Corporate governance mechanisms and voluntary intellectual capital disclosures : evidence from the Netherlands

Amsterdam Business School

Corporate Governance Mechanisms and Voluntary Intellectual

Capital Disclosures: Evidence from the Netherlands

Final version

Name: Salih Alpay Kelkitli Student number: 11147334

Thesis supervisor: Dr. S. Bissessur Second supervisor: Dr. W. Janssen Date: 22/01/2016

Word count: 14744, 0

MSc Accountancy & Control, specialization Accountancy Faculty of Economics and Business, University of Amsterdam

Statement of Originality

This document is written by student Alpay Kelkitli who declares to take full responsibility for the

contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

Abstract

The purpose of this study is to investigate the relationship between corporate governance mechanisms and voluntary disclosure of intellectual capital in AEX listed firms. The dependent variable, intellectual capital disclosure, is measured using the category score for workforce/training and development, product innovation, and measure of brand value in the Thomson Reuters ESG ASSET4 database. These variables are chosen based on the intellectual capital classification scheme of Guthrie & Petty (2000). The independent variables comprise various forms of corporate governance mechanisms, namely, independent directors, board size, audit committee independence and ownership concentration, and are also obtained from the same database. The research opts to test the relationship between intellectual capital disclosure and corporate governance by using three separate regression models, each model using a separate measure for intellectual capital disclosure and controlling for firm profitability, size and leverage. Results of the multiple regression models indicate that with the exception of the variable for ownership concentration in one of the models, there seems to be no statistical significance of the corporate governance variables with voluntary intellectual capital disclosure.

Contents 1 Introduction ... 6 1.1 Research question ... 7 1.2 Contribution ... 8 2 Theoretical Framework ... 9 2.1 Information Asymmetry ... 9 2.2 Voluntary Disclosure ... 9 2.3 Corporate governance ... 10 2.4 Reporting standards ... 12

2.4.1 IAS 38 – Intangible Assets ... 12

2.4.2 Internally generated Intangible Assets ... 13

2.4.3 Usefulness of Financial Information... 14

3 Literature Review ... 15

3.1 What is IC? ... 15

3.2 Voluntary IC disclosure ... 17

3.3 Determinants of voluntary disclosure of IC ... 18

3.4 Corporate governance mechanisms and (IC) disclosure ... 18

4 Development of hypotheses ... 20 4.1 Board composition ... 20 4.2 Audit Committee ... 21 4.3 Ownership structure ... 21 5 Methodology ... 23 5.1 Institutional setting ... 23

5.2 Development of Corporate Governance in the Netherlands ... 23

5.3 Legal regulation ... 24 4

5.4 Board structure in the Netherlands ... 25

5.5 Sample selection and data source ... 25

5.6 Measurement of Dependent and Independent variables ... 26

5.6.1 Evaluating the quantity and quality of IC disclosure ... 26

5.6.2 Corporate Governance variables ... 28

5.6.3 Control Variables ... 29

5.7 Data analysis ... 32

6 Results... 37

6.1 Descriptive analysis of intellectual capital disclosure ... 37

6.2 Descriptive Statistics ... 37

6.3 Regression Results ... 38

6.4 Examination of Hypotheses ... 39

6.5 Summary and Conclusion ... 43

7 References ... 45

1 Introduction

The agency theory, coined by Jensen and Meckling (1976), argues that a misalignment of interests exists between shareholders and managers of firms. The theory stems from the assumption that the self-interest of managers within the firm can compromise the best interest of the shareholders. Assuming that perfect contracts are not realizable, shareholders will control the behavior of the managers by engaging in corporate governance mechanisms, such as monitoring and other control activities. Besides corporate governance, voluntary disclosure of information can also be considered a mechanism used to help reduce agency conflicts (Jensen and Meckling, 1976).

The purpose of this study is to investigate the relationship between corporate governance mechanisms and voluntary disclosure of intellectual capital in annual reports. The modern economy has converged from a labor and capital intensive industrial economy, to one that generates tangible and intangible assets through the use of knowledge. Organizations operating in the so-called “knowledge economy” no longer depend on the production of material goods for their economic value (Guthrie et al., 2004). Rather, organizations gain shareholder value and maintain a competitive advantage in their sector through the creation and manipulation of intellectual capital (hereafter: IC), such as knowledge, experience, good relationships and technological capacities (Brüggen et al. 2009). There is no consensus on the definition of IC. Sveiby (1997) categorizes IC into human, structural and organizational capital, while Guthrie and Petty (2000) classify IC into internal structure, external structure, and human capital. Sullivan (2000) defines intellectual capital as knowledge that can be converted into profit, highlighting it as potentially the core element for corporate performance. Within an agency context, the disclosure of IC can reduce the information asymmetry between investors and firms, by reducing the uncertainty of future prospects and facilitating a more accurate valuation of the firm. Despite the increasing significance of IC as the main driver for the value of organizations today, traditional accounting framework and financial reporting fail to reflect this value in the financial statements. Facebook, for example, has a market to book ratio of approximately 600%, and similar conclusions can be made for Google, LinkedIn, and so forth. According to Brennan (2001), this gap between the book value and market value of firms is explained by IC or more specifically intangible assets. The gap in question leads to increasing information asymmetry between investors and firms, and ultimately to an inefficient resource allocation within capital markets (Li et al., 2008).

Market participants attribute high value to the performance and valuation of firms. Thus, it is of importance that the annual report provides investors the ability to accurately predict and understand what course management has set out and how this will create value for them. Aside from the very limited accounting requirements and standard on intangible assets, there is no financial disclosure requirement for IC. As IC is a very broad term encompassing numerous elements, it is difficult to incorporate it into the current accounting framework. The current traditional accounting framework thus fails to reflect the economic value generated by an organization’s IC in the financial statements (Brüggen et al. 2009). As a result of this value gap, one can challenge to what extent the financial statements of an organization provide new and relevant information towards its shareholders and potential investors.

1.1 Research question

The focus of most previous studies is on the extent to which firms and managers voluntarily disclose information on IC (Brennan, 2001; Bontis, 2003; Goh and Lim, 2004; Ismail and Hamzah 2008). Relatively few studies research the specific reasons behind the variation across firms with respect to the disclosure of IC. Several studies provide evidence for firm size and industry type as being key determinants for the voluntary disclosure of IC in annual reports (Beaulieu et al., 2002; Bozzolan et al., 2003; Garcia-Meca et al., 2005; Brüggen et al. 2009;) , other research identify social, political and cultural factors as determinants of IC disclosure (Chaminade and Roberts, 2003; Abeysekara, 2007). Thus, limited research has addressed the relationship between corporate governance mechanisms and IC disclosure (Cerbioni and Parbonetti, 2007; Li et al., 2008; Hidalgo et al., 2011).

This research analyses the relationship between corporate governance mechanisms and IC

disclosure. The research hypothesis states that significant relationships exist between IC disclosure in annual reports and board size, audit committee size, the frequency of audit committee meetings, ownership concentration, and institutional shareholding, controlling for firm size, firm performance, and industry type.

1.2 Contribution

The majority of the previous literature with respect to the relationship between corporate governance and voluntary disclosure has focused on the Anglo-Saxon (UK/US) model of corporate governance and control. No empirical research has been conducted in the Netherlands, as such this study will extend current literature.

The Netherlands is a European country with a communitarian corporate governance system. The Dutch civil law has French origins, and can be considered part of the broad legal tradition Civil law. Contrary to common law, civil law is characterized by low minority shareholder protection rights (La Porta et al., 1998). Moreover, the Dutch stock market is characterized by significant shareholders. According to Becht and Roel (1999) there is a significant difference between shareholder ownership in the USA compared to European countries: the median largest block holder in the Netherlands is 43.5% and below 5% in the USA.

Furthermore, Garcia-Meca and Sanchez-Ballesta (2010) suggest from their meta-analysis on corporate governance and voluntary disclosure that differences in legal and institutional environments explain why the implementation of similar corporate governance mechanisms results in different outcomes in different countries. Since the study of the Dutch context has not been researched previously, the results of this study will provide interesting findings to compare the effectiveness of governance code recommendations.

The remainder of this study proceeds as follows: in the next section the theoretical background will be developed. In section 3 the literature review will be discussed. In section 4 the research objective, along with the development of the hypotheses, will be discussed. Section 5 will discuss the institutional and legal setting of the research, along with the operationalization of the variables and analysis of the data used. In section 6 the results of the regression models will be discussed, along with a summary and conclusion.

2 Theoretical Framework

The theoretical background of this paper is based on two research areas. Namely, voluntary disclosure and corporate governance. The agency theory is routed in both research areas. This section will discuss these areas of research in an agency context.

2.1 Information Asymmetry

Jensen and Meckling (1976) define an agency relationship as a contracting relationship where one or more principals engage an agent to perform a specific purpose on their behalf. Assuming that both the agent and the principal act rationally and aim to maximize their own utility, a divergence in interests may exist between both parties. The principal can engage in several activities to align his own interests with that of the agent. Firstly, the principal can monitor the actions and behavior of the agent. Furthermore, the principal can limit the divergence of his interests with that of the agent by means of formal contracts (bonding costs). Since it is not economically efficient to fully align the interests of both the agent and the principal, in addition to the bonding and monitoring costs incurred by the principal, there will also be a residual loss as a result of the resulting misalignment in interests. The sum of the monitoring costs of the principal, bonding costs by the agent, and the residual loss are defined as agency cost.

2.2 Voluntary Disclosure

Voluntary disclosure can be defined as disclosures in excess of those required by law and can be considered as ex post decisions to provide information after having observed its content. Whereas required disclosure can be considered as a commitment to disclosure regardless of its content (Leuz and Verrecchia, 2000). Within the context of agency theory, voluntary disclosure can be seen as a means for managers to avoid bonding and monitoring activities by the shareholders. The higher the information asymmetry is between managers and shareholders, the more shareholders will have an incentive to control the behavior of managers through bonding and monitoring activities. However, managers may opt to inform shareholders that they are operating optimally by means of voluntary disclosure of information. If credible, this disclosure will alleviate the information asymmetry and agency conflict, consequently requiring shareholders to engage in less monitoring activities and/or mechanism (Watson et al., 2002).

Besides providing with relevant and reliable information that aids shareholders and board of directors in effective monitoring of management, voluntary disclosure, and the disclosure of financial information, in general, can play a role in reducing agency costs with respect to debt contracting (Healy and Palepu, 2001). Firstly, firms have an incentive to lower their cost of capital when seeking to raise capital in debt markets or renegotiation terms with their current lenders. If the firm’s financial reporting system provides the lender with insufficient relevant, reliable, and timely information, the information asymmetry between the firm (agent) and the lender (principal) will be relatively high. Consequently, the lender will have a higher uncertainty with respect to the firm’s current and future performance, and as such require a higher risk premium. Thus, managers and shareholders alike have an incentive to provide high-quality financial reporting to portray to their lenders that they are committed to providing reliable and timely information, in order to lower their cost of capital (Armstrong et al. 2010). Secondly, accounting figures may be used as input for debt covenants. Accounting based covenants in debt contracts or performance-pricing provisions (i.e. adjustment to the interest rate based on accounting information) can be used to reduce the agency costs.

This research focuses on a particular type of voluntary disclosure, namely IC or intangible assets. Generally, one of the factors a lender considers when assessing a credit approval or revision, is the likelihood the debtor will have sufficient net assets to cover the loan. According to Watts (2003) similar to liquidators, bankers and other lenders employ conservative accounting. Accordingly, intangible assets are often not included in the net assets as the value is not verifiable by an active market. This paper will therefore assess voluntary disclosure of IC as a means to create a transparent information environment for monitoring purposes by shareholders.

Important to consider here is that the decision to disclose voluntary information can be explained in terms of a cost-benefit framework. As discussed above, positive effects of voluntary disclosure are the reduction in information asymmetry and the cost of capital (Leuz and Verecchia, 2000). Obvious costs associated with voluntary disclosure of information would be the possible deterioration of the firm’s competitive position in the market.

2.3 Corporate governance

As defined by Larker et al. (2007, p. 964), corporate governance comprises of the “set of mechanisms that influence the decisions made by managers when there is separation of ownership and control”. The relationship between shareholders and managers of a widely held firm can be explained in the 10

context of an agency relationship. While shareholders (principal) supply the firm with financial capital and ultimately hold control rights, they are usually not involved in the day-to-day activities and decision making process. This responsibility is delegated to the board of directors, who likewise further delegate their responsibilities to managers (agent). As a result of the separation of ownership and control, managers acquire tacit knowledge and consequently have an information advantage over the shareholders and the board of directors. Managers possess information concerning the present and future performance of the firm, unknown to the shareholders and possibly the board of directors. The extent of the information asymmetry between the principal and the agent determine the magnitude of the agency conflict and the mechanisms needed to mitigate those (Armstrong et al., 2010). As a result of this separation of ownership and control, the need arises for shareholders to engage in bonding and monitoring activities by implementing corporate governance mechanisms.

To summarize, the agency conflicts arising from the separation of ownership and control are mediated by the following mechanisms of accountability: voluntary disclosure and corporate governance. Shareholders implement corporate governance mechanisms to monitor the behavior of the managers, and on the other side managers disclose more information. Based on agency theory, the relationship between corporate governance and the voluntary disclosure may be complementary or substitutive. If the relationship is complementary, the theory predicts that implementation of a corporate governance mechanism will lead to less opportunistic behavior and less information asymmetry, as such to more voluntary disclosure. In an increased monitoring environment, managers are less inclined to withhold information for their own benefit (Ho and Wong, 2001; Cerbioni & Parbonetti, 2008). However, the relationship could also be substitutive. This would mean that the implementation of a corporate governance mechanism would lead to less disclosure. Firms will not be inclined to voluntarily disclose information if one accountability mechanism would substitute for another one. Moreover, managers and shareholders alike consider proprietary costs in their decision making with respect to voluntary disclosure. Specifically, voluntary disclosure of IC could potentially lead to competitive disadvantages for the firm. If the relationship is substitutive, managers and shareholders would be less inclined to voluntarily disclose information if the firm’s corporate governance system sufficiently mitigates information asymmetry (Ho and Wong, 2001; Cerbioni & Parbonetti, 2008).

2.4 Reporting standards 2.4.1 IAS 38 – Intangible Assets

This study will examine AEX listed firm’s IC disclosure, and as such will focus on the European setting. Pursuant to Regulation (EC) No 1606/2002 of the European parliament, all listed EU firms are required to report their consolidated financial statements as of January 1, 2015, in accordance with IFRS as adopted by the International Accounting Standards Board (IASB). As of yet, there is no standard for reporting IC. However, taking into account the nature and structure of IC, IAS 38 – Intangible assets can be considered as the most relevant standard.

IAS 38 is applicable for all intangible assets other than financial assets, exploration and evaluation assets, expenditure on the development and extraction of minerals, oil, natural gas, and similar resources, intangible assets arising from insurance contracts issued by insurance companies, and intangible assets covered by another IFRS, such as IFRS 5, IAS 12, IAS 17, IAS 19, or IFRS 3 (IAS 38.1).

The standard states several criteria concerning the classification of intangible assets. IAS 38 requires the intangible asset to be identifiable non-monetary asset without physical substance, controlled by the entity as a result of past events, and from which future economic benefits are expected.

Furthermore, IAS 38 allows recognition of internally and externally generated intangible assets only when it is probable that future economic benefits that are attributable to the asset will flow to the entity and if the cost of the asset can be measured reliably. Besides the two criteria mentioned, internally generated assets have to meet additional requirements in order to be presented on the balance sheet. The probability of future economic benefits must be based on reasonable and tolerable assumptions on the conditions concerning the existing life of the asset (IAS 38.22). If the previously mentioned criteria are not met, the expenditures of this asset should be recognized as an expense when they are incurred (IAS 38.68).

According to Mouritsen, Bukh, and Marr (2004) IC often fails to meet requirements set in IAS 38, due to its high uncertainty, low seperability, and hard identifiability, and hence is quite often not recognized as an asset. Failing to meet the requirements means that the firm will have to recognize expenses associated with IC in the income statement, while future cash flows generated by these “assets” are not accounted for. As such, the current standard fails to provide investors and other users of the financial statements with relevant information.

However, IC that does not meet the criteria set in IAS 38 can under certain circumstances be allowed to be recognized as an asset. Such a circumstance arises in the acquisition of these assets. Generally speaking, firms do not tend to trade IC on the market, however, in the case of an acquisition, where one company acquires another company, the value of IC indirectly finds its way on the balance sheet. More often than not, the acquirer pays a premium on top of the net assets of the company it acquires. This premium is reported on the balance sheet as goodwill. Under IFRS 3, the relevant standard for goodwill, the probability recognition criterion is always considered to be satisfied for intangible assets that are acquired separately or in a business combination, and the fair value of an intangible asset acquired in a business combination can normally be measured with sufficient reliability to be recognized separately from goodwill. Thus, under mergers and acquisitions, intangible asset are easily reported on the balance sheet.

2.4.2 Internally generated Intangible Assets

Under the current accounting framework of IFRS, there is a discrepancy between internally generated and externally acquired intangible assets. Intangible assets are separated into two phases, namely the research phase and the development phase (IAS 38.52). Intangible assets designated at the research phase are not recognized, and are recorded as an expense when incurred. An entity cannot demonstrate that an intangible asset at this phase exists and will generate probable future economic benefits (IAS 38.54). Intangible assets arising from the development phase need to fulfill certain criteria before they can be recognized. According to IAS 38.57 intangible assets arising from the development phase of an internal project can be recognized if the firm can demonstrate: (1) technical feasibility, (2) intention to complete and use or sell the asset, (3) ability to use or sell the asset, (4) how the intangible asset will generate probable future economic benefits, (5) availability of adequate resources to complete the development and, use or sell the intangible asset, and finally (6) the reliable measurement of the expenses attributable to the development of the intangible asset.

The criteria set forth in IAS 38.57 leads to conservative accounting of intangible assets. While a firm incurs the expenses spent on generating, maintaining, and improving intangible assets, it is not able to financially portray the potential benefits of this investment to investors, or other users of its financial statements. Furthermore, there are certain internally generated intangible assets that cannot meet the criteria set in IAS 38. For example, corporate culture cannot be separated from the firm and cannot be sold. Same could be said for human capital, advertising costs (as specifically mentioned in IAS 38.69), training expenses, and research costs. The main arguments against the recognition of internally 13

generated intangible assets are the difficulty in reliably measuring the value such assets, and the uncertainty regarding the control of the future economic benefits arising from these assets. Additionally, IAS 18.63 specifically prohibits the recognition of internally generated brands, mastheads, publishing titles, customer lists and items similar in substance.

2.4.3 Usefulness of Financial Information

Lev and Zamorin (1999) provide evidence indicating that the usefulness of financial information has deteriorated over the past 20 years. They partially attribute this change to the failure of financial reporting to adequately reflect the impact of large investments that drive a firms operations and economic conditions. Specifically, Lev and Zamorin (1999) argue that the fundamental accounting principle that ensures costs are periodically matched with revenues is distorted, as R&D investments are recorded as expenditures when incurred, and are not matched with the benefits of this investment in the future. This has created an information gap between the user and the issuer of information. Moreover, Bukh and Johanson (2003) state that this information gap can be attributed to the increased request for non-financial information regarding intangible assets, and the lack of an accounting framework to organize and structure this non-financial information. Additionally as stated in Cerbioni and Parbonetti (2007), Lambert (2001) also consider that firms have shown more interest in non-financial information, mainly with respect to non-non-financial performance measures.

Lev and Zamorin (1999) suggest that financial statements should provide more timely and reliable information on intangible assets. However, as stated in Li et al. (2008), Canibano et al. (2000) argue that the costs associated with reworking the current accounting standards in order to provide more grounds for the disclosure of intangible assets will be very costly for IC intensive firms, and that the overall benefit of such a change will not offset the associated costs. Canibano et al. (2000) suggest that voluntary disclosure is considered a more appropriate approach to tackle the current limitation in accounting standards for intangible assets.

3 Literature Review

As highlighted previously, modern firms and economies are driven by intellectual and human capital. The existing literature provides a wide range of definitions for IC. The definitions range from very broad ones such as: the difference between market and book value, termed as hidden value by Edvinsson (1997), to specific ones such as: “intellectual capital is intellectual material – knowledge, information, intellectual property, experience that can be put to create wealth” (Stewart, 1997). The definition of intangible assets provided by IAS 38 is very limited compared to these. IAS 38 provides a very narrow definition, elements such as human capital, customer loyalty, and reputation are not included due to obvious measurement complications.

The shortcoming of financial reporting standards and the need for a better understanding of IC has led to the development of several IC frameworks. Brennan and Connell (2000) provide an overview of the main frameworks developed to classify IC from a managerial perspective. These classification schemes will be shortly discussed.

Kaplan and North (1992) have created the balanced scorecard framework. This framework provided organizations with a more complete measurement of performance by combining financial performance indicators with operational measures on “customer satisfaction, internal processes, and the organizations innovation and improvement activities operational measures that are the drivers of future financial performance”.

Skandia, a Swedish firm providing financial services, disclosed “hidden” IC assets in the business in a supplement to their 1995 interim report (Edvinsson, 1997). The aim of this additional report was to take a step toward a balanced annual report, where financial measures are completed with non-financial information. Skandia has taken a strategic path to successfully manage IC, in order to maximize its value adding potential. This strategic decision eventually led to the development of the Skandia value scheme framework, where IC is defined as structural capital and human capital (Edvinsson and Malone, 1997). According to Skandia, a company’s market value is driven by financial capital and intellectual capital. The framework they developed considers intellectual capital as a result of combining human capital and structural capital. Structural capital is further broken down into customer capital and organizational capital.

Sveiby (1997) developed the so called intangible asset monitor, one of the most popular frameworks used in IC literature. This framework distinguishes IC into the following categories, namely internal structure, external structure, and individual competence.

The internal structure category refers to internally generated or brought in items, such as patents, concepts, models research and development, and computed and administrative systems. Since these assets are generated internally (by employees in the firm) or acquired, firms can decide to replace or invest these assets with reasonable certainty (Guthrie and Petty, 2000).

The external structure encompasses the relationships of the firm with its customers and suppliers. One can think of assets such as brand names, trademarks, and firm reputation. Unlike internal structures, external structures are relatively difficult to manage. A firm’s reputation and relationship with its clients is susceptible to change over time.

Employee competence encompasses elements such as education, skills, experience etc. While employee competence is a significant driver in the value and competitive edge of an organization, it cannot be owned by anyone except the individual who possesses it. Sveibby (1997) argues that employee competence should be measured and placed on the balance sheet, as in the so called “knowledge organizations” human capital has replaced machines and hence should be part of the firm’s assets (Guthrie and Petty, 2000).

Although both the balanced scorecard and the intangible asset monitor divide IC into three categories and emphasize the need to combine financial and non-financial measures, according to Guthrie and Petty (2000) there are major differences between these different frameworks used to classify IC. While Sveiby (1997) attempts to redefine the firm from a knowledge perspective, the balanced scorecard maintains the classical definition of the firm. Also, Sveiby (1997) assumes people as the sole sources of value and profit for the firm, this assumption is however not shared by the balanced scorecard framework.

3.2 Voluntary IC disclosure

In 1994 IC disclosure became an interesting research topic, when Swedish financial services firm Skandia disclosed “hidden” IC assets in the business in their annual report. Additionally, Sveiby’s framework set the foundation of IC disclosure measurement, which eventually led to the emergence of IC disclosure research. Guthrie and Petty (2000) further developed this framework by creating of shortlist of IC indicators within the three IC classifications in Sveiby’s framework. This framework is highly accepted and frequently used in content analysis of IC disclosure in annual reports (Brennan, 2001; Guthrie and Petty, 2000; Vandemaele et al., 2005; Bozzolan et al., 2003; April et al., 2003; Goh and Lim, 2004; Abeysekara and Guthrie, 2005; Bruggen et al., 2009).

One of the first studies on IC disclosures in annual reports was published by Guthrie et al. (1999). The study focused on the IC disclosure in the annual reports of publicly listed Australian companies in 6 industry groups. Guthrie et al. (1999) conclude that there is no accounting for IC. Utilizing the same methodology as Guthrie and Petty (2000), Brennan (2001) examined 11 knowledge-based Irish listed companies. Despite the fact that these companies had a substantial level of non-physical, intangible IC assets (deduced from the fact that firms in their data had significant differences in market and book values), Brennan (2001) concluded that the level of IC disclosure was low. Goh and Lim (2004) studied the IC disclosure practices of the top 20 profit-making listed

companies in Malaysia, by quantitatively and qualitatively analyzing their annual reports of 2001. The study concluded that the IC disclosure in the annual reports were qualitatively high, but

quantitatively low.

Contrary to the above mentioned studies, Bontis (2003) performed a large scale content analysis on the annual reports of 10,000 Canadian corporations. Despite the strong impact of IC on the drivers of future earnings, limited disclosure of IC was identified in the annual reports. Out of the 10,000 annual reports examined, merely 68 corporations reported on IC.

Ismail and Hamzah (2008) examined the voluntary disclosure of IC reporting in Egyptian

companies’ annual reports. The study concluded that the level of voluntary disclosure was low, and like Goh and Lim (2004), the disclosures are more qualitative than quantitative.

3.3 Determinants of voluntary disclosure of IC

According to Williams (2001), leverage, industry exposure, and listing status have an effect on the level of IC disclosure in the annual report of 31 FTSE 100 listed firms in the UK from 1996-2000. Beaulieu et al. (2002) and Garcia-Meca et al. (2005) find that company size has a positive relationship on the amount of IC disclosure.

Using the framework of Guthrie and Petty (2000), Bozzolan et al. (2003) have performed a content analysis on 30 non-financial firms listed on the Italian Stock exchange in 2001. The study concludes that firm size and industry have an influence on the level of IC disclosure in the annual report of Italian firms. However, the study notes that the determinants do not explain the content of IC disclosed.

Contrary to the above mentioned indicators influencing the level of IC disclosure in annual reports, Chaminade and Roberts (2003) argue that culture might have an effect on IC reporting and that differences in culture need to be assessed before establishing international guidelines for IC

disclosure. The study performs a comparative analysis between the implementation of IC disclosure systems in Spain and Norway. In line with the findings of this study, Abeysekara (2007) suggest that IC disclosures systems are influenced by social, political, and economic factors.

Bruggen et al. (2009) examined the determinants of the decision to disclose IC by performing a content analysis on the annual reports of 125 publicly listed Australian firms. The study concludes that industry type and firm size are key determinants of IC disclosure in annual reports. The study also tests whether information asymmetry plays a role in determining the level of IC disclosure. Contrary to the theoretical predictions (as discussed in the theoretical framework), Bruggen et al. (2009) find no significant relationship between the level of information asymmetry and the level of IC disclosure in annual reports.

3.4 Corporate governance mechanisms and (IC) disclosure

Previous literature has mainly focused on the relationship between corporate governance and voluntary disclosure. However, the research conducted on the relationship between corporate governance and IC disclosure in particular, is limited. The following studies provide a good overview of research examining this relationship between these two control mechanisms that are supposed to mitigate the agency conflict between shareholders and managers.

.

Cheng and Courtenay (2006) examine the relationship between board monitoring and the level of voluntary disclosure. The study examines the association between the corporate governance variables: board independence, board size, and CEO duality, and voluntary disclosure. A significant positive relationship is observed for board independence, while no evidence is provided for a significant relationship for the variables CEO duality and board size.

Cerbioni and Parbonetti (2007) investigate the association between the corporate governance variables and voluntary IC disclosure for biotechnology firms in 10 EU countries. They extend previous literature by additionally analyzing board dimension and board structure (nominating, audit and compensation committees are considered) as independent variables. The findings conclude that there is a significant relationship between corporate governance variables and voluntary IC disclosure, specifically for the independent variables board independence, CEO duality, and board structure. Furthermore, Li et al. (2008) examine the relationship between corporate governance mechanisms and IC disclosure for 100 UK listed public companies. The study builds on previous literature by investigating more variables for corporate governance. Specifically, the corporate governance factors considered are Board composition, ownership structure, internal auditing mechanism (size of audit committee and frequency of meetings), and role duality. The study concludes significant relationships with all independent variables except for role duality.

Hidalgo et al. (2011) enrich the literature further by examining even more corporate governance mechanisms, and by focusing on the Mexican context they deviate from the conventional framework of the Anglo-Saxon corporate governance structure which dominates previous literature on this topic. The study corroborate the view that institutional investor shareholding have a negative impact on the level of voluntary disclosure of IC. Furthermore, the study concludes that a board size of up to 15 members has a positive relationship with on disclosure of IC.

4 Development of hypotheses

The board of directors are a form of internal control mechanism designed to monitor the behavior and activities of the management team, and are responsible for advising and monitoring managers. The structure of the board of directors plays a key role in the monitoring of managers. To be effective in their monitoring role, directors need to be sufficiently knowledgeable and independent from the CEO and the management team. A distinction is made between inside and outside directors. Inside directors are considered to have a tie to the firm and are the people the firm employs, while outside directors have no other affiliation to the firm besides being on the board. Independent directors can act as a control mechanism to alleviate the agency conflict between shareholders and managers by means of voluntary disclosure of information (Fama and Jensen, 1983). As inside directors possess valuable firm-specific information, they can act as an important factor in effective decision making. However, inside directors lack independence, as such they may not be inclined to share this specific information if they believe it will lead to interference of outside directors with the strategic decisions of the CEO (Armstrong et al., 2010). Evidently there is a tradeoff to be made between inside and outside directors (referentie?).

Prior research suggests mixed empirical results. Eng and Mak (2003) examine the relationship between ownership structure and board composition on voluntary disclosure, and suggest a positive relationship. Furthermore, Chen and Jaggi (2000) find a positive relationship between independent directors and the comprehensiveness of information in disclosures. Likewise, Li et al. (2008) and Patelli and Prencipe (2007) also find that independent directors are positively related to voluntary disclosure of IC. On the other hand, Haniffa and Cooke (2005) find a negative relationship, and others no relationship at all (Ho and Wong, 2001; Brammer and Pavelin, 2006). The study argues that both corporate governance and voluntary disclosure of IC will coexist, therefore:

H1: An increase in the proportion of independent directors to the total number of directors in the

board is positively related to voluntary IC disclosure.

Board size is an important measure on the efficiency and supervision of management conduct. However, academic literature is unclear with respect to the superiority of certain board compositions and structures. A higher board size may be considered beneficial in the sense that more members will increase the expertise and resources available to the organization. Moreover, a larger board size may enrich an organizations solutions to the challenges it faces by providing a larger diversity of perspectives. However, larger boards may be lead to inefficiency costs through poorer communication and increased decision making time (Hidalgo et al., 2011). Therefore the hypothesis is as follows:

H2: Board size is associated with voluntary IC disclosure

4.2 Audit Committee

Besides the non-executive members of the board, and members of the supervisory board in a two tier board structure, subcommittees of the (supervisory) board are also responsible for the monitoring of important processes and decisions made by the board. An important subcommittee is the audit committee. The audit committee comprises financial literate members of the board or the supervisory board in a two-tier structure. The audit committee is responsible for ensuring that the board acts in the interest of the shareholders by monitoring the board’s judgments related to the organizations financial reporting and internal control. Additionally, the audit committee is responsible for the appointment of the external auditor, and receipt of both the internal and external audit reports. According to Li et al. (2008) an effective audit committee can diminish agency costs, and increase the value relevance of IC disclosure. Likewise, Barako et al. (2006) conclude that the presence of an audit committee is a significant factor explaining the level of voluntary disclosure of firms listed on the Nairobi stock exchange (NSE).

Therefore the research constructs the following hypothesis:

H3: Audit committee independence is positively related to voluntary intellectual capital disclosure.

4.3 Ownership structure

Agency theory suggests that the ownership structure of a firm will influence the level of monitoring and level of voluntary disclosure, as the power of shareholders on the board is a function of the resources they control. Higher ownership diffusion will lead to higher pressure from shareholders for the disclosure to reduce agency costs as a result of information asymmetry. Whereas, in a setting with high shareholder concentration less information asymmetry exists between the shareholders and

management (Hidalgo et al., 2011). Shareholders with concentrated ownership generally have access to the information they need in order to actively monitor the management. As a consequence, less information will be disclosed to the market, as such, the research hypothesizes that:

H4: Concentrated ownership is negatively related to voluntary IC disclosure.

5 Methodology

The study aims to examine the relationship between two control mechanisms, namely corporate governance and voluntary IC disclosure, aimed to mitigate the agency conflict arising from the separation of ownership and control. The study will conduct an empirical analysis on a sample of companies listed on the Amsterdam Euronext Stock exchange. No similar empirical study has been carried out with respect to the Netherlands. As such the effect of the institutional setting in the Netherlands can provide the current literature with insights for future research. In order to gain a better understanding of the empirical results, it is important to provide a general overview of the legal regulation and the development of corporate governance in the Netherlands.

5.2 Development of Corporate Governance in the Netherlands

Awareness of Corporate Governance issues and the Dutch Corporate Governance environment has undergone a remarkable development over the past two decades. During the 1990s a public debate with respect to measures against hostile takeovers sparked the need for best practice corporate governance recommendations for listed organizations in the Netherlands. The increased public awareness with reference to corporate governance eventually led to the development of self-regulatory initiatives in the form of corporate governance codes (Hooghiemstra and van Ees, 2011).

The first corporate code in the Netherlands is the 40 recommendations of the Peters Committee, named after its chairman. The aim of this committee was “to increase the effectiveness of management, supervision and accountability to investors of Dutch companies” (Akkermans et al., 2007 p. 1107). Despite the promising goal of this code, its reliance on self-enforcement formed an impediment to its success. The Peters Committee was optimistic in assuming that self-enforcement through market forces would lead to high compliance by Dutch organizations. As concluded in De Jong and Roosenboom (2002) the disclosure of compliance in reference to the 40 recommendations for Dutch firms in the period 1997-2002 was scarce. Furthermore, according to De Jong et al. (2005) the introduction of corporate governance code, specifically the Peters committee in the Netherlands, did not have any influence on the relationship between firm value and corporate governance code. De Jong et al. (2005) remain skeptical and suggest that initiatives that rely on self-enforcement generally

do not succeed, and the 40 recommendations of the Peters Committee provides evidence for this statement.

In light of several Dutch corporate scandals such as the misleading financial statements of Royal Ahold’s US subsidiary and the overstatement of oil reserves by Royal Dutch Shell, and the low compliance of the 40 recommendations of the Peters committee, provided grounds for improvement of the corporate governance framework in the Netherlands and eventually led to several changes in Book 2 of the Dutch Civil Code (DCC). The general tendency of the changes were aimed at strengthening the shareholders rights (Cziraki et al., 2013).

One of the changes is the establishment of a new corporate governance code. On the initiative of the Dutch ministry of Finance, Euronext Amsterdam, listed firms, a delegation of shareholders, and institutional investors, the Tabaksblat Committee was established to create a new corporate governance code in 2003. The new code is based on a set of 21 best-practice principles concerning the conduct of the management board and supervisory board, general meeting of shareholders, and financial reporting and the auditing process. Contrary to the previous code, this one is based on the comply-or-explain principle. The listed firm is required to include a compliance statement in its annual report; if the firm deviates from the code it needs to explain its reasoning for non-compliance. Despite the legal requirement on the comply-or-explain principle, the legal enforcement remains ineffective (Hooghiemstra and van Ees, 2011).

Furthermore, several changes were implemented in Book 2 DCC. Firstly, the shareholders meeting received the authority to approve transactions with a material impact on the company. Secondly, shareholders (including holders of depository receipts) received the right to provide input to the agenda of the general meeting of shareholders, provided that they own at least 1% of the total share capital or shares with a market value of EUR 50 million. Thirdly, the general meeting received the right to appoint supervisory board members and dismiss the whole board given a majority of votes, and the right to adopt remuneration policies for directors (Bekkum et al., 2010).

5.3 Legal regulation

The Netherlands is classified as a civil law (more specifically French civil law) country. Contrary to common law countries, civil law countries are associated with weaker investor protection rights. According to La Porta et al. (1998) French civil law countries, such as the Netherlands, provide the weakest protection to shareholders independent of per capita income. According to La Porta et al 24

(1998), the Netherlands is given a score of 2 in a scale ranging from 1 to 5 for Anti-director rights. The strong legal enforcement system in the Netherlands partially compensates the weak legal protection rights for shareholder (La Porta et al., 1998). Companies in French civil law countries characterized by weak legal investor rights generally have a higher concentrated ownership of shares. Increased ownership concentration can be seen as a response by shareholders to the weak legal rights they have. According to La Porta et al. (1998) this characteristic is also valid for the Netherlands. 5.4 Board structure in the Netherlands

There are two types of board systems, namely the tier system and the two-tier system. In the one-tier system members responsible with the day-to-day activities (executive directors) and those responsible for the independent supervision of the former group (non-executive directors) are seated on the same board of directors. In a two-tier system, the executive directors and non-executive directors are split into the management board (executives) and supervisory board (non-executives). Listed companies in the Netherlands predominately have a two-tier system.

Companies that have more than 100 employees, with a legally installed works council, and equity in excess of 16 million are required to have a two-tier structure. As such, the majority of listed firms in the Netherlands have a two-tier board structure. According to the Dutch Corporate Governance Code the supervisory board needs to consist of independent members.

5.5 Sample selection and data source

The research has used the Thomson Reuters ASSET4 ESG database for the independent variable. The independent variables have also been extracted from the ASSET4 database, except for one variable measuring ownership concentration. Datastream is used for the control variables and the independent variable measuring ownership concentration.

The Thomson Reuters ASSET4 ESG Data provides environmental, social, economic and governance information, based on key performance indicators (250) and individual data points (approximately 700). Analysts gather data using publicly available sources, such as annual reports, CSR reports, and NGO websites. The key performance indicators are distributed into eighteen categories, which are then grouped under the four pillars, namely economic performance, environmental performance, social performance, and corporate governance performance. The database allows one to rate and compare firms by providing a category score (taking into account the score of each key performance 25

indicator), pillar score (taking into account each category scores), and an overall score (taking into account each pillar score). The ASSET4 database provides a relative scoring measure; each score is calculated through equally weighting and z-scoring the underlying data points and benchmarking it to all companies in the ASSET4 database.

The research aims to provide evidence for the relationship between intellectual capital and corporate governance mechanisms in the Dutch context. Data for a sample of 46 firms listed on the Euronext Amsterdam stock exchange over the period 2002-2015 was extracted from the Thomson Reuters ESG ASSET4 database.

5.6 Measurement of Dependent and Independent variables 5.6.1 Evaluating the quantity and quality of IC disclosure

The dependent variable of this study is IC disclosure. The majority of prior research examining the determinants of IC disclosure have opted to use content analysis to measure the extent of IC disclosure in annual reports (Beattie and Thomson, 2007). According to Haniffa and Cooke (2005, p. 404) content analysis is “a method of codifying the text (or content) of a piece of writing into various groups (or categories) depending on selected criteria”. As such, the selection and

development of categories into which content units can be placed is essential in content analysis (Haniffa and Cooke, 2005; Li et al., 2011; Cerbioni and Parbonetti, 2007).

When choosing content analysis, many authors have adopted the IC framework developed by Sveiby (1997). This framework classifies IC into three distinct categories, namely internal capital, external capital, and human capital. The first column contains internal capital terms. These are IC created internally by the firm and its employees, which the firm has of disposal. The second column contains indicators for external capital. Finally, column three contains indicators for employee competence. This is IC which the firm holds temporarily. When an employee leaves the firm, the knowledge that the individual possesses also partially leaves the firm. Refer to table 1 for a modified classification scheme of this framework as used by Guthrie and Petty (2000), Bozzolan et al. (2003), and Cerbioni and Parbonetti (2007). By means of a scoring methodology, IC disclosure is measured based on the frequency and disclosure of the 24 IC indicators/content elements categorized under the three classifications.

Figure 1

1. Internal Capital: Organizational (Structural) 2. External Capital: Customer (Relational) 3. Human Capital: Employee competence

1.1. Intellectual Property · Brands · Know-how

· Patents · Customers · Education

· Copyrights · Customer loyalty · Vocational qualification · Trademarks · Company names · Work-related knowledge 1.2. Infrastructure Assets · Distribution channels ·Work-related competencies · Management philosophy · Business collaborations · Entrepreneurial spirit,

innovativeness, proactive and reactive abilities, changeability.

· Corporate culture · Licensing agreements · Management processes · Favourable contracts · Information systems · Franchising agreements

· Networking systems

· Financial relations

Note: This is the IC classification scheme as derived from Guthrie & Petty (2000) Deviating from prior literature, this research has opted to measure the quality and quantity of IC disclosure by selecting appropriate variables in the ASSET4 database. The classification scheme above is used as a guideline for selecting appropriate variables. The social category score SOTD – Workforce/Training and Development (there are in total seven categorical scores under the social pillar) is selected as an appropriate measure for IC disclosure. The ASSET4 ESG data glossary provides the following description for this variable:

“The workforce/training and development category measures a company's management commitment and effectiveness towards providing training and development (education) for its workforce. It reflects a company's capacity to increase its intellectual capital, workforce loyalty and productivity by developing the workforce's skills, competences, employability and careers in an entrepreneurial environment.”

Based on this description the SOTD variable hits several content elements and IC indicators in the classification scheme presented in figure 1, namely the following elements under the category human capital: education, know-how, work-related knowledge, work-related competencies and entrepreneurial spirit. The variable provides a measure in percentages, with the lowest being 0% and the highest being 100%.

Furthermore, the variable ENPI – product innovation is selected as the second variable to measure the quantity and quality of IC disclosure. The ASSET4 ESG data glossary provides the following description for this variable:

“The product innovation category measures a company's management commitment and effectiveness towards supporting the research and development of eco-efficient products or services. It reflects a company's capacity to reduce the environmental costs and burdens for its customers, and thereby creating new market opportunities through new environmental technologies and processes or eco-designed, dematerialized products with extended durability.”

Based on this description the variable ENPI hits several content elements and IC indicators in the classification scheme presented in figure 1, namely the following elements under the category human capital: entrepreneurial spirit, innovativeness, proactive and reactive abilities, changeability.

Moreover, the variable BV – total value of the company’s brands is selected as a third proxy variable to measure the quantity and quality of voluntary IC disclosure. The BV variable hits the elements brands under the external capital category, and intellectual property under the internal capital category. 5.6.2 Corporate Governance variables

Similar to the dependent variable, the independent variables for corporate governance mechanisms were selected from the ASSET4 ESG Database. Board independence (IndBoardscore) was measured as the percentage of independent board members as reported by the company. This variable provides a relative score measure in percentages, with the lowest being 0% and the highest being 100%. Board size (Boardsize) was measured as the total number of board members in excess of 10 or below eight using a relative score variable. Audit committee independence (IndACscore) was measured as the percentage of independent board members on the audit committee as stipulated by the company using a score variable. Concentrated ownership (OwnConc) was measured using a variable measuring closely held shares.

5.6.3 Control Variables

Prior research studying the relationship between corporate governance mechanisms and voluntary disclosure is not consistent on the selection of control variables. The following control variables have been used in previous studies: firm size, leverage, profitability, firm age, length of listing, legal enforcement, ownership structure and audit firm size. According to the meta-analysis conducted by Ahmed and Courtis (1999) on the relationship between corporate characteristics and disclosure levels in annual reports, firm size and listing status are significantly associated with disclosure levels, while profitability, leverage, and audit firm size had mixed results.

A widely used control variable is firm size, measured either as total assets (Cerbioni and Parbonetti, 2007; Ho and Wong, 2001; Hidalgo, 2010), sales revenue (Li et al., 2008; Patelli and Prencipe, 2007) or market value of common stock (Cheng and Courtenay, 2006). Agency theory predicts that there is a positive relationship between firm size and disclosure. Larger firms tend to have a higher proportion of capital held by outsiders, thus leading to higher agency costs and likewise higher incentives for corporate disclosures (Hossain et al., 1995; Watts and Zimmerman, 1981). Furthermore, according to Beaulieu et al. (2002) and Garcia-Meca et al. (2005) firm size has a positive relationship on the amount of IC disclosure.

Moreover, larger firms tend to be more leveraged and consequently, incur higher monitoring costs. Shareholders and managers, have a higher incentive to disclose information in order to reduce the information asymmetry with their debt holders (Jensen and Meckling, 1976). Prior research has widely used leverage as a control variable, however several different measurement methods have been used. Leverage is measured as total debt over total assets (Hidalgo et al., 2010), debt over total book value of equity (Cerbioni and Parbonetti, 2007; Ho and Wong, 2001), long-term debt over equity (Cheng and Courtenay, 2006), and equity over total assets (Patelli and Prencipe, 2007).

Furthermore, profitability is also related to voluntary disclosure. When firms have a higher profitability, managers have an incentive to disclose detailed information to support their growth (Raffournier, 1995). Likewise, when firms have a higher profitability managers are more inclined to disclose less information to conceal the reasons for the lower growth. Additionally, the corporate control content hypothesis suggests that managers who risk losing their job as a result of low earnings

performance, use corporate disclosures to “lower the likelihood of undervaluation and explain away poor earnings performance” (Healy & Palepu, 2001, p. 421). Prior research have measured profitability as return on equity (Cerbioni and Parbonett, 2007; Hidalgo et al., 2010), return on assets (Cheng and Courtenay, 2006; Li et al., 2008), return on capital employed (Ho and Wong, 2001) and operating income divided by total assets (Patelli and Prencipe, 2007).

Based on prior research and the above considerations, the study has selected firm size, leverage and profitability as control variables in order to isolate the relationship between corporate governance mechanisms and IC disclosure. Following Cerbioni and Parbonetti (2007), Ho and Wong (2001) Hidalgo, 2010) this study uses total assets as a measure for firm size. Furthermore, the study measures leverage as total debt over total book value of equity. Finally, following Cerbioni and Parbonetti (2007) and Hidalgo et al. (2010) the study uses return on equity as a measure for profitability.

Table 1

Measurement of dependent and independent variables

Variable Proxy Measurement1

Dependent variables - Voluntary Intellectual Capital Disclosure

1 SOTD Workforce/Training and Development

The workforce/training and development category measures a company's management commitment and effectiveness towards providing training and development (education) for its workforce. It reflects a company's capacity to increase its intellectual capital, workforce loyalty and productivity by developing the workforce's skills, competences, employability and careers in an entrepreneurial environment.

2 ENPI Product Innovation

The product innovation category measures a company's management commitment and effectiveness towards supporting the research and development of eco-efficient products or services. It reflects a company's capacity to reduce the environmental costs and burdens for its customers, and thereby creating new market opportunities through new

environmental technologies and processes or eco-designed, dematerialized products with

extended durability.

3 LnBV Brand Value Natural logarithm of the total value of the company's brands in US dollars.

Independent variables - Corporate governance mechanisms

1 Board Composition Board independence Percentage of independent board members as reported by the company. 2 Board Composition Board Size Total number of board members which are in excess of ten or below eight. 3 Internal Auditing Mechanism Audit Committee independence (IndACscore) Percentage of independent board members on the audit committee as stipulated by the company. 4 Ownership Structure Ownership Concentration (OwnCon) Closely held shares - shares held by insiders (WC05475).

Control Variables

1 Firm Size Total assets (Size) Natural logarithm of total assets (WC02999). 2 Profitability Return on Equity (RoE) Net income (WC01551) over common equity (WC03501). 3 Indebtness Leverage Ratio Total debt (WC03255) over common equity (WC03501). Note: This table presents an overview of all variables.

1 _{Measurement descriptions have been deducted from the Thomson Reuters ESG Asset4 Glossary.}

31

5.7 Data analysis

The study made use of the ASSET4 database and Datastream. Using excel, data obtained from both databases were combined and organized in the long format. The data used for this study can be described as panel data: observations over multiple governance and IC disclosure related variables, obtained over the period 2002-2015 for AEX listed firms. The panel data used in this study does not contain observations for all variables for each firm in each year as such the data can be defined as unbalanced panel data.

To test the hypotheses stated above with respect to the relationship between IC disclosure and corporate governance mechanisms the study opted for a multivariate OLS regression model. The study opted to run the multivariate regressing using three different variables as a measure for IC disclosure. To ensure the validity of the OLS model tests for homoscedasticity, linearity and normality assumptions were conducted. Furthermore, tests for multicollinearity were conducted by means of a Spearman and Pearson correlation matrix, and computing the variance inflation factors (VIF). Firstly, testing individually the linearity between the dependent variables and the independent variables revealed that only the independent variable Boardsize portrayed a linear relationship. Secondly, plotting the residuals of the model in a histogram revealed that the residuals are not normally distributed. Additionally, numerical tests for normality of residuals were used, such as the Shapiro-Wilk. These tests also indicated signs of non-normality in the distribution of the residuals, nonetheless, the residuals do fulfil the condition mean zero assumption. Thirdly, the homogeneity of the residuals were tested by plotting the residuals of the regression model against predicted values. This visual test revealed that the residuals do not have a constant variance and that the data points get narrower towards the right end. This is an indication of hetereoskedasticity.

To correct for the OLS assumption violations appropriate transformations were conducted to improve the distribution of the residuals and the linearity of the dependent variable with the independent variables. Despite the transformations, the plot of the residuals in a histogram together with a normal distribution density line reveals that the distribution of the residuals is slightly skewed to the left and kurtosis still appears to be present. Furthermore, the transformations do seem to increase the linearity of certain independent variables with the dependent variable, but this is not the case for all of the independent variables. Considering the sample size of the model the study relies on

the central limit theorem and considers the non-normality of the residuals and non-linear relationship of certain independent variables with the dependent variable as appropriate for using OLS.

To adjust for heteroskedasticity the regression analysis will use robust standard errors. As the pattern of heterekoskedasticity is not known this seems a more efficient choice compared to using a weighted least squares (WLS).

Table 3a presents the correlation matrices for both Spearman and Pearson. All independent variables showed significance at 0.05 level for at least one IC disclosure measure, except for LnRoE under the Pearson correlation. After adjusting for size, leverage and profitability, the Spearman correlation matrix concludes that the correlation of the independent variables are equal to or below 0.35, with the exception of the high correlation between Size and BV. This high correlation is logical as Size is a measure of a company’s total assets and BV is a measure of brand value. In the BV regression model Size will not be include as a control variable in order not to inflate the coefficients and the R-squared of the model. Overall, the results of the Spearman correlation matrix suggests that there is no sign of significant multicollinearity of the independent variables.

Furthermore, based on the Pearson correlation matrix the study concludes that there is no sign of multicollinearity as the correlations of the independent variables do not describe a strong relationship. Independent variable associations are all equal to or below 0.30, with the exception for the relationship between the independent variables Size and BV. As discussed earlier, this high correlation is logical and will not be included as a control variable in the BV regression model.

As a final test for multicollinearity table table 4 presents the variance inflation factor’s (VIF) for each independent variable. The VIF’s for the independent variables are all lower than 2, with the exception for LnIndBoardscore that has a VIF of 2.15. According to Haniffa and Cooke (2005) multicollinearity can be seen as an issue in a regression model when the VIF exceeds 10. Based on this threshold, no significant sign of multicollinearity is apparent between the independent variables.

Table 2a

Descriptive statistics for dependent and independent variables (untransformed)

Obs Mean Median Std. Dev. Min Max

Dependent variables

SOTD (%) 503 69.7 82.9 27.4 5.6 97.4 ENPI (%) 503 61.9 68.1 30.3 12.4 99.7 Brand value (BV) (in million US dollars) 504 22,800 6,530 57,300 179 470,000

Independent variables

Corporte governance mechanisms

Board composition (%) (IndBoardscore) 477 66.9 85.2 34.5 1.3 95.0 Boardcomposition (%) (Boardsize) 503 57.1 60.3 18.2 1.4 78.1 Audit committee independence (%) (IndACscore) 480 57.4 69.8 23.1 0.0 71.3 Ownership concentration (# of closely held shares in thousands)

(OwnCon) 427 113 26 227 - 1,931

Control variables

Leverage (Lev) 501 1.3 0.7 3.1 -14.8 44.1 Profitability (RoE) 502 0.1 0.1 0.7 -9.5 4.6 Firm Size (Total Assets in million US dollars) (Size) 502 65 6 199 0 1,320

Notes (in brackets the datastream codes are noted):

This table presents descriptive statistics for proxies for voluntary IC disclosure, and corporate governance mechanisms. The proxy variables for voluntary IC disclosure comprises of SOTD, ENPI and BV. SOTD (SOTD) = category score variable for workforce/training and development. ENPI (ENPI) = category score variable for product innovation. BV (ECCLDP033) = datapoint measure for total value of the company's brands in US dollars. All three proxies for voluntary IC disclosure are retrieved from the Thomson Reuters ASSET4 ESG Database.

The proxy variables for corporate governance comrpise of, IndBoardscore, Boardsize, IndACscore and OwnCon. IndBoardscore (CGBSO07S) = a score variable measuring the percentage of independent board members as reported by the company. Boardsize (ASSET4 code: CGBSO01S) = score variable measuring the total number of board members which are in excess of ten or below eight. IndACscore (CGBFO01S) = score variable measuring the percentage of independent board members on the audit committee as reported by the company. OwnCon = is a proxy variable for corporate governance mechanism, specifically ownership concentration and measures the number of closely held shares in thousands. With the exception of ownership concentration, the proxy variables for corporate governance mechanisms are also derived from the Thomson Rueters ASSET4 ESG Database and the respective variable codes are included in brackets in this note. OwnCon is derived from Datastream, with reference WC05475. The control variables comprise of Lev, RoE and Size. Lev = a measure for firm indebtness/leverage, and is computed by dividing total debt (WC03255) over common equity (WC03501). RoE = a measure for firm profitability and is computed by dividing Net income (WC01551) over common equity (WC03501). Size = a measure for firm size and is computed by taking the natural logarithm of total assets (WC02999).

Table 2b

Dependent variables

ENPI SOTD

Score bins (%) observations % of total cumulative % observations % of total cumulative %

0-10 0 0% 0% 20 4% 4% 10-20 60 12% 12% 37 7% 11% 20-30 77 15% 27% 12 2% 14% 30-40 24 5% 32% 17 3% 17% 40-50 30 6% 38% 30 6% 23% 50-60 27 5% 43% 20 4% 27% 60-70 47 9% 53% 43 9% 36% 70-80 31 6% 59% 47 9% 45% 80-90 59 12% 71% 141 28% 73% 90-100 148 29% 100% 136 27% 100% Total 503 503

Notes: This table provides a further breakdown of the observations for the dependent variables ENPI and SOTD. Refer to table 2a above for a detailed description of these variables.