BOARD DIVERSITY OF MNC-PARENT COMPANIES AND FINANCIAL REPORTING QUALITY OF THEIR SUBSIDIARIES

BOARD DIVERSITY OF MNC-PARENT

COMPANIES AND FINANCIAL REPORTING

QUALITY OF THEIR SUBSIDIARIES

Faculty of Economics & Business Msc Accountancy & Controlling

Track Accountancy

Course Master Thesis

Course Code EBM869B20

Thesis Coordinator S. Rusanescu

Date 21-06-2020

Words 12.226

S2907844

Manon van der Woning

Admiraal de Ruijterweg 319-1 1055 LX Amsterdam

+31648783801 manon-vdw@hotmail.com

2

Abstract

This study investigates whether parent’s board diversity, regarding gender and age, is associated with the financial reporting quality of their foreign subsidiaries. Higher board diversity leads to enhanced group dynamics, which increases the board’s monitoring role, and to more different beneficial resources, which enhances their advisory role. Therefore, as prior research shows for standalone firms, board diversity increases their ability to restrain managers from engaging earnings management. However, this has not been investigated within multinational corporations. Parent boards may be concerned with subsidiary financial reporting quality, because parent executives engage in earnings management in the individual subsidiary statements, which are all included in the consolidated reports of the multinational corporation. Since female directors are less tolerant of executives’ opportunistic behaviour than male directors and since the coexistence of different generations leads to a better division of labour in boards, this study expects both gender diversity and age diversity of parent boards to be positively associated with subsidiary financial reporting quality. To test the hypotheses, a sample of subsidiaries of US multinational corporations that are domiciled in 20 European countries during the period 2011-2017 is used. Results show that parent boards that are more gender diverse are associated with better subsidiary financial reporting quality, whereas parent boards that are more age diverse are associated with lower financial reporting quality of their foreign subsidiaries. Thus, parents’ board diversity influences the financial reporting quality of foreign subsidiaries, however, the effect is not always positive.

Keywords: multinational corporation; subsidiary; financial reporting quality; earnings

3

Table of Content

Abstract ... 2

Table of Content ... 3

Introduction ... 4

Theoretical Framework and Hypotheses Development ... 8

FRQ in MNCs ... 8

Board Diversity ... 11

Hypotheses Development ... 13

MNC-parent’s board gender diversity and subsidiary FRQ. ... 13

MNC-parent’s board age diversity and subsidiary FRQ. ... 15

Methodology ... 17

Sample Selection ... 17

Measurement of Variables ... 18

Empirical Model and Control Variables ... 20

Results ... 22

Descriptive Statistics ... 22

Correlations ... 23

Regression Results ... 26

Discussion and Conclusion ... 28

Findings and Conclusion ... 28

Theoretical and Practical Implications ... 30

Limitations and Directions for Future Research ... 32

References ... 33

Appendices ... 40

Appendix I ... 40

4

Introduction

Over the last decades, the economic importance of foreign subsidiaries of multinational corporations (henceforth MNCs) has risen. An OECD Report (2018, p. 5) states that, at global level, “MNEs and their foreign affiliates are estimated to produce 33% of global output”, of which 12% is produced by the foreign affiliates. Furthermore, an increasing percentage of total earnings of MNCs has its source in foreign operations (Fan, 2008) and 31% of the world exports is traded by foreign affiliates of MNCs (OECD, 2018). Thus, MNCs and their foreign subsidiaries deserve attention. From a financial reporting perspective, a MNC-parent company is obliged to prepare consolidated financial statements that reflect the interests in its controlled subsidiaries. The purpose of this consolidation process is to eventually present the financial statements of a parent and its subsidiaries as if the group were a single entity (Beuselinck et al., 2019). Because all the individual subsidiary financial statements are included in the consolidated statements of the MNC, MNCs have the distinctive ability to take advantage of their articulated group structure, by ordering their subsidiaries to make adjustments in the reporting of their accounts, enabling the group as a whole to meet earnings targets (Stewart and Kinney, 2013). Consequently, the financial reporting quality (henceforth FRQ) of MNCs and their subsidiaries is an interesting topic.

Prior literature has focused on how different organizational structures affect dividend pay-out policy, investment efficiency and capital structure, and concluded that MNC-parent companies have substantial power over their subsidiaries, manifesting itself in pressure on the subsidiaries’ decisions about operating, investing and financing (Bartlett and Ghoshal, 1989; Beuselinck et al., 2019; Graham et al., 2015; Robinson and Stocken, 2013; Stein 2002). Typically, MNC-parents constantly monitor performance objectives for all of their subsidiaries and issue targets for them (Busco et al., 2008; Dossi and Patelli, 2008). Often, at the end of the fiscal period, MNCs determine how close their consolidated financial statements would be to their targets, communicate new earnings goals to their subsidiaries, and, if needed, request them to make further adjustments prior to the closing of their accounts (Beuselinck et al., 2019). Such pressure is the reason why MNC-parent companies have an influence on the financial reporting decisions of subsidiaries (Dyreng et al., 2012) and it implicates that there is a connection between the MNC-parent and the FRQ of their subsidiaries.

The role of boards is to monitor and give advice to the executives to ensure that they act in the shareholders’ interests. A part of this role is that boards are charged with overseeing the firm’s

5 financial reporting process. If boards better fulfil this role, they may enhance the firm’s FRQ. Prior research has found relationships between certain board characteristics and FRQ. For example, Abbott et al. (2012) find a negative association between female board presence and the likelihood of a financial restatement, Ferris and Liao (2019) conclude that firms with a higher proportion of busy independent directors have a poorer FRQ and Xie et al. (2003) state that board members with corporate or financial backgrounds are associated with firms that engage less in earnings management, a proxy for FRQ. However, most of these studies are based on the FRQ of consolidated financial statements of MNCs, and do not investigate the direct association between MNC-parent board characteristics and the FRQ of their subsidiaries. It is shown that MNC headquarters make sure that the financial reporting objectives of the firm as a whole are carried out in all the lower levels of the organization (Prencipe, 2012), and that MNC-parent companies even use their subsidiaries when they seek to beat a threshold (Bonacchi et al., 2018). Moreover, the MNC-parent board is likely to be concerned about the subsidiary FRQ, because the executives of the parent use the individual financial statements of the subsidiaries to draw up the consolidated reports. Therefore, it is reasonable to argue that there is a relationship between the MNC-parent board and the FRQ of their subsidiaries. This study aims to fill this research gap by analysing the association between two different board characteristics of MNC-parents, namely board gender diversity and board age diversity, and the FRQ of their foreign subsidiaries. This leads to the following research question:

How is the board diversity of MNC-parent companies, regarding gender and age, associated with the FRQ of their foreign subsidiaries?

This study uses two theories to build the hypotheses. The Agency Theory focuses on the board’s monitoring role, thus on the fiduciary duty to protect the interests of the shareholders (Eisenhardt, 1989; Jensen and Meckling, 1976). This theory argues that possible groupthink will be diminished in a heterogeneous group, which leads to enhanced group dynamics and consequently to an increased monitoring effectiveness of boards. Additionally, the Resource Dependence Theory emphasizes the advisory role of the board, and suggests that more diverse boards will provide different beneficial resources to the firm and thus give better advice to the executives (Pfeffer, 1972; Pfeffer and Salancik, 1978). According to these theories, more diverse boards may be better in restraining the executives to conduct earnings management.

6 more gender diverse: firms with female board participation have less opportunistic earnings management (Srinidhi et al., 2011); female board presence is significantly associated with a lower likelihood of a financial restatement (Abbott et al., 2012); and there is evidence that the proportion of women on board reduces accounting aggressiveness, and therefore has a negative relation with earnings management (Gavious et al., 2012). Taking into consideration that female directors are less tolerant of executives’ opportunistic behaviour than male directors and that men and women possess different sets of skills (e.g. Adams and Ferreira, 2009; Ali et al., 2014; Krishnan and Parsons, 2008; Powel and Ansic, 1997), including more gender diversity in boards will enhance the functioning and efficiency of the board, which increases their ability to restrain the executives from engaging earnings management. Since the consolidated reports include the individual subsidiary statements, this study expects that there is a positive association between MNC-parent board gender diversity and FRQ of their foreign subsidiaries.

Age diversity is the second board characteristic and has also been researched several times before in standalone companies. Cox and Blake (1991), for example, state that age diversity in a board provides a deep mixture of skills, perspectives and insights that can further improve the creativity and problem-solving capabilities of the group. Moreover, higher board age diversity enables a better labour division within the board and leads to more different values due to different generations (e.g. Ali et al., 2014; Handajani et al., 2014; Houle, 1990; Mishra and Jhunjhunwala, 2013). Thus, board age diversity may result in boards giving better advice and being more effective in monitoring the executives to keep them from committing earnings management. Since MNC-parents exercise their power over their subsidiaries to make sure that all their financial reporting objectives are met (Dyreng et al., 2012), this study expects that the positive effects on FRQ of consolidated financial statements, caused by a more age diverse MNC-parent board, will also apply for the subsidiary FRQ. Thus, I expect that there is a positive association between age diversity of the MNC-parent board and the FRQ of their foreign subsidiaries.

To test my hypotheses, I use a sample of 1256 subsidiary-year observations, with 410 unique foreign subsidiaries from 66 unique MNC-parents from the US, during the period 2011-2017. The information is collected from the following databases: Orbis, Compustat Global and BoardEx. Additional subsidiary information is hand-collected from the Edgar Portal. Subsidiary FRQ is the dependent variable of this study. Since FRQ is a complex concept, the level of earnings management is often used as a proxy for this and this study will do so as well.

7 A popular way of measuring earnings management is by using the magnitude of discretionary accruals (Dechow et al., 1995). This study uses the widely applied Modified Jones Model to calculate the discretionary accruals (Jones, 1991). The first independent variable, board gender diversity, is calculated using Blau’s index of heterogeneity (Blau, 1977). Concerning the second independent variable, board age diversity, the coefficient of variation will be calculated, measured by the standard deviation of the age of board members divided by the age mean. To test the two hypotheses, an OLS regression is conducted on the FRQ of the foreign subsidiaries.

The findings of this study show that higher board gender diversity of MNC-parents is related to less subsidiary earnings management, thus to better FRQ of their foreign subsidiaries. However, the results show that MNC-parent companies with higher board age diversity are associated with more subsidiary earnings management, and consequently with worse FRQ of their foreign subsidiaries. A possible explanation for this unexpected result is that too much age diversity may cause negative effects, like potential generation gaps and the possibility of communication issues that could lead to worse board performance. Therefore, too much board age diversity may be detrimental to the monitoring role boards, thereby increasing the likelihood of executives conducting earnings management. Since the individual subsidiary financial statements are included in the consolidated financial statements of the MNC, this study’s results imply that MNC-parent board gender diversity increases and board age diversity decreases the FRQ of the consolidated reports.

In addition to addressing the aforementioned research gap, this study makes three contributions to the literature. First of all, whereas most studies regarding board gender diversity effects on FRQ focus on standalone companies (Abbott et al., 2012; Gavious et al., 2012; Heminway, 2007; Srinidhi et al., 2011), this study contributes to the literature by investigating whether board gender diversity of a MNC-parent company directly influences the FRQ of their subsidiaries. Since the economic significance of subsidiaries is still growing (OECD, 2018), their FRQ matters to a lot of external users of consolidated reports given that the individual subsidiary statements are included in these consolidated reports. This study shows that board gender diversity is related to less subsidiary earnings management, suggesting that it is also associated with better FRQ of consolidated financial statements, because more gender diverse parent boards are able to curb earnings management at the subsidiary level. Secondly, this study contributes to the strand of literature on age diversity within the corporate structure. The association between FRQ and age diversity of the top management team has been researched

8 before (Huang et al., 2012; Qi et al., 2018). However, to my knowledge, there are no prior studies that investigate the association between FRQ and age diversity of the board. Next to that, the studies that do investigate board age diversity effects on (the more general) firm performance only consider standalone companies and do not make a distinction between the MNC-parent and their subsidiaries (Ferrero-Ferrero et al., 2015; Li et al., 2011; Talavera, 2018; Taljaard et al., 2015). Thus, this study contributes by analysing the effect of MNC-parent board age diversity on the FRQ of their subsidiaries. The findings imply that higher MNC-parent board age diversity is associated with more subsidiary earnings management. Since this is inconsistent with the general agreement in this strand of literature, this study provides new insights. Due to the growing financial importance of subsidiaries (OECD, 2018), investigating the determinants of their FRQ matters not only to the subsidiary investors, but also to investors of the MNC-parent, because of the inclusion of the subsidiary financial statements into the consolidated reports. Thirdly, this study contributes to the growing body of literature on earnings management within MNCs. For example, Beuselinck et al. (2019) and Dyreng et al. (2012) investigate where MNCs manage their earnings and find that MNCs engage more earnings management in subsidiaries from countries with a weak rule of law or with more lenient regulations. Whereas these studies focus on why MNCs manage their earnings via their subsidiaries, this study contributes by looking at the board diversity of the MNC-parent company as a potential cause for the subsidiary earnings management. Boards monitor managers to ensure that they act in the shareholders’ interest. Since the top executives may influence the subsidiaries’ financial reporting (Prencipe, 2012), this study provides additional evidence about the parent’s board ability to curb earnings management.

The remainder of this paper is organized as follows. In section 2, the existing literature is reviewed and a theoretical framework is used to develop the hypotheses. Section 3 describes the research methodology. In section 4 the results are presented. Section 5 is the final chapter and contains the discussion and the conclusion of this study.

Theoretical Framework and Hypotheses Development

The Agency Theory states that information published by companies is used as a mechanism for monitoring the actions of executives (Jensen and Meckling, 1976). In addition, Eisenhardt

9 (1989) claims that, in many organizations, there is a division between the company’s owners (i.e. the principals) and the management of that company (i.e. the agents).Because both groups have different needs, a conflict of interest arises. Additionally, both groups do not have the same information, which is known as information-asymmetry. To avoid ineffectiveness as a result, companies provide information in their annual reports (Martinez-Ferrero and Garcia-Sanchez, 2016). In the Statement of Financial Accounting Concepts No. 1, the FASB (1978, p. 5-6) formulates that “Financial reporting is expected to provide information about an enterprise’s financial performance during a period and about how management of an enterprise has discharged its stewardship responsibility to owners”. The FRQ is defined as the quality of the numbers that are reported in these financial statements. Higher quality provides more information about the firm’s financial performance features that could be relevant to a specific decision made by a specific decision-maker (Dechow et al., 2010).

However, the credibility of financial reporting is often threatened, because companies engage in so-called ‘earnings management’. According to Healy and Wahlen (1999, p. 368), earnings management occurs “when managers use judgement in financial reporting and in structuring transactions to alter financial reports to either mislead some stakeholders about the underlying economic performance of the company or to influence contractual outcomes that depend on reported accounting numbers”. Consequently, FRQ of firms is reduced when managers engage earnings management. Since MNCs are big international companies with a lot of stakeholders that are interested in their financial statements, a change in the FRQ of these firms has many different effects. To name a few, Cutillas Gomariz and Sanchez Ballesta (2013) state that higher FRQ increases investment efficiency. Moreover, Park (2018) finds that, due to the importance of a high-quality information system, there is a positive association between FRQ and future innovation. Thus, the FRQ of MNCs has far-reaching consequences for society.

A MNC is a business group with multiple daughter companies, also known as subsidiaries, that are legally independent and function as single economic entities (Beuselinck et al., 2019; La Porta et al., 1999). The subsidiaries are obliged to prepare standalone financial statements, but they will do this according to the group guidelines, to enable the MNC-parent to prepare the consolidated statements for the group as a whole (Prencipe, 2012). Apart from some specific adjustments (e.g. the elimination of intercompany transactions or the alignment of different accounting policies), the consolidation process consists of the line-by-line aggregation of the group’s assets, liabilities, revenues and expenses to form a consolidated balance sheet and

10 income statement (Sutton, 2004). The consequence of this process implies that an increase in subsidiary earnings (i.e. net of intercompany transactions), all other things equal, directly translates into an increase in the consolidated profits of the MNC (Beuselinck et al., 2019). Because of this group structure, MNCs may take advantage by organizing their financial reporting in such a way that the earnings management happens at the subsidiary level (Stewart and Kinney, 2013). Typically, MNC-parent companies issue targets for their subsidiaries and constantly monitor the performance objectives (Busco et al., 2008). Concerning financial reporting objectives, MNC-parent’s executives normally set the tone, because they ensure that the objectives are carried out in all the lower levels of the organization (Prencipe, 2012). Since MNC-parents frequently make use of consolidation software that show real-time monitoring of the performance of their subsidiaries, they can determine whether the targets are likely to be met, and, if not, they can instantly communicate new earnings goals to the subsidiaries prior to the closing of their accounts (Beuselinck et al., 2019).

The fact that the individual financial statements of the subsidiaries are consolidated into the financial statements of the group as a whole, creates a possibility for the MNC-parent to manage the earnings via their subsidiary financial statements, instead of directly in the consolidated reports. Some studies already examined how and why MNC-parents manage their earnings through their subsidiaries. For example, Dyreng et al. (2012) find that foreign earnings management is performed more at firms that have subsidiaries in weak rule of law countries, than at firms with subsidiaries in locations where the rule of law is strong. The authors argue that the subsidiaries’ local legal systems affect the MNC’s propensity to engage there in earnings management and that, if the earnings management in foreign operations is discovered, the consequences are likely to be less stringent when there is a weak rule of law. Moreover, the study of Bonacchi et al. (2018) finds evidence in line with their expectations: Italian parent firms coordinate earnings management in their subsidiaries, through parent directors also having a position on the subsidiary’s board. They conclude that MNC-parents drive the earnings management of their Italian non-listed subsidiaries in order to meet or beat their own benchmarks (Bonacchi et al., 2018). Furthermore, Dearborn (2009) formulates a reason why MNCs engage earnings management at the subsidiary level. When the wrongdoing is discovered, the author states, the MNC-parent could deny their knowledge of the subsidiary misreporting and therefore may incur lower legal and reputational penalties than if the misreporting took place at the MNC-parent level. Furthermore, Beuselinck et al. (2019) find that MNCs rebalance their financial reporting strategies by clustering earnings management in

11 subsidiaries operating in countries with more lenient regulations. All in all, MNC-parent companies can, and do, influence the FRQ of their (foreign) subsidiaries to be able to meet group benchmarks. Earlier evidence shows that MNCs engage in earnings management in their subsidiaries to face less stringent consequences if the earnings management is discovered.

Board Diversity

The role of boards is two-folded; boards have a monitoring and an advisory role. This study uses the Agency Theory and the Resource Dependence Theory to explain them. As was formulated in the earlier section, the Agency Theory states that in many organizations there is a division between the company’s owners and the management of that company (Eisenhardt, 1989). Since both groups have different needs and different information available for them, boards of directors have been appointed. They are charged with monitoring the firm’s activities in order to protect the stakeholders’ interests (Xie et al., 2003). Next to that, the Resource Dependence Theory argues that boards are the mechanism for managing external dependencies (Pfeffer and Salancik, 1978), and are responsible for reducing environmental uncertainty (Pfeffer, 1972). Furthermore, this theory argues that boards can potentially bring other resources to the firm, such as information, skills, access to key constituents (e.g. suppliers, buyers, social groups), and legitimacy (Gales and Kesner, 1994). This highlights the advisory role of boards. Both monitoring and giving advice are, thus, two important benefits boards could bring to firms. Bhagat and Bolton (2008) even state that the board is a key corporate governance mechanism and that it is ultimately responsible for the success of a company. Moreover, boards are charged with overseeing the firm’s financial reporting process and therefore they affect the FRQ (Anderson et al., 2004). This specific role of boards is regularly emphasized by regulators like the SEC and the FASB (Anderson et al., 2004). Furthermore, the likelihood that an organization will perform earnings management is related to the general composition of the board (Xie et al., 2003).

There are different board characteristics that influence the financial performance of companies; the focus of this study is board diversity. Board diversity is defined as the variety in the composition of the board (Kang et al., 2007) and one can think of the heterogeneous composition in terms of age, gender, race, education, nationality, culture, experience, and many other aspects that are unique per individual (Mishra and Jhunjhunwala, 2012). The Resource Dependence Theory provides some convincing theoretical arguments why board diversity is

12 good for firms. Namely, Hillman et al.’s (2000) extension of this theory suggests that the individual members of the board, who differentiate in terms of age, gender, education, nationality etc., are able to provide different beneficial resources to the firm. And since a more diverse board provides more valuable resources, the board activities will be more effective (Carter et al., 2010). On top of that, Fondas and Sassalos (2000) argue that diverse boards bring different opinions and perspectives into discussions and are, therefore, more effective than homogenous boards. Because boards are responsible for making strategic decisions, for developing links with important external stakeholders and for helping talents from the labour market to get involved in the organization, companies benefit from a diverse board (Ali et al., 2014). Furthermore, board diversity creates a wider knowledge base and, therefore, the competitive advantage is greater for organizations that have a diverse board (Erhardt et al., 2003). Additionally, the Agency Theory argues how board diversity positively influences the board’s monitoring role. Since possible groupthink will be diminished in a heterogeneous group, the group dynamics will be enhanced and consequently, the monitoring effectiveness of boards will increase (Abbott et al., 2012).

Even though these results indicate numerous advantages of a diverse board, Ali et al. (2014, p. 508) formulate that “the impact of diversity on performance may not be uniform across different dimensions of diversity”. For instance, the results of Taljaard et al. (2015) show that racial diversity within boards is not associated with financial performance, however, younger average board age and increased gender diversity are shown to have strong associations with enhanced share price performance. Moreover, Solal and Snellman (2019) state that, since a gender-diverse board is interpreted as revealing preference for diversity and a weaker commitment to shareholder value, the firms that increase board diversity suffer a decrease in market value. Additionally, Tarigan et al. (2018) study the influence of board diversity on financial performance in Indonesian context and the results reveal that diversity in terms of nationality is beneficial for firms, while gender and education heterogeneity is proven otherwise. Thus, the results on the influence of board diversity are either positive or negative. Nevertheless, since board diversity is proven to significantly influence firms, it is an interesting construct to investigate.

13

Hypotheses Development

MNC-parent’s board gender diversity and subsidiary FRQ. The first aspect of this

study is gender diversity of MNC-parent boards. Both the earlier mentioned theories (i.e. the Agency Theory and the Resource Dependence Theory) can explain positive effects of board gender diversity for companies. The Agency Theory emphasizes the board’s role in monitoring and controlling the managers (Eisenhardt, 1989; Jensen and Meckling, 1976) and suggests that gender diverse boards may help reduce agency problems between the managers and shareholders. Indeed, females place less importance on personal interests and common practices and are, therefore, less tolerant of executives’ opportunistic behaviour than men (Krishnan and Parsons, 2008). Moreover, female directors are more likely to raise more questions than male directors and they might also be tougher and more active monitors (Adams and Ferreira, 2009; Carter et al., 2003).

The Resource Dependence Theory focuses on how the external resources of organizations affect the behaviour of the organization (Pfeffer, 1972; Pfeffer and Salancik, 1978). The presence of female directors helps companies maximize access to critical resources, because men and women possess different sets of skills, knowledge and perspectives (Rogelberg and Rumery, 1996). By integrating both perspectives, an organization may be better positioned to critically weigh risks and benefits of strategic decisions (Ali et al., 2014). Next to that, Powel and Ansic (1997) find that women are more cautious and less risk-seeking than men, therefore both sexes adopt different strategies when having to make financial decisions. In sum, broadening the talent pool by including women directors in the board will enhance the functioning and efficiency of the board (Peni and Vahamaa, 2010), and significantly contribute to board performance (Mishra and Jhunjhunwala, 2012).

Prior research has studied the effects of board gender diversity on FRQ. Abbott et al. (2012) state that female board presence will enhance the board’s ability to maintain an attitude of independence and diminishing the extent of groupthink, and consequently enhance the board’s ability to monitor financial reporting. Furthermore, Ho et al. (2014) argue that the ethical leadership of females and their conservative mindset contributes to a better internal control environment with stronger emphasis on ethical and conservative financial reporting. Because women are highly sensitive to the risk of lawsuits and to reputational loss, Srinidhi et al. (2011) find that they act more decisively than men to enhance earnings quality. Krishnan and Parsons (2008) also study, and find, that the earnings quality is higher for firms with more female

14 directors. Moreover, because of the higher ethical level and because they are more trustworthy than men, Gull et al. (2017) argue that the presence of female directors deters managers from managing earnings. Furthermore, the paper of Gavious et al. (2012) finds evidence that there is a negative relation between the presence of female board members and earnings management due to the fact that accounting aggressiveness is affected by the proportion of women on the board.

However, there are also findings of previous studies that do not support the conclusion of a negative association between gender diversity and earnings management. For example, Sun et al. (2011) were unable to find any correlation between female participation in audit committees and earnings management. Possible causes for the observed results, that are given by the authors, are that some female audit committee directors may believe that not all earnings management is unethical or that women are not uniform in their ability to influence other audit committee members. Furthermore, Peni and Vahamaa (2010) find no significant relationship between CEO gender and earnings management. However, the authors argue that the lack of significance may be related to the extremely low number of female CEOs in their sample. Also, the study of Ge et al. (2011) does not find a significant gender effect on earnings management for CFOs, but this is possibly due to the fact that they have only 82 female CFOs in their sample of 914 CFOs. Since the studies that could not find a positive association between gender and FRQ, are about audit committees, CEOs or CFOs, this study assumes that this will not be the case for boards.

To put it all together, in the context of MNCs, boards are charged with overseeing the financial reporting process and consequently may influence the FRQ of the group. MNC-parent companies often take advantage of their group structure by ordering their subsidiaries to make adjustments in the reporting of their accounts, enabling the group as a whole to meet earnings targets. Moreover, board gender diversity helps companies maximize access to critical resources, because men and women possess different perspectives. Since they are more cautious, less risk-seeking, and have a conservative and ethical mindset, female board members are less tolerant of executives’ opportunistic behaviour then men. Additionally, more gender diversity diminishes possible group think in boards. Therefore, the board raises more questions and there will be more active monitoring of the managers. Thus, board gender diversity increases the board’s ability to give advice to the executives, and also helps them monitor these executives to limit the amount of earnings management within the MNC. Since the consolidated

15 financial statements of the MNC-parent include the individual statements of the subsidiaries, the MNC-parent board does not only concern about earnings management in the consolidated financial statements, but also about the amount of earnings management that happens in the subsidiary financial statements. So, board gender diversity not only increases the board’s ability to curb earnings management in the consolidated statements of the MNC, but most probably also in the subsidiary financial statements. Therefore, I expect that a more gender diverse parent board will result in a higher subsidiary FRQ.

H1: There is a positive association between the gender diversity of the MNC-parent

board and the FRQ of their foreign subsidiaries.

MNC-parent’s board age diversity and subsidiary FRQ. Besides board gender,

board age remains one of the most central observable backgrounds of diversity issues for boards (Kang et al., 2007). Age can be seen as a dynamic proxy of an individual’s life experience and can be used when profiling someone (Mannheim, 1949). The age of an individual includes a wide range of factors that has an influence on the formation of personal values during their lifespan (Rhodes, 1983). From a theoretical perspective, the earlier mentioned Agency Theory can explain positive effects of board age diversity for companies. The Agency Theory suggests that age diverse boards may help reduce agency problems between the management and the owners or other stakeholders, by emphasizing the board’s role in monitoring and controlling the managers (Eisenhardt, 1989; Jensen and Meckling, 1976). Indeed, Handajani et al. (2014) state that board age is highly considered in determining the board composition, especially in formulating policies and strategies to safeguard the diversity interest of the stakeholders. Additionally, agency problems might be reduced because age-heterogeneous boards can enable a more efficient division of labour at the board level, because the older group is able to provide experience, a network and financial resources; the middle level is responsible for the main executive tasks; and the youngest group will learn from the other two groups and develop its knowledge of the business (Houle, 1990). Furthermore, the younger directors seem to be more active in driving business success and in focussing on the firm’s future, while the older group forms an accumulation of skill-based competencies (Handajani et al., 2014).

The Resource Dependence Theory views an organization as an open system that depends on external organizations and environment contingencies (Pfeffer, 1972; Pfeffer and Salancik, 1978). It suggests that board age diversity helps organizations maximize access to critical

16 resources through the coexistence of different generations and consequently of different values, habits, experiences, motivational goals and cultural norms that influence the intuitive decision-making approach of the board members (Cucari et al., 2018). These different values, beliefs and attitudes that are caused by age differences, are commonly referred to as the ‘generation gap’ (Taveggia and Ross, 1978). Mishra and Jhunjhunwala (2013) state that boards that consist of different age groups automatically have a diverse set of perspectives and skills, and this way a balanced board is created. However, as the authors argue, it is important that the board’s decision-making process is not dominated by one age group. Since retired-age members often form a large proportion of the board, the younger board members should not feel intimated by their senior counterparts and must still participate independently (Mishra and Jhunjhunwala, 2013). Moreover, Hambrick et al. (1996) argue that age diversity could induce short-term issues of communication and trust, arising from generation gaps. However, there is evidence that the different features of younger and older board members complement each other, and that these differences can, therefore, be leveraged to upgrade the strategic decision-making of the organization (Ali et al., 2014).

Although the evidence from the above section seems to contain only positive effects of board age diversity, the findings on the relationship between age diversity in boards and firm performance are inconsistent, with some studies reporting positive effects (e.g. Kilduff et al. 2000; Mahadeo et al., 2012) and some studies reporting negative effects (e.g. Murray, 1989; Milliken and Martins, 1996). None of the studies regarding board age diversity are investigating the effect on FRQ though. However, prior research focused on the relation between the age effect of executives and FRQ. For instance, Qi et al. (2018) found that, due to being more ethical, conservative and risk-averse, top executives near retirement age will engage in earnings management less likely than top executives that are not. Furthermore, Huang et al. (2012) apply the same line of reasoning to conclude that there is a positive association between the CEO age and the firm’s FRQ. Since the age of executives seems to have an influence on FRQ, this study assumes that the age of board members also affects the FRQ of an organization.

All in all, in the context of MNCs, boards may influence the FRQ of the group, because they are charged with overseeing the financial reporting process. To enable the group as a whole to meet earnings targets, MNC-parent companies often take advantage of the group structure by commanding their subsidiaries to make adjustments in the reporting of their accounts. Additionally, prior literature documents that age diverse boards are beneficial for companies

17 because they maximize access to critical resources through the coexistence of different generations. Moreover, age diversity creates a balanced board, due to a diverse set of perspectives and skills. Age-heterogeneous boards can enable a more efficient division of labour at the board level, because different features complement each other and can be leveraged to upgrade the strategic decision-making. Hence, board age diversity increases the board’s ability to give advice to the managers as well as the ability to monitor the managers to ensure that earnings management is reduced. Since the consolidated financial statements of the MNC include all the individual subsidiary statements, MNC-parent boards are not only concerned with reducing earnings management in the consolidated statements, but may also be interested in the subsidiary financial statements. Board age diversity of the MNC-parent board thus may decrease the subsidiary earnings management as well. Therefore, I expect that a more age diverse board will result in a higher subsidiary FRQ.

H2: There is a positive association between the age diversity of the MNC-parent

board and the FRQ of their foreign subsidiaries.

Methodology

The first step in the sample selection is identifying all listed firms from the US in the Compustat database. Next, I discard the firms in the banking and financial sector from the sample due to a different set of regulations and capital structure requirements for these firms (Aggarwal et al., 2019). Firms that do not have the necessary financial information to carry out the research are also excluded, as well as firms with missing board related information. Financial information of the MNC-parent companies is obtained from Compustat, while the information about the MNC-parent’s board is gathered from BoardEx. Then, for these listed non-financial firms, the names and jurisdiction of all their material subsidiaries included in exhibit 21.1 of the 10-K filings are hand-collected from the Edgar Portal. The hand collection of material subsidiaries results in a data file of 210,663 unique observations. In the following step, the names and countries of the material subsidiaries are matched with those of the firms covered by the Orbis database. Given the coverage of Orbis, I only select subsidiaries from 20 European countries, namely Austria, Belgium, Bulgaria, Czech Republic, Germany, Denmark, Spain, Finland, France, Croatia, Hungary, Italy, Luxembourg, Netherlands, Norway, Poland, Portugal,

18 Romania, Sweden, and Slovakia. Then, subsidiaries operating in the financial industry are deleted for the same reason as stated above. Moreover, subsidiaries with unavailable financial information are discarded from the sample. The financial information of the subsidiaries is collected from the Orbis database. After merging the information from these three databases, the final sample consists of 1256 subsidiary-year observations, over period 2011-2017, with 410 unique foreign subsidiaries and 66 unique MNC-parents.

Measurement of Variables

The dependent variable of this study is the FRQ of foreign subsidiaries of MNCs. FRQ is a complex concept and the level of earnings management is often used as a proxy for this; this study does so as well. A popular way of measuring earnings management is by calculating the magnitude of discretionary accruals (Dechow et al., 1995). Jones (1991) models a linear relationship between total accruals and change in sales and property, plant and equipment. Since revenues can be manipulated through an increase of sales on accounts, Dechow et al. (1995) argue that changes in accounts receivable, representing a change in the amount of credit sales, should be deducted from the change in revenues in the Jones (1991) model. This study uses this widely applied Modified Jones Model to calculate the discretionary accruals (Dechow et al., 1995; Jones, 1991). The following equation is used to estimate the discretionary accruals, i.e. DISC_ACC_SUB:

TACit / TASit–1 = α1 ( 1 / TASit–1 ) + α2 [ ( ΔREVit - ΔRECit ) / TASit–1 ] + α3 ( PPEit / TASit–1 ) + εit (1)

In this equation, TACit is total accruals for firm i in year t (i.e. income before extraordinary items less operating cash flows), TASit-1 is the total assets for firm i in year t-1, ΔREVit is the change in revenue for firm i in year t (i.e. revenues in year t less revenues in year t-1), ΔRECit is the change in receivables for firm i in year t (i.e. net receivables in year t less net receivables in year t-1), PPEit is property, plant and equipment for firm i in year t. Additionally, α1, α2 and α3 are the OLS estimates. The Modified Jones Model is estimated cross-sectionally, for each year and industry combination. The residual of the equation (i.e. εit) shows the discretionary accruals for firm i in year t. The dependent variable of this study, DISC_ACC_SUB, is the absolute value of the discretionary accruals, since the absolute value captures both income-increasing and income-decreasing earnings management.

19 The explanatory variables of this study are MNC-parent board gender diversity, BOARD_GEN_DIV, and MNC-parent board age diversity, BOARD_AGE_DIV. Even though many studies define board gender diversity as the number of women on the board divided by the total number of board members, this percentage may not be an appropriate measure of diversity, as a high number of women directors actually shows a high degree of homogeneity in terms of gender (Campbell and Minguez-Vera, 2008). Therefore, this study uses the Blau Index (Blau, 1977) to measure gender diversity. This index takes into account two attributes of diversity, namely the number of gender categories, i.e. ‘variety’, and the evenness of the distribution of board members among them, i.e. ‘balance’ (Campbell and Minguez-Vera, 2008). The Blau Index ranges from 0 to a maximum of 0,5. The maximum is reached when the proportion of each category is at a maximum; this means when the board has an equal number of female and male directors. The variable BOARD_GEN_DIV is calculated using the following formula for the Blau Index:

Blau Index = 1 - ∑𝑛𝑛_𝑖𝑖=1𝑃𝑃_𝑖𝑖2 (2)

where Pi is the percentage of the board members in each category and n represents the number of categories used.

For the second independent variable, BOARD_AGE_DIV, the Blau Index is not used, because this index is not suitable for measuring diversity corresponding to continuous attributes, like age (Tsui et al., 1995). Moreover, categories that correspond to generations would have been created and since a lot of board members belong to the same generation, the so-called ‘Boomers’ (1946-1964), the Blau Index would misrepresent the board age diversity (Ferrero-Ferrero et al., 2015). Rather than the occurrence of different generation categories, or even the range between the youngest and oldest board member, this study is, by contrast, interested in the variation of board members’ ages. Therefore, the first step in computing BOARD_AGE_DIV is calculating the standard deviation of the age of board members (Ferrero-Ferrero et al., 2015; McIntyre et al., 2007). Then, the coefficient of variation is calculated, because this is a scale-invariant and direct measurement of diversity (Aggarwal et al., 2019). The following expression is used:

Coefficient of variation = 𝑆𝑆𝑆𝑆𝑆𝑆𝑛𝑛𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝑆𝑆𝑑𝑑𝑑𝑑𝑖𝑖𝑆𝑆𝑆𝑆𝑖𝑖𝑑𝑑𝑛𝑛

20

Empirical Model and Control Variables

This study uses an Ordinary Least Squares (OLS) regression model, shown below, for analyzing the effect of gender and age diversity of MNC-parent boards on the FRQ of their foreign subsidiaries. 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷_𝐴𝐴𝐷𝐷𝐷𝐷_𝐷𝐷𝑆𝑆𝑆𝑆𝑖𝑖𝑆𝑆 = β0 + β1𝑆𝑆𝐵𝐵𝐴𝐴𝐵𝐵𝐷𝐷_𝐺𝐺𝐺𝐺𝐺𝐺_𝐷𝐷𝐷𝐷𝐷𝐷_{𝑖𝑖𝑆𝑆} + β2 𝑆𝑆𝐵𝐵𝐴𝐴𝐵𝐵𝐷𝐷_𝐴𝐴𝐺𝐺𝐺𝐺_𝐷𝐷𝐷𝐷𝐷𝐷_{𝑖𝑖𝑆𝑆} + β3 𝐷𝐷𝐷𝐷𝑆𝑆𝐺𝐺_𝐷𝐷𝑆𝑆𝑆𝑆𝑖𝑖𝑆𝑆 + β4 𝑃𝑃𝐺𝐺𝐵𝐵𝑃𝑃_𝐷𝐷𝑆𝑆𝑆𝑆_{𝑖𝑖𝑆𝑆} + β5 𝐿𝐿𝐺𝐺𝐷𝐷_𝐷𝐷𝑆𝑆𝑆𝑆_{𝑖𝑖𝑆𝑆} + β6 𝐿𝐿𝐵𝐵𝐷𝐷𝐷𝐷_𝐷𝐷𝑆𝑆𝑆𝑆_{𝑖𝑖𝑆𝑆} + β7 𝐺𝐺𝐵𝐵𝐵𝐵𝐺𝐺𝐺𝐺𝐺𝐺_𝐷𝐷𝑆𝑆𝑆𝑆𝑖𝑖𝑆𝑆 + β8 𝐷𝐷𝐷𝐷𝑆𝑆𝐺𝐺_𝑃𝑃𝐴𝐴𝐵𝐵_{𝑖𝑖𝑆𝑆} + β9 𝑃𝑃𝐺𝐺𝐵𝐵𝑃𝑃_𝑃𝑃𝐴𝐴𝐵𝐵_{𝑖𝑖𝑆𝑆} + β10 𝐿𝐿𝐺𝐺𝐷𝐷_𝑃𝑃𝐴𝐴𝐵𝐵𝑖𝑖𝑆𝑆 + β11 𝐿𝐿𝐵𝐵𝐷𝐷𝐷𝐷_𝑃𝑃𝐴𝐴𝐵𝐵_{𝑖𝑖𝑆𝑆}+ β12 𝐺𝐺𝐵𝐵𝐵𝐵𝐺𝐺𝐺𝐺𝐺𝐺_𝑃𝑃𝐴𝐴𝐵𝐵_{𝑖𝑖𝑆𝑆} + ∑𝑛𝑛−1_{𝑘𝑘=1 k}µ 𝐷𝐷𝐷𝐷𝐷𝐷_𝑖𝑖𝑘𝑘 + ∑𝑛𝑛−1_{𝑥𝑥=1 x}𝜆𝜆 𝐷𝐷𝐵𝐵𝑆𝑆𝐺𝐺𝐺𝐺𝐵𝐵𝐶𝐶_𝑖𝑖𝑥𝑥 + ∑_{𝑦𝑦=2011 y}2017 𝜔𝜔 𝐶𝐶𝐺𝐺𝐴𝐴𝐵𝐵_𝑖𝑖𝑦𝑦 + 𝜀𝜀_{𝑖𝑖𝑆𝑆} (4)

This equation examines the association of board diversity with subsidiary FRQ after controlling for factors documented in previous literature. The dependent variable, DISC_ACC_SUB, represents the absolute value of discretionary accruals of the subsidiary, measured using the Modified Jones Model (Dechow et al., 1995; Jones, 1991). The explanatory variables, BOARD_GEN_DIV and BOARD_AGE_DIV, represent the gender and age diversity of the MNC-parent board. Based on the two hypotheses of this study, I expect the coefficients of the main independent variables to have a negative sign with the discretionary accruals. The equation controls for firm size (SIZE_SUB and SIZE_PAR), firm performance (PERF_SUB and PERF-PAR), financial leverage (LEV_SUB and LEV_PAR), loss (LOSS_SUB and LOSS_PAR) and sales growth (GROWTH_SUB and GROWTH_PAR) of both the foreign subsidiary and the MNC-parent. The definitions of the variables used in Equation (4) are presented in Table 1 in Appendix I. Additionally, the model uses dummy variables to control for potential industry (SIC), country (COUNTRY) and time (YEAR) effects.

Previous studies indicate that the above mentioned firm characteristics are useful in predicting earnings management (Cheng and Warfield, 2005; Chih et al., 2008; Francis et al., 2005; Geiger and North, 2006; Hong and Anderson, 2011; Kim et al., 2012; Meek et al., 2007). Most of these studies are based on standalone companies, so the empirical model includes a set of controls for these subsidiary characteristics. But besides that, it has been documented that MNC-parents have the power to influence the financial reporting decisions of their subsidiaries (Beuselinck et al., 2019; Dyreng et al., 2012). Therefore, the model also includes control variables for parent characteristics.

21 Firm size is included in the regression model as a control variable, however, there is no agreement in the literature regarding the direction of the effect of firm size on earnings management. Meek et al. (2007) state, for example, that larger firms may have stronger governance and are subject to better monitoring by auditors and financial analysts, so firm size is expected to be negatively associated with earnings management. This should be the case for SIZE_SUB. However, for these reasons it can also be possible that large parent firms especially engage earnings management in their foreign subsidiaries (instead of in their own entity), resulting in a positive association for SIZE_PAR. Furthermore, Watts and Zimmerman (1990) find that larger firms are more likely to perform income-decreasing earnings management. Consequently, firm size can be either positively or negatively associated with earnings management. SIZE_SUB and SIZE_PAR are measured by the natural logarithm of their total assets.

Secondly, PERF_SUB and PERF_PAR are included in the model to control for the financial performance of the subsidiary and the MNC-parent company. Firms with higher financial performance are expected to manage their earnings downwards to pay lower taxes (Arun et al., 2015; Watts and Zimmerman, 1990). To the contrary, DeFond and Jiambalvo (1994) and Keating and Zimmerman (1999) conclude that, if firms have poorer performance, the managers have stronger incentives to manage their earnings. This evidence indicates both positive and negative associations between firm performance and earnings management. The control variables PERF_SUB and PERF_PAR are measured by the ROA (i.e. return on assets).

Thirdly, LEV_SUB and LEV_PAR are control variables that are associated with the financial condition of both the foreign subsidiary and the MNC-parent company. DeAngelo et al. (1994) report that troubled companies have strong incentives to use income-decreasing accruals, because the managers’ willingness to take write-offs and reduce dividends can help convincing lenders that they are serious about streamlining the operations. Moreover, accruals models may overestimate the accruals for poorly performing firms (Dechow et al., 1995). Thus, this study expects a positive association between financial leverage and earnings management. The leverage is measured as total liabilities divided by total assets.

Another factor about the financial condition of the firm that is included as a control variable is financial distress (i.e. loss), defined by LOSS_SUB and LOSS_PAR. It is expected that firms

22 facing financial problems tend to engage in income-decreasing earnings management (Healy, 1985). This is because managers have incentives to further reduce current earnings when earnings are so low that no matter which accounting procedures are selected the target earnings will not be met; this strategy is known as ‘taking a big bath’. This strategy indicates a positive association between a loss and earnings management. The variables LOSS_SUB and LOSS_PAR are dummy variables taking the value of 1 if the firm reported a negative net income and 0 otherwise.

Lastly, GROWTH_SUB and GROWTH_PAR are included to control for a firm’s growth in sales. Chih et al. (2008) argue that high-growth firms tend to manage discretionary accruals upwards to be able to report increased earnings and Meek et al. (2007) state that firms with high growth are typically less transparent and thus may have greater opportunities for opportunistic earnings management. Based on this prior research, this model expects a positive association between sales growth and earnings management. The sales growth is measured at the sales growth ratio.

The expected signs of the coefficients for above mentioned variables are shown in Table 2 in Appendix II. Additionally, since the extent of earnings management may also differ across industries and countries and over time, the model includes industry, subsidiary-country and year fixed effects to control for this. SIC is a dummy variable according to standard industry classification (SIC) codes. This study uses 2 digits SIC codes. COUNTRY is a dummy variable which indicates one of the 20 countries that the foreign subsidiaries could be incorporated in. YEAR is a dummy variable that indicates the fiscal year. Hence, following Peni and Vahamaa (2010), throughout the panel regressions, a three-way fixed-effects specification is used, which controls for the possible change in earnings management over time and also allows for a different intercept for each industry and each country in the sample.

Results

The descriptive statistics of the variables of this study are presented in Table 3. The mean of DISC_ACC_SUB is 0.045, which means that discretionary accruals represent, on average, 4.5% of the prior year’s total assets reported by the subsidiaries. The 25%-75% range of DISC_ACC_SUB is between 0.012 and 0.052. Both the explanatory variables have low means; 0.277 for BOARD_GEN_DIV and 0.122 for BOARD_AGE_DIV respectively. In the case of

23 gender diversity, this indicates that the average percentage of male directors in the board is sufficiently higher than the percentage of female directors. Concerning age diversity, this shows that all the ages of the directors in the board are highly clustered together around the age-mean. Moreover, Table 3 shows that the mean of the natural logarithm of the assets from subsidiaries is 10.336, while for parent companies the mean is 16.104. Furthermore, the mean ROA is 4.6% for the subsidiaries and 6.3% for the parent companies. The financial leverage of both subsidiaries and parent companies is close together, namely 57.0% versus 61.6%. Furthermore, 17.5% of the subsidiaries and 8.0% of the parent companies reported negative net income in their financial statement during the given period. Lastly, the mean growth ratio of subsidiaries is 11.4%, while for the parent companies it is only 1.2%.

Table 3: Descriptive statistics

Variable N Mean St.Dev. P(25) Median P(75)

DISC_ACC_SUB 1260 0.045 0.059 0.012 0.024 0.052 BOARD_GEN_DIV 1260 0.277 0.098 0.208 0.291 0.346 BOARD_AGE_DIV 1260 0.122 0.035 0.099 0.115 0.138 SIZE_SUB 1260 10.336 1.826 9.105 10.303 11.501 PERF_SUB 1260 0.046 0.201 0.012 0.046 0.098 LEV_SUB 1260 0.570 0.504 0.294 0.522 0.753 LOSS_SUB 1260 0.175 0.380 0 0 0 GROWTH_SUB 1260 0.114 1.282 -0.053 0.026 0.112 SIZE_PAR 1260 16.104 1.638 14.916 15.902 17.387 PERF_PAR 1260 0.063 0.060 0.027 0.058 0.102 LEV_PAR 1260 0.616 0.182 0.501 0.603 0.727 LOSS_PAR 1260 0.080 0.271 0 0 0 GROWTH_PAR 1260 0.012 0.103 -0.041 0.020 0.063

Notes: This table reports the descriptive statistics for the dependent (DISC_ACC_SUB), the main independent

(BOARD_GEN_DIV and BOARD_AGE_DIV) and the control variables (SIZE_SUB, PERF_SUB, LEV_SUB,

LOSS_SUB, GROWTH_SUB, SIZE_PAR, PERF_PAR, LEV_PAR, LOSS_PAR and GROWTH_PAR) in the sample. The

variable definitions are given in Table 1.

Correlations

The strength of the relationships between all the variables of interest is also measured. This is done by using the Pearson correlation and the outcomes are shown in Table 4. The correlation between DISC_ACC_SUB and BOARD_GEN_DIV is negative and significant at the 5%-level, suggesting that the magnitude of subsidiary discretionary accruals decreases with the gender diversity of the parent’s board. Instead, the coefficient of the correlation between DISC_ACC_SUB and BOARD_AGE_DIV is positive and significant at the 5%-level, suggesting that subsidiary discretionary accruals increases with the parent’s board age diversity. Since the

24 hypotheses expected the independent variables to have a negative association with discretionary accruals, the correlations support hypothesis 1 for gender diversity and do not support hypothesis 2 for age diversity.

When looking at the correlations of the dependent variable with the control variables, we observe that discretionary accruals are higher in more leveraged and loss-making subsidiaries. In turn, subsidiaries report lower discretionary accruals when they are larger and more profitable, and also when their parent is a larger corporation. The other correlations indicate that subsidiary discretionary accruals are not associated with the subsidiary growth, the performance of the parent, how leveraged the parent company is, whether the parent reports a loss, and with the parent growth.

Among the control variables, there are a few strong and significant correlations observed. The correlation between the size of the parent company and the size of the subsidiary is, for example, 0.384 and is significant on the 5%-level, indicating that larger parents have larger subsidiaries. Additionally, as one would expect, the subsidiary performance is negatively correlated both with the subsidiary financial leverage (i.e. 0.405) and with the subsidiary reported loss (i.e. -0.459). The highest observed significant correlation is between the performance of the parent company and their reported loss (i.e. -0.589). However, the correlations between each pair of explanatory variables, in general, are fairly small. Therefore, multicollinearity is not an issue in the model.

Table 4: Pearson correlation matrix Variable 1 2 3 4 5 6 7 8 9 10 11 12 13 1 DISC_ACC_SUB 1 2 BOARD_GEN_DIV -0.143* 1 3 BOARD_AGE_DIV 0.072* -0.065* 1 4 SIZE_SUB -0.076* 0.234* -0.044 1 5 PERF_SUB -0.138* -0.014 0.082* 0.037 1 6 LEV_SUB 0.165* -0.026 0.009 -0.128* -0.405* 1 7 LOSS_SUB 0.153* -0.020 -0.020 -0.108* -0.459* 0.225* 1 8 GROWTH_SUB 0.012 -0.021 -0.048 0.008 0.025 -0.014 -0.022 1 9 SIZE_PAR -0.095* 0.245* -0.020 0.384* -0.019 0.021 -0.157* -0.043 1 10 PERF_PAR 0.008 0.003 0.078* 0.096* 0.079* -0.009 -0.117* -0.041 0.195* 1 11 LEV_PAR -0.024 0.246* 0.015 0.012 0.032 0.002 0.031 0.018 0.043 -0.122* 1 12 LOSS_PAR 0.007 -0.129* 0.005 -0.081* -0.030 -0.041 0.041 0.056* -0.229* -0.589* 0.144* 1 13 GROWTH_PAR 0.026 0.011 0.127* -0.011 0.128* -0.033 -0.105* -0.029 -0.028 0.208* -0.177* -0.201* 1

Notes: This table reports the matrix of Pearson correlation coefficients for all the variables used in this study. The definitions of the variables are presented in Table 1. * Indicates significance

Regression Results

The estimation results of the OLS panel regression of Equation (4) are presented in Table 5. The panel regression was estimated with three-way fixed effects (i.e. industry, country and year fixed effects). Including these fixed effects removes the effect of unobservable time-invariant characteristics, and ensures that the net effect of the explanatory variables on the dependent variable is assessed properly. Moreover, clustering standard-errors is important, because a failure to do so can cause heavily underestimated standard errors (Cameron et al., 2009). Consequently, it can cause over-rejection of the two hypotheses of this study. Therefore, to correct for heteroscedasticity, this study clusters standard errors by subsidiary and year. Table 5 includes four columns. Column 1 presents the results of the model from Equation (4) including only the control variables. Columns 2 and 3 show the results of the model from Equation (4) including separately each main independent variable and the controls; board gender diversity for Column 2 and board age diversity for Column 3, respectively. Lastly, Column 4 shows the results of the full model from Equation (4). The adjusted R2 of the estimated models vary between 13.9% and 15.6%, thus are relatively low. However, it should be noted that low adjusted R2’s are common in this type of accrual regression models (Arun et al., 2015; Gavious et al., 2012; Srinidhi et al., 2011; Peni and Vahamaa, 2010).

Based on the results reported in Column 4 in Table 5, it can be concluded that board gender diversity of the MNC-parent company is negatively and significantly associated with earnings management of the foreign subsidiary (β1=-0.076, p<0.05). Every 1 percentage point increase in the board gender diversity, measured by the Blau index, decreases the discretionary accruals of the subsidiary by 0.076 percentage points. This means that subsidiary FRQ is better when the parent boards are more gender diverse. Hence, hypothesis 1 is supported.

Furthermore, the results show that board age diversity of the MNC-parent company is positively and significantly associated with earnings management of the foreign subsidiary (β2=0.140, p<0.05). Every increase in board age diversity, measured by the coefficient of variation, by 1 percentage point increases the discretionary accruals of the subsidiary by 0.140 percentage points. This means that a greater age diversity of the parent’s board relates to higher levels of discretionary accruals reported by the foreign subsidiaries, thus to a lower FRQ. Even though these results indicate a significant association, hypothesis 2 expected a positive association between board age diversity and subsidiary FRQ. Therefore, hypothesis 2 is not supported.

Table 5: Regression results

Variable Model 1 Model 2 Model 3 Model 4

Coefficient t-value Coefficient t-value Coefficient t-value Coefficient t-value

Intercept 0.077 1.93 0.064* 2.04 0.046 1.42 0.050 1.57 (0.040) (0.031) (0.032) (0.032) BOARD_GEN_DIV -0.077** -3.52 -0.076** -3.65 (0.022) (0.021) BOARD_AGE_DIV 0.143** 2.78 0.140** 2.82 (0.051) (0.050) SIZE_SUB -0.001 -0.48 -0.000 -0.05 -0.001 -0.62 -0.000 -0.19 (0.002) (0.002) (0.002) (0.002) PERF_SUB 0.004 0.20 0.004 0.18 0.018 0.13 0.003 0.12 (0.021) (0.021) (0.005) (0.021) LEV_SUB 0.018** 3.06 0.018** 3.02 0.018** 3.08 0.018** 3.05 (0.006) (0.006) (0.005) (0.006) LOSS_SUB 0.016* 2.40 0.017* 2.54 0.016* 2.41 0.016* 2.55 (0.007) (0.007) (0.007) (0.006) GROWTH_SUB -0.000 -0.14 -0.000 -0.09 0.000 0.12 0.000 0.13 (0.001) (0.001) (0.001) (0.001) SIZE_PAR -0.002 -1.56 -0.002 -1.42 -0.002 1.35 -0.002 -1.20 (0.002) (0.001) (0.002) (0.002) PERF_PAR 0.057 1.34 0.045 1.20 0.051 1.13 0.040 1.00 (0.043) (0.038) (0.045) (0.040) LEV_PAR 0.000 0.00 0.011 0.60 -0.002 -0.10 0.009 0.48 (0.019) (0.018) (0.019) (0.018) LOSS_PAR 0.003 0.31 -0.001 -0.14 0.002 0.14 -0.003 -0.28 (0.011) (0.010) (0.011) (0.011) GROWTH_PAR 0.003 0.13 0.005 0.27 -0.003 -0.13 0.000 0.02 (0.020) (0.021) (0.021) (0.022)

SIC Included Included Included Included

COUNTRY Included Included Included Included

YEAR Included Included Included Included

N 1256 1256 1256 1256

Adj. R2 _{0.139 0.151 0.145 0.156}

F-value 8181.88*** 21756.75*** _1834.45*** 3074.78***

Notes: This table reports the results of the three-way fixed effects panel data regression on the FRQ of foreign subsidiaries.

Industry, country and year fixed effects are included, but not individually reported. The study clusters standard errors for subsidiary and year. The regression results are based on Equation (4):

𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷_𝐴𝐴𝐷𝐷𝐷𝐷_𝐷𝐷𝑆𝑆𝑆𝑆𝑖𝑖𝑆𝑆 = β0 + β1 𝑆𝑆𝐵𝐵𝐴𝐴𝐵𝐵𝐷𝐷_𝐺𝐺𝐺𝐺𝐺𝐺_𝐷𝐷𝐷𝐷𝐷𝐷_{𝑖𝑖𝑆𝑆} + β2 𝑆𝑆𝐵𝐵𝐴𝐴𝐵𝐵𝐷𝐷_𝐴𝐴𝐺𝐺𝐺𝐺_𝐷𝐷𝐷𝐷𝐷𝐷_{𝑖𝑖𝑆𝑆} + β3 𝐷𝐷𝐷𝐷𝑆𝑆𝐺𝐺_𝐷𝐷𝑆𝑆𝑆𝑆_{𝑖𝑖𝑆𝑆} + β4 𝑃𝑃𝐺𝐺𝐵𝐵𝑃𝑃_𝐷𝐷𝑆𝑆𝑆𝑆_{𝑖𝑖𝑆𝑆} + β5 𝐿𝐿𝐺𝐺𝐷𝐷_𝐷𝐷𝑆𝑆𝑆𝑆𝑖𝑖𝑆𝑆 + β6 𝐿𝐿𝐵𝐵𝐷𝐷𝐷𝐷_𝐷𝐷𝑆𝑆𝑆𝑆_{𝑖𝑖𝑆𝑆} + β7 𝐺𝐺𝐵𝐵𝐵𝐵𝐺𝐺𝐺𝐺𝐺𝐺_𝐷𝐷𝑆𝑆𝑆𝑆_{𝑖𝑖𝑆𝑆} + β8 𝐷𝐷𝐷𝐷𝑆𝑆𝐺𝐺_𝑃𝑃𝐴𝐴𝐵𝐵_{𝑖𝑖𝑆𝑆} + β9 𝑃𝑃𝐺𝐺𝐵𝐵𝑃𝑃_𝑃𝑃𝐴𝐴𝐵𝐵_{𝑖𝑖𝑆𝑆} + β10 𝐿𝐿𝐺𝐺𝐷𝐷_𝑃𝑃𝐴𝐴𝐵𝐵_{𝑖𝑖𝑆𝑆} + β11 𝐿𝐿𝐵𝐵𝐷𝐷𝐷𝐷_𝑃𝑃𝐴𝐴𝐵𝐵𝑖𝑖𝑆𝑆+ β12 𝐺𝐺𝐵𝐵𝐵𝐵𝐺𝐺𝐺𝐺𝐺𝐺_𝑃𝑃𝐴𝐴𝐵𝐵_{𝑖𝑖𝑆𝑆} + ∑_{𝑘𝑘=1 k}𝑛𝑛−1µ 𝐷𝐷𝐷𝐷𝐷𝐷_𝑖𝑖𝑘𝑘 + ∑_{𝑥𝑥=1 x}𝑛𝑛−1𝜆𝜆 𝐷𝐷𝐵𝐵𝑆𝑆𝐺𝐺𝐺𝐺𝐵𝐵𝐶𝐶_𝑖𝑖𝑥𝑥 + ∑2017_{𝑦𝑦=2011 y}𝜔𝜔 𝐶𝐶𝐺𝐺𝐴𝐴𝐵𝐵_𝑖𝑖𝑦𝑦 + 𝜀𝜀_{𝑖𝑖𝑆𝑆}

Column 1 shows the outcomes of the -control variables only- Equation (4). Column 2 and Column 3 present the outcomes of Equation (4) when only one main independent variable is included in the equation: board gender diversity in Column 2 and board age diversity in Column 3. Column 4 presents the outcomes for the whole Equation (4). The variable definitions are presented in Table 1. The standard errors are given in parentheses. The *, **, and *** denote statistical significance at the p <.10, p <.05 and p <.01 level (two-tailed), respectively.

The results of the estimation of Equation (4) in Column 4 hold when including each main independent variable separately and the coefficients are fairly similar (β1=-0.076, p<0.05 in Column 4, while β1=-0.077, p<0.05 in Column 2; β2=0.140, p<0.05 in Column 4, while β2=0.143, p<0.05 in Column 3). Moreover, Equation (4) was also estimated using control variables only. The coefficients of the control variables indicate that subsidiary discretionary accruals increase when subsidiaries are more leveraged and prone to report losses. This is consistent with prior literature and with my expectation on the sign of these two controls. However, the coefficients of the other control variables are not statistically significant, suggesting that these variables are not related to the discretionary accruals of the subsidiary.

Since the regression results force us to reject hypothesis 2, it is important to look at possible causes underlying these results. My results suggest that age diversity may be detrimental to the monitoring role of the board, thereby increasing the likelihood of earnings management. Namely, too much age diversity in boards could lead to a big ‘generation gap’. In line with this argument, Taveggia and Ross (1978) mention that age difference can cause differences in values and beliefs among group members, which increases the likelihood of misunderstanding and elevates the level of conflict. It will be time-consuming to reduce this counterproductive conflict and, consequently, the efficiency of the board’s task performance (i.e. monitoring the executives) will decrease (Murray, 1989). Moreover, too much age diversity could induce communication- and trust-issues, which negatively impacts board performance as well (Hambrick et al., 1996). Namely, as Mishra and Jhunjhunwala (2013) state, younger board members could feel intimidated by their older peers and thus would not participate independently anymore. All these reasons are likely to explain why age diversity of the parent’s board members could impair effective board monitoring, resulting in a lower FRQ.

Discussion and Conclusion

The role of boards is to make sure that executives act in the best interest of shareholders. An important part of this role is monitoring the managers to stop them from engaging earnings management. There are multiple studies documenting that board characteristics are related to a firm’s earnings management and thus their FRQ (e.g. Abbott et al., 2012; Gavious et al., 2012; Ho et al., 2014; Huang et al., 2012; Srinidhi et al., 2011). However, to my knowledge, there are