

The main function of a CSR committee is aligning the business priorities of an organisation with the sustainability priorities of the organisation (Shah et al., 2014). According to Mackenzie (2007) CSR committees play an important role in advising the management of an organisation on the formulation of the CSR strategy and on sustainability disclosure. The presence of a CSR committee could indicate a higher quality organisation because a CSR sustainability committee has a critical coordinating function, which is accountable for the CSR strategy and CSR performance of an organisation (Shah et al., 2014). Therefore, the following hypothesis is proposed:

Hypothesis 3: The presence of an internal sustainability committee enhances the

negative association between quality of sustainability disclosure and information asymmetry.

(15)

Page 15

III. Research Method

In this chapter the research method is discussed. First, the sample design is discussed and thereafter the research design is presented. This chapter ends with presenting the analytical models that were used to test the hypotheses.

Sample Design

The original sample included 45.135 firm-year observations with a time span of 7 years, ranging from 2009 to 2015. The initial reason for this time period, was to capture recent developments in the field of sustainability reporting. Initially, the data from 2016 and 2017 was also going to be used. However there was not sufficient data available for the years 2016 and 2017. This was especially caused by the limited availability of data for measuring information asymmetry. Therefore these years were not included in the analyses. Another potential issue regarding the data was the financial crisis of 2007 and 2008. The financial crisis could bias the results, therefore firm-year observations starting from 2009 are used.

In total, 45.135 firm-year observations were obtained from the Thomson Reuters Datastream and the I/B/E/S database. First, the observations were checked for duplicates and missing values in the descriptions. This reduced the amount of firm-year observations by 1.512. Second, all firm-year observations with no sustainability disclosure were deleted from the sample, causing a decrease in firm-year observations to 18.561. Further, 25 firm-year observations were excluded because of missing ESG score values.

Another important criterion for the observations that are suitable for analysis is the presence of information on analyst forecast accuracy from the I/B/E/S database, so as to determine the level of information asymmetry. The data from the Thomson Reuters Datastream was linked to the I/B/E/S database, which led to a significant drop in the amount of observations; this information was only available for 3.068 firm-year observations with a sustainability report. Table 1 shows the total amount of firm-year observations for all hypotheses. The data available for Hypothesis 2 was lower compared to the other hypothesises, because not all observations contained information on whether an external sustainability auditor was attracted.

(16)

Page 16 The dataset includes, after removing the above mentioned observations, a global set of organisations. The reason for focusing on a global set of firms is because previous research is mostly done on a national level (Cormier et al., 2005; Frost, et al., 2008). Appendix 1 shows the number of observations per country for the final sample. It shows that there is a wide variety in the number of observations per country. The US is overly represented with 2643 observations; this is mainly caused by the availability of data from the I/B/E/S database. The sample also includes major countries with relatively few observations. For example, Sweden and Portugal are only covered with one firm-year observation.

In addition to inter-country differences in sustainability disclosure, it is also expected that there will be inter-industry differences, as the relevance of certain information might differentiate between industries (Hahn & Kühnen, 2013). Table 2 shows the industry distribution of the firm-year observations of the final sample. This table shows an equal distribution of the total sample. Exceptions are the (1) telecommunications observations, with a relatively low total of 47 firm-year observations, and (2) industrials, with a total of 532 observations, which is a relatively high amount of observations compared to other industries.

Table 1 Firm-year observations

Total collected 45.135

Total after removing firm-year with missing data 43.623

Total with sustainability report 18.561

Total with ESG scores information 18.536

Total with IBES information (H1&3) 3.068

Total with sustainability auditor information (H2) 2.006

Total with bid ask spread data (Robustness test) 2.347

Table 2 Observations per industry and year

Sector 2009 2010 2011 2012 2013 2014 2015 Total

Oil & Gas 25 29 25 29 34 32 33 224

Basic Materials 28 41 38 39 38 39 35 279 Industrials 47 69 71 75 78 83 71 532 Consumer Goods 42 43 48 57 55 60 58 395 Healt Care 28 30 30 31 36 35 33 243 Customer Services 42 49 50 54 55 54 55 387 Telecommunications 5 5 6 7 7 6 7 47 Utilities 28 37 36 38 36 35 33 268 Financials 36 35 45 57 62 67 65 396 Technology 26 35 39 44 43 45 42 297 Total 307 373 388 431 444 456 432 3068

(17)

Page 17

Research Design

Table 3 provides an overview of the variables used in this research. All measurements of the variables will be discussed in detail. The robustness check measure, the bid ask spread, will be further discussed in Chapter 4.

Information Asymmetry

Information asymmetry is not directly observable, and so a measure needs to be selected to act as a proxy (Petersen and Plenborg, 2006). A commonly used and generally accepted method for measuring information asymmetry is the accuracy of forecasts by analysts (Shroff et al., 2013; Marquardt and Wiedman, 1998; Martínez-Ferrero et al., 2016; Lang and Lundholm, 1996). Therefore, this research also used this approach as a proxy for measuring information asymmetry.

According to Hope & Pope (2003) and Barron et al. (1998), organisations that are more transparent about their disclosures have lower information asymmetry between the organisations and the analysts, which leads to higher accurate earnings forecasts by analysts. When the earnings forecasts by the analysts are inaccurate, there will be a greater difference between the expected value and the actual value of earnings per share, which indicates a higher level of information asymmetry.

The data with regard to the analyst forecast error (AFE) was obtained from the I/B/E/S database available from DataStream. To determine the AFE, the yearly forecast accuracy was calculated. Lang and Lundholm (1996) argue that there is no specific point in time to identify when the disclosed information influences the market and the forecasts by analysts. Complementary, Espahbodi et al. (2015) and Hutira (2016) found that there is not a substantial difference between forecast horizons. Therefore, the yearly average of the forecasted earnings per share was used.

Name of variable Mnemonics Role Description Source

ESG Score ESGTOTAL Independent Overall ESG score with an ESG Controversies overlay for firm i in year t ASSET4 Environmental score ENSCORE Independent Score environmental pillar of ESG score for firm i in year t ASSET4 Social Score SOSCORE Independent Score social pillar of the ESG score for firm i in year t ASSET4 Corporate Governance Score CGSCORE Independent Score corporate governance pillar of the ESG score ASSET4 Analyst Forecast Error AFE Dependent Difference between forecasted and expected EPS for firm i in year t I/B/E/S CSR Sustainability Committee CSRSUSCOM Interacting Binary variable which takes 1 if present and 0 otherwise for firm i in year t ASSET4 External Sustainability Auditor SUSAUDIT Interacting Binary variable which takes 1 if present and 0 otherwise for firm i in year t ASSET4

Industry Type INDUSTRY Control ICB industry codes Datastream

Firm Size SIZE Control Log of total assets of firm i in year t Worldscope

Leverage LEV Control Total debt to total assets of firm i in year t Worldscope

Country CNTRY Control ISO country code Datastream

Year YEAR Control Financial year t Datastream

Bid-Ask spread BAS Robustness check Difference between bid and ask prices for firm i in year t Datastream

(18)

Page 18 This research applied one method suggested by Hutira (2016) to measure the analyst forecast error. This concerns an absolute measure, which focuses on the difference between the actual earnings per share (AEPS) and the yearly average of the forecasted earnings per share (FEPS). The forecast error serves as a measure of bias of analysts indicating information asymmetry. There are two types of AFE, namely negative and positive AFE. If the error is negative (indicating that the forecast is higher compared to the actual earnings per share), analysts are opportunistic. If the error is positive (i.e., if the forecast is lower compared to the actual earnings per share), the analysts are pessimistic.

It is important to note that this research is concerned with the information asymmetry between organisations and stakeholders. According to Shehata (2014), analysts serve as an intermediary between an organisation and its stakeholder(s). Analyst forecasts thus serve as a mechanism of communicating sustainability information to stakeholders (Shehata, 2014).

However, an often-heard criticism of using the AFE as a proxy for information asymmetry is that this measure is biased (Easterwood & Nutt, 1999). Second, Hilary and Hsu (2013) found that the consistency of analyst forecasts is of greater importance with regard to the quality of information as compared to the accuracy of analyst forecasts. To determine the appropriateness of the use of analyst forecast error, a robustness check was performed by using an additional measure for determining the amount of information asymmetry: the bid ask spread. This robustness check is further elaborated on in Chapter 4.

Quality of Sustainability Disclosure

To measure the quality of sustainability disclosure, the ESG score developed by Thomson Reuters, available from ESG ASSET4, was used as a proxy. The ESG score measures both the performance and the quality of sustainability disclosure of an organisation. The ESG score consists of three pillars, namely the environmental, social and governance pillar. Each pillar is divided into three or four categories consisting of multiple equally weighted indicators. Table 4 gives an overview of the ESG score. All pillars were both individually and jointly tested to determine their impact on information asymmetry. In this research, the new ESG score methodology was used, which was developed by ESG ASSET4 in 2017. This new methodology represents the new strategic ESG framework of Thomson Reuters. The score is formed by 178 indicators, covering an organisation’s environmental, social and governance performance. The ESGTOTAL variable is the combination of the score on the three pillars and is discounted for ESG controversies. In case companies are involved in ESG controversies, they will get a lower combined score for the latest closed fiscal year.

(19)

Page 19 Controversies serve as an indicator of low quality and could thus be valuable for determining the quality of sustainability disclose (Aoudi & Marsat, 2016). The controversies score methodology developed by Thomson Reuters is calculated on the basis of 23 recent topics on controversy. These topics of controversies are benchmarked on industry groups.

Presence of a CSR Sustainability Committee

Organisations that have a CSR sustainability committee are more transparent with regard to CSR performance compared to organisation with no CSR sustainability committee (Adams, 2002; Cowen et al., 1987). Both CSR committees and CSR teams will be seen as CSR sustainability committees. The data used for these analyses was obtained from ASSET4, available from Datastream. This binary interacting variable was measured by looking at whether a CSR sustainability committee is present or not.

Presence of an External Sustainability Auditor

The presence of an external sustainability auditor for sustainability information could possibly influence the effect of sustainability disclosure on information asymmetry as elaborated in the previous chapter. Data from ASSET4 was used to determine whether or not an organisation makes use of an external sustainability auditor for their sustainability report.

Organisations with no information available on whether an external sustainability auditor for sustainability was attracted were excluded for testing Hypothesis 2. In total, 2.006 firm year observations were used for testing the effect of the presence of an external sustainability auditor on the association between the quality of sustainability disclosure and

Pillar Category Indicators in Rating Weights

Environmental Resource Use 20 11%

Emissions 22 12% Innovation 19 11% Social Workforce 29 16% Human Rights 8 4% Community 14 8% Product Responsibility 12 7% Governance Management 34 19% Shareholders 12 7% CSR Strategy 8 4% Total: 178 100%

(20)

Page 20 information asymmetry. This external sustainability auditor variable will be incorporated as a binary variable.

Control Variables

This research controlled for multiple organisational characteristics and settings that could potentially bias the findings. Cuadrado-Ballesteros et al. (2017) studied the effect of sustainability assurance on information asymmetry. Cuadrado-Ballesteros et al. (2017) used multiple organisational characteristics as control variables, this present research followed their approach and also incorporated these control variables. The following control variables were therefore included: country, industry, firm size, leverage and year. Dummies were used for the industry, country and year variables.

Country and industry control variables were used to take into account the country-level and industry-country-level differences with regard to sustainability disclosure. Especially on a national level it was expected that there were differences between the extensiveness of sustainability reports. One should, for example, expect that organisations in highly developed countries will disclosure more extensive on sustainability information compared to organisation in developing countries. The industry and country control variables were controlled for as fixed effects in the regression model. Industry-related fixed effects were separated into ten categories following the Industry Classification Benchmark (ICB). Country-related fixed effects were indicated on the basis of ISO country codes. Year-related effects were also included as control variables, due to the pace at which sustainability disclosure is developing. It was expected that the quality of sustainability disclosure was higher in more recent years.

Furthermore, it was expected that there are differences between certain types of industries with regard to sustainability disclosure, as not all topics of sustainability disclosure are relevant for all types of industries. An example of such industry differences with regard to sustainability disclosure is that organisations that operate in industries closer to society and that are thus more easily observable for society will disclose more information on sustainability (Branco et al., 2008).

Further, this research controlled for firm size, which was measured by taking the logarithm of the total assets, because of both the size and spread of the values. Prior research found that firm size can affect the availability and quality of information (Atiase, 1985; Gray et al., 1995; Guidry and Patten, 2012; Brammer and Pavelin, 2006; Gu and Wang, 2005). It

(21)

Page 21 was expected that organisations that are more sizable also have more complex information available, which, according to Gu and Wang (2005), leads to more information asymmetry.

Leverage was measured as the amount of debt to total assets. Gao and Zhu (2015) found that organisations with high information asymmetry tend to use more debt capital. Therefore, it was expected that there is a positive relation between the leverage of an organisation and the amount of information asymmetry.

Data analysis

This archival research used multiple regression model to test the expected relations. The regression models included the following control variables: (1) the log of the total assets, indicating the firm size, (2) second leverage, indicating how an organisation finances its assets, and (3) third year dummies were included as control variables. Moreover, industry and country effects were controlled for in all the regression models.

The time span of this research is 7 years. Therefore, panel data was used for all models. The reason for choosing panel data was to overcome unobservable firm heterogeneity, which limits possible endogeneity problems. A fixed effects model was used, as the Hausman test resulted in a p-value of 0.001, indicating that the fixed effects model would be most appropriate to use.

Three different research models were used for testing the relation between the quality of sustainability disclosure and information asymmetry. All samples include panel data, whereby the variables represent company i in year t. The first hypothesis was concerned with the effect of the quality of sustainability on information asymmetry, whereby all components of the quality of sustainability disclosure were tested separately and combined. In model 1, both the component scores and combined score are referred to as ESG-scores. Model 2 & 3 both represent the interacting variables, namely respectively external sustainability auditor and CSR sustainability comittee. The control variables country, industry and year were included as respectively 35, 10, and 8 dummy variables.

Model 1 (main relation)

AFEit= β0 + β1ESG-scoresit +β2SIZEit + β3LEVit +∑𝟒𝟏𝒂=𝟕 βaCNTRYait + ∑𝟓𝟏𝒃=𝟒𝟐 βbINDUSTRYbit + ∑𝟓𝟗𝒄=𝟓𝟐

(22)

Page 22

Model 2 (interacting effect external sustainability auditor for sustainability information)

AFEit= β0 + β1ESGTOTALit + β2ESGTOTALit * SUSAUDIT +β3SIZEit + β4LEVit +∑𝟒𝟏𝒂=𝟕 βaCNTRYait +

∑𝟓𝟏

𝒃=𝟒𝟐 βbINDUSTRYbit + ∑𝟓𝟗_{𝒄=𝟓𝟐} βcYEARcit + αi + µit

Model 3 (interacting effect CSR sustainability committee)

AFEit= β0 + β1ESGTOTALit +β2ESGTOTALit * CSRSUSCOM + β3SIZEit + β4LEVit +∑𝟒𝟏𝒂=𝟕 βaCNTRYait +

(23)

Page 23

IV. Results

First, the descriptive statistics are presented. After presenting these descriptive statistics, the results of the pairwise correlations are discussed. Third, the regression results are presented for all models, including some additional relevant analyses. This chapter ends with a robustness test of the obtained results by replacing the AFE measure with the bid ask spread measure for information asymmetry.

Descriptive Statistics

Table 5 provides an analysis of the variables used in this research. To eliminate the effect of outliers, the AFE variable and leverage control variable were winsorized. Leverage was winsorized at the top and bottom one percent and AFE was winsorized at the top and bottom five percent. Both the ESGTOTAL variable and component variables allowed scores ranging from 0 to 100. Therefore these variables were not trimmed.

The mean ESGTOTAL score is 51,5, which almost equals the initial dataset, excluding the firm-year observations with no information available on the ESG score. An analysis was conducted on the development of the ESGTOTAL score for the sample used in this research (2008-2015). This analysis showed a yearly increase of the ESGTOTAL score for all subsequent years, which is in line with research conducted by KPMG (2013).

The average AFE was 1,7, indicating that there is on average a difference of 1,7 between the expected and actual earnings per share per firm-year observation, which could either be positive or negative. The error value presented here is absolute and was not scaled by, for example, the actual earnings per share.

The main sample included more than 3.000 observations. The average of the variable SUSAUDIT is 0,40, indicating that 40 percent of the firm year observations had an external

Variable N Mean Std. Dev. Min Max

AFE 3068 1,699 1,243 0,280 4,310 ESGTOTAL 3068 51,497 15,404 10,570 95,110 CGSCORE 3068 79,161 19,902 0,000 97,580 SOSCORE 3068 71,967 20,930 4,610 97,680 ENSCORE 3068 74,322 21,806 0,000 95,110 SUSAUDIT 2006 0,404 0,491 0 1 CSRSUSCOM 3068 0,786 0,411 0 1 SIZE (LOG) 3068 7,290 0,782 4,584 11,187 LEV 3068 0,277 0,175 0,000 2,193

(24)

Page 24 sustainability auditor. The variable CSRSUSCOM had an average of 0,79, which means that 79 percent of the firm year observation had a CSR sustainability committee. Appendix 2 summarizes the interacting values per industry.

The sample used for the second hypothesis included more than 2.000 firm year observations with information available with regard to the presence of an external sustainability auditor for sustainability information. Hypothesis 2 concerned the interacting effect of the presence of a sustainability auditor on the impact of the quality of sustainability disclosure on information asymmetry. Appendix 3 shows the distribution of the dependent and independent variables of this research on an industry level. This overview indicates that the average ESGTOTAL score for the telecommunications industry is 10 points lower compared to the average average of the other industries. However it is not expected that this will influence the overall results due to the relatively low number of observations (47) for this industry. No specific analysis was done on country-level, due the relatively high difference in amount of firm-year observations per country.

Correlations

The Pearson correlation results between the variables are presented in Table 6. First, it was expected that the component scores and total ESG score are correlated to each other. The obtained results are in line with this expectation. Significant correlations (at 1%) were found between all separate components and the ESGTOTAL score. The correlation coefficients between the components were all in the range of 20,5% - 57,8%. This indicates that organisations with a high (or low) score on one component were also likely to have a high (or low) score on the other components. The correlation coefficients between the total ESG score and the environmental, governance and social component scores were respectively 25,7%, 5,7% and 35,3%. These correlations were all significant at the 1% significance level.

The correlation coefficients for the measurement of the independent variables were relatively high, especially between the environmental and the social score (r = 0,578; p <,001). To limit the issue of multicollinearity, all models after model 1 only included the ESGTOTAL variable. A negative significant correlation (r = 0,104; p <,001) was found between the ESGTOTAL score and AFE, this was in line with hypothesis 1, because it indicates that a high (or low) quality of sustainability disclosure is correlated with a low (high) level of information asymmetry. Further, a significant negative correlation was found

(25)

Page 25 between the governance score and AFE (-25,7%). However, the social score and environmental score were not significantly correlated with AFE.

Size (r=0.347; p <,001)) and SUSAUDIT (r=0,134) ; p <,001) were both positively and significantly correlated with AFE. This indicates that bigger organisations and/or organisations with a sustainability auditor have more analyst forecast errors. This result therefore contradicts the expected negative correlation. LEVERAGE and CSRSUSCOM were both not significantly correlated with AFE.

Findings

Panel data was used for all models and all models included the same control variables. Two regression models were significant and these were consistent with the expected relation. All regression models included AFE as dependent variable. All regression models, except for model 2, included 3.068 firm-year observations. Model 2, concerned with the interacting effect, included in total 2.006 firm-year observations. The results of all the regression models are presented in Table 7.

Model 1 was used for the main relation concerning the impact of the quality of sustainability disclosure on information asymmetry. According to the first hypothesis, the level of information asymmetry should be lower (higher) if the quality of sustainability disclosure is higher (lower), so a negative relation was expected between these two variables. First, the ESGTOTAL score was used as measure for the quality of sustainability disclosure. Second, the components of the ESGTOTAL score were separately tested according to model 1. Model 1 indicated a significant negative relation between the ESGTOTAL score and the AFE, with a R-squared of 0,108, which implies that the model explains 10.8% variance of the response data around the mean. The coefficient of ESGTOTAL was -0.008 and was highly significant (p <,01). Therefore, it is possible to state, that when ESGTOTAL score is higher

Table 6 Pairwise correlation matrix

1 2 3 4 5 6 7 8 9 1 AFE 1,000 2 ESGTOTAL -0,104*** 1,000 3 CGSCORE -0,257*** 0,057*** 1,000 4 SOSCORE -0,022 0,353*** 0,205*** 1,000 5 ENSCORE 0,022 0,257*** 0,270*** 0,578*** 1,000 6 SUSAUDIT 0,134*** 0,100*** -0,075*** 0,210*** 0,270*** 1,000 7 CSRSUSCOM 0,004 0,118*** 0,160*** 0,325*** 0,346*** 0,203*** 1,000 8 SIZE 0,347*** -0,059** -0,260*** 0,161*** 0,158*** 0,336*** 0,134*** 1,000 9 LEV 0,003 -0,008 0,013 -0,054*** -0,024 0,056 0,002 -0,018 1,000 **Indicates p<0.01 ***Indicates p<0.001

(26)

Page 26 (lower), the analyst forecast error will be lower (higher). This outcome showed that higher (lower) quality of sustainability disclosure therefore lead to a lower (higher) level of information asymmetry. Therefore, the first hypothesis is accepted.

In addition to ESGTOTAL, the correlation between all three component scores and AFE were tested separately. No significant correlations were found between both the governance and environmental score and the dependent variable AFE. However, a significant negative relation was found between the social score and AFE, with a R-squared of 0,103, which explained 10,3% of the variance of the response data around the mean. The coefficient of the social score was -0,005 and was significant at a p-value of 5%. This finding indicates that when the social score is higher (lower), the analyst forecast error will be lower (higher). Therefore, hypothesis 1b is also supported.

Model 2 included external sustainability auditor as an interaction term. However no significant influence of presence of an external sustainability auditor on the effect of the quality of sustainability disclosure on information asymmetry was found. Similar to model 2, model 3 therefore did not support the expectation. Model 3 was concerned with the influence of a CSR sustainability committee on the relation between quality of sustainability disclosure and information asymmetry. However, no significant correlation was found for the interacting role of this variable on the effect of the quality of sustainability disclosure on information asymmetry. Both hypotheses 2 and 3 were therefore not supported by the regression models.

Table 7 Regression models

Model 1 Model 1a Model 1b Model 1c Model 2 Model 3 Additional 1 Additional 2

ESGTOTAL -0,008*** -0,007*** -0,010*** CGSCORE -0,002 0,004 SOSCORE -0,005** ENSCORE 0,000 SUSAUDIT 0,187 CSRSUSCOM -0,341* 0,121 -0,254*** ESGTOTAL*SUSAUDIT 0,002 ESGTOTAL*CSRSUSCOM 0,002 CGSCORE*CSRSUSCOM -0,008*** SIZE (LOG) 0,464** 0,464** 0,480** 0,465** 0,375 0,449** 0,426** 0,436** LEV 1,014*** 1,017*** 1,055*** 1,019*** 1,104*** 1,002*** 1,010*** 1,055*** YEAR Y Y Y Y Y Y Y Y

INDUSTRY fixed effects Y Y Y Y Y Y Y Y

CNTRY fixed effects Y Y Y Y Y Y Y Y

Constant Y Y Y Y Y Y Y Y

Overall R2 0,108 0,101 0,103 0,110 0,123 0,105 0,104 0,095

N 3068 3068 3068 3068 2006 3068 3068 3068

*p<0,1 **p<0,05 ***p<0,01

(27)

Page 27 Furthermore the coefficients of the control variables SIZE and LEVERAGE were both positively and significantly (respectively at p <,05 and p <,01) correlated with all models, except for SIZE in regression model 2. These relatively high coefficients indicated that organisations that are more (less) sizable or higher (lower) leveraged have higher (lower) levels of information asymmetry. This is in line with the expectations.

Additional Analyses

Multiple additional regression analyses were performed with the variables included in this research. Two significant relations were found. These can also be found in Table 7, identified as ‘additional 1’ and ‘additional 2’. First, the presence of a CSR sustainability committee influenced the association between the governance score and AFE. The coefficient was -0,008 and was highly significant (p <,01). The R-squared had a value of 10,4 %. So, 10,4% of the variance in the governance score was explained by this additional analysis. This additional analyses implies that the presence of a CSR sustainability committee negatively influences the association between the governance score and the amount of information asymmetry.

The second additional regression that was significant concerned the presence of a CSR sustainability committee. This presence of a CSR sustainability committee was tested as an independent variable, instead of an interacting variable. This additional analysis showed a significant negative relation (-0,254, p <,01) between the CSRSUSCOM and the AFE, with a R-squared of 0,095, which explained 9.5% of the variance of the response data around the mean. The coefficient of CSRSUSCOM was -0,254 and was highly significant at a p-value of 1%. This indicates that the presence of a CSR sustainability committee could reduce the amount of information asymmetry. Although there was no influence of CSR sustainability committee on the relation between the quality of sustainability disclosure and the amount of information asymmetry, it is possible to conclude that organisations should implement a CSR sustainability committee to reduce information asymmetry in general.

Robustness Check Information Asymmetry

To determine the appropriateness of the use of analyst forecast error, a robustness check was performed by using an additional measure for determining the amount of information asymmetry: the bid ask spread.

The bid-ask spread is a measure that is widely used for determining the amount of information asymmetry. The bid-ask spread is the difference between what a buyer wants to

(28)

Page 28 pay for a share and the price at which a seller wants to sell the share. According to Clarke and Shastri (2001), the bid-ask spread is a more appropriate measure for information asymmetry compared to the analyst forecast error. They found that the bid-ask spread relates to firm characteristics that are associated with information asymmetry. In case of no information asymmetry, the bid price should be equal to the ask price.

Due to the limited availability of data on bid and ask prices, the total amount of firm-year observations for this robustness check was lower compared to the AFE measure of information asymmetry. In total 2.347 firm-year observations were used for this robustness check. The regression results are presented in Table 8. A negative significant was found between ESGTOTAL and the bid ask spread. The coefficient was -0,004 and was significant (p <,1). The R-squared had a value of 27,5%. So, 27,5% of the variance in bid ask spread was explained by model 1. Furthermore, a negative significant was found between the social component score and the bid ask spread. The coefficient was -0,002 and was significant (p <,1). The R-squared has a value of 25,8%. So, 25,8% of the variance in bid ask spread was explained by model 1

The second additional analysis was concerned with the relation between the presence of a CSR sustainability committee and information asymmetry. Significant results were found for this relation when using the AFE as measure for information asymmetry. Consistent with this outcome, a negative relation was found when using the bid ask spread as measure for information asymmetry. The coefficient was -0,186 and was significant (p <,1). The R-squared has a value of 22,0%. So, 22,0% of the variance in bid ask spread was explained by the additional analysis. Thus, the results of the robustness check were consistent with the results when using AFE as measure for information asymmetry. Therefore, AFE is a robust measure for information asymmetry.

(29)

Page 29

Table 8 Regression models Robustness Check

Model 1 Model 1a Model 1b Model 1c Model 2 Model 3 Additional 1 Additional 2 ESGTOTAL -0,004* -0,005* -0,004* CGSCORE -0,000 -0,002 SOSCORE -0,002* ENSCORE 0,001 SUSAUDIT 0,011 CSRSUSCOM 0,002 -0,023 -0,186* ESGTOTAL*SUSAUDIT 0,001 ESGTOTAL*CSRSUSCOM -0,000 CGSCORE*CSRSUSCOM 0,002 SIZE (LOG) 11,531 10,287 10,739 10,783 11,958 11,728 10,999 9,645 LEV 0,619** 0,626 0,630** 0,621** 0,614** 0,606** 0,606* 0,611* YEAR Y Y Y Y Y Y Y Y

INDUSTRY fixed effects Y Y Y Y Y Y Y Y CNTRY fixed effects Y Y Y Y Y Y Y Y

Constant Y Y Y Y Y Y Y Y

Overall R2 0,275 0,243 0,258 0,265 0,280 0,254 0,238 0,220 N 2347 2347 2347 2347 2347 2347 2347 2347 *p<0,1 **p<0,05 ***p<0,01

(30)

Page 30

V. Conclusion and discussion

Sustainability disclosure is subject to a decrease in trustworthiness. Currently, organisations are not providing sufficient and multifaceted information towards their stakeholders. Previous research was especially focussed on the presence and degree of sustainability disclosure. However, these measures insufficiently take into account the trustworthiness of sustainability disclosure. Therefore, this research used the quality of sustainability disclosure. This research takes the negative point of view and found results in support of this view. Multiple hypotheses were tested and supporting evidence was found the main hypothesis (H1) and one sub hypothesis (H1b).

According to Hypothesis 1, there should be a negative association between the quality of sustainability disclosure and information asymmetry. This research found evidence for a negative relationship between the quality of sustainability disclosure and information asymmetry, which was in line with Hypothesis 1. This highly significant relation showed that when there was a higher (lower) quality of sustainability disclosure, there was a lower (higher) level of information asymmetry. Past research concerned with the quality of sustainability disclosure is scant. This research provided new insights for both practice and science by focussing on the quality aspect of sustainability disclosure. Organisations could decrease information asymmetry by increasing the quality of sustainability disclosure.

Furthermore, a significant negative relation was found between the social score and the amount of information asymmetry. Therefore, Hypothesis 1b is accepted. A higher (lower) score on the social pillar will thus lead to a lower (higher) amount of information asymmetry.

Second, the aim of this study was to examine whether the presence of a CSR sustainability auditor or the presence of a CSR sustainability committee would affect the relation between the quality of sustainability disclosure and information asymmetry. No evidence was found for the supposed influence of the presence of a sustainability auditor or a CSR sustainability committee. Therefore, organisations do not necessary need to implement a sustainability auditor or CSR sustainability committee to impact the relation between the quality of sustainability disclosure and information asymmetry. Contrary, an additional analysis showed a significant negative relation between the presence of a CSR sustainability committee and the amount of information asymmetry. Although there was no interacting effect, it is still recommended for organisations to incorporate a CSR sustainability committee, because the presence of a CSR committee reduces the amount of information asymmetry directly.

(31)

Page 31 The aim of this study was to use a global sample. However, some countries were only represented by less than ten firm-year observations. Therefore it was not possible to draw conclusions about inter-country differences. Furthermore, Appendix 1 shows that the US is overly represented in the sample; more than 80% of the observations were from the US. This is mainly caused by the availability of the data concerning the analyst forecast error. So, the validity of this research was negatively affected. Future research should therefore use a more evenly distributed sample, to increase external validity.

Second, it is questionable whether the ESGTOTAL measure for the quality of sustainability disclosure was sufficient for measuring the quality of sustainability disclosure. It is difficult to determine this, due to different quality perspectives and the differences between organisations. To measure the quality of sustainability disclosure, the ESGTOTAL incorporated the ESG controversies score, which penalizes organisations that are involved in scandals related to sustainability reporting. Controversies will lead to lower ESGTOTAL score for organisations, and serve as an indicator that reflects the quality of sustainability disclosure. However, only 23 controversy topics were identified. Therefore it is possible that some scandals were not included in this measure, which could lead to positively biased total scores. Future research should therefore incorporate other measures that better reflect the quality of sustainability disclosure, whereby these studies take into account the organisational differences and the differences in quality perspectives.

Furthermore, future research should take a qualitative approach instead of a quantitative approach. Quality is a concept that is hard to define, which is especially caused by the multiple lenses through which quality can be observed. Favourable, future research should observe quality from the stakeholder perspective.

This research aimed to identify the effect of the quality of sustainability disclosure on information asymmetry and aimed to answer whether organisations should make use of an external sustainability auditor or CSR sustainability committee to moderate this relation. This research found evidence for a negative relation between the quality of sustainability disclosure and information asymmetry. Furthermore, an additional analysis revealed that the presence of a CSR sustainability committee reduces information asymmetry. To conclude, organisations should increase the quality of sustainability disclosure and implement a CSR sustainability committee to reduce information asymmetry.

(32)

Page 32

References

Adams, C.A. (2002) Internal organisational factors influencing corporate social and ethical reporting: beyond current theorising. Accounting, Auditing & Accountability Journal, 15(2), 223-250.

Adams, C. A. & Evans, R. (2004). Accountability, completeness, credibility and the audit expectations gap. Journal of Corporate Citizenship, 14, 97-115.

An, Y., Davey, H. & Eggleton, I. R. C. (2011). Towards a comprehensive theoretical framework for voluntary IC disclosure. Journal of Intellectual Capital, 12(4), 1469-1930.

Arjaliès, D. L. & Mundy, J. (2013). The use of management control systems to manage CSR strategy: A levers of control perspective. Management Accounting Research, 24(4), 284-300.

Atiase, R. K. (1985). Predisclosure information, firm capitalization, and security price behavior around earnings announcements. Journal of Accounting Research, 21-36. Bagnoli, M., & Watts, S. (2017). Voluntary assurance of voluntary CSR disclosure. Journal

of Economics & Management Strategy, 26(1), 205-230.

Bain, M. (2016, April 16) Is H&M misleading customers with all its talk of sustainability? Retrieved from: https://qz.com/662031/is-hm-misleading-customers-with-all-its-talk-of-sustainability/

Barret, P. & Philips, M. (2016, September 2016) Can ExxonMobil Be Found Liable for Misleading the Public on Climate Change? Retrieved from:

https://www.bloomberg.com/news/articles/2016-09-07/will-exxonmobil-have-to-pay-for-misleading-the-public-on-climate-change

Barron, O. E., Kim, O., Lim, S. C., & Stevens, D. E. (1998). Using analysts' forecasts to measure properties of analysts' information environment. Accounting Review, 73(4), 421-433.

Brammer, S. & Pavelin, S. (2006). Voluntary environmental disclosures by large UK companies. Journal of Business Finance & Accounting, 33(7–8), 1168-1188.

Quality of Sustainability Disclosure and Information Asymmetry Master’s thesis MSc A&C Accountancy 2017/2018 University of Groningen