The effects of governance characteristics on disclosure quality: evidence from Dutch pension funds

(1)

Master Thesis, MSc Accountancy

University of Groningen, Faculty Economics and Business

The effects of governance

characteristics on disclosure quality:

evidence from Dutch pension funds

CHRIS FENNEMA Student number: s4170733 Jensinghelaan 5 9591 EB Onstwedde tel.: +31683701544 e-mail: c.fennema@student.rug.nl

Supervisor: drs. W. (Wilfred) Kevelam RA

ABSTRACT

The goal of this paper is to give insights into the recent developments of the disclosure qual ity of Dutch pension funds after the regulatory changes in 2014 and 2015. As governance plays a vital role in the quality of disclosure, various governance attributes are examined in this paper. Specifically, the board model, board size, presence of a supervisory board, board activity and board independence and their relationship with disclosure quality are assessed. By examining the disclosure quality of pension fund, a contribution to the existing literature of disclosure quality and governance is made, as pension funds, or any other financial institution, are usually left out due to their specific characteristics. Based on a sample of 238 pension fund-year observations for 80 different pension funds in the Netherlands between 2015 and 2019, statistically significant relations are found between the presence of a supervisory board, board independence and disclosure quality. Also, board size plays a significant moderating role in the relationship between board model and disclosure score. No significant results were found for the remaining variables.

Key words: Pension funds, governance, pension fund governance, board of directors, supervisory board, disclosure quality, Agency Theory, Stakeholder Theory, Legitimacy Theory, Signaling Theory

Word count: 11.287 (excl. tables) January 18, 2021

(2)

1. INTRODUCTION

1.1 Introduction

The Netherlands has been awarded for the best pension system in the world in 2018 and 2019 consecutively by Mercer (2019). Most of the Dutch citizens are dependent on their pensions, that they receive from the pension funds, for their retirement. Yet, the pension funds are in difficult times (Dagblad De Limburger, 2020). In the aftermath of the financial crisis and the European debt crisis, the pension funds have to deal with a low coverage ratios (Apostolakis et al., 2016; Kevelam & Laning, 2013). Consequently, the participants may need to pay more contributions next year to restore the coverage ratio of these pension funds. Also, the risk on cutting on the pension promises to pensioners, which is the ultimum remedium to restore the coverage ratio, is also a real possibility. These measures might impair the trust participants have in the pension funds. In order to restore this trust good (corporate) governance is required (Stewart & Yermo, 2008).

The topic of corporate governance has received a lot of attention by both academics and practitioners after the occurrence of corporate scandals at the beginning of this century (Cohen et al., 2004). Consequently, the international community started to question the quality of financial reporting (Klai & Omri, 2011; Agrawal & Chadha, 2005). Twenty years in the century and corporate scandals still seem to happen, as became apparent earlier this year that 2 billion euros went missing on Wirecard’s, a German fintech company, balance sheet. With around 730 employees fired and a remaining debt of 2.8bn euro’s, stakeholders of the company are seriously affected. With the former director on Interpol’s international most-wanted list as a fugitive, corporate governance and the quality of financial reporting remains undoubtedly an important topic to address.

Due to these corporate scandals, the financial crisis and the European debt crisis have resulted in initiatives to change regulations (Leuz & Wysocki, 2016). For the pension funds in the Netherlands this meant regulatory changes in 2014 and 2015. Specifically, this meant a new governance code in 2014 as well as a new law to strengthen the corporate governance for pension funds. Additionally, in 2014, the Financial Assessment Framework (FTK) of the pension funds was renewed to create more stability as an answer to the high volatility in the financial market, in which pension funds operate a lot. Finally, regulation regarding the transparency about the costs pension funds make was also introduced in 2015.

Based on the regulatory changes in 2014 and 2015 the goal of this study is to determine how the disclosure quality of Dutch pension funds developed in recent years. To determine this, approximately 240 annual reports from 80 different pension funds in the period between 2015 and 2019 are being analysed. Furthermore, as governance is still an important topic and some of the regulations focus on this topic, the purpose of this study is to assess the impact governance characteristics has on the disclosure quality of Dutch pension funds, resulting in the following research question of the study:

How did the disclosure quality of Dutch pension funds develop in recent years and how do pension fund governance characteristics influence this disclosure quality?

Effective governance is as important for pension funds as it is for normal corporations (Ambachtsheer et al., 2006). One of the most important roles of an organization’s governance system is to make sure that the quality of financial disclosure is sufficient (Cohen et al., 2004). Additionally, the role of the governance structure,

(5)

with respect to financial reporting, is to make sure that the financial statements are in compliance with the Generally Accepted Accounting Principles (GAAP) and that they are credible (Lin & Hwang, 2010). However, in contemporary times the quality of voluntary disclosure is also rising in importance (García-Meca & Sánchez-Ballesta, 2010). Voluntary disclosure is defined by the FASB as disclosures, primarily outside the

financial statements, that are not explicitly required by GAAP or specific country rules (FASB, 2001), such

as environmental, social and other types of disclosures (García-Meca & Sánchez-Ballesta, 2010).

The most important governance mechanism is the board (Baysinger & Butler, 1985; Fama & Jensen, 1983). The board remains responsible for monitoring and ensuring the quality of the information content in the financial statements (Jensen, 1993). In this study we will therefore examine the effects various board aspects have on the disclosure quality. Specifically, the following board characteristics of Dutch pension funds will be addressed: board model, board size, and whether board size has an moderating role, board activity, board independence. In addition, we will examine whether the presence of a supervisory board will have an influence on the disclosure quality, as the supervisory board supervises the board (Ran et al., 2015).

The mechanism of corporate governance and its significance in ensuring high quality financial reporting arises from agency theory (Habib & Jiang, 2015). Previous studies that have examined the relationship between board characteristics and disclosure have therefore often been using the agency theory (Vitolla et al., 2020; Frias‐Aceituno et al., 2013; Donnelly & Mulcahy, 2008; Barako et al., 2006). A key concept of the agency theory is information asymmetry that arises from the separation of ownership and control (An et al., 2011; Fama & Jensen, 1983). Information asymmetry occurs when some persons have more information than others. Consequently, these persons have an information advantage over others (Salehi et al., 2014). One way to reduce this information asymmetry is by disclosing more information (Healy & Palepu, 2001; Diamond & Verrecchia, 1991). Another theory that deals with the reduction of information asymmetry is the signaling theory (Connelly et al., 2011). This theory proposes that when organizations have superior quality over others, they have the incentive to signal their superior quality (An et al., 2011). One way to signal their superior quality is by disclosing it. However, according stakeholder theory, it is critical for organizations, and especially pension funds, to consider all stakeholders to keep legitimacy in order to operate (Michelon & Parbonetti, 2012). Stakeholder theory expands the agency theory by focusing on the relationship between the organization and its stakeholders (An et al., 2011). This theory is often used when researching the quality of disclosure (Dienes & Velte, 2016; Khan, 2010; Brammer & Pavelin, 2008), as it presents arguments for firms to be transparent in their communication with the variety of stakeholders they have (Winter & Zülch, 2019). Another theory that deals with the relationship between the organization and its stakeholders is the legitimacy theory (An et al., 2011). Disclosing information can be considered as a communication strate gy for an organization to maintain their legitimacy and influence society’s perceptions (Cho & Patten, 2007).

1.2 Contribution

The association between governance and disclosure in Europe is relatively new and thus scarce (Ali, 2010). By providing empirical evidence of the effects various governance attributes have on the disclo sure quality of Dutch pension funds I contribute to the scarce literature available on these topics. Usually, the association is examined in an American setting. Therefore, it is also interesting to examine what the association is in a European setting. Furthermore, this study will contribute to the limited research that has been done about the

(6)

role corporate governance has in reducing information asymmetry (Elbadry et al., 2015). By examining the effects governance characteristics has on the disclosure quality a contribution to the literature will be made, as the central objective of disclosure is to improve transparency and thus reduce information asymmetry and agency costs to satisfy all stakeholders (Jizi, 2017). A reduction in information asymmetry between the principal and agent will lead to less opportunity for the agent to behave opportunistically (Elbadry et al., 2015). This will result in less agency problems. Consequently, the confidence participants have in pension fund will rise.

This study focusses specifically on the disclosure quality of Dutch pension funds. Where the majority of the literature focus on governance in for-profit organizations, the attention paid to non-profit organizations (such as pension funds) is rather limited (Ambachtsheer et al., 2006; O’Regan & Oster, 2005). A pension fund can also be considered as a financial institution (de Lange, 2019), which are often omitted from research due to their specific characteristics (Ali & Abdelfettah, 2019). Examining them is therefore interesting, as they are usually left out. Furthermore, examining the effects of governance aspects on the disclosure quality is extra relevant for Dutch pension funds due to the regulatory governance changes in 2014. By examining the governance characteristics of the pension fund we also have a social contribution, as the quality of disclosure and the corresponding relevant determinants are of high interest to the beneficiaries (Stewart & Yermo, 2008).

By examining the effects of board activity, measured by frequency of meetings, we contribute to the existing knowledge as we provide insights in the processes board members use to exert their control (Beasley, 1996). However, despite its ease measuring it, the effect of frequency of board meetings has been studied very rarely (Dienes & Velte, 2016). According to Lagasio & Cucari (2019) there are still some hesitations regarding the frequency of board meetings. With examining the effects of frequency of board meetings on disclosure quality we fill up this void. Furthermore, the majority of the literature examine the effectiveness of governance in an Anglo-American setting, where the focus is on one-tier board structure, whereas the two-tier board structure is usually ignored (Jungmann, 2006). This inevitably results in limited literature available on the effectiveness of a supervisory board (Ran et al., 2015). By examining the effect of a supervisory board has on the quality of disclosure we contribute to this stream of literature.

1.3 Structure of the paper

The remainder of this paper is as follows. In Section 2 the theoretical structure, which serves as an basis for our empirical work and analysis, will be provided. This section will include a background of the literature, description of the underlying theories, description of the Dutch pension fund and their governance and finally a set of hypotheses. Section 3 will describe the research design of our study. In here the process o f data gathering, the sample and the measurement of our variables will be discussed. Thereafter, the descriptive statistics and the regression results are displayed and described in Section 4. The results will be interpretated in Section 5. This section includes a discussion, implications, limitations of the study and suggestions for future research.

(7)

2. THEORETICAL FRAMEWORK

In this section of the study the theoretical framework is addressed. First, a background of the existing literature is discussed. After that, a brief description of the underlying theories is given. Thereafter, the Dutch pension system and pension funds’ governance is described. Finally, based on the existing literature, underlying theories, and logical reasoning a set of hypotheses are formulated in the end of this section.

2.1 Background

Prior studies have shown that some aspects of the board can have an impact on the financial reporting quality of an organization (Johl et al., 2013). But rather defining “financial reporting quality” these studies focus on factors such as earnings management, earnings quality, financial restatements and fraud etc. as they break down the financial reporting process and therefore affect the quality of f inancial statements (Cohen et al., 2004). For example, Beasley (1996) showed, in his well-cited paper, that the board of fraud firms consist out of less outside and independent members than boards of no-fraud firms. In some cases the board delegates the oversight of the financial reporting process to the audit committee (Lin & Hwang, 2010). However, Beasley (1996) additionally showed that the composition of the board, rather than the presence of audit committee, is more important for reducing the occurrence of fraud. Consistent with this is the study from Klein (2002), Xie et al., (2003) and Kao & Chen (2004) who all concluded that board independence is negatively associated with earnings management. Contrary to this, Vafeas (2000) shows that earnings informativeness tends to decline when the proportion of outside board members increases. In another study by Bajra & Cadez (2018) the relationship between multiple board characteristics and earnings management were examined. They only found that board independence and board rotation are negatively associated with earnings management at significant levels. Other characteristics, such as board size, frequency of meetings and financial expertise were negative, but not statistically significant. However, Kao & Chen ( 2004) and Vafeas (2000) found positive relationship between board size and earnings management. Meaning that earnings management tends to increase when board size becomes larger. In addition, Xie et al., (2003) showed that more active boards (proxied by the number board meetings) are negatively associated with earnings management. Contrary to this, is the study from Johl et al., (2013) who concluded that variables: board independence, board size, board expertise and board meetings are not significantly related to earnings management. These results do suggest that boards might be ineffective in restraining earnings management.

In two French studies the relationship between board of director characteristics and the quality of disclosure have been examined. Ali (2010) concluded that firms with good disclosure are characterized by a higher proportion of outside directors. Ali & Abdelfettah (2019) came to similar conclusions, as they also conclude that the proportion of independent directors is positively related to the level of disclosure. In addition, they concluded that firms with smaller sized boards experience better quality in their financial disclosures. Consistent with this is the study from Chen & Jaggi (2000), who reported a positive relationship between independent directors and the comprehensiveness of financial disclosures. Cai et al., (2006) concluded as well that smaller sized boards are associated with a lower level of information asymmetry.

Other studies have also examined the relationship between several board characteristics and other types of disclosure, such as environmental and voluntary reporting (Healy & Palepu, 2001; Lagasio & Cucari, 2019). For example, with respect voluntary reporting, Haniffa & Cooke (2002) hypothesized that the proportion of

(8)

outside directors has a positive influence on the extent of voluntary disclosure, contrary to what they have hypothesized, they found a significant negative relationship. They found similar outcomes for corporate social disclosure (Haniffa & Cooke, 2005). The results of Eng & Mak (2003) are consistent with these studies as they also found a negative relationship between outside directors and voluntary disclosure. However, Cheng & Courtenay (2006) and Donnelly & Mulcahy (2008) found a positive relationship between non-executive directors and voluntary reporting. Consistent with this, with respect to environmental reporting, Khan (2010) found a significant and positive relationship between non-executive directors and the extent of CSR reporting and Iatridis (2013) found that quality of environmental disclosure increases as the percentage of independent directors on the board increases. Additionally, Cheng & Courtenay (2006) found that there is no relationship between board size and voluntary reporting. This means that the board size does not affect the level of voluntary disclosure. With respect to environmental reporting, Rao et al., (2012) reported a positive relationship between board size and environmental reporting. However, in other studies no such relationship is found (Halme & Huse, 1997).

2.2 Underlying theories

2.2.1 Agency Theory

The agency theory is directed at the agency relationship (Eisenhardt, 1989), which is defined by Jensen & Meckling (1976) as “a contract under which one or more persons (the principal(s)) engage another person

(the agent) to perform some service on their behalf which involves delegating some decision making authority to the agent.”. The most discussed agency relationship is the one between the shareholders (as the principal)

and the management (as the agent) of an organization (An et al., 2011). The agency theory is about resolving two agency problems (a) the problem that the desires or goals of the principal and agent conflict and (b) it is difficult or expensive for the principal to verify what the agent is actually doing (Eisenhardt, 1989). The latter agency problem can also be defined as information asymmetry. Information asymmetry occurs when someone (the agent) has an information advantage over the other (the principal) (Salehi et al., 2014; An et al., 2011).

Due to these agency problems the agents have incentives and opportunities to maximize their own utility at the expense of the principal. Therefore, the principals have reasons to introduce governance mechanisms to reduce these agency problems (Dey, 2008). The board is such a mechanism, and is recognized as the most important internal governance mechanism of a company (Kent & Stewart, 2008; Fama & Jensen, 1983). According to agency theory the main task of the board is to monitor the agent(s) on behalf of the principals (Hillman & Dalziel, 2003). The governance mechanism are designed to limit agent(s)’ opportunistic behavior and reduce the information asymmetry, thus improving disclosure policy (Ashbaugh-Skaife et al., 2006; Botti et al., 2014).

In the context of pension funds, the participants of a pension fund can be seen as the principals and the board of a pension fund can be seen as the agent. This means that the participants of a fund delegate the decision-making authority towards the board of a pension fund. Consequently, it is difficult for the participants to observe the actions of the board. Therefore, the board may have an information advantage over the participants which gives them the incentive, and opportunity, to maximize their own utility rather than that of the participants of the funds. This risk can be minimized by governance mechanism, such as internal supervision in the form of non-executives, supervisory board, or a visitation committee.

(9)

2.2.2 Signaling Theory

Based on the concept of information asymmetry the signaling theory cannot be ignored. As the signaling theory is mainly about the reduction of information asymmetry between two parties, in any social setting (An et al., 2011; Connelly et al., 2011; Spence, 2002). The signaling theory is often used in why organizations commit to voluntary disclosure (Pivac et al., 2017). This theory is found by (Hughes, 1986). The essence of this theory is that the signalers are insiders who have more information regarding an individual, product, or organization that is not available for outsiders, the receivers (Connelly et al., 2011). In a general business setting, outsiders do not know the quality of the insiders (e.g. organizations). Consequently, the organization of high quality incurs an opportunity loss because it’s higher quality is not perceived by outsiders, while organizations of low quality obtain an opportunity gain (An et al., 2011). Within this situation the organization with high-quality gets the incentive to inform the outsiders about their superior quality. Disclosing this information in annual reports is one way to ‘signal’ the superior quality to outsiders. Signaling could bring many benefits to a firm, such as improving corporate image, creating an understanding of its products or services and improving the relationships with various stakeholders (An et al., 2011; Rodgers, 2007; Singh & Mitchell Van der Zahn, 2008; Vergauwen & van, 2005).

Within a pension fund the directors of the board can be considered as insiders who possess more information regarding the pension fund, and its performance, than the outsiders, in this case the participants. Therefore, when a pension fund performs well, compared to other funds, they have the incentive to signal this superior quality to their participants. However, when their performance is worse than other funds, the fund might not signal this towards their participants as this may be part of the information advantage the board has.

2.2.3 Stakeholder Theory

The previously discussed agency theory is primarily occupied with the relationship between the firm and its stockholders (Hill & Jones, 1992). The stakeholder theory expands on this theory by looking at the relationship(s) between the firm and all stakeholders (An et al., 2011). The term stakeholder is defined by (Freeman, 1984) as “any group or individual who can affect or is affected by the achievement of the firm's

objectives”. Within this term a distinction can be made between an ethical branch and managerial branch of

stakeholder theory (Deegan, 2002). According the ethical branch, all stakeholders should be provided with the information regarding how the organizations activities affect them even if they do not use the information. Within the managerial branch they focus only on the stakeholders that are most important to the firm (An et al., 2011).

The stakeholder theory can be considered as an “system-oriented theory” (Deegan, 2002). Essentially, a system-oriented-theory focuses on the role of information and disclosure in the relationship between organizations, the State, individuals and groups (Gray et al., 1996). But according (Gray et al., 1995) the stakeholder theory can better be seen together with legitimacy theory, as there is overlap between the two.

The stakeholder theory is an extended version of the agency theory, as it considers all stakeholders and not just the stockholders. Based on the definition of stakeholder, the participants of the pension fund can be considered as stakeholders as they are affected by the achievements (or failures) of the pension fund. The

(10)

pension fund has, just like any organization, multiple stakeholders. However, based on the managerial branch the participants of the fund can be considered as the most important ones.

2.2.4 Legitimacy Theory

Legitimacy theory is just like stakeholder theory, as it is focusing on the relationship between the organization and the society at large (An et al., 2011). Suchman (1995) gave the following definition to legitimacy theory: “Legitimacy is a generalized perception or assumption that the actions of an entity are desirable, proper, or

appropriate within some socially constructed system of norms, values, beliefs, and definitions”. Based on this

definition there is a social contract between the organization and the society in which it operates. The organization needs to operate in accordance with this social contract, that i s in accordance with the socially constructed system of the society at large (An et al., 2011). However, this socially constructed system of norms, values and beliefs is dynamic. Therefore, the actions of the entity may differ from the socially constructed system at one point, which inevitably results in a so-called ‘legitimacy gap’. The legitimacy of an organization is at risk when there is a legitimacy gap (Sethi, 1979). There are two responses towards the legitimacy gap, either the gap can be ignored or alternatively the gap can be reduced by disclosing more information (Haniffa & Cooke, 2005).

Pension funds do not create life threatening products nor do they engage in activities that are unsafe to the society, but being legitimate is deemed imperative (Khan, 2010). According (Lindblom, 1994) there are four strategies that entities can use to secure their legitimacy, and in the context of this study the most important one is: informing stakeholders about intended improvements in performance.

A pension fund may not create life threatening products or engage in activities that are unsafe to the society, but they can invest in companies that do. And when a pension fund invests in such a company, the pension fund might breach the so-called “social contract” which inevitably results in a legitimacy gap. Consequently, the legitimacy of the fund is at risk. To reduce this gap, pension fund can disclose more information. For example, some funds provide information that they do not invest in companies that are involved in the production, distribut ion or selling of weapons, as this is probably undesirable by the society.

2.3 Pension funds in the Netherlands

This study focusses on Dutch pension funds. Therefore, discussing the Dutch pension system and pension fund’ governance is necessary. Firstly, we will explain briefly how the pension system works and then we will describe the pension fund governance and finalize this subsection with an elaboration of the five different types of board models.

2.3.1 The Dutch system

Statistics Netherlands (CBS) defines pension as “a periodic payment that people receive after reaching a predetermined age or due the death of a partner”. The pensions in The Netherlands exists out of three pillars: the pension received from the state (AOW), additional pension from employment conditions and the pensions provisions individuals can arrange individually (de Lange, 2019). The summation of all three pillars determines the height of the pension of an individual.

When it comes to pension there are usually three parties involved, namely the employee(s), the employer and the pension fund (Maatman, 2005). Where the employee(s) and employers come to an pension

(11)

agreement (i.e. agreements regarding pension), which is part of the collective labour agreement (de Lange, 2019). Whereas the pension fund is responsible for executing the pension agreement. Once an employee enters into a pension agreement it will need to pay contributions to the pension fund, which will be invested (de Lange, 2019). Based on the contributions and the returns from their investments the pension funds will be able to pay the pensions to employees now and in the future. This means that a fiduciary relationship exists between the pension fund and its stakeholders, as the pension fund (as fiduciary) exerts control over the financial resources while the stakeholders (the beneficiaries) carry the economic risks (Maatman, 2005).

In general there are three types of self-governing pension funds, namely the institutional type, the contractual type and the trust (which is a hybrid version of the former two) (Stewart & Yermo, 2008). The Dutch pension funds can be classified as the institutional type, in here the pension fund is an ‘independent entity with its own legal personality and capacity and hence it has its own internal governing board (Stewart & Yermo, 2008). The Dutch pension funds are responsible for most of the pension that individuals receive from the second pillar. In The Netherlands there are three types of pension funds a) industry-wide pension funds, which focus on an entire sector, b) company pension funds, which are assigned towards a single organization and c) pension funds for liberal professions, for example lawyers and doctors (de Lange, 2019).

2.3.2 Pension fund governance

Corporate governance is defined by the (OECD, 2004, p.11) as corporate governance involves a set of relationships between a company’s management, its board, its shareholders and other stakeholders. Corporate governance also provides the structure through which the objectives of the company are set, and the means of attaining those objectives and monitoring performance are determined. Based on the Dutch corporate governance code, governance is about ‘management and control, about responsibility and influence, and about supervision and accountability’ (MCCG, 2016). The Dutch code of corporate governance is usually used as a starting point for making governance codes for the industry specific (de Lange, 2019). The Federation of the Dutch Pension funds combined forces with the Labour Foundation to create a governance code specifically for pension funds in The Netherlands. In here they focus on three functions of ‘good pension fund governance’ which are 1) management 2) supervision and 3) accountability (Federation of the Dutch Pension Funds & The Labour Foundation, 2013).

A distinction can be made between internal governance mechanism and external governance mechanisms (Brown et al., 2011). The main vehicle of internal governance mechanism is the governing board, who are responsible for protecting the interests of stakeholders of a firm through directing its operations and by supporting its decision-making (Naciti, 2019). The governing board of a pension fund, the board, is the equivalent of a board in a corporation (Stewart & Yermo, 2008; Federation of the Dutch Pension Funds & The Labour Foundation, 2013; Clark, 2004; Ambachtsheer et al., 2006). Within the pension funds there are five different board models (paritarian, independent, inverse mixed, paritarian mixed, and independent mixed), which will be described subsequently. The internal supervision of the pension fund is related to the chosen board model. Based on the principles of good pension fund governance by (The Labour Foundation, 2005) and (Maatman, 2005) the following forms of internal supervision can be present: 1) visitation body 2) supervisory board 3) non-executives in a one tier board structure

(12)

Besides the internal governance mechanism, the main external governance mechanism for Dutch pension funds are the central bank of The Netherlands (DNB) and the Dutch Authority for the Financial Markets (AFM). Based on the Pension Act the Dutch Central Bank is charged with the supervision over pension funds regarding prudential oversight and material oversight. Prudential oversight is oversight regarding the financial solidity of the pension fund. Material oversight is concerned with everything else that is not part of the prudential and behavioural oversight. The monitoring of behavior and communication is delegated to the AFM (de Lange, 2019; Pension Act, 2006). As this study focuses on the effects of board characteristics on disclosure quality, we will not go into further detail on the external governance mechanisms.

2.3.3 Pension fund board models

In here the five different board models will be elaborated. Either the board is paritarian, independent or one of the three mixed models (paritarian mixed, independent mixed and inverse mixed) (de Lange, 2019). The type of internal supervision depends on the nature of the pension fund and its chosen board model.

Within the paritarian model the board exists out stakeholders, being the employer/employees and the pensioners. Where the board exists out of 50% of representatives of the employees and representatives of the pensioners. The remaining 50% of the board exists out of the representatives of the employers. A maximum of two experts (not being direct representatives of stakeholders) can be added to the board in order to have some expertise within this board (de Lange, 2019; Pension Act, 2006). The independent board is a board that does not consist out of stakeholders. As a matter of fact, this board consist out of, at least, two directors. The directors are, compared to a paritarian model, not direct representatives of any stakeholder (i.e. employees, employer, or pensioners). This board exists out of only external, independent, directors (de Lange, 2019).

When the pension fund is an industry-wide one (e.g. metal industry) and has a paritarian or independent board model, the internal supervision is entrusted to a supervisory board. In case of a corporate pension fund (e.g. Shell) with a paritarian model or independent model the internal oversight is entrusted to a supervisory body or a visitation body, unless the invested capital is not more than 1 billion euro or when the pension fund is entirely insured (de Lange, 2019).

The mixed boards exists out of executives and executives directors (Pension Act, 2006). The non-executive directors are responsible for the internal supervision over the non-executive directors. Therefore, in a mixed model, there is no presence of an internal supervisory body, such as a supervisory board or a visitation body. A mixed board is therefore also referred to as a one-tier board structure (de Lange, 2019). There are three forms of mixed models. In the paritarian mixed model the executive directors are direct representatives of the stakeholders, which is no different than the paritarian model. These executives are accompanied with at least, three independent non-executives. Within the independent mixed model, the executives as well as the non-executives are independent. This means that they are not direct representatives of the stakeholders of the pension fund. In addition, the board exists, again, out of at least three independent non-executives. In case of the inverse mixed model the executive directors are independent, and the non-executive directors are not. The non-executive directors exist out of direct representatives of the stakeholders (i.e. employees, employers and pensioners). Because the non-executive directors’ independence is questionable, this type of mixed model requires a chairman that is independent (Pension Act, 2006). Also, this type board model is accompanied with

(13)

an audit committee who oversees the risk management, investment policy and financial information provision by the pension fund (de Lange, 2019).

2.4 Hypotheses development

2.4.1 Board model

Within the Dutch pension funds, as mentioned before, a distinction can be made between five types of board models (de Lange, 2019). For practical reasons we will classify the board either as a one-tier board model (i.e. a mixed model) or as a two-tier board model (i.e. paritarian or independent model).

In the principal-agent relationship the principal (i.e. the board) will either be monitored by a supervisory board or by non-executive directors. In a two-tier board model the internal supervision is delegated towards a supervisory board or visitation body (Bezemer et al., 2014; de Lange, 2019). In case of a one-tier board model the responsibility of monitoring rests with the non-executives (de Lange, 2019).

Not many studies spent their attention towards the different types of board structures, as most countries only allow one type (Millet-Reyes & Zhao, 2010). There are some studies where a comparison between the one-tier and two-tier board structures has been made (Block & Gerstner, 2016; Jungmann, 2006; Millet-Reyes & Zhao, 2010). The main advantages the one-tier board has over the two-tier board model is that there is a superior flow of information, faster decision making and a better understanding and greater involvement (Bezemer et al., 2014; Jungmann, 2006). The biggest disadvantage of a one-tier board is that it has to make and monitor the same decision (Jungmann, 2006). Also, the non-executives are perceived to be less independent (Dienes & Velte, 2016), which will have an influence on their monitoring role. The main advantages the two-tier board model has over the one-tier board is that there is a separation between managing and control (Jungmann, 2006). However, the two-tier board model also has its disadvantages, such as a lack of information or knowledge needed to actually be able to properly supervise the executives (Block & Gerstner, 2016). In addition, the two-tier board might lead to excessive monitoring by the supervisory board, as might be the case in a mixed model that is dominated by non-executives (Baysinger & Butler, 1985). Surprisingly, the independence in a two-tier system is generally better but their supervision seems to be less effective (Dienes & Velte, 2016). Regardless its disadvantages, the two-tier board model is favoured by many reformers of corporate governance (Millet-Reyes & Zhao, 2010) and from an agency perspective the presence of an independent supervisory board is also preferred (Dienes & Velte, 2016). Therefore, we expect that a two-tier board model will be more effective than a one-tier board model, resulting in the following hypothesis:

Hypothesis 1: The two-tier board model(s) is positively related with respect to the disclosure quality

2.4.2 Board size

Based on the agency- and stakeholder theory, the characteristics of the board can have an impact on the quality, and level, of disclosure (Ali & Abdelfettah, 2019). A characteristic that often has been examined with respect to the performance of for-profit organization is the size of the board (Ali, 2010; Bajra & Cadez, 2018; Dienes & Velte, 2016; Gois, 2009; Johl et al., 2013; Vafeas, 2005). However, the effect of size on board behavior is rather less clear in the non-profit sector (O’Regan & Oster, 2005).

(14)

From an agency perspective small boards are more effective than large boards (Jensen, 1993). Lipton & Lorsch (1992) also argue that the board should exist out of a maximum of ten board members, as they prefer eight or nine board members. Empirical evidence is provided by Yermack (1996), who concluded that smaller boards are more effective as the value of the firm rises. Consistent with this are the results from (Klein, 2002), who reported that firms with smaller sized boards are more effective and have, in general, better quality of information disclosure. Others favour larger boards, as they provide better supervision and monitoring (Coles et al., 2008). As Peasnell et al. (2005) shows that board size is positively related to the quality of accounting information. However, larger boards are often associated with poor cohesiveness, an inability to initiate strategic change, poor communication, and a lack of control (Hillman et al., 2011). Empirical evidence suggest that when boards are larger the likelihood of earnings management and fraud increases (Beasley, 1996; Kao & Chen, 2004). Therefore, we hypothesize:

Hypothesis 2: Board size is negatively related to the disclosure quality

Based on the first hypothesis an positive relationship between a two-tier board structure and the disclosure score is expected. Previous hypothesis states that board size is negatively associated with the disclosure quality. Therefore, we expect that the positive relationship between a two-tier board structure and the disclosure score weakens as the size of the board increases, resulting in the following hypothesis:

Hypothesis 2a: The positive effect of a two-tier board model on disclosure quality becomes weaker as size of the board increases

2.4.3 Presence of a supervisory board

The presence of a supervisory board is closely intertwined with the chosen board model, since having a supervisory board distinguishes a two-tier board model from a one-tier board model. However, most studies examine the effectiveness of corporate control in an American setting. Therefore, the focus is usually on the one-tier board structure and little attention is spent to the two-tier board structure (Millet-Reyes & Zhao, 2010). As mentioned before, the presence of a supervisory board is not mandatory for all pension funds and is only required when the invested capital is more than one billion euros or when the pension fund is entirely insured (de Lange, 2019). This means that the presence of a supervisory board in other circumstances is voluntary.

With the majority of the countries operating under a one-tier board structure (Firth et al., 2007), the effects of supervisory boards are not often examined, resulting in limited literature available (Ran et al., 2015). However, there are some studies that indicate that the supervisory boards do have value. As the results Tušek et al. (2009) indicate that the presence of a supervisory board can improve the financial reporting quality and Firth et al. (2007) show that the supervisory board is positively associated with the informativeness of earnings. Therefore, we hypothesize:

Hypothesis 3: The permanent presence of a supervisory board is positively related to the disclosure quality

(15)

2.4.4 Board activity

Boards need to be active in order to ensure high-quality, transparent reporting in annual reports (Kent & Stewart, 2008). Boards that meet more frequent are more active which will result in a better performing board (Lipton & Lorsch, 1992). Therefore, the frequency of board meetings is being used as a proxy for board activity, which is in line with previous study (Vafeas, 1999). He suggests that the frequency of meetings is an important dimension of board operations. Despite its ease measuring it, the relationship between the frequency of board meetings and firm performance is not often examined (Dienes & Velte, 2016). Based on agency- and stakeholder theory the frequency of board meetings is in the interest of all stakeholders, as the extra time spent on meeting would reduce agency costs (Lipton & Lorsch, 1992). From an legitimacy theory point of view an increased frequency of board meetings is necessary in a dynamic environment, in which pension fund operate, to ensure legitimacy (Dienes & Velte, 2016).

Usually the financial reporting responsibilities are delegated to an audit committee (Kent & Stewart, 2008). A study from (Hoque et al., 2013) reports a positive relationship between the frequency of audit committee meetings and firm’s financial performance. However, the function of the audit committee, if present, in a Dutch pension fund is to oversees the risk management, investme nt policy and the disclosure of financial information (de Lange, 2019). That means that the responsibility of financial reporting remains with the board. Firth et al., (2007) find that when the supervisory board meets more frequent the informativeness of earnings tends to increase as well.

Previous studies provide evidence of a positive relationship between the frequency of board meetings and CSR reporting (Kent & Stewart, 2008; van Staden). While Dienes & Velte (2016) found a negative, but non-significant, relationship between frequency of meetings and CSR reporting. Consequently, we expect that when a board is more active, by meeting more frequently, the disclosure quality will increase. Therefore, we hypothesize:

H4: The frequency of board meetings is positively related to the disclosure quality

2.4.5 Board independence

Independence is found to be of prime importance in most European codes (Zattoni & Cuomo, 2008). Yet it is not explicitly emphasized in the code of the Dutch pension funds for good pension fund governance (Federation of the Dutch Pension Funds & The Labour Foundation, 2018). Based on agency- and stakeholder theory the presence of non-executive directors (i.e. independent, outside, directors) is necessary to properly monitor and control the actions of the executive directors due to their opportunistic behavior (Jensen & Meckling, 1976). Nevertheless, there could also be some down-sides of boards being dominated by non-executives such as, excessive monitoring (Baysinger & Butler, 1985) and perhaps a lack of knowledge to be effective (Block & Gerstner, 2016).

Comprehensiveness of the disclosures and increased quality of the disclosure could also be explained from a signaling theory perspective. According to Fama & Jensen (1983, p.315) the non-executive directors have the incentive to develop their reputation as experts in decision control. These outside, indepe ndent, directors use their directorship to signal that ‘a) they are decision experts, b) they understand the importance of diffuse and separate decision control and c) they can work with such decision control systems’. And when the

(16)

board is dominated by non-executives it could give them more power to force onto more, and better quality of disclosure by the organization (Haniffa & Cooke, 2002).

As mentioned in the background there are several studies that show positive effects of the presence independent directors on the board. Such as, Beasley (1996) who showed that firms with independent directors on the board experience no financial statement fraud. Chen & Jaggi ( 2000) find a positive association between the proportion of independent directors and the comprehensiveness of financial disclosures. Consistent with these studies are Iatridis (2013) and Khan (2010) who both find a positive relation between the proportion of independent directors and the quality of environmental disclosure and CSR reporting. Ali (2010) provides evidence of a positive relationship between independent directors and the quality of disclosure, while Haniffa & Cooke (2002) and Eng & Mak (2003) found a significant, but negative relationship between the proportion of non-executive directors and voluntary disclosure. However, Rupley et al. (2012) found no relationship between independent directors and voluntary disclosure.

In general, it is assumed that board independence contributes to effective governance and financial reporting (Ianniello, 2013) and the majority of the studies show positive effects of the presence of independent directors. Therefore, we hypothesize:

H5: The presence of independent directors is positively associated with the disclosure quality

Based on the hypotheses above, the following conceptual model has been assembled:

(17)

3. METHODOLOGY

In this section the research design of the study will be elaborated. First, the data and the sample of the research will be described. Thereafter, the dependent variable, independent and control variables will be operationalized. Also, the measurement of the variables of interest will be described. Finally, the section will end with the regression model of this study.

3.1 Data and sample

In this study a trend analysis will be held over the years 2015-2019. As mentioned before, this period is relevant due to the regulatory changes with respect to pension funds’ governance. At the beginning of this century there were approximately 1.000 pension funds in the Netherlands (de Lange, 2019). Currently, there are approximately 224 pension funds left. In this study a sample of 80 pension fund, from the 224 that exists, was taken. Pension funds were selected at random to ensure a representative sample. Due to the fact that two funds terminated during 2019, the final sample size for 2019 is 78. In the other years, a sample of 80 was used. The complete sample of pension funds is shown in Appendix A.

In order to test the relationship between governance attributes and the quality of disclosure for Dutch pension funds efficiently, the annual reports of 2015, 2017 and 2019, rather than all years in the timespan, have been assessed. Meaning that all the data has been hand-collected from the annual reports. This would result in 240 fund-year observations. Due to the termination of two funds in 2019 the total fund-year observations is 238. In order to test the fourth hypothesis only 183 observations was used due to missing observations.

3.2 Dependent variable

The dependent variable of this empirical study is the disclosure quality of Dutch pension funds. For this study the following definition of disclosure quality is used: “the extent to which financial reporting conveys information about the firm’s financial position that is complete, neutral, and free from error” (FASB, 2010; Garrett et al., 2014). However, the concept quality (of disclosure) is hard to define due to the dependency of the context and its subjective nature (Beattie et al., 2004). Therefore, we define disclosure quality in this study as follows: “the extent to which pension fund disclose information that is complete, accurate and reliable”. (Singhvi & Desai, 1971)

To measure the quality of disclosure, two main approaches have been used (Beattie et al., 2004). The first approach is the subjective way of analysts who rate the quality of disclosure. The other approach is semi -objective. Within this approach either a disclosure index or textual analyses can be used. Previous studies have widely adopted the disclosure index to measure disclosure quality (Ali & Abdelfettah, 2019; Eng & Mak, 2003; Haniffa & Cooke, 2002; Iatridis, 2013; Scaltrito, 2015). Therefore, a disclosure index was used in this study to measure the disclosure quality of Dutch pension funds.

For this research a disclosure index was developed to compute the disclosure score for each pension fund in the sample. The index is mainly based on Dutch Accounting Standards Board 610 (RJ 610). The index consists out of 62 items and are divided over eight categories, namely diversity (7 items), pensions (9 items), investments (8 items) and cost transparency (11 items), contributions (7 items), financial position of the fund

(18)

(9 items), indexation (4 items), governance code for Dutch pension funds (3 items) and other (4 items). Appendix B presents the items that are included in the disclosure index.

In order to reduce the subjectivity, every item is considered to be equally important, which is in line with the study of (Ali & Abdelfettah, 2019). Therefore, a simple binary coding scheme was used, assigning a value of “1” when an item is present and “0” otherwise, instead of a coding scheme that incorporates ordinal measures. This means that the index is unweighted (Marston & Shrives, 1991). The maximum score that can be achieved on the index is thus 62 (minus items that are not applicable). The total score for each pension fund will be subsequently divided by the maximum points available to determine a percentual score for each pension fund. The score will vary between 0% and 100%. Whereby a higher percentage is indicative for a higher disclosure quality as more items are disclosed.

3.3 Independent variables

In this research five independent variables are being used, namely: board model, presence of a supervisory board, board size (also used as moderating variable), board activity and board independence. All the data regarding the independent variables are also hand-collected from the annual reports of the pension funds. The definition and the measurement of these variables will be described subsequently.

The board model refers to the type of board model used within the pension fund. As mentioned before, there are five types of board models, namely paritarian, independent or one of the three mixed models (de Lange, 2019). Board model is measured with a dichotomous variable, coded “1” if the fund has a two-tier board model (i.e. paritarian model or independent model) and “0” if the fund has a one-tier board model (i.e. one of the mixed models). The supervisory board contains out of a group of individuals who are responsible for monitoring the board (Ran et al., 2015). As mentioned in the theory section, whether a pension fund has a supervisory board or not depends on the chosen board model and the nature of the fund. The presence of a supervisory board is also measured as a dichotomous variable, coded “1” if a supervisory board is present and coded “0” if not.

The third independent variable, which also serves as a moderating variable, is the board size of the pension fund and is measured by the total numbers of directors sitting on the bo ard. This is consistent with other studies (Bajra & Cadez, 2018;Ali, 2010; Vafeas, 2005). The activity of the board is measured by the frequency of meetings by the board held during a specific year. This is consistent with (Bajra & Cadez, 2018; Vafeas, 1999). Finally, board independence refers to the percentage of independent directors sitting on the board, which is in line with (Iatridis, 2013; Chen & Jaggi, 2000; Rupley et al., 2012). Independent directors can also be classified as outsiders, as they have no further ties to the pension fund besides being a board member (Klein, 2002).

3.4 Control variables

In order to improve the goodness of the regression model, we control for a number of pension fund-specific factors which have been shown to have an effect on disclosure quality (Archambault & Archambault, 2003; Johl et al., 2013). The control variables for this research are pension fund size and leverage.

Several studies have found a positive association between disclosure and firm size (Chavent et al., 2006; Ho & Taylor, 2007). There are multiple reasons why size is related to the reporting practices of an

(19)

organization. For instance, from the agency perspective (Jensen & Meckling, 1976) it can be argued that larger sized organizations have greater agency costs. Consequently, these organizations have the inc entive to disclose more. Also, larger sized organizations have more resources to disclose more (Buzby, 1975). It can also be argued that a larger sized pension fund must serve a greater audience. Size of the pension fund will be measured as the natural logarithm of total assets (LNASSETS), which is in line with most prior studies.

In prior studies there is a relationship found between a firms leverage and quality of disclosure (Barako et al., 2006; Chavent et al., 2006). One reason for this is that highly leveraged organizations disclose more in order to reduce their agency costs of debt (Wallace et al., 1994). Therefore, we will need to control for this variable, which is consistent with prior studies (Chen & Jaggi, 2000; Eng & Mak, 2003; Johl et al., 2013). A pension funds leverage will be measured by a pension funds´ coverage ratio, which is the equivalent of a firms leverage ratio.

The measurements of all variables are summarized in the Table I below. TABLE I Operationalization & measurement of variables

Variable Label Measurement

Dependent variable

Disclosure Index Score DSCORE The percentual score achieved on the disclosure index by each pension fund in the sample. Total numbers of points achieved divided by the maximum amount of points available.

Independent variables

Board Model BMODEL Whether the board model is a two-tier board model or a one-tier board model, coded as 1 if the fund has a two-tier board and 0 if the fund has a one-tier board model. Board size BSSIZE The total numbers of directors on the board.

Presence supervisory board SBOARD Whether there is a supervisory board present in the pension fund or not. Coded as 1 if there is a supervisory board present and 0 if not.

Board activity BMEETINGS The total number of board meetings held by the board in a year.

Board independence BINDP The ratio of independent to total number of directors on the board.

Control variables

Pension fund size LNASSETS Size of the pension fund measured by natural logarithm of total assets.

(20)

3.4 Empirical model

To test the hypotheses that are presented in the previous section, univariate, bivariate and multivariate analyses have been performed. The univariate analysis provides descriptive statistics of the variables of interest such as the number of observations, the mean, standard deviation, minimum observation and the maximum observation per variable. The Pearson correlation matrix has been used for the bivariate analysis, to check for potential multicollinearity problems that might be present.

Finally, for the multivariate analysis, the pooled Ordinary Least Square (OLS) regression method is used in the regression on disclosure quality. In regression analysis it is common that the first model includes only the dependent and control variables. Thereafter, each hypothesis is tested separately with the control variables included. Resulting in multiple models (i.e. models 2 till 7). Combining all models into a full model (i.e. model 8) results in the following empirical model:

DSCORE = 𝛽0 + 𝛽1𝐵𝑀𝑂𝐷𝐸𝐿𝑦𝑖, 𝑡 + 𝛽2𝐵𝑆𝐼𝑍𝐸𝑖, 𝑡 + 𝛽3(𝐵𝑀𝑂𝐷𝐸𝐿𝑦𝑖, 𝑡 ∗ 𝐵𝑆𝐼𝑍𝐸𝑖, 𝑡) + 𝛽4𝑆𝐵𝑂𝐴𝑅𝐷𝑖, 𝑡 + 𝛽 + 𝛽5𝐵𝑀𝐸𝐸𝑇𝐼𝑁𝐺𝑆𝑖, 𝑡 + 𝛽6𝐵𝐼𝑁𝐷𝑃𝑖, 𝑡 + 𝛽7𝐿𝑁𝐴𝑆𝑆𝐸𝑇𝑡𝑖, 𝑡 + 𝛽8𝐿𝐸𝑉𝑖, 𝑡 + 𝜖𝑖, 𝑡.

Definition of each variable is displayed in Table I, βi represents the coefficients of each variable and ϵi,t is the error term

(21)

4. RESULTS

In this section the results of the research will be elaborated. First, the descriptive statistics will be summarized and displayed. This includes the Pearson correlation matrix. Thereafter, the regression results will be displayed and discussed.

4.1 Descriptive statistics

The descriptive statistics for the dependent, independent and control variables in the empirical model, are displayed in Table II.The descriptive statistics (number of observations, mean, standard deviation, minimum and maximum) for dependent variable (disclosure score) are given for each year. By doing this, the trend of the disclosure score over recent years can be easily traced. With respect to the independent and control variables the number of observations, means, standard deviations, minima and maxima are given for the entire timespan.

As can be observed from the descriptive statistics, the mean of disclosure scor es for 2015, 2017 and 2019 are 72,2% 77,8% and 79,3% respectively. With a lot of regulatory changes in 2014/2015, the average disclosure score increased quite a lot from 2015 to 2017. However, the increase from 2017 to 2019 is rather marginal. The lowest disclosure score achieved was 50,8% in 2015 and the highest disclosure score of 92% was reached, surprisingly, in 2017. In addition, approximately 88% of the pension fund operate under a two-tier board structure and 44% of the funds’ supervision is delegated towards a supervisory board. Furthermore, the average board consist out of 8 members. However, the amount of board members ranges from 3 to 15. In addition, on average, the board meets 4 times a year. While one board managed to meet 23 times in one year .

TABLE II Descriptive Statistics

Variable Observations Mean Std. Dev. Min Max

Dependent variable Disclosure score 2015 80 .722 .085 .508 .890 Disclosure score 2017 80 .778 .078 .542 .920 Disclosure score 2019 78 .793 .081 .520 .919 Independent variables Board model 238 .882 .323 0 1

Presence supervisory board 238 .441 .498 0 1

Board size 238 7.782 2.279 3 15 Board activity 183 10.94 4.448 2 23 Board independence 238 .087 .153 0 1 Control variables LNASSET 238 14.345 1.696 10.331 19.393 Leverage 238 1.111 .096 .903 1.39

The Pearson correlation matrix can be found in Table III below. Consistent with my hypotheses, the presence of a supervisory board and a more independent board are both positively significant (at the 5% and 10% respectively) correlated with disclosure score. However, to the contrary of our hypotheses we find a positive and significant, at the 5% level, correlation between board size and the disclosure score. Furthermore, we found no statistically significant correlation between board model, board activity and the disclosure score.

There is a rule of thumb if the correlation between independent variables is equal to or greater than .800 then this can be problematic. This problem is called (multi)collinearity. The correlation between board

(22)

independence and board model is -0.537. However, this is still below the .800 mentioned above. Therefore, (multi)collinearity seems not to be an issue in this research.

TABLE III Pearson Correlation matrix statistics

Variables (1) (2) (3) (4) (5) (6) (7) (8)

1. Disclosure score 1

2. Board model 0.006 1

3. Presence supervisory board 0.301*** _0.308*** ₁

4. Board size 0.169** _-0.281*** _0.250*** ₁ 5. Board activity 0.019 0.143* _0.217*** _-0.111 ₁ 6. Board independence 0.169** _-0.537*** _-0.010 _0.019 _0.071 ₁ 7. LNASSET 0.182** _-0.310*** _0.318*** _0.707*** _0.100 _0.233*** ₁ 8. Leverage -0.030 -0.268*** _-0.296*** _-0.069 _-0.229*** _0.106 _0.013 ₁ *_{p < 0.10,}**_{p < 0.05,}***_{p < 0.01}

Additionally, a variance inflator factor (VIF) test was performed to check whether multicollinearity is present between the variables. Whenever the test has an outcome of 10 or higher it means multicollinearity is present, which will be problematic for the regressions later. However, the test indicates no outcome of 10 or higher so, multicollinearity is not a problem in this research. The outcomes of this test are tabulated in Appendix C.

4.2 Regression results

The results of the regression analysis between the various governance attributes and the disclosure index score of Dutch pension funds are displayed in Table IV. As mentioned in the previous section, the first model only contains the dependent variable and control variables, which is common in regression analysis. Thereafter, each independent variable is tested separately, resulting in the various models. Finally, model 8 refers to the entire empirical model, meaning that all variables of interest in this research are included.

As can be seen from the regression table, there are significant relationships between the interaction term, presence of a supervisory board, board independence and the disclosure score. In the second model, the first hypothesis is tested which predicts that a two-tier board is positively related with the disclosure quality. The direct effect of board model on disclosure score is positive, as hypothesized, but not statistically significant (β = 0.010, not significant). Therefore, no evidence was found that supports this hypothesis, and thus the first hypothesis is being refuted.

Furthermore, the second hypothesis argues that a larger board is negatively related with disclosure quality. This relationship is tested in the third model of the regression table. This model shows an insignificant result (β = 0.004, not significant). However, the result also shows a positive rather than negative relationship between board size and disclosure score, which is contrary to what is hypothesized. Due to its insignificance and opposing direction hypothesis 2 could not be accepted. However, in model 4 hypothesis 2a is tested which argues that a greater board size weakens the positive relationship between board model and disclosure quality. In this model the relationship between board model is statistically significant, but negative ( β = -0.176 p < 0.05). Additionally, the interaction term board model * board size is positive and statistically significant ( β 0.020 p < 0.05). This result does suggest that the negative relationship between board model and disclosure score weakens as board size increases. This is contrary to what is hypothesized in H2a. Therefore, hypothesis 2a could not be accepted.

(23)

In the fifth model in the regression table represents the third hypothesis (H3). In this hypothesis it is argued that the permanent presence of a supervisory board is positively related with disclosure quality. Model 5 shows a positive, statistically significant, relationship between the presence of a supervis ory board and the disclosure score (β = 0.038 p < 0.01). This is consistent with what is hypothesized, therefore hypothesis 3 is accepted.

In the sixth model the fourth hypothesis is tested. This hypothesis predicts a positive relationship between the board activity, measured by the frequency of meetings, and disclosure quality. Model 6 shows that there is no significant relationship (β = -0.000, not significant) between board activity and the disclosure score. Also, the coefficient is close to zero. This indicates that there is no relationship between board activity and disclosure quality. Therefore, hypothesis 4 could not be accepted as no evidence was found for a relationship.

Finally, model 7 represents the fifth hypothesis which argues that board inde pendence positively related with disclosure quality. The model shows a positive, and statistically significant, relationship (β = 0.077 p < 0.05), which is in line with what is hypothesized and thus leads to the acceptance of the fifth hypothesis (H5).

TABLE IV Regression results for Disclosure score

Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8

Hypotheses H1 H2 H2a H3 H4 H5 VARIABLES Board model 0.010 -0.176** -0.124 (0.018) (0.085) (0.104) Board size 0.004 -0.014 -0.010 (0.003) (0.009) (0.011)

Board model * Board size 0.020** 0.016

(0.009) (0.011)

Presence of a supervisory board 0.038*** 0.046***

(0.012) (0.016) Board activity -0.000 -0.001 (0.002) (0.002) Board independence 0.077** 0.106** (0.037) (0.051) LNASSETS 0.007** 0.007** 0.003 0.004 0.003 0.010** 0.006* 0.000 (0.003) (0.003) (0.004) (0.004) (0.003) (0.004) (0.003) (0.006) LEV -0.034 -0.028 -0.028 -0.039 0.010 -0.029 -0.042 0.030 (0.058) (0.059) (0.058) (0.060) (0.059) (0.066) (0.058) (0.068) R-squared 0.020 0.021 0.026 0.051 0.060 0.034 0.037 0.148

Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

The effects of governance characteristics on disclosure quality: evidence from Dutch pension funds