CEO power and IFRS 7 disclosure quality: Evidence from UK listed companies

(1)

CEO power and IFRS 7 disclosure quality:

Evidence from UK listed companies

Master Thesis, MSc Controlling

University of Groningen, Faculty of Economics and Business

22 June 2020

Sander Osinga S2964783

Frans van Mierisstraat 48 9718 SV Groningen

+31 621651374 s.osinga.3@student.rug.nl

Supervisor: dr. Y. Karaibrahimoglu

(2)

2

Abstract

Following the introduction of IFRS 7, being effective from 1 January 2007 onwards, both financial and non-financial firms are required to provide disclosures in their financial

statements about the significance, nature and extent of risks arising from financial instruments (risk exposure) and how they manage those risks (assurance). As the ultimate objective is to improve the disclosure quality in the firm’s annual reports, this study examines the impact of CEO power on the IFRS 7 disclosure quality, as well as the moderating role of board

independence. Using a sample of 1,852 non-financial premium listed companies in the United Kingdom between 2007 and 2016, the results show a positive influence of CEO power on the quality of IFRS 7 disclosures. Additionally, after splitting the disclosure quality into the risk exposure and assurance component, the findings reveal that a substantial part of this positive association is attributed to the assurance component, as this refers to favourable information. Considering the moderating effect, we find that an independent board in combination with a powerful CEO could improve the disclosure quality. These findings contribute to the literature of mandatory disclosures in accordance with IFRS 7, as it focuses on different, novel

determinants that influence disclosure quality, using a comprehensive index.

Keywords: IFRS 7 disclosure quality, CEO power, board independence, non-financial premium firms, UK

(3)

3

Introduction

From 1 January 2007 onwards, the International Accounting Standards Board (IASB) introduced the ‘IFRS 7 Financial Instruments: Disclosures’, in order to improve the risk disclosure quality and quantity in annual reports of firms. IFRS 7 requires entities to provide disclosures in the entity’s financial statements that enables users to evaluate the significance of financial instruments for the entity's financial position and performance, as well as the nature and extent of risks arising from financial instruments to which the entity is exposed during the period, and how the entity manages those risks (IFRS 7, paragraph 1). The overall objective is to provide more transparency to the users of financial statements on the entity’s risk exposure on the one hand (risk exposure component) and on how the entity manages or hedges those risks on the other hand (assurance component). Hence, IFRS 7 consists of two distinctive components that differ from each other in terms of the information they provide. In fact, the firm’s risk exposure relates to negative information, whereas the hedging activities to address the risk that is associated with financial instruments reveal positive information. Furthermore, the standard applies to all entities that hold financial instruments, both financial and non-financial. Thereby, it supersedes ‘International Accounting Standard (IAS) 30, Disclosures in Financial Statements of Banks and Similar Financial Institutions’, which was only applicable for banks and other financial institutions, and ‘IAS 32, Financial Instruments: Presentation’ (Muthupandian, 2008; PricewaterhouseCoopers, n.d.). Thus, IFRS 7 impacts a lot more entities than the IAS standard did, also because simple financial instruments, like borrowings, cash and investments, are now included as well, which increases the relevance for studying this topic.

Consistent with the agency theory, the information asymmetry between the managers - including the CEO - and shareholders is expected to be reduced by complying to IFRS 7, meaning that both interests are more aligned, which results in a reduction of agency costs (Adznan & Nelson, 2014; Bischof, 2009; Hassan, 2014). However, previous literature well documented that the disclosure quality depends on multiple factors. Among all, management attitude plays a significant role (Gibbins, Richardson & Waterhouse, 1990). More specifically, CEOs could influence the quality of disclosed information (Ben-Amar & Boujenoui, 2007; Matsunaga & Yeung, 2008). For instance, Kothari, Shu and Wysock (2009) find evidence that CEOs have a tendency to withhold bad news from investors relative to good news, which is dependent on their incentives and ability to withhold bad news. For firms with high litigation risk, this tendency is reduced for instance. However, CEOs that have more personal wealth at

(5)

5 stake, face greater information asymmetry and greater career concerns, have a higher

tendency to delay the disclosure of bad news to investors (Kothari et al., 2009). Given this influence of CEOs over disclosures, the empirical question arises whether and how the quality of IFRS 7 disclosures has been met in case of a powerful CEO. Therefore, the aim of this study is to examine the impact of CEO power on the IFRS 7 disclosure quality of non-financial premium listed companies in the United Kingdom.

Donaldson and Davis (1991) argue that CEO power could have a negative impact on the alignment between the interests of the shareholder and the CEO. Furthermore, powerful CEOs will have a negative influence on shareholder benefit, as “any superiority in

shareholder returns […] would be explained away by agency theory as being due to the spurious effects of financial incentives” (Donaldson & Davis, 1991). Daily and Johnson

(1997) address the influence of CEO power as well, by suggesting that more powerful CEOs could have more discretion to act in their own interest and pursue objectives which are inconsistent with maximizing shareholder wealth. Nevertheless, CEOs with enhanced power do not necessarily engage in self-interested behaviour, as they may be protected from

different organisational pressures, which enables them to behave in accordance with the interests of the company and the shareholders (Daily & Johnson, 1997). So, IFRS 7 aims at improving the disclosure quality of its risk exposure and assurance components, by reducing information asymmetry and enhancing the alignment between shareholders’ interests and CEOs’ interests. Therefore, it is very interesting to examine the impact of CEO power on the IFRS 7 disclosure quality, considering the aforementioned different insights. This leads to the following research question:

“What is the impact of CEO power on the quality of IFRS 7 disclosures in the UK?”

By answering this research question, one could argue whether the results differ when the impact of CEO power is examined on the risk exposure component relative to the

assurance component of IFRS 7. In accordance with this standard, management could either disclose the firm’s risk arising from their financial instruments (risk exposure) or provide information about their hedging activities meant to address the risks that are associated with the use of financial instruments (assurance). As already mentioned, the risk exposure

component refers to negative or unfavourable information, while the assurance component indicates positive or favourable information. Given the self-interested discretionary actions of CEOs in combination with the CEO’s stake in the company and career concerns, I predict a

(6)

6 negative association between CEO power and the quality of risk exposure items, but a

positive association between powerful CEOs and the quality of hedging items.

As mentioned before, the discretionary actions of powerful CEOs could negatively impact the alignment between the interests of shareholders and CEOs, hence, increasing information asymmetry. One way to mitigate these actions, is by means of an independent board, as independent non-executive directors enhance the monitoring role of the board of directors (Duru, Iyengar & Zampelli, 2016). Since an independent board is more likely to engage in transparent and higher quality financial reporting and disclosure practices, this paper examines the moderating influence of board independence, a corporate governance mechanism, in the association between CEO power and IFRS 7 disclosure quality.

In order to answer the research question above, as well as the additional test of the IFRS 7 distinctive components and the moderating effect of board independence, this research focuses on a sample of non-financial premium listed firms in the United Kingdom between the year 2007 and 2016. This is based on the working paper of Karaibrahimoglu and Porumb (2019), which is used to gather data regarding IFRS 7 disclosure quality. The governance data is obtained from BoardEx, whereas the data regarding the financial control variables are obtained from Compustat. After eliminating the missing values, 1,852 firm-year observations are remaining in the final sample.

The main findings of this study are that there is evidence, in contrast to the prediction, that CEO power is positively associated with IFRS 7 disclosure quality. In other words, when CEOs also serve as a chairman of the board, serve the company’s board for a longer period of time, and/or have a higher proportion of shares, they are likely to increase the disclosure quality. This implies that CEOs do not abuse their power, the interests between the CEO and shareholders do not diverge substantially, and/or the CEOs do not pursue its own objectives at the expense of the shareholders and thereby, reducing the disclosure quality. As already mentioned, one explanation could be that powerful CEOs do not necessarily engage in self-interested behaviour. The rationale behind this, is that these CEOs may be protected from different organisational pressures, which enables them to behave in accordance with shareholders’ interests and the interests of the firm (Daily & Johnson, 1997).

However, after splitting the disclosure quality into the risk exposure and assurance component, we find that CEO power is positively associated with both dimensions as well. Since the assurance related disclosures refer to positive information, hence, are perceived

(7)

7 favourably by stakeholders, CEOs with enhanced power might increase the disclosure quality, taken into consideration their stake in the company as well as their pay and tenure prospects. This provides an additional explanation for the observed positive relationship between CEO power and the overall IFRS 7 disclosure quality.

Besides, this research proves that the proportion of independent non-executive

directors strengthens this association, although the effect is incrementally lower for firms with a high board independence, in relation to firms with a low board independence. The

interpretation is that board independence limits the discretion of CEOs. Overall, the

implication of this finding, is that an independent board in combination with a powerful CEO could benefit the disclosure quality, since in that case, the board could have more capacity to provide valuable resources to the company (Duru et al., 2016).

This paper has multiple contributions. First of all, to the best of my knowledge, the association between CEO power and IFRS 7 disclosure quality has not been investigated yet, which raises the need for a study that could fill this literature gap. Above all, there is extant literature regarding CEO power and its impact on different firm characteristics, such as firm performance (Boyd, 1995; Daily & Johnson, 1997; Kang & Zardkoohi, 2005). Likewise, the existing literature with regard to the mandatory disclosures in financial statements that are prepared in accordance with IAS or IFRS is quite extensive. However, these studies mainly focus on the overall compliance or compliance based on a specific IFRS standard

(Atanasovski, 2015). Considering IFRS 7, i.e., the disclosure requirements for financial instruments, the available papers are more limited, while most of them examine the

compliance with IFRS again (Bischof, 2009; Amoako & Asante, 2013; Pucci & Tutino, 2013; Zango, Kamardin & Ishak, 2015). Besides, they are focusing on different companies, markets and perspectives. For instance, Bischof (2009) studied the effects of the IFRS 7 adoption on disclosure quality in the European banking sector, while Pucci and Tutino (2013) investigated the compliance with IFRS 7 requirements in an Italian context. Similarly, Amoako and Asante (2013) and Zango et al. (2015) investigated the compliance with IFRS 7 of listed banks in Ghana and Nigeria, respectively. Atanasovski (2015) and Hassan (2014) investigated another perspective within this research area, by looking at different firm characteristics that influence the quality of disclosures in accordance with IFRS 7, which is more in line with this research. However, within the range of potential determinants that could influence the (components of) IFRS 7 disclosure quality, CEO power has not been examined yet. Moreover, this study concentrates on non-financial entities in the UK, in contrast to most research above who focus

(8)

8 particularly on banks. Furthermore, it examines the quality of risk disclosures rather than the compliance with IFRS or the quantity of disclosures. According to Miihkinen (2012) who investigated this quantitative aspect, the latter only partially explains the total risk disclosure quality. Therefore, this research provides a more comprehensive understanding of IFRS 7 disclosure quality.

In addition, the novelty of this study also stems from the inclusion of the moderating influence of an independent board on the association between CEO power and IFRS 7 disclosure quality. Prior literature documented that board independence could have a moderating impact on the relationship between (a measure of) CEO power and some dependent variables, such as firm performance or earnings announcement tone (Duru et al., 2016; DeBoskey, Luo & Zhou, 2019). Therefore, this evidence provides an opportunity to examine whether this moderating role is also present in case of IFRS 7 disclosure quality as dependent variable, which makes it a valuable contribution to the literature.

Lastly, within the strand of literature regarding CEO power and other (non-risk) disclosures, the results are somewhat mixed. For instance, Li, Gong, Zhang and Koh (2018) investigated the moderating effect of CEO power on the relationship between environmental, social, and governance (ESG) disclosures and firm value. They found that a higher level of ESG disclosure was associated with a higher firm value and that CEO power enhanced this effect. The paper of Jizi et al. (2014) shows similar evidence, albeit with regard to CSR disclosures. On the contrary, Muttakin, Khan, and Mihret (2018) show that CEO power is negatively associated with CSR disclosures. Accordingly, there is a lack of unequivocal evidence with regard to CSR disclosures. This raises the question of whether and why the impact of CEO power on the quality of IFRS 7 disclosures is positive or negative and whether there is a difference between the risk exposure component and the assurance component. Therefore, this paper contributes to the existing, although limited, literature of IFRS 7 by examining which impact CEO power has on the quality of IFRS 7 disclosures, including the moderating role of an independent board, for non-financial premium listed companies in the UK, using recent data.

The remainder of this study is structured as follows. The next section provides a theoretical background with regard to risk disclosures and CEO power for the foundation of multiple hypotheses. The subsequent section discusses the methodology of the research, whereas the results are shown in section four. Finally, the conclusion and discussion are presented in the last section.

(9)

9

Theoretical background and hypothesis development

2.1 CEO power and disclosure quality

Based on the agency theory, we could predict the main relationship, as mentioned above, and provide possible answers to the research question. Agency theory has been used extensively among a lot of different domains, like accounting, economics, marketing, and sociology (Eisenhardt, 1989). Essentially, agency theory is directed at relationships in which principles (shareholders) delegate the responsibility to agents (managers) who perform the work, similar to a contract. However, this separation between ownership and control of the firm could potentially result in a conflict of interest (Daily & Johnson, 1997). Managers, including CEOs, are assumed to be self-interested and might not always act in the best interests of the shareholders, which leads to information asymmetry. With regard to this, agency theory explains how information asymmetry between the managers and shareholders can be mitigated through cost-effective monitoring tools, one of them being financial

reporting disclosures like IFRS 7 (Adznan & Nelson, 2014; Bischof, 2009). This way, IFRS 7, which aims at increasing the disclosure quality of risk exposure and assurance that is associated with the company’s financial instruments, reduces the information asymmetry and better aligns the interests of managers and shareholders.

According to Linsley and Shrives (2006), disclosures are defined as risk disclosures if they inform the reader of “any opportunity or prospect, or of any hazard, danger, harm,

threat or exposure, that has already impacted upon the company or may impact upon the company in the future or of the management of any such opportunity, prospect, hazard, harm, threat or exposure”. However, managers, including CEOs, might not want to report on risk

because of three reasons (Dobler, 2008). Firstly, the manager may not disclose risk information, since he/she does not or pretends not to have any. Secondly, a manager may withhold available non-verifiable risk information, because he/she chooses to misreport or is not able to credibly disclose the information. Thirdly, the threat of economic disadvantages could be a cause for a manager to not report risk information properly. All these incentives are reducing the quality of risk disclosures. However, the CEO has the highest say in the

organization, thus it is likely that the CEO could, ultimately, decide which information to disclose regarding risk reporting.

Consequently, the power of the CEO could have major implications for the IFRS 7 disclosure quality. “CEO power refers to the potential for the CEO to leverage ownership or

(10)

10

position to pursue her or his own goals” (Combs, Ketchen, Perryman & Donahue, 2007).

According to the agency theory, CEOs have considerable power over the resources of the company, because no shareholder could get direct control over the company and because shareholders are highly dispersed (Jensen & Meckling, 1976). Although there is evidence that powerful CEOs could have a positive influence on the company, like quicker response times and clear authority lines, it is likely that powerful CEOs abuse their power to pursue their own goals at the expense of shareholders and other stakeholders when their interests deviate

(Combs et al., 2007).

In order to measure CEO power, I propose three different variables, CEO duality, CEO tenure, and CEO ownership, as they are the three most widely used variables to measure CEO power (Cannella & Shen, 2001). All three variables, if positive and high, improve the power of the CEO. That is, a CEO with a dual role, long tenure, and large ownership, has greater power to actually make decisions and pursue goals that do not take into account the shareholders’ interests. In fact, their enhanced power might stimulate them to not disclose risk information with regard to financial instruments adequately because of the three reasons mentioned above, thereby reducing the IFRS 7 disclosure quality. Therefore, the private interests of powerful CEOs are likely to have an opposite effect to the reduction of

asymmetric information and the improved alignment of the interests of managers (CEOs) and shareholders as a result of the adoption of IFRS 7. Accordingly, I expect that CEO power has a negative impact on the quality of IFRS 7 disclosure practices, hence the first hypothesis:

Hypothesis 1: CEO power is negatively associated with IFRS 7 disclosure quality.

2.2 CEO power and the two distinctive components of disclosure quality

In addition, the IFRS 7 disclosure quality will be split into a risk exposure component and an assurance component, to examine whether CEO power has a different impact on each of the two dimensions of IFRS 7. The first component is an indication of negative or

unfavourable information, as it comprises the disclosure of the firm’s risk arising from financial instruments, whereas the second one provides assurance about the how the management addresses the risks and uncertainties associated with the use of financial instruments, which is related to their hedging activities. Hence, the assurance dimension of IFRS 7 refers to positive or favourable information.

(11)

11 With regard to risk exposure on the one hand, Karaibrahimoglu and Porumb (2019) report that risk disclosures are related with negative information about future performance of firms as a result of potential uncertainties. Furthermore, risk disclosures are positively associated with investor’s risk perception (Kravet & Muslu, 2003). In their paper, they also provide evidence for a higher stock return volatility, as well as a higher dispersion of earnings forecasts following an increase in risk disclosures. On the other hand, assurance-related disclosures reduce the risk perception of investors, thereby providing a positive signal to the firm’s stakeholders (Karaibrahimoglu & Porumb, 2019).

Concerning the association between CEO power and the two distinctive components of IFRS 7, two additional hypotheses are formulated. Kothari et al. (2009) suggest that CEOs who have more personal wealth at stake and greater career concerns are more likely to delay the disclosure of bad news. Therefore, I expect that, due to the self-interested behaviour and discretionary actions of CEOs, a more powerful CEO will reduce the IFRS 7 disclosure quality in terms of risk exposure items, but will increase the IFRS 7 disclosure quality with regard to hedging items. This results in the following two hypotheses:

Hypothesis 2: CEO power is negatively associated with the IFRS 7 disclosure quality of risk exposure items (risk exposure component).

Hypothesis 3: CEO power is positively associated with the IFRS 7 disclosure quality of hedging items (assurance component).

2.3 The moderating role of board independence

Referring to the agency theory, an independent board of directors, that is, a board with a high proportion of independent non-executive directors, is expected to be more effective in monitoring and controlling financial information, while excess inside (and thus, dependent) directors on the board are associated with lower financial reporting and disclosure quality (Vafeas, 2005; Jizi et al., 2014). This is because independent non-executive directors are not influenced by corporate insiders, resulting in a higher level of disclosure (Oliveira, Rodrigues, & Craig, 2011). In fact, Armstrong, Core, and Guay (2014) suggest that board independence leads to higher firm transparency, thereby reducing information costs. Additionally, Duru et al. (2016) argue that firms that endeavour to attract independent directors, engage in more transparent financial reporting and disclosure practices, leading to lower information asymmetry. Furthermore, as effective monitors, independent directors will limit managerial

(12)

12 opportunism and entrenchment (Jensen & Meckling, 1976; Duru et al., 2016). Besides,

effective board monitoring will prevent the CEO from abusing its power and will ensure that CEO power is used in such a way that it will benefit the firm (Combs et al., 2007). In other words, the self-interested discretionary actions of powerful CEOs, as well as their urge to pursue objectives that are not in accordance with that of the firms’ shareholders, will be mitigated in case of an independent board. Therefore, the increased monitoring effectiveness of independent board members might weaken the predicted negative relationship between CEO power and overall IFRS 7 disclosure quality, hence the fourth hypothesis:

Hypothesis 4: The negative association between CEO power and IFRS 7 disclosure quality is weaker for firms with a more independent board of directors.

2.4 Conceptual model

The conceptual model in Figure 1 graphically displays the above-mentioned hypotheses.

(13)

13

Research Methodology

3.1 Sample selection

The sample consists of UK non-financial companies with a premium listing on the London Stock Exchange during the period 2007-2016. This is based on the working paper of Karaibrahimoglu and Porumb (2019), which is used to gather data regarding IFRS 7

disclosure quality. The IFRS 7 disclosure data is hand-collected from the financial reports of each company in the sample. This data includes information about qualitative as well as quantitative disclosures of risk exposures of financial instruments and the way in which these risks are managed, which is the assurance component. The data regarding CEO power, board independence, and the governance-related control variables are obtained from the BoardEx database. In this research, only individuals who serve the role as a CEO are used, which results in a panel dataset that includes CEO’s information about their compensation, as well as their wealth, time in role, and whether they also are the chairman of the board. Furthermore, the dataset includes data about the number of independent non-executive directors on the company’s board and on the audit committee to measure board independence and audit committee independence, respectively. Additionally, BoardEx provided the total number of directors on the company’s board which corresponds to the size of the board. The remaining financial control variables are obtained from the Compustat database.

After matching the governance and the financial data with the IFRS 7 disclosure quality data using the ISIN codes and eliminating missing values of all variables of interest, the final sample consists of 1,852 firm-year observations.

3.2 Research design

Research model 1

In order to test the hypotheses, an ordinary least squares (OLS) regression with robust standard errors is used in this research. With regard to the first hypothesis, in which I

hypothesize that CEO power has a negative influence on IFRS 7 disclosure quality, the following model is developed:

(14)

14

IFRS7_DQ_All = β0 + β1CEO_Powerit + β2Board_Independenceit +

β3CEO_Compensationit + β4Independent_ACit + β5Board_Sizeit + β6ROAit +

β7Leverage + β8Firm_Sizeit + β9Year_Dummyit + β10Industry_Dummyit + εit (1)

Where IFRS7_DQ_All is an index of the overall disclosure quality calculated as the equal weighted average of the normalized value of 13 items that measures the extent to which firms’ disclosures are in line with the IFRS 7 best practices. The selection of items used in this index is based on the items that have the highest relevance for the users of financial statements, provided by IFRS 7, as well as on a pilot study performed in Karaibrahimoglu and Porumb (2019).

The main variable of interest in Model 1 is CEO_Power. In this study, an index is derived from three variables in order to measure CEO power, which are duality, tenure, and ownership. The index variable is computed as the equal weighted average of the three items mentioned above, using normalized values.

First of all, CEO duality occurs when the CEO also serves as a chairman of the board of directors (Daily & Johnson, 1997). If this is the case, a CEO could exert more influence over the board and hide crucial information from other directors, as the chairman has the ability to influence information provided to other board members, controls the agenda of board meetings, and guides the nomination of new board members and (Imhoff, 2003; Jizi et al., 2014). CEO duality is a dichotomous variable which equals 1 if the CEO is also the board chairman (Cannella & Shen, 2001).

Secondly, CEO tenure is another important determinant to measure the power of CEOs. According to Muttakin et al. (2018), CEOs who have held their position for a long period, are more likely to have more managerial power, because they are becoming

entrenched. Therefore, CEOs with long tenures are expected to have more control over the board, hence, they tend to place more emphasis on their own interests at the expense of the interests of shareholders (Chen, Lu, & Sougiannis, 2012). CEO tenure is defined as the number of years that the CEO has served the board (Combs et al., 2007).

The third variable that measures how powerful CEOs are, is CEO ownership. CEOs with a large portion of shares, are prone to be more powerful than CEOs without an

(15)

15 process, and they could prevent their own involuntary dismissal (Fredrickson, Hambrick & Baumrin, 1988). In this paper, CEO ownership is measured as the value of CEO's total wealth divided by the firm’s market capitalization (de Andrés, Arranz-Aperte & Rodriguez-Sanz, 2017). The market capitalization data is obtained from the Compustat database and calculated by multiplying the share price with the total number of shares outstanding. As the governance data of each firm from BoardEx is related to one particular month of a year, the market capitalization for each firm is calculated using data of the last available day of the corresponding month.

I expect the coefficient of CEO_Power to be negative, as CEOs with a dual role, long tenure, and/or large ownership, tend to have more discretion to act in their own interest and pursue goals that are inconsistent with shareholders’ interests. Hence, they may not disclose risk information adequately, resulting in a lower quality of IFRS 7 disclosures.

Furthermore, I have included several control variables in order to strengthen the relationship between CEO power and the IFRS 7 disclosure quality. Consistent with prior literature, the control variables are: CEO compensation, the natural logarithm of CEO

compensation, audit committee independence, board size, profitability, leverage, and firm size (Cannella & Shen, 2001; Combs et al., 2007; Miihkinen, 2012; Muttakin et al., 2018).

Combs et al. (2007) argue that one way to employ CEO power is when CEOs pressure the board to increase their compensation. Therefore, I control for CEO compensation

(CEO_Compensation), which is calculated as the natural logarithm of the total direct compensation (salary & bonus) plus the total equity linked compensation of the CEO.

Furthermore, firms that have an independent audit committee are more likely to have a higher disclosure quality, as the audit committee is a monitoring mechanism that improves the disclosure quality and reliability of corporate reporting, while it reduces the information asymmetry between management and shareholders as well as management opportunism (Akhtaruddin & Haron, 2010; Cormier, Ledoux Magnan & Aerts, 2010; Jizi et al., 2014). This is why the variable Independent_AC is included in the empirical model. The main function of the audit committee is to oversee and monitor the corporate financial reporting process of an organization (Xie, Davidson & DaDalt, 2003; Rahman & Ali, 2006). However, in order to become effective, it should be independent and include non-executive directors (Oliveira et al., 2011). Therefore, the variable Independent_AC is defined as the ratio of independent non-executive in the audit committee directors (excluding any chairman that is considered

(16)

16 independent and non-executive) to the total number of audit committee members (Bukit and Iskandar, 2009).

Board size is another control variable that is taken into account, as directors of larger boards could engage in more free-riding or opportunistic behaviour compared to small boards (Muttakin et al., 2018). In that sense, their discretionary actions could negatively influence the disclosure quality. Board_Size is defined as the total number of directors on the firm’s board.

In addition, ROA is defined as the ratio of earnings before interest and taxes to the total assets to control for firm profitability, whereas Leverage is used to control for financial risk and agency costs. Firm leverage is measured by calculating the ratio of the firms’ total debts (current liabilities plus long-term debt) to total assets. Lastly, Firm_Size is measured as the natural logarithm of the firm’s total assets to control for firm size. According to Miihkinen (2012) and Hassan (2014), profitability, leverage and firm size are important drivers for the quality of disclosures, which is why these control variables are included as well.

Research model 2

For the second and third hypothesis, the overall index of IFRS 7 disclosure quality is split into two distinctive components, risk exposure and assurance, as thoroughly discussed in the sections above. Model 2 is used to assess the association between CEO power and the disclosure quality of the risk exposure component of IFRS 7:

IFRS7_DQ_Risk = β0 + β1CEO_Powerit + β2Board_Independenceit +

Where IFRS7_DQ_Risk is an index variable of the risk exposure related disclosure quality that consists of 3 items on risk exposure, risk concentration and impairment

(Karaibrahimoglu & Porumb, 2019). These are selected from the 13 items used for the IFRS 7 disclosure quality index. Subsequently, the quality index is computed as the equal weighted average of the 3 items, using normalized values.

(17)

17 The main variable of interest, CEO_Power, is expected to have a negative coefficient, as the disclosure of risk exposure items refers to negative information about future firm performance and increases the risk perception of investors. Given the CEO’s career concerns and stake in the company, CEO power is likely to be associated with a reduction in the risk exposure related disclosures.

Research model 3

In accordance with the previous model, model 3 depicts the assurance component of IFRS 7 disclosure quality, in order to examine the third hypothesis, i.e., the relation between CEO power and the disclosure quality of the hedging activities arising from the use of financial instruments. Hence, the following model:

IFRS7_DQ_Assurance = β0 + β1CEO_Powerit + β2Board_Independenceit +

Where IFRS7_DQ_Assurance is an index of IFRS 7 disclosure quality related to the assurance component. The variable is computed as the equal weighted average of the normalized value of 6 pre-determined items out of the 13 items of the overall disclosure quality index. These 6 items include disclosures of hedging activities with regard to interest rate risk, liquidity risk, currency risk, credit risk, and other price risks, as well as disclosures of the existence of collateral for credit risk exposure (Karaibrahimoglu & Porumb, 2019).

Again, the main variable of interest is CEO_Power, although the predicted sign is positive rather than negative for this model. The reasoning is similar to the above, however, because the assurance related disclosures involve items about hedging activities arising from financial instruments, it will reduce the investor’s risk perception. Hence, this information is perceived positively by stakeholders and therefore, the coefficient is expected to be positive as well.

(18)

18

Research model 4

To test the fourth hypothesis, whether board independence has a positive moderating influence on the predicted negative effect between CEO power and the overall IFRS 7 disclosure quality, I estimated the following model:

IFRS7_DQ_All = β0 + β1CEO_Powerit + β2Board_Independenceit + β3CEO_Powerit x

Board_Independenceit + β4CEO_Compensationit + β5Independent_ACit +

β6Board_Sizeit + β7ROAit + β8Leverage + β9Firm_Sizeit + β10Year_Dummyit +

β11Industry_Dummyit + εit (4)

Where Board_Independence is defined as the ratio of the number of independent non-executive directors to the total number of directors on the board. First of all, the responsibility of the board is to scrutinize the financial reporting and disclosure of the firm, to protect shareholders’ interests (DeBoskey et al., 2019). In addition, Vafeas (2005) argues that a board with a higher number of independent directors will be more effective monitors with regard to financial information, while too many inside directors will impair financial reporting and disclosure quality. Therefore, I expect that the coefficient of Board_Independencewill be positive.

The main variable of interest in Model 2 is the interaction term CEO_Power x

Board_Independence. Drawing upon research of Duru et al. (2016), who provide evidence for

a negative association between CEO duality and firm performance that is positively moderated by board independence, I expect the coefficientof CEO_Power x

Board_Independence to be positive as well. That is, board independence is expected to have a

positive moderating influence on the association between CEO power and disclosure quality based on IFRS 7. This is because Duru et al. (2016) suggest that due to their effective monitoring, an independent board is more likely to engage in transparent financial reporting and disclosure practices, thereby reducing information asymmetry, which ultimately will limit the self-interested behaviour and discretionary actions of powerful CEOs. Similarly, but with the opposite effects, DeBoskey et al. (2019) find that board independence negatively

(19)

19 example of voluntary disclosure. Although it is a negative moderating effect, the rationale behind it is the same.

Research model 5 and 6

Apart from the additional tests of the two separate IFRS 7 components with regard to the main relationship between CEO power and IFRS 7 disclosure quality, model 5 and 6 are constructed in order to examine the above-mentioned moderating effect in respect of the risk exposure related disclosures and assurance related disclosures, respectively. In essence, the two models are similar to model 4, though the minor difference is the dependent variable it contains. Furthermore, as board independence increases the monitoring ability of directors, which results in more transparent financial reporting and higher disclosure quality in general, I predict that the coefficient of the main variable of interest, CEO_Power x

Board_Independence, will be positive for both scenarios. The corresponding models are

presented below:

IFRS7_DQ_Risk = β0 + β1CEO_Powerit + β2Board_Independenceit + β3CEO_Powerit

x Board_Independenceit + β4CEO_Compensationit + β5Independent_ACit +

β6Board_Sizeit + β7ROAit + β8Leverage + β9Firm_Sizeit + β10Year_Dummyit +

β11Industry_Dummyit + εit (5)

IFRS7_DQ_Assurance = β0 + β1CEO_Powerit + β2Board_Independenceit +

β3CEO_Powerit x Board_Independenceit + β4CEO_Compensationit +

β5Independent_ACit + β6Board_Sizeit + β7ROAit + β8Leverage + β9Firm_Sizeit +

β10Year_Dummyit + β11Industry_Dummyit + εit (6)

(20)

20

Results

4.1 Descriptive statistics

Table 1 shows the descriptive statistics of the variables used in this study. The mean of the disclosure quality index (IFRS7_DQ_All) is about 0.364, which indicates that on average, firms comply to the 13 selected items that are important to financial statement users by 36.4% of the disclosures. Regarding the risk exposure and assurance components of IFRS 7, this percentage is 39.8 and 32.1, and refers to the 3 selected risk exposure items and 6 selected hedging items, respectively.

With regard to the CEO power index (CEO_Power) the average is 0.104. The average for the independent audit committee (Board_Independence) is close to 50% (0.569), meaning that independent non-executive directors comprise approximately half of the board of

directors, on average. Furthermore, the average logarithm of the compensation of CEOs is 7.256, whereas the average audit committee of the firms in the sample consists of a relatively high number of independent non-executive directors (0.936). Moreover, the board of directors consists of approximately 8 members on average (8,339). In addition, the mean of the

profitability (ROA) of the firms used in the sample is 0.096, their leverage (Leverage) amounts, on average, to 0.211, while their average size (Firm_Size) is 6.988.

Due to outliers that are present for IFRS7_DQ_All, IFRS7_DQ_Risk,

IFRS7_DQ_Assurance, and the financial control variables CEO_Compensation, Leverage and Firm_Size, these variables are winsorized at the 1% and 99% level.

(21)

21 Table 1. Descriptive statistics

4.2 Correlation matrix

In Table 2, the Pearson correlation matrix is presented which shows the correlation coefficients for all variables. Regarding the correlations with the dependent variable

IFRS7_DQ_All, all variables are statistically significant at the 5% level and are relatively low

correlated. This means that no variable is moving severely in the same or opposite direction of IFRS 7 disclosure quality. However, the aforementioned low correlation does not hold for

IFRS7_DQ_Risk and IFRS7_DQ_Assurance. Nevertheless, these dependent variables are

examined separately from IFRS7_DQ_All, so we do not need to take this into account. Moreover, similar to the overall index, the index variables of the two distinct components show relatively low correlation coefficients compared to the other independent variables, Yet, in contrast to the overall IFRS 7 disclosure quality, not all independent variables are

statistically significant at the 5% level.

Interestingly, with respect to IFRS7_DQ_All, CEO power shows a significant negative correlation (r = -0.073), providing initial evidence for hypothesis 1. Furthermore, board independence is positively correlated with the dependent variable (r = 0.254), meaning that as board independence goes up, the IFRS 7 disclosure quality increases as well. Apart from the profitability (ROA), all other control variables are positively correlated as well. As regards

Variables Obs Mean Std. Dev. Min Max

Dependent variables IFRS7_DQ_All 1,852 0.364 0.173 0 0.769 IFRS7_DQ_Risk 1,852 0.398 0.265 0 1 IFRS7_DQ_Assurance 1,852 0.321 0.240 0 0.917 Independent variable CEO_Power 1,852 0.104 0.108 0.001 0.560 Moderator variable Board_Independence 1,852 0.569 0.137 0 0.889 Control variables CEO_Compensation 1,852 7.256 0.908 4.804 9.568 Independent_AC 1,852 0.936 0.151 0 1 Board_Size 1,852 8.339 2.159 3 17 ROA 1,852 0.096 0.076 -0.919 0.450 Leverage 1,852 0.211 0.164 0 0.705 Firm_Size 1,852 6.988 1.692 3.691 11.754

The variables IFRS7_DQ_All, IFRS7_DQ_Risk, IFRS7_DQ_Assurance, CEO_Compensation, Leverage & Firm_Size are winsorized at the 1% and 99% level

(22)

22

IFRS7_DQ_Risk and IFRS7_DQ_Assurance, the correlation with CEO power is insignificant,

but the remaining significant variables are nearly similar to the overall index in terms of (the direction of) their coefficients. Overall, the correlation coefficients (except for the assurance component of IFRS 7, as already explained) are below the threshold of 0.7, indicating that multicollinearity will not be an issue. In addition, Table 3 shows that the variance inflation factors for all independent variables are below 10, which supports the argument stated above.

Table 2. Pearson correlation matrix

Variables (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) IFRS7_DQ_All (1) 1,000 IFRS7_DQ_Risk (2) 0.638* 1,000 IFRS7_DQ_Assurance (3) 0.794* 0.344* 1,000 CEO_Power (4) -0.073* 0.005 -0.033 1,000 Board_Independence (5) 0.254* 0.134* 0.200* -0.268* 1,000 CEO_Compensation (6) 0.204* 0.078* 0.149* -0.242* 0.385* 1,000 Independent_AC (7) 0.120* 0.016 0.081* -0.105* 0.441* 0.321* 1,000 Board_Size (8) 0.164* -0.019 0.168* -0.075* 0.164* 0.504* 0.257* 1,000 ROA (9) -0.049* 0.005 -0.024 0.009 -0.049* 0.102* 0.010 -0.015 1,000 Leverage (10) 0.174* 0.081* 0.104* -0.178* 0.097* 0.187* 0.084* 0.196* -0.013 1,000 Firm_Size (11) 0.293* 0.093* 0.202* -0.213* 0.393* 0.697* 0.360* 0.662* -0.112* 0.345* 1,000 *significant at the 5% level

All variables are standardized; they have a mean of zero and a standard deviation of one

The variables IFRS7_DQ_All, IFRS7_DQ_Risk, IFRS7_DQ_Assurance, CEO_Compensation, Leverage & Firm_Size are winsorized at the 1% and 99% level

Table 3. Variance Inflation Factors

Variables VIF 1/VIF

Firm_Size 3.22 0.310392 CEO_Compensation 2.21 0.452709 Board_Size 1.86 0.536499 Board_Independence 1.47 0.680006 Independent_AC 1.33 0.754493 Leverage 1.17 0.855169 CEO_Power 1.13 0.882375 ROA 1.09 0.914668 Mean VIF 1.69

(23)

23 4.3 Regression results for IFRS 7 disclosure quality

In Table 4, the results of the OLS regression using robust standard errors are

presented, in order to test the first three hypotheses. The first model (1) shows the results for the first hypothesis, while the second (2) and third (3) model present the findings in respect of hypothesis 2 and 3, respectively. The R-squared of model 1 is 17.3%, which means that 17.3% of the variation of the dependent variable is explained by the other variables in the regression models. For model 2, this R-squared amounts to 12.3%, while model 3 shows a value of 11.3%.

Hypothesis 1 predicted that CEO power is negatively related with IFRS 7 disclosure quality. The agency theory literature anticipated that, in case of interest deviations between managers and shareholders, CEOs are likely to abuse their power to pursue their own objectives which are inconsistent with maximizing shareholders’ wealth. Therefore, the expectation was that CEOs with enhanced power would counteract the expected reduction in information asymmetry and improvement in the alignment of interests of both managers and shareholders, when firms comply with the IFRS 7 standard. Hence, the predicted negative association between CEO power and IFRS 7 disclosure quality. Surprisingly, the main variable of interest, CEO_Power is significantly positively associated with IFRS7_DQ_All (P<0.05), which indicates that the overall disclosure quality will increase when the CEO is more powerful. That is, when CEOs have a dual role, serve the company’s board for a longer period of time and/or have a higher proportion of shares. Thus, hypothesis 1 is not supported.

Nevertheless, the additional tests of the two components of IFRS 7 disclosure quality, risk exposure and assurance, could provide more insights about the positive relationship between CEO power and the overall disclosure quality. Above all, the results of model 2 illustrate that CEO power is positively associated with the disclosure quality of the risk exposure items (P<0.01), which is in contrast with hypothesis 2, that predicted a negative association due to its negative perception. However, the results of the third model are in line with the corresponding hypothesis 3, implying that CEO power positively affects the

assurance component of the disclosure quality (P<0.05). Hence, a more powerful CEO is associated with higher disclosure quality of items about a firm’s hedging activities that arise from the use of financial instruments, which could be explained by the positive information it provides to investors and other stakeholders. So, according to the regression results of Table 4, hypothesis 2 is rejected, while the third hypothesis can be accepted.

(24)

24 Furthermore, in line with prior evidence of Abraham & Cox (2007) and Oliveira et al. (2011), Table 4 shows that the coefficient of the moderator variable, Board_Independence, is positive and significant for all three models (P<0.01). This result implies that if a board consists of a relatively high number of independent non-executive directors, the IFRS 7 disclosure quality will increase. Accordingly, this result is predicted by the agency theory, as board independence reduces the agency costs due to more transparent financial reporting and increased monitoring effectiveness. As a result, the information asymmetry diminishes, leading to a higher quality of IFRS 7 disclosures, which is confirmed by the findings.

With regard to the control variables, Table 4 shows that Leverage has a significant positive effect on both IFRS 7 disclosure quality in general, as well as for the two distinct components (P<0.01). The variable Firm_Size is only significantly positive for the overall disclosure quality (P<0.01) and for the assurance component of IFRS 7 (P<0.10). For the aforementioned significant results, this indicates that larger firms with an increased level of leverage tend to disclose higher quality information on the risk exposure items and hedging items arising from financial instruments. Though, higher leveraged firms are not significantly associated with a higher disclosure quality on the risk exposure component alone, which may be due to the negative information it reveals in combination with the higher threat of financial distress for these firms. This interpretation is supported by the fact that the relationship is positive and significant for the assurance related disclosures, which refer to positive information.

Nonetheless, the results above are in accordance with the agency theory, which suggests that because larger and highly leveraged firms have more agency costs (due to increased monitoring costs and difficulties) than smaller firms with low levels of leverage, they disclose more and higher quality information in order to reduce these agency costs (Jensen & Meckling, 1976; Oliveira et al., 2011; Hassan, 2014; Atanasovski, 2015). In addition, with regard to Leverage, the results imply that highly leveraged firms, which are characterized by a higher degree of financial risk, may seek to provide higher disclosure quality information because of the following reasons: to avoid litigation risk, because they experience pressure to explain the causes of this increased risk, or to provide a signal about the way the firm manages these risks effectively (Hassan, 2014). With regard to Firm_Size, larger firms are able to incur more costs to obtain disclosure quality information compared to smaller firms. Hence, they have more resources and therefore, they can pay more attention to improve the IFRS 7 disclosure quality (Hassan, 2014).

(25)

25 Table 4. Regression results for IFRS 7 disclosure quality

(1) (2) (3)

Variables IFRS7_DQ_All IFRS7_DQ_Risk IFRS7_DQ_Assurance

CEO_Power 0.047** 0.065*** 0.060** (0.024) (0.025) (0.026) Board_Independence 0.163*** 0.107*** 0.159*** (0.027) (0.028) (0.028) CEO_Compensation 0.003 0.060* -0.006 (0.033) (0.034) (0.036) Independent_AC -0.022 -0.038 -0.044* (0.024) (0.029) (0.026) Board_Size 0.008 -0.092*** 0.112*** (0.031) (0.031) (0.032) ROA -0.013 0.026 0.000 (0.023) (0.022) (0.032) Leverage 0.094*** 0.088*** 0.052** (0.027) (0.025) (0.026) Firm_Size 0.200*** 0.061 0.078* (0.041) (0.041) (0.045) Constant -0.100 0.290** -0.179 (0.115) (0.123) (0.116)

Year dummies Yes Yes Yes

Industry dummies Yes Yes Yes

Observations 1,852 1,852 1,852

R-squared 0.173 0.123 0.113

Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

All variables are standardized; they have a mean of zero and a standard deviation of one The variables IFRS7_DQ_All, IFRS7_DQ_Risk, IFRS7_DQ_Assurance, CEO_Compensation, Leverage & Firm_Size are winsorized at the 1% and 99% level

4.4 Regression results for the moderating effect of board independence

The results for hypothesis 4 are shown by the fourth model (4) of Table 5. The continuous by continuous interaction variable (CEO_Power x Board_Independence) is

statistically significant at the 1% level (P<0.01), but has a negative coefficient. At first glance, this indicates that board independence negatively moderates the relationship between CEO power and IFRS 7 disclosure quality, which is opposite to what was predicted by hypothesis

(26)

26 4. In fact, I anticipated a positive moderating influence, as the increased monitoring

effectiveness of independent board members would be likely to mitigate discretionary actions of CEOs, including their urge to pursue objectives at the expense of shareholders’ wealth, which decrease the disclosure quality.

Table 5. Regression results for the moderating effect of board independence

(4) (5) (6)

Variables IFRS7_DQ_All IFRS7_DQ_Risk IFRS7_DQ_Assurance

CEO_Power 0.201*** 0.233*** 0.132** (0.055) (0.057) (0.067) Board_Independence 0.208*** 0.156*** 0.180*** (0.032) (0.034) (0.034) CEO_Power x Board_Independence -0.158*** -0.172*** -0.074 (0.057) (0.058) (0.071) CEO_Compensation 0.007 0.064* -0.005 (0.033) (0.034) (0.036) Independent_AC -0.017 -0.032 -0.042 (0.024) (0.029) (0.026) Board_Size 0.007 -0.093*** 0.112*** (0.031) (0.031) (0.032) ROA -0.013 0.026 0.000 (0.023) (0.022) (0.033) Leverage 0.095*** 0.090*** 0.052** (0.027) (0.025) (0.026) Firm_Size 0.196*** 0.056 0.076* (0.041) (0.041) (0.045) Constant -0.101 0.289** -0.179 (0.115) (0.122) (0.116)

Year dummies Yes Yes Yes

Industry dummies Yes Yes Yes

Observations 1,852 1,852 1,852

R-squared 0.175 0.126 0.113

Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

All variables are standardized; they have a mean of zero and a standard deviation of one

The variables IFRS7_DQ_All, IFRS7_DQ_Risk, IFRS7_DQ_Assurance, CEO_Compensation, Leverage & Firm_Size are winsorized at the 1% and 99% level

(27)

27 The additional models (5) and (6), in which the IFRS 7 is split into two components, show a similar outcome, although the coefficient of the assurance related disclosures is not significant. Therefore, the results of model 6 are not taken into consideration with respect to the moderating influence of board independence. Concerning the remaining two models, the effect of board independence on the association between CEO power and IFRS 7 disclosure quality can be visualized by using an interaction plot or marginsplot, in order to obtain a better understanding of the effect of the interaction variable. In the process of creating this interaction plot, the simple slopes of the dependent variables, IFRS7_DQ_All and

IFRS7_DQ_Risk, on the independent variable, CEO_Power, will be calculated, while holding

the moderator variable Board_Independence constant at different combinations of values, ranging from very low to very high. Afterwards, the simple slopes are visualized in an interaction plot by using regression lines. Each line represents the effect on the dependent variable as a result of a one-unit change in the independent variable, while

Board_Independence is held constant at different values. To recall, these values represent the

ratio of independent non-executive directors to the total number of directors on the board.

Figure 2 presents the interaction plot with respect to IFRS_DQ_All, while Figure 3 illustrates an interaction plot as well, albeit with regard to IFRS_DQ_Risk. For simplicity, both figures only show the effect of board independence on the relationship between CEO power and (the risk exposure component of) IFRS disclosure quality for the minimum and maximum value (as a matter of fact, the 1 and 99 percentile values) of board independence. From both figures, as they are comparable to each other, we can derive that the overall IFRS 7 disclosure quality as well as its risk exposure component becomes higher when the power of the CEO is stronger. Evidently, relative to firms with low board independence, firms with a high proportion of independent non-executive directors experience an incrementally lower increase in the disclosure quality, since the line that represents low board independence is increasing, while the line for high board independence is slightly decreasing.

To summarize, the interaction plots presented in Figure 2 and Figure 3 indicate that the positive association between CEO power and IFRS 7 disclosure quality, and between CEO power and the risk exposure component of IFRS 7, is stronger for firms that have a more independent board of directors. Hence, board independence has a positive moderating effect. However, this effect is less positive for firms with high board independence, in relation to firms with low board independence.

(28)

28 Figure 2. Interaction plot IFRS7_DQ_All

(29)

29 With regard to hypothesis 4, the results do not support the predicted relationship, as board independence was expected to weaken the negative association between powerful CEOs and the overall disclosure quality index. Instead, the main association appeared to be positive, and the moderating influence is positive as well. Nevertheless, I predicted that board independence would have a positive moderating influence, which is in line with the results. In essence, a weakening of the negative association, as predicted by hypothesis 4, is similar to a strengthening of the positive association, as is shown in the interaction plots. So, the

hypothesis in itself is rejected, though the rationale behind it is in accordance with the findings.

Considering the other independent variables, Table 4 and Table 5 show that all variables that were significant in model 1, are still significant in model 4. This provides even stronger evidence for all the above-mentioned associations regarding CEO_Power,

Board_Independence, Leverage, and Firm_Size and the dependent variable IFRS7_DQ_All.

The same holds for the two separate components of IFRS 7. The remaining variables are either not significant at all, or not significant for both the overall disclosure quality index and its two components. Therefore, these results are not taken into consideration in the remainder of this paper.

(30)

30

Conclusion and discussion

The aim of this research was to examine the impact of CEO power on the IFRS 7 disclosure quality by using a sample of non-financial premium listed firms in the United Kingdom, between the period 2007-2016. Moreover, the moderating effect of an independent board of directors, that is, a board that consists of independent non-executive directors, on the association between CEO power and IFRS 7 disclosure quality is examined. Additional tests have been performed in order to examine the impact of the main independent variable on each of the two components of IFRS 7 disclosure quality separately, as the standard incorporates disclosures that refer to both positive and negative information.

5.1 Effect of CEO power on IFRS 7 disclosure quality

According to the literature review, I predicted a negative association between the power of CEOs and the disclosure quality of both the risk exposure and assurance component of IFRS 7, in this paper referred to as ‘IFRS 7 disclosure quality’. However, if only one component was taken into account, the association was expected to be negative in respect of the risk exposure related disclosures of IFRS 7 as it refers to unfavourable information, while it would be positive for the assurance related disclosures, since that indicates favourable information.

The rationale for the main relationship was mainly based on the agency theory, which assumes that CEOs are self-interested and therefore, might not always act in the best interests of the shareholders. Furthermore, the theory suggests that CEOs have considerable power over the resources of the company, because no shareholder could get direct control over the company and because shareholders are highly dispersed (Jensen & Meckling, 1976). In fact, powerful CEOs could have more discretion to act in their own interest and are more likely to abuse their power to pursue objectives at the expense of shareholders when their interests deviate (Daily & Johnson, 1997; Combs et al., 2007). Therefore, the private interests of CEOs were likely to have an opposite effect on the presumed reduction in information asymmetry and the improved alignment of the interests of CEOs and shareholders, as a result of the adoption of IFRS 7.

In contrast to hypothesis 1, the results of the OLS regression provide evidence for a positive association between CEO power and IFRS 7 disclosure quality. This implies that CEOs who also serve as a chairman of the board, who have a longer tenure, and/or a larger

(31)

31 ownership, tend to have a positive impact on the quality of IFRS 7 disclosures, thereby

meeting the objectives of IFRS 7. So, considering the agency theory literature, CEOs appear not to abuse their power to pursue their own objectives at the expense of shareholders, or their interests do not considerably deviate from each other in such a way that the CEO misuses its power and as a result, impairs the disclosure quality. Besides, despite the fact that CEOs have the highest say in the organization and decide which information to disclose (and which not), they might not be influenced by various incentives, as suggested by Dobler (2008), that lead to inadequate disclosures of risk information, and hence, reduce the quality of risk

disclosures.

The interpretation of these findings can be found in the research of Daily and Johnson (1997), who suggest that powerful CEOs do not necessarily engage in self-interested

behaviour. The rationale behind this insight, is that these CEOs may be protected from different organisational pressures which enables them to behave in accordance with

shareholders’ interests and the interests of the firm. In addition, Combs et al. (2007) state that there is evidence that powerful CEOs have a positive influence on the company, like clear authority lines and quicker response times. This could also be the case for the CEOs in the sample, which would explain the positive association. Another interesting implication of this result is related to CSR disclosures. Jizi et al. (2014) report that CEOs with enhanced power may improve transparency about their CSR activities in order to become more successful, or to increase their pay or tenure prospects. With regard to IFRS 7 disclosures, we could infer the same. Specifically, powerful CEOs might want to increase the quality of disclosures, because they aim for more success, in terms of salary, bonus or ownership or because they want to stay in the company for a little while longer.

Until now, the focus has been on the implication of the findings with regard the overall disclosure quality, whereas the results of the additional tests with respect to the two separate components of IFRS 7 shed a different light on how to interpret the findings of the main relationship. Namely, the results showed that CEO power is positively associated with the assurance component of IFRS 7 disclosure quality, in accordance with the prediction, as this type of information will reduce the risk perception of investors. Besides, the risk exposure component proved to have a positive association with the power of CEOs as well, although this effect was not anticipated. The logic for this finding may correspond with the notion of Daily and Johnson (1997) as stated above. That being said, a significant part (6 out of 13 items) of the positive association between CEO power and the overall IFRS 7 disclosure

(32)

32 quality is attributed to the assurance related disclosures. Therefore, an additional explanation why CEO power is positively related to the disclosure quality in accordance with IFRS 7, is because a substantial part of the disclosures is perceived positively by stakeholders, rather than negatively, which could be an incentive for powerful CEOs to capitalize on their ability to pursue their own objectives, considering their stake in the company and career concerns. Referring to the work of Jizi et al. (2014) as mentioned before, the explanation why CEOs might aim to improve the disclosure quality, in order to increase their pay and tenure

prospects or to become more successful, is because the assurance related disclosures refer to favourable information, which is considered as being positive in the eyes of investors and other stakeholders.

To conclude, one major implication is that powerful CEOs do not always harm the company, nor have a negative impact on certain firm characteristics or elements, as the observed IFRS 7 disclosure quality for instance. This could be particularly important for investors, stakeholders and other users of financial statements, since risk disclosures assist them in making economic decisions, and provide information which enables them to assess the risks that affect the future economic performance of firms (Hassan, 2014). In addition, it is always important to focus on both dimensions of such an element if applicable, since in this research, the results are largely driven by the positive dimension of IFRS 7. Similar to the above, this implication could be valuable as well for users of financial statements in order to make more effective decisions.

5.2 Effect of board independence on association between CEO power and IFRS 7 disclosure quality

In respect of the moderator variable, that represents the proportion of independent non-executive directors on the board, the expectation was that it would positively impact the main relationship, because of the increased monitoring effectiveness of independent board

members. In fact, an independent board, ultimately, leads to a higher IFRS 7 disclosure quality due to more effective monitoring, because they are not influenced by corporate insiders (Vafeas, 2005; Oliveira et al., 2011; Jizi et al., 2014). This leads to a higher transparency with regard to financial reporting and disclosure practices (Duru et al., 2016). Consequently, this reduces information asymmetry, which will limit discretionary actions of powerful CEOs and prevents that the CEO will abuse its power (Combs et al., 2007).

(33)

33 although with regard to a different dependent variable (Duru et al., 2016; DeBoskey et al., 2019).

According to the regression results, board independence has indeed a positive moderating influence on the association between CEO power and (the exposure component of) IFRS 7 disclosure quality, but this effect is slightly decreasing for firms with a high proportion of independent non-executive directors, in relation to firms that have a low board independence. This implies that board independence limits the CEO’s discretion used over the IFRS 7 disclosures, which is in accordance with the expectation that independent directors are likely to limit managerial opportunism and prevent the CEO from abusing its power (Combs et al., 2007; Duru et al., 2016). However, this expectation is based on the belief that CEOs are

not using their power in such a way that it will benefit the firm and that an independent board

increases the disclosure quality. On the contrary, the main regression results indicated that more powerful CEOs tend to increase IFRS 7 disclosure quality, thereby being beneficial to the firm, and the results of the interaction plots show an incrementally lower increase in disclosure quality as board independence goes up. So, the restricted discretion of CEOs used over the disclosures as a result of higher board independence does not necessarily result in a higher disclosure quality which in turn will benefit the firm, as is suggested by prior research.

Aside from the notion above, the general managerial implication of these results is that, given the power of the CEO, firms should have some independent non-executive

directors on the board if they aim for a higher IFRS 7 disclosure quality. As stated by Duru et al. (2016), “a powerful CEO enhances the board’s capability to provide valuable resources to

the firm, including providing advice …”. Thus, regardless of the fact that the CEO’s private

interests and discretionary actions do not appear to have a negative impact on the IFRS 7 disclosure quality, the monitoring role of an independent board of directors seems to have a positive influence on the disclosure quality, as it strengthens the positive relation between CEO power and disclosure quality based on IFRS 7.

5.3 Research limitations and further research

In performing a research like this, it is inevitable to not encounter any limitations. Therefore, this research includes the following limitations. First of all, the internal validity of the calculation of the CEO power index could be an issue. According to Cannella and Shen (2001), the three variables duality, tenure, and ownership are the most widely used variables

(34)

34 to measure CEO power. However, Daily and Johnson (1997) use four different sources of CEO power, each with one, two or three measures. Likewise, Pathan (2009) measures CEO power by CEO duality and whether the CEO is internally-hired. Therefore, the results could be different when using an index of CEO power that consists of different measures, or when using another method in order to compute the index, such as PCA analysis. A second

limitation is that this research uses proxy variables, like the composition of the board of directors, for generating the board independence of firms, while these proxy variables do not capture all elements of the complex network of the social relationships between CEOs and directors (Combs et al., 2007). An example is that different type of roles, such as the control or resource provider role, that outside directors fulfil could have influenced the results. Thirdly, the external validity can be questioned, as this study only focuses on a sample of premium listed firms in the UK. So, the results are potentially not generalizable across other countries or smaller companies that are not listed on the stock exchange.

Since the results of this research were not completely consistent with the findings in prior literature, there could be some avenues for further research. For instance, other sources of CEO power could be examined, or another method could be used in order to derive an index of CEO power, to check whether this will change the outcome. Furthermore, as the proxy variable used for board independence might not be fully representative, additional measures could be included, for instance the financial expertise or background of the

directors, in order to arrive at a more reliable index variable. In addition, the time frame could be adjusted, by looking into the years beyond 2016 as well, or the sample could include both financial and non-financial firms from other countries, as IFRS 7 is adopted globally (Who uses IFRS Standards?, 2018).

CEO power and IFRS 7 disclosure quality: Evidence from UK listed companies