1 MSc Accountancy & Control, track Control
Faculty of Economic and Business, University of Amsterdam
Master Thesis:
CFO Gender and Financial Restatements
Final version
Danceney Domacasse (10456996)
Date:
22 June 2015
First supervisor:
dr. B. Qin
2
Statement of Originality
This document is written by student Danceney Domacasse who declares to take
full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and
that no sources other than those mentioned in the text and its references have
been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision
of completion of the work, not for the contents.
3
Contents
1 Introduction... 5
2 Literature Review and Hypotheses ... 8
2.1 The Changing Role of the CFO: From Accounting to Accountable... 8
2.2 CFO, Financial Reporting Quality and Restatements ... 9
2.2.1 CFO and Financial Reporting Quality ... 9
2.2.2 CFO and Restatements ... 10
2.3 Gender effect in Accounting Research ... 11
2.4 Theory and Hypotheses ... 13
2.4.1 Female CFO and Financial Restatement ... 13
2.4.2 CFO gender, Financial Restatement and CFO Discretion ... 15
2.4.3 Financial Restatement and Male to Female CFO Transition ... 16
3 Research Design ... 18 3.1 Variable measurement ... 18 3.2 Empirical model ... 20 3.3 Sample selection ... 23 4 Results ... 24 4.1 Descriptive statistics ... 24 4.2 Regression Results ... 30 4.2.1 Baseline Regression ... 30 4.2.2 Regression analysis ... 31
4.2.3 Additional Tests Hypothesis 1... 32
4.3 Moderating Effect CFO Discretion ... 33
4.4 Male to Female and Financial Restatements ... 36
5 Discussion and Limitations ... 38
4
Abstract
This thesis examines the influence of CFO gender on financial restatement based on CFO gender transitions while using a sample of S&P 1500 firms over the period 1997 to 2014. The results presented in this thesis suggests that no evidence has been found that support a relationship between CFO’s gender and financial restatement. Three different settings are used to provide statistical evidence for the relationship between CFO’s gender and financial restatement. The first sample contains only firms that experienced a change in the gender of their CFO. The second sample consists of firms that did and did not experience a change in the gender of their CFO. The main focus on the second sample was on the female CFO’s. However, the use of this sample was also inconclusive in providing significant evidence for the relationship between CFO gender and financial restatement. In addition, this thesis also studied whether CFO discretion, proxied by CFO tenure, has an influence on the relationship between CFO gender and financial restatement. So no evidence is found of the relationship between female CFO’s and financial restatement. In a sample containing transition and non-transition firms, this thesis has provided evidence that financial restatements while the CFO contains a higher level of power, measured by CFO tenure in years, is more likely to happen. With regards to the expectation that financial restatements increases the likelihood of male to female CFO transition, this thesis has failed to find significant evidence. Overall, the findings in this thesis can help future research to
investigate this gender differences and financial reporting quality while using a different approach, considering the fact that this thesis has failed to find significant evidence for the suggested relationships.
5
1 Introduction
The objective of this study is to provide empirical evidence regarding the relationship between a CFO’s gender and the likelihood of financial restatement. The reason I have been motivated to study this topic is because of the still small, but increasingly percentage of women who holds executives positions, especially in the US. Catalyst (2004) was one of the first research that brought gender diversity and female representation in top management and boardrooms to the attention. Moreover, Bloomberg (2013) indicated that only 54 CFO’s positions was held by female. Both these studies indicate that executives’ position is still a section dominated by men. Furthermore, a study by Scholz (2008) provided evidence that since the passage of Sarbanes-Oxley (2002), restatements have increased tremendously. Which means that the passage of Sarbanes-Oxley, which was introduced to restore confidence in the integrity of the company’s financial accounting was a success.
After the passage of Sarbanes-Oxley, the role of the CFO regarding his responsibilities towards the financial reporting has changed. If before, the CFO was mainly involved in
bookkeeping, preparing tax statements and creating budget, nowadays companies expect CFO to not only cut costs but also play an active role in the company’s financial accounting, internal control systems, and risk management. Because of the introduction of SOX not only the CEO is legally responsible for the company’s actions, but sections 304 and 906 have legislatively elevated the CEO and CFO to the same level of financial oversight. Furthermore, these sections also makes the CEO and CFO personally responsible for the financial information and
disclosures released by the company (Leech, 2003; Fabozzi, Drake and Polimeni, 2008; Collins et al., 2009).
Prior literature regarding the effects of the passage of SOX are mixed. Most studies find evidence that SOX and the resultant SEC certification requirement may have altered
management’s discretionary reporting behavior to make it more conservative, because firms report lower discretionary accruals after SOX than in the period preceding SOX (e.g. Lobo and Zhou, 2006; Chen, Elder and Hung, 2014; Cohen, Dey and Lys, 2008; ). The results from prior literature might suggest that the financial reporting quality has increased since the passage of SOX. However, the net costs and benefits of SOX is very difficult to assess according to Coates and Srinivasan (2014). Scholz (2008) studied the changing nature and consequences of public financial restatements between 1997 and 2006. As she indicates, a restatement event begins with
6 the announcement of an accounting problem or potential accounting problem and concludes with the filing of the amended results.
Most US based research on gender diversity and female representation focus on the relationship between gender in senior management, board rooms and firm performance. One of the first researches to emphasis the issue of enhancing firm performance while having gender diversity in boardrooms or in senior management was Catalyst (2004). Since then, several research and advisory organizations are investigating this issue. Among them was Carter et al. (2003), Frink et al. (2003) and Adams and Ferreira (2009) who found a positive relationship between board diversity and firm value. They find that a higher percentage of women and minorities in boardrooms can increase firm value. The argument that gender diversity can enhance firm performance is a business case argument suggesting that more diversity translates into better decisions because of the heterogeneous characteristics of a group. Even though most of this research studied gender diverse boardrooms, they all suggest that the variable gender has influence on firm performance.
Most of the studies that look at female CEO’s and CFO’s linked gender to earnings management while looking at their risk taking behavior. The theory behind this is mostly that women are more cautious than men and therefore report lower absolute abnormal accruals (Barua et al, 2010). They also provide evidence for this theory by showing that female CFO’s report lower level of absolute accruals and lower accrual estimation errors. Furthermore, Francis et al. (2014) researched differences in risk-taking behavior based on CFO’s gender. They
concluded that female CFO’s are associated with less tax aggressiveness as compared to their male counterparts, indicating that female CFO’s are more risk averse.
Causes and consequences of financial restatements have been widely researched
according to Abbott et al. (2012). Most of this studies have linked financial restatements with the role of board of directors and audit committee due to the fact that they have a monitoring role (Abbott, et al. 2012). They find a significant association between the presence of at least one women on the board and a lower likelihood of restatement. Aier et al. (2005) also looked at financial restatements but from a different viewpoint. The financial expertise of CFO’s was among the CFO’s characteristics that they looked at. They concluded that restatements is
negatively associated with the CFO’s financial expertise. Other financial restatement research is related to CEO/CFO turnover Feldmann et al. (2009) and Hennes et al. (2008). In summary,
7 most of the abovementioned studies have studied gender diversity and financial restatements from the other viewpoint than the CFO transitions and financial restatements. Gender diversity research was mostly based on gender diversity in boardrooms and financial restatement research looked at CEO/CFO turnover or other CFO characteristics.
According to Habib and Houssain (2013) very little research has been done on the role of CFO’s gender on financial reporting quality. Barua et al. (2010) also indicates that using other proxies such as restatements might be fruitful avenue for future research in gender diversity. Finally, in order to enhance the power of the test Francis et al. (2014) has motivated me to investigate gender differences from a transition point of view. For this reason, the approach of this thesis will be focused mainly on the transition of male CFO’s to female CFO’s considering the CFO’s role in financial decision making. The way I have chosen to express the quality of the CFO’s role in financial reporting quality is by linking CFO transitions with financial
restatements.
This study contributes to the existing literature in three ways. Firstly, my research contributes to the financial restatement literature because this study provides empirical evidence regarding the effect of CFO transitions on financial restatements. Secondly, the way this study differs from the previously mentioned studies is that it contributes to the CFO gender and financial restatements literature by proving empirical evidence related to the relationship
between CFO transitions and financial restatements. To my knowledge, this has been one of the first studies to provide evidence for this relationship. Thirdly, although managerial turnover has been studied before, to my knowledge, not one of these studies have examined the likelihood of a male to female CFO transition after a financial restatement. For this reason, this will be the first study to contribute in such a way to the restatement literature.
This paper is structured as follows, chapter 2 will discuss the relevant literature regarding the key topics of the paper and hypotheses. Chapter 3 addresses the research method. Chapter 4 provides the results obtained. The paper concludes in chapter 5 with a discussion of the results, limitations of this thesis and areas for future research.
8
2 Literature Review and Hypotheses
The literature review will address three areas related to the CFO, financial restatements and gender effect in accounting research. The first section will address research related to the changing role and responsibilities of the CFO in the financial reporting process. The second section will focus on research studies about the importance of financial reporting quality and financial restatements. Finally, the third section will discuss research related to studies about gender effect in accounting research.
2.1 The Changing Role of the CFO: From Accounting to Accountable
CFO´s are the key players in financial reporting process. The rise of the CFO, not only in corporate America, but globally has been extraordinary. They have managed to transform their position as corporate treasurer to become one of the most important players in corporate America today (Zorn, 2004; Geiger and North, 2006; Ge et al., 2011). Previously, their tasks mainly involved bookkeeping, preparing tax statements and creating budget which was usually after production decisions has been made. Nowadays, they have a monitoring role and are in charge of financial planning, budgeting, internal control and financial reporting (Feng et al., 2011).
Besides, they are responsible for establishing and maintaining internal controls and reporting of deficiencies to the audit committee and the external auditors. Therefore, they must work closely with internal controls in order to identify any potential internal control material weaknesses (Aier et al., 2005).
Prior literature suggests that the role and responsibilities of a CFO should have the following skills to become successful: develop a strategic vision; become an information
specialist; possess well developed communication skills and be a leader. To mention a few of his specific tasks which falls under the previously mentioned skills are: pursue shareholder value; construct reliable control systems; comply with laws, rules and regulations; monitor and evaluate internal controls; develop and execute strategic plans and goals; evaluate performance relative to the strategic goals; participate in long-term and short-term budgeting; raise capital and manage the firms’ capital structure while creating shareholders’ value; understand and mitigate risks; minimize cost of capital; and communicate with the firms’ Board of Directors, shareholders, creditors, and credit rating agencies (Fabozzi, Drake and Polimeni, 2008; Favaro, 2001; Bragg, 2007). Based on the abovementioned responsibilities of the CFO, it is obvious that the CFO
9 should not only tell the CEO how businesses did go, but also provide guidance and direction to accomplish their contribution to the company’s goals. Therefore, it can be concluded that the CFOs´ role has changed over the years to be much more comprehensive and to include decision-making that extends beyond the accounting and treasury functions (Fabozzi, Drake and Polimeni, 2008).
As the role and responsibilities of the CFO is becoming much more complex, financial scandals such as Enron and Worldcom are not contributing to the publics’ and investors’ confidence in the CFO’s monitoring responsibilities, the Sarbanes-Oxley Act of 2002 has been enacted by the Congress to help restore their reputation (Geiger and North, 2006; Hirschey, Smith and Wilson, 2012; Chen, Cheng and Lo, 2014). Its main purpose is to protect investors by improving the accuracy and reliability of corporate disclosures (SOX, 2002) and to restore confidence in the integrity of the company’s financial accounting, internal control systems, and risk management (Fabozzi, Drake and Polimeni, 2008). According to Collins et al. (2009): “The key CFO accountability provisions of SOX can be summarized as follows: section 302 requires CFOs to certify, among other things, the fairness of the presentation of the financial statements; section 906 imposes on CFOs potential criminal sanctions for misstated financial statements; and section 304 requires that CFOs disgorge any bonuses or stock market gains that result from misstated earnings.” Not only does this means that CFO’s have now been
legislatively elevated to the same level of financial oversight as the CEO. But together with sections 302 and 906, it also means that the CEO and CFO are personally responsible for the financial information and disclosures released by the company (Leech, 2003; Fabozzi, Drake and Polimeni, 2008; Collins et al., 2009).
2.2 CFO, Financial Reporting Quality and Restatements
2.2.1 CFO and Financial Reporting Quality
To guarantee the integrity of a company´s financial statement, several departments together with the audit committee, CEO and CFO are responsible for the underlying processes. An external auditor reports its findings to the audit committee regarding tests he has performed in the financials for material misstatement based on prevailing accounting rules. The internal audit department is responsible for separately testing accounting processes and controls. Finally, the management is required to certify that financial reports do not contain misleading information
10 (Larcker and Tayan, 2011). As a result of the CFO’s role and responsibilities regarding the financial statements, it can be concluded that the financial reporting quality is one way to measure how he has performed his job. However, the measures for determining the financial reporting quality differs from research to research.
Prior literature regarding the financial reporting quality is extensive. However, most studies overlap regarding the measures for financial reporting quality. According to Hirschey, Smith and Wilson (2012) the academic research has demonstrated that some improvements in the reporting quality occurred since the introduction of the SOX Act (Lobo and Zhou, 2006; Cohen, Dey and Lys, 2008; Bartov and Cohen, 2009). Lobo and Zhou (2006) reports an increase in conservatism in financial reporting following SOX, while looking at management’s
discretionary reporting behavior. According to Cohen, Dey and Lys (2008) this suggests that firms switched from accrual-based to real earnings management methods after the passage of SOX. Meanwhile, Scholz (2008) demonstrated that there has been a rise in restatement announcements between 1997 and 2006. This suggests that management might manipulate or overstate results to meet targets (Larcker and Tayan, 2011).
2.2.2 CFO and Restatements
Financial restatements can cause sleepless nights for executives at home or in prison and it can have a negative effect on the company. Financial statements not prepared in accordance with generally accepted accounting principles are presumed to be misleading or inaccurate. In case the financial statement contains a material error or irregularity, it should be restated.
Even though SOX was implemented to improve the accuracy and reliability of corporate disclosures, a research conducted by Scholz (2008) reported that there has been a rise in
restatement announcements between 1997 and 2006. Financial Executives International and Wu (2001) reported that revenue recognition caused 33 percent of the restatements between the periods of 1995 to 2000.
Companies are aware of the effects of having to announce a restatement and therefore are taking actions to restore the confidence by issuing earnings guidance more frequently after having to announce a restatement and being more transparent while doing so (Ettredge, Huang and Zhang, 2012). Firms are also more transparent in mentioning which actions they are taking to restore investors’ credibility, such as engaging in share repurchases and announcing changes
11 in internal control mechanisms (Chakravarthy, deHaan, and Rajgopal, 2014). It remains difficult to assess the direct and indirect benefits of the SOX, but it is worth mentioning that companies are doing their best in order to regain and restore the publics’ and investors’ credibility in their financial reporting. Prior literature suggests that CFO only engage in material accounting manipulations, because they are pressured by the CEO to do so (Feng et al., 2011), contrary to what people might think that CFO’s seek immediate personal financial benefits.
As was the case with prior literature regarding financial reporting quality, research regarding financial restatements is extensive. Academic research shows that following a restatement announcement, the company´s experience a negative market reaction. These
companies also tend to be tend to be smaller, less profitable, and slower growing than their peer in the same industry and they face more serious financial uncertainties (Kinney and McDaniel, 1989; GAO-03-138; GAO-06-678; Hribar and Jenkins, 2004; Palmrose and Scholz, 2004; Wu, 2002; Chen, Elder and Hung, 2014).
Several studies also describe the effects of management turnover for restating companies as the management is responsible for the companies’ performance. Restatement firms were
significantly more likely to experience turnover of their CEOs, CFOs, especially for restatements due to the use of aggressive accounting (Arthaud-Day et al., 2006; Desai, Hogan and Wilkins, 2006; Hennes, Leone and Miller, 2008; Collins et al., 2009; Burks, 2011).
2.3 Gender effect in Accounting Research
The number of female executives has been increasing and therefore gender differences in
accounting related to their decision-making behavior has begun to attract attention in accounting research (Francis et al., 2014). In the business ethics research, where ethical differences between genders have been examined extensively, the literature suggests that there is a difference
between ethical behavior of men and women. Men are more driven to gain economic benefits and successful careers therefore they are more prone to break rules. On the contrary, women lean more towards harmonious relationships and helping others, and are less likely to behave
unethical (Habib and Houssain, 2013, p.95). Women tend to be less risky in gambling (Levin, Snyder, and Chapman, 1988) and while placing their bets (Johnson and Powell, 1994). All these studies suggests that women are more risk averse than men.
12 Byrnes et al. (1999) and Charness and Gneezy (2012) finds that women, on average, are more cautious and less aggressive than men in a variety of decision contexts. Women tend to be less likely to hold most of their assets in stocks (Sundén and Surette, 1998) and they tend to allocate their pension more conservatively than men (Bernasek and Shwiff, 2001). Barua et al. (2010) finds enough evidence that supports their hypothesis for firms that have female CFO’s will have higher accruals quality than firms with male CFO´s. Furthermore, Peni and Vähäama (2010) provide considerable evidence implying that female CFO’s are following more
conservative earnings management strategies because they are associated with
income-decreasing discretionary accruals. On the contrary to the previously mentioned studies, Ge et al. (2011) do not find a CFO gender effect on reporting choices, whereas they find significant CFO fixed effects on accounting choices; this suggests that these common and observable
characteristics capture only a small portion of CFO styles. Because of this contradiction in the literature, it will be important to provide empirical evidence to see whether there is a relationship between a CFO’s gender and the likelihood of financial restatements.
Prior literature regarding the effect of gender diversity in boardrooms suggest that gender does play a role in accounting. For example, Adam and Ferreira (2009) find that gender-diverse boards allocate more effort to monitoring. Carter, Simkins and Simpson (2003) and Deso and Ross (2012) finds that while the representation of women and minorities in boardrooms increases, the firms’ value also increases. Abbott et al. (2012) find a significant association between the presence of at least one woman on the board and a lower likelihood of restatement. Gul, Srinidhi, and Ng (2011) find that stock prices of firms with gender-diverse boards reflect more firm-specific information, which makes these stock prices more transparent. Based on the previously mentioned studies, it can be concluded that gender diversity in boardrooms has a significant impact on corporate governance, firm performance and stock price.
Recently, the economics literature regarding the effect of corporate executives’ gender on corporates’ decision-making has been growing. For example, Huang and Kisgen (2013)
compared female executives financial and investment decisions with male executives. Their results suggests that men shows overconfidence behavior while taking significant corporate decisions compared to women, because they undertake more acquisitions and issue debt more often. Women also tend to place wider bounds on earnings estimates and are more likely to exercise stock options early. Levi, Li and Zhang (2014) examined the impact of the CEO´s
13 gender or corporate directors on the pricing of and returns on mergers and acquisitions. They find that the bid premium paid over the pre-announcement target share price when the CEO of the bidding company is a woman is, over 70 percent smaller than when the bidding company CEO is a man. On the other hand, Dwyer, Gilkeson and List (2002) show that while women exhibit less risk-taking than men in their mutual fund investment decisions, the impact of gender on risk taking is significantly weakened when they control for investor knowledge of financial markets and investments. These results indicate that it remains unclear what effect the variable gender has on corporate decision-making. Finally, Francis et al. (2014) compared the gender differences in financial reporting decisions from an accounting conservatism viewpoint. They compared the CFO’s tax aggressiveness in firms that had experienced a male-to-female CFO transition. They find that CFOs are more conservative in their financial reporting, but that this relationship varies with the levels of various firm risks such as litigation risk, default risk, systematic risk, and CFO specific risk such as job security risk.
2.4 Theory and Hypotheses
Economics, psychology and sociology theories can be used to make predictions and describe the results. For this reason I think it would be interesting and contributing to the literature to use theories from both fields. In this paper, I will focus on CFO gender, instead of gender diversity of the top management team because as was pointed out by Ge et al. (2011) and Francis et al. (2014). “The CFO typically oversees the firm’s financial reporting process and therefore he/she likely has the most direct impact of all the senior managers on the accounting related decisions of the firm.”
2.4.1 Female CFO and Financial Restatement
A study by Barua et al. (2010) has provided evidence for the existence of a relationship between CFO’s gender, especially female CFO’s, and the quality of accruals. In this research they
conclude that firms with female CFO’s have higher quality of accruals. Moreover, financial restatements have been widely used as a proxy for low earnings quality (Kuang, Liu and Qin, 2015). Therefore, an indirect relationship between the CFO’s gender and financial restatements has already been established through academic research. Upper echelons theory also suggests that top managers’ individual characteristics affect how they assess or interpret their situations and therefore impact their decisions. For this reasons, both prior research and theory have
14 provided evidence for the fact that individual characteristics of executives’ plays and important role in certain corporate decisions (Ge et al., 2011).
According to psychology studies, females are more risk averse than men because females were more cautious in their gambling decisions than males (Levin, Snyder, and Chapman, 1988). A meta-analysis by Byrnes et al. (1999) also supported this by concluding, in generaal, that male are more likely to take risks than female participants. However, they also conclude that the gender differences varied according to context and age level. This means that the gender differences does not seem to manifest in a simple or constant way. Therefore, the evidence provided in psychology literature is not always consistent with the notion that female are risk averse compared to men. Meanwhile, gender differences in economics research has provided more conclusive evidence regarding executives’ gender differences than psychology studies. For instance, Peni and Vähämaa (2010) conclude that female CFO’s show a more conservative earnings management strategy. In addition, Barua et al. (2010) report that earnings quality is positively associated with gender diversity in senior management and female CFO. Furthermore, corporate governance research also provide evidence that firms with female directors exhibit higher earnings quality (Krishnan and Parsons, 2008; Srinidhi, Gul and Tsui, 2011). Based on the prior literature, it can be concluded that female CFO and directors has a positive influence on decision-making. One possible explanation for the reason why economics literature has managed to provide more consistent evidence is the fact that economic literature studies gender
differences from a more simplistic approach compared to psychology studies. Therefore, economic studies link the significant results they have obtained, more easily to gender differences rather than personal differences.
While looking at the influence of CFO gender on the likelihood of financial restatement from an overconfident perspective –as one of the variables of judgement and decision-making literature (hereafter, JDM)-, this provides a base for the possibility of the existence of a negatively related relationship. The JDM focuses on how individual characteristics affects judgement and decision-making (Bonner, 2008). The individual characteristics are: (1) confidence, (2) risk attitudes and (3) intrinsic motivation. Huang and Kisgen (2013) provide evidence in line with this theory by reporting that men exhibit relative overconfidence in significant corporate decision making compared to women. This overconfidence behavior has been detected by Goel and Thakor (2008) in the CEO selection process and managers who
15 successfully climb the corporate ladder to become CEO’s are also likely to become
overconfident. In addition, Hribar and Yang (2010) report that overconfident managers make more income-increasing accruals choices and that they engage in fraudulent activities (Schrand and Zechman, 2012). Even though, the previous literature regarding overconfidence behavior has focused on managers and CEO’s behavior, based on Huang and Kisgen (2013) and because the CFO is the top-level executive most directly involved in accounting choices, the findings in this line of research lend support to the fact that female CFO’s might have follow a more
conservative strategy in their financial reporting. Formally, I predict that:
H1: Female CFO’s and the likelihood of a financial restatement is negatively
related.
2.4.2 CFO gender, Financial Restatement and CFO Discretion
The previous paragraph focused on prior literature that established a relationship between CFO gender, especially female CFO, and financial restatements. Aforementioned was that a call to investigate causes of financial restatements has been made. This was among the reasons that motivated Scholz (2008), Abbott, Parker and Presley (2012), Aier et al. (2011) and Ge et al. (2011) to contribute to the restatement research by studying possible causes of financial restatements while focusing on the effect of various individual characteristics.
Ge et al. (2011) reports that when CFO discretion is higher CFO style reflects more in accounting choices. This study was based on the upper echelons theory by Hambrick and Mason, 1984; Hambrick, 2007 which states that organizational outcomes-strategic choices and
performance levels are partially predicted by managerial background characteristics. This means specifically that that top managers’ individual characteristics affect how they assess or interpret their situations and therefore impact their decisions. Following this theory and JDM literature, it is clear that individual characteristics -specifically CFO’s gender- might influence the accounting decisions, in this case financial reporting. Prior literature has managed to provide evidence that gender is a significant variable in CFOs’ managing styles and risk attitudes (Barua et al., 2010; Peni and Vähämaa, 2010; Francis, Hasan and Wu, 2014; Francis et al., 2014). The moderator CFO discretion was introduced by Hambrick and Mason, 1984; Hambrick, 2007 in order to enhance the validity of upper echelons theory. Hambrick (2007) states that “discretion exists when there is an absence of constraint and when there is a great deal of means-ends
ambiguity-16 that is, when there are multiple plausible alternatives.” In other words, in case there are multiple alternatives, a choice made by a CFO will be made based on the CFOs’ individual
characteristics. In addition, Feng et al. (2011) found that CFO’s become involved in accounting manipulations under pressure from CEO’s, rather than instigating such manipulations for
immediate personal financial gain. Together with the fact that CFO’s discretion can influence the CFO’s accounting choices, the evidence provided by Feng et al. (2011) strengthen this theory because CEO’s use their power to push CFO’s to become involved in accounting manipulations. Therefore, it can be concluded that given the fact that CEO power influences CFO’s accounting decisions, CFO’s individual characteristics becomes much more important variable in
determining the strength of the relationship between a CFO’s gender and their accounting decisions.
The moderating effect of CFO discretion in this case I expect to be by making the relationship between CFO gender and financial restatements stronger. Ge et al. (2011) was one of the first studies to provide empirical evidence for CFO style being reflected more in
accounting choices when CFO discretion is higher. Therefore, managerial discretion was not only present in CEOs managing styles, but also in CFOs’ accounting choices. Moreover, the moderating effects of CFO discretion on the relation between CFOs’ gender and financial restatements contributes to the literature of upper echelons theory. For this reason, I expect to provide evidence supporting the following hypothesis:
H2: The relationship between CFO gender and financial restatements is stronger when CFO discretion is higher.
2.4.3 Financial Restatement and Male to Female CFO Transition
Much of the existing research on restatements has examined the process by which firms deal with repairing their damaged imaged resulting from a financial restatement (Karpoff et al. 2008a, 2008b), the implications of relationship damage such as higher cost of capital (Hribar and
Jenkins, 2004), legal consequences (Palmrose and Scholz, 2004), negative market reactions (Audit Analytics, 2008; Scholz, 2008) and higher reputation actions after restatement
(Chakravarthy et al., 2010). A number of papers have also examined leadership and governance changes after a restatement without explicitly linking the actions to gender differences (Agrawal and Cooper, 2009; Arthaud-Day et al., 2006; Collins et al., 2009; Karpoff, Lee and Martin,
17 2008b). However, only Abbott et al. (2012) examined the possible effect of gender on the
likelihood of financial restatements.
The reason why a financial restatement increases the likelihood of a male to female CFO transition is because management turnover increases following instances of GAAP violations and fraudulent reporting (Arthuad-Day et al., 2006; Feldmann et al, 2009; Agrawal and Cooper, 2009; Burks, 2010), females are less likely to behave unethical (Habib and Houssain, 2013) and female are risk averse (Byrnes et al., 1999; Charness and Gneezy, 2012). In addition, from a legitimacy perspective, firms tend to distance itself from the management team associated with the restatement (Arthuad-Day et al., 2006; Feldmann et al., 2009; Chakravarthy et al., 2010). Finally, the gender differences in accounting choices for the development of the previous hypotheses, which has been discussed in greater detail previously, also apply in this case. Therefore, it can be concluded that prior literature has provided evidence that turnover is an effective means to repair legitimacy and restore confidence in the financial reporting process (Feldmann et al., 2009). Moreover, institutional theory suggests that a firms’ management should adopt strategies that help them establish, maintain or defend their organization’s legitimacy. As the firm distance itself from the management team associated with the restatement, the message will be send that the firm is taking steps to regain its legitimacy. Therefore, hiring a new CFO whose individual characteristics are perceived to be having a less risky behavior than its predecessor is mostly searched by a firm by hiring a female CFO. Formally, I predict:
H3: Financial restatements increases the likelihood of a male-to-female CFO transition.
18
3 Research Design
This chapter will discuss how the variables to test the hypothesis presented in paragraph 2.4 will be measured. Next, together with the empirical model the description of the variables will be presented. Finally, the sample selection will be described.
3.1 Variable measurement
Measures of financial restatements
Prior literature has generated a relatively simple measure for restating companies (e.g. Abbott et al. 2012; Aier et al. 2005; Agrawal and Cooper, 2009; Arthaud-Day et al. 2006; Kuang et al. 2015). In the empirical model, for the variable RESTATE a dummy will be created which equals 1 if the firm has had a financial restatement, and 0 otherwise. In this case the variable RESTATE will not make a distinction between financial restatements based on accounting, fraud or clerical error.
Measure of gender differences
Following Francis et al. (2014) I adopt the same measure to capture possible gender effect on financial restatements. A dummy variable POST will be used that equals 1 if a year is after a male to female CFO transition year and 0 if a year is before the CFO transition year. This measure helps me to examine the gender effect on financial restatements by comparing the degrees of restatements between pre- and post-transition periods for male to female CFO.
Measure of CFO discretion
CFO discretion, which has been introduced in hypothesis 2, will not be measured the same way as Ge et al. (2011) because this method would be too complex and indirect. For this reasons, a simpler method will be adopted to capture CFO discretion which is by using CFO’s tenure in years as a proxy for CFO discretion. A dummy variable POWER will be created based on the CFO tenure measured in years. POWER is coded 1 if CFO tenure is above the median, and 0 if the CFO tenure is lower than the median.
19 Control variables
To alleviate omitted variable problem, I included two categories of control variables based on Abbott et al. (2012), Francis et al. (2012) and Kuang et al. (2015). Following their study, a wide range of opportunity- and incentives- based control variables are used.
The first category of variables which are being controlled for is when the lack of controls provide an opportunity for a restatement to occur. Therefore, this category of control variables are opportunity-based (Abbott et al., 2012). Abbott et al. (2004) find that firms with audit
committees that are independent meet at least four times annually, and have at least one financial expert on the audit committee have a reduced incidence of restatement. For this reason, ACE is being defined in a manner consistent with Abbott et al. (2004) and Abbott et al. (2012), and expect a negative relation between ACE and the incidence of restatement. The following opportunity-based control variable are consistent with Beasley (1996). BOARDSIZE and is defined as the number of board directors. According to Beasley (1996) larger boards may result in ineffectual monitoring due to a proclivity toward communication breakdowns and
inefficiencies. Therefore, I expect a positive relationship between BOARDSIZE and incidence of restatement. OUTSIDER is defined as the percentage of board members who are non-employee directors. A negative relation between the percentage of outside directors on the board and the incidence of restatements is expected, because outside directors have reputational capital incentives to minimize the incidence of restatements. Finally, consistent with Beasley and Dechow et al. (1996), I expect a positive relation between a CEO that is also chairman of the board and the incidence of restatement, because while the CEO is also chairman of the board it reduces the effectiveness of the board’s ability to monitor the CEO. CEOCHAIR is coded 1 in instances where the two positions are combined, 0 otherwise. In addition, corporate governance control variables are being introduced based on Kuang et al. (2015). Bedard et al. (2014) proved that CFO’s on the board have more effective internal control over financial reporting, higher accruals quality, and lower likelihood of restatements. Based on this, they conclude that accounting choices made by a CFO serving on the board are more aligned with shareholder interest. Hence, I expect CFOONBOARD to have a positive effect on the likelihood of restatement.
The second category of variables which are being controlled for are in terms of economic determinants. According to Abbott et al. (2012) prior restatement research has documented that
20 capital market pressures may lead to more aggressive accounting and eventual restatement. Various restatement research (e.g. Aier et al, 2005; Abbott et al, 2012; Francis et al, 2014; Kuang et al, 2015) has used LEVERAGE as a control variable, because increased incidence of debt suggests the presence of debt covenants, resulting in increased incentives for earnings management. Following Kuang et al. (2015) MTB will represent the market value of equity divided by the book value of equity, because the desire to maintain a firm’s rate of growth can create an incentive to manipulate earnings (Beasley, 1996). According to Abbott et al. (2012) the desire to maintain a firm’s rate of growth can create an incentive to manipulate earnings. Finally, M&A firms undertaking merger and acquisition activities are more likely to restate their annual reports (Kuang et al. 2015). In addition, economic determinants such as research and
development expenditures (R&D) and determinants related to the desire of long-term debt of firms are also being used. FIN and ISSUANCE will be measured based on Kuang et al. (2015). FIN will take the value of 1 if the sum of long-term debt issuance and sale of common stock scaled by a firms total assets is greater than 0.2 or 0 otherwise. ISSUANCE will also be a dummy variable taking the value of 1 if a firm’s long-term debt issuance and sale of common and preferred stock is greater than 0, and 0 otherwise. Finally, following Francis et al. (2012) it is likely that restatements are non-random across time and industries, and therefore, consistent with them I estimate my model with year and industry fixed effects to control for the idiosyncratic effects of time and industry.
3.2 Empirical model
To examine the gender effect on financial restatement the same method as was used by Francis et al. (2014) will be adopted. The primary research design is to compare the degrees of financial restatements between the pre- and post-transition periods from male-to-female CFO turnover firms. This method mitigates the criticism other empirical studies often face that the observed differences are not attributable to gender, but instead to some omitted factors (Francis et al. 2014).
21 The main empirical model is the following:
RESTATEi,t = β0 + β1 POSTi,t + β2 ACEi,t + β3 BOARDSIZEi,t + β4 OUTSIDERi,t +
β5 CEOCHAIRi,t + β6 CFOONBOARDi,t + β7 LEVERAGEi,t +
β8 MTBi,t + β9 M&Ai,t + β10 R&Di,t + β11 LOSSi,t + β12 SIZEi,t + (1)
β13 FINi,t + β14 ISSUANCEi,t + β15 YEAR + β16 INDUSTRYi,t + ɛi,t
POST will be the primary variable of interest and is a dummy variable that equals one if a year is after the CFO transition and zero if a year is before the CFO transition year, consistent with Francis et al. (2014).
22 TABLE 1 VARIABLE DESCRIPTION VARIABLE Description RESTATE FEMALE FEMALECFO MALETOFEMALE POST ACE BOARDSIZE OUTSIDER CEOCHAIR CFOONBOARD LEVERAGE MTB M&A R&D LOSS SIZE FIN ISSUANCE YEAR INDUSTRY
An indicator equal to 1 for all misstated years, and 0 otherwise. An indicator equal to 1 if the transition is from male to female CFO, 0 otherwise (i.e., either male to male, female to female, or female to male).
An indicator equal to 1 if the firms’ CFO is female, and 0 otherwise. An indicator equal to 1 if the firms CFO in year t has male to female transition, and 0 otherwise (i.e., either male to male, female to female, or female to male or no transition).
An indicator equal to 1 if a year is after the CFO transition year, and 0 if a year is before the CFO transition year.
An indicator equal to 1 if the audit committee is comprised entirely of independent directors, has at least one director with financial expertise, and meets at least four times annually, and 0 otherwise.
The number of directors on the board.
The percentage of the board members who are non-employee directors. An indicator equal to 1 if the chairperson of the board holds the
managerial positions of CEO or president, and 0 otherwise. An indicator equal to 1 if a CFO is a member of the board, and 0 otherwise.
The ratio of total liabilities to total assets.
The market value of equity divided by the book value of equity. An indicator equal to 1 if the firm engaged in a merger or acquisition or the firm had an acquisition that contributed to sales during the current year, and 0 otherwise.
Research and developments expenditures scaled by total assets An indicator equal to 1 if a firm reported negative net income in current year, and 0 otherwise.
Natural logarithm of total assets for firm.
An indicator equal to 1 if the sum of long-term debt issuance and sale of common stock scaled by a firms total assets is greater than 0.2 or 0 otherwise.
An indicator equal to 1 if a firms’ long-term debt issuance and sale of common and preferred stock is greater than 0, and 0 otherwise. Year dummies.
23
3.3 Sample selection
Following Francis et al. (2014) the gender information used is primarily from the ExecuComp database, which covers most of the S&P 1500. Execucomp generated a number of 17,565 observations for CFO’s gender. The CFO construction sample is based on the following filters: (1) both pre- and post-transition CFO’s must be in office consecutively for at least three years excluding the transition year; (2) financial firms and utility companies are being excluded (SIC codes between 6000 and 6999 and between 4900 and 4999, and the sample period is from 1997 to 2014. The reason why this sample period was selected was because finding a sufficiently large sample was expected to be problematic to capture most transitions as possible. The resulting sample is then merged with Compustat, Audit Analytics and Risk Metrics to obtain financial and corporate governance information. The final CFO transition sample consists of 2,124 firm-year observations with 33 cases of male-to-female CFO transitions, 24 cases of female-to-male CFO transitions and 257 cases of male-to-male CFO transitions.
24
4 Results
This chapter will present the results of the predictions proposed in chapter 2. The first paragraph will describe the descriptive statistics and correlations based on the male to female CFO
transition. The second paragraph will present the regression results based on two sample, a sample containing only firms that experienced a CFO gender transition and a sample containing firms that did and did not experience a CFO gender transition. This regressions were run to tes hypothesis 1. The third paragraph reports regression results for hypothesis 2, which is regarding the effect of CFO discretion. Finally, this chapter tests for the predictions of hypothesis 3 which is reports the relationship between male to female CFO transition and financial restatements.
4.1 Descriptive statistics
Table 2 panel A shows the industry distribution for the sample containing only transition firms. The manufacturing industry represents 52% of the total firm year observations. Panel B reports how the amount of observations was determined.
Table 2
Panel A: Observations per Industry POST
(1) (2) (3)
Industry SIC Number of Observations
Agriculture, Forestry & Fishing 0100 - 0999 6
Mining 1000 - 1499 109
Construction 1500 - 1799 55
Manufacturing 2000 - 3999 1111
Transportation & Public Utilities 4000 - 4999 173
Wholesale Trade 5000 - 5199 73
Retail Trade 5200 - 5999 191
Services 7000 - 8999 406
Public Administration 9100 - 9999 0
25
TABLE 2
Panel B: Sample Selection
Observations about Gender
15,929 Observations dropped based on SIC
between (4900 - 4999 and 6000 - 6999)
4,121 CFO tenure <3 years
1,455 Observations FEMALECFO
10,353 Observations dropped because of no
transitions 8,229 Observations POST 2,124
Panel C describes the amount of financial restatements and male-to-female CFO transitions that took place in the sample period. It appears that the proportion of firms female CFO’s has been increasing over time. Notable is the fact that the financial instability because of the financial crisis may have increase in female CFO’s quite gradually. This increase is quit notable in 2007 compared to the previous years where there has not been a male-to-female CFO transition. However, this could have been influences by the selections made to construct the sample in this research or because of the completeness of the gender information regarding the database.
26
TABLE 2
Panel C: Distribution of Observations
Year RESTATE Male to Female CFO Transition 1997 0 0 1998 0 0 1999 0 0 2000 0 0 2001 0 0 2002 0 0 2003 0 0 2004 0 0 2005 0 0 2006 1 0 2007 12 1 2008 19 1 2009 15 4 2010 16 6 2011 25 16 2012 33 5 2013 30 0 2014 4 0 --- ---- --- 155 33
Table 2 panel D reports sample statistics containing the number of observations, mean, standard deviation, median and the 25% and 75% quartile of male-to-female CFO transition sample. This sample consists only of firms that experienced a CFO transition during the period 1997 and 2014. Different amount of observations are reported because of the availability of data after merging data from different databases. The mean value of RESTATE is 0.0730, which is lower than was reported by Kuang et al. (2015). The reported mean values of BOARDSIZE 9.1199, FIN 0.6317 and ISSUANCE 0.9980 are similar to the means reported by Kuang et al. (2015). OUTSIDER is 0.7992, the mean value of BOARDSIZE is 9.1199, the mean value of ACE is 0.0186, the mean value of CEOCHAIR is 0.3649.
27 Table 3 in provides Pearson correlations matrix between independent and dependent variables. Based on these results, RESTATE is correlated positively with ACE, and negatively with ISSUANCE and SIZE at a significant level of 0.01 and 0.05 respectively. Correlations of -0.5 and below and 0.5 and higher influences the reliability of the mode. Based on this I conclude that all the correlations reported are outside of the previously mentioned range and therefore the variables do not influence the reliability of the model.
TABLE 2
Panel D: Summary Statistics
Obs. Mean Std. Dev. P25 Median P75
RESTATE 2124 0.0730 0.2602 0.0000 0.0000 0.0000 POST 2124 0.0155 0.1237 0.0000 0.0000 0.0000 FEMALE 2124 0.0918 0.2888 0.0000 0.0000 0.0000 ACE 592 0.0186 0.1352 0.0000 0.0000 0.0000 BOARDSIZE 592 9.1199 2.0709 8.0000 9.0000 10.0000 OUTSIDER 592 0.7992 0.1056 0.7500 0.8333 0.8889 CEOCHAIR 592 0.3649 0.4818 0.0000 0.0000 1.0000 CFOONBOARD 592 0.0270 0.1623 0.0000 0.0000 0.0000 LEVERAGE 2103 0.2480 0.3144 0.0330 0.2022 0.3414 MTB 2070 3.1344 39.3680 1.2785 2.0582 3.5890 M&A 1137 1.0000 0.0000 1.0000 1.0000 1.0000 R&D 1380 0.0484 0.0808 0.0048 0.0208 0.0601 LOSS 2104 0.1991 0.3995 0.0000 0.0000 0.0000 SIZE 2107 7.7284 1.6956 6.6150 7.6335 8.7734 FIN 1974 0.6317 0.4825 0.0000 1.0000 1.0000 ISSUANCE 1975 0.9980 0.0450 1.0000 1.0000 1.0000
28 TABLE 3 Pearson Correlations 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 1. RESTATE 1 2. POST -0.0352 1 0.1044 3. ACE 0.0883** 0.0485 1 0.0317 0.2388 4. BOARDSIZE -0.0353 -0.0293 -0.0442 1 0.3908 0.477 0.2824 5. OUTSIDER 0.0143 0.0011 -0.0242 0.0165 1 0.7285 0.9791 0.5565 0.6892 6. CEOCHAIR 0.0081 0.0701*** 0.0256 0.0697 -0.1073* 1 0.8434 0.0881 0.5336 0.0902 0.009 7. CFOONBOARD -0.0523 0.0312 -0.0229 0.086** -0.2506* 0.0468 1 0.2042 0.4493 0.5776 0.0365 0 0.2557 8. LEVERAGE 0.0016 -0.0178 -0.0057 0.1426* 0.0046 -0.047 0.0136 1 0.9411 0.4155 0.8891 0.0005 0.9104 0.2538 0.7407 9. MTB 0.0005 -0.0022 -0.0084 0.0276 0.0392 0.0568 0.0095 0.028 1 0.9828 0.9218 0.8397 0.5047 0.3427 0.1693 0.8191 0.2026
29
TABLE 3
Pearson Correlations (continued)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 10. M&A . . . . 11. R&D -0.0227 0.0241 0.0187 -0.2887* -0.0433 0.0096 -0.0424 -0.0581 0.0807* . 1 0.3997 0.3702 0.7085 0 0.3852 0.8467 0.3951 0.0311 0.0029 . 12. LOSS 0.0234 0.0061 0.0253 -0.1664* -0.0674 -0.0351 -0.0301 0.0637* -0.0199 . 0.3006* 1 0.2837 0.7797 0.5383 0 0.1013 0.3938 0.4641 0.0035 0.3652 . 0 13. SIZE -0.0464 -0.0216 -0.0306 0.5330* 0.1903* 0.0343 -0.02 0.1375* 0.0127 . -0.2975* -0.1782* 1 0.033 0.3223 0.4573 0 0 0.4045 0.6273 0 0.5644 . 0 0 14. FIN -0.0259 0.0035 0.0529 0.088** -0.031 -0.0378 0.062 0.2477* 0.0156 . -0.0219 -0.0283 0.2309 * 1.0000 0.2501 0.8757 0.2136 0.0381 0.4666 0.3745 0.1444 0 0.4927 . 0.4292 0.2093 0 15. ISSUANCE -0.0727* 0.0057 0.006 0.0215 0.0207 0.0323 0.0073 0.0176 0.002 . -0.0052 -0.0055 0.0218 0.0590* 1 0.0012 0.8005 0.8871 0.6124 0.6258 0.4481 0.8634 0.4344 0.9311 . 0.8502 0.8068 0.3318 0.0087 * significant at 0.01 level ** significant at 0.05 level *** significant at 0.10 level
30
4.2 Regression Results
4.2.1 Baseline Regression
Table 4 presents’ baseline regression results on how the transition from male CFO’s to female CFO’s affects financial restatements, in which RESTATE was used as the dependent variable and the sample consisted only of CFO transition firms (i.e. male to female, male to male, female to female, or female to male). RESTATE is coded 1 if a firm’s financial statement was restated and 0 otherwise. POST is coded 1 if the firm year is after the CFO transition, and 0 if the firm year is after the CFO transition). The coefficient on POST is -0.1299 and is statistically not significant, although it is in the expected negative direction. The expectation was that female CFO’s and the likelihood of a financial restatement is negatively related. The marginal effect of POST -0.4469, which is reported in column 3, but as POST turns out to be not significant, therefore this marginal effect is not meaningful for this model. The result is economically not meaningful and therefore is not consistent with the hypothesis that female CFO’s is negatively related with financial restatements as compared to their male counterparts.
TABLE 4 Baseline Regression
1 2 3
Predicted Sign Coefficient t-value POST - -0.1299 -1.160 ACE - 0.0000 BOARDSIZE + -0.0551 -2.080 OUTSIDER - 0.3486 0.960 CEOCHAIR + 0.1091 1.180 CFOONBOARD + 0.0136 0.120 LEVERAGE - 0.2012 1.030 MTB + 0.0139 0.720 MA + 0.0000 RD + 0.5899 0.780 LOSS + -0.0097 -0.060 SIZE + 0.0647 1.600 FIN + 0.0779 1.090 ISSUANCE + 0.0000
Industry and year effects Yes
Observations 124
31 4.2.2 Regression analysis
Following Francis et al. (2012) a differences-in-differences regression was employed. To remove the effect of contemporaneous changes on the previously reported results, a differences-in-differences regression was employed. The changes could be driven by unobservable time series changes contemporaneous with CFO changes. Their method was used however some
adjustments was introduced to make the model more fitting for this thesis. The sample used consisted of firms observations that had experienced a CFO transition in the period between 1997 and 2014. A dummy variable (FEMALE) was constructed that equals 1 if a firm is a male to female CFO transition, and 0 had experienced another type of CFO transition (e.g. male to female, male to male, female to female, or female to male). An interaction term (POST * FEMALE) was created with the dummy variable and POST. If female CFO’s are less likely to get a financial restatement, I expect the coefficient on the interaction variable to be significantly negative.
Table 5 reports the results for the differences-in-differences regression in which RESTATE was used as the dependent variable. RESTATE is coded 1 if a firm experienced a financial restatement and 0 otherwise. The results are not different compared to the results from the baseline regression, meaning that the differences-in-differences regression shows that the interaction term POST*FEMALE is statistically not significant. This suggests that the model used shows no significant evidence for it to support the central claim (H1) in this thesis that female CFO are negatively related to financial restatement.
32
TABLE 5
Differences-in-Differences Regression
1 2 3
Predicted Sign Coefficient t-value
POST - -0.0846 -0.6800 FEMALE - -0.0877 -0.8300 POST*FEMALE ? 0.0000 ACE - 0.0000 BOARDSIZE + -0.0593 -2.1900 OUTSIDER - 0.3338 0.9100 CEOCHAIR + 0.1198 1.2800 CFOONBOARD + 0.0313 0.2600 LEVERAGE - 0.2027 1.0400 MTB + 0.0133 0.6900 M&A + 0.0000 R&D + 0.7804 0.9900 LOSS + -0.0138 -0.0900 SIZE + 0.0676 1.6600 FIN + 0.0971 1.2900 ISSUANCE + 0.0000
Industry and year effects Yes
Observations 124
Pseudo/Adjusted R² -0.1201
4.2.3 Additional Tests Hypothesis 1
Table 6 shows results after running a logit regression while using FEMALECFO as independent variable in a sample of 10,353 firm observations with 863 cases of female CFO. Compared to the sample used for regression in table 6, this sample contains transition- and non-transition firms. This gives the opportunity to investigate a larger sample size. However, a limitation of this approach is that there is no distinction between firms that experiences a CFO gender transition or not. FEMALECFO is an indicator equal to 1 if there is a female CFO, 0 otherwise.
The results shown in table 6 column 3 suggests that FEMALECFO is negatively
correlated with RESTATE. This is consistent with the predicted expectation that FEMALECFO reduces the chances of a firm to experience a restatement. Even though the sample used in this logit regression differs from the previous regression, the direction of the results regarding
FEMALECFO is consistent with the predicted direction, with a coefficient of -0.0278. However, these results does not provide significance evidence that male to female CFO transition and financial restatement is negatively correlated. Therefore, hypothesis 1 is rejected. In summary,
33 this mean that based I did not find evidence for supporting the prediction that male to female CFO transition is negatively related to financial restatements.
TABLE 6
Logit Regression FEMALECFO
(1) (2) (3) Predicted Sign Coefficient z-value
FEMALECFO - -0.0278 -0.06 ACE - 0.3727 0.47 BOARDSIZE + -0.1387 -1.60 OUTSIDER - 0.8595 0.65 CEOCHAIR + -0.0695 -0.24 CFOONBOARD + 0.4732 1.05 LEVERAGE - -0.5175 -0.70 MTB + -0.0167 -0.68 M&A + 0.0000 R&D + -6.2394 -1.65 LOSS + 0.5989 1.37 SIZE + 0.1897 1.50 FIN + 0.0245 0.08 ISSUANCE + 0.0000
Industry and year effects Yes
Observations 786
Pseudo/Adjusted R² 0.0436
4.3 Moderating Effect CFO Discretion
The measure which is used to proxy for CFO discretion is POWER. According to Kuang et al. (2015) and Ng and Feldman (2010) CFO’s accumulate firm-specific and job-related knowledge over the course of their careers. Therefore, the firm-specific and task-related knowledge both accrue hierarchical influences to CFOs. If female CFO’s risk aversion and CFO discretion is the underlying reason that female CFO’s is less likely to get financial restatements, then an even higher correlation for female CFO’s should be detected, when their tenure is high.
To test this, CFOTENURE is constructed using Execucomp information. First the CFOTENURE in years is being calculated based on the reported CFO and executive’s company id number combination, because Execucomp did not yield enough information about the date of the CFO became and left the company. A dummy variable POWER was created based on the CFOTENURE. POWER is coded 1 if CFOTENURE is above the median, which is reported in
34 table 7 panel A, and 0 is CFOTENURE is below the median. Based on the previous selection of CFO transition sample, the minimum CFOTENURE in years is 3. I rerun the empirical model while introducing the new variable POWER. Descriptive statistics for CFOTENURE is reported in table 7 panel A, for POWER in panel B and regression results in panel C.
The results from running the empirical model included with the new variable POWER is
reported in panel C. The sample used was the sample containing only firms that had experienced a CFO gender transition. Based on the previously mentioned dummy variables a three-way interaction term was created, POST*FEMALE*POWER. For completeness the three two-way interactions POST*FEMALE, FEMALE*POWER and POST*POWER were also included in the model.
According to Ge et al. (2011) when CFO discretion is higher CFO style reflects more in accounting choices. Therefore, the higher the CFO tenure, the higher the CFO discretion. The results indicate that POST is in this case positively related to RESTATE, but it is not statistically significant. FEMALE, which is an indicator equal to 1 if the transition is from male to female CFO, 0 otherwise (i.e., either male to male, female to female, or female to male) is not
statistically significant related to RESTATE, indicated by the t-value of 0.15. The newly introduced interaction term in the model POST*FEMALE*POWER is negative but not
statistically significant. However, POWER is statistically significant with a coefficient of 0.3015.
TABLE 7 Descriptive Statistics Panel A: CFOTENURE
Obs. Mean Std. Dev. P25 Median P75
CFOTENURE 2,124 4.0876 1.0680 3 5 5
TABLE 7
Panel B: Summary Statistics POWER
Frequency Percent Cum.
LOW POWER 734 34.56 34.56
HIGH POWER 1,390 65.44 100
35 This results suggests that the higher the power of the CFO, the probability for not receiving a financial restatement increases with 30.15%. In addition, BOARDSIZE and SIZE are also statistically significant with coefficient of -0.0936 and 0.1005 respectively. In summary, this result possibly suggest that CFO discretion, proxied by CFOTENURE, does not strengthen the relationship between CFO gender and financial restatements, because the coefficient of POST changed from being negative to being positive, compared to the results reported in table 5. This might suggests that the proxy used does not capture CFO discretion.
TABLE 7
Panel C: Interaction POWER
(1) (2) (3) Predicted Sign Coefficient t-value
POST - 0.0823 0.43 FEMALE - 0.0222 0.15 POWER ? 0.3015 2.83 POST*FEMALE*POWER ? -0.2008 -0.79 POST*FEMALE ? 0.0000 FEMALE*POWER ? -0.0578 -0.26 POST*POWER ? 0.0000 ACE - 0.0000 BOARDSIZE + -0.0936 -3.39 OUTSIDER - 0.4257 1.15 CEOCHAIR + 0.1173 1.29 CFOONBOARD + -0.0172 -0.14 LEVERAGE - 0.1453 0.79 MTB + 0.0087 0.48 MA + 0.0000 RD + 1.5975 1.84 LOSS + -0.0086 -0.06 SIZE + 0.1005 2.48 FIN + 0.0755 1.06 ISSUANCE + 0.0000
Industry and year effects Yes
Observations 124
36
4.4 Male to Female and Financial Restatements
A logit regression has been conducted to test the likelihood that financial restatements increases a male-to-female CFO transition. In this case RESTATE was replaced by MALETOFEMALE as the dependent variable. RESTATE in this case is considered to be the independent variable, followed by the previously used control variables. MALETOFEMALE is an indicator equal to 1 if there is a female to male CFO transition in year t, and 0 if there is no transition or male to male transition or male to female transition in year t. RESTATE and control variables are all one year lagged therefore they represent year t-1.
The model tested is as follows:
MALETOFEMALEi,t = β0 + β1 RESTATEi,(t-1) + β2 ACEi,(t-1) + β3 BOARDSIZEi, (t-1) +
β4 OUTSIDERi,(t-1) + β5 CEOCHAIRi,(t-1) + β6 CFOONBOARDi,(t-1) +
β7 LEVERAGEi,(t-1) + β8 MTBi,(t-1) + β9 M&Ai,(t-1) + β10 R&Di,(t-1) + (2)
β11 LOSSi,(t-1) + β12 SIZEi,(t-1) + β13 FINi,(t-1) + β14 ISSUANCEi,(t-1) +
β15 YEAR i,(t-1) + β16 INDUSTRYi,(t-1) + ɛi
The model above has been tested on sample containing CFO’s whose tenure was bigger than 2. The reason why this model was chosen was because the previous model containing contained CFO’s whose tenure was 3 years or more. While running equation 2 with the sample used for testing the effect of CFO discretion, STATA omitted numerous variables (e.g. ACE,
CEOCHAIR, CFOONBOARD, FIN, ISSUANCE, M&A and R&D) based on the fact that STATA could not perform a reasonable test. For this reason I have chosen to incorporate in this sample CFO’s whose tenure was 2 years or more. This has resulted in a sample containing 88 cases of male to female CFO transitions.
Table 8 reports the results for testing hypothesis 3. The coefficient RESTATE is positive, coefficient value of 0.9708, however this is not statistically significant. Therefore, this sample and the chosen model does not manage to provide statistically evidence for the prediction that financial restatements increases the likelihood of a male-to-female CFO transition.
37
TABLE 8
Panel B: Logit Regression MALETOFEMALE (1) (2) Coefficient z-value RESTATE 0.97080 1.20 ACE 0.00000 BOARDSIZE 0.08954 0.46 OUTSIDER 0.10429 0.04 CEOCHAIR -0.23874 -0.36 CFOONBOARD 0.28453 0.25 LEVERAGE 1.79279 1.51 MTB -0.00150 -0.11 M&A -0.30130 -0.45 R&D 1.16124 0.19 LOSS -0.16922 -0.15 SIZE -0.14933 -0.53 FIN 0.74602 0.91 ISSUANCE 0.00000
Industry and year effects Yes
Observations 1291
38
5 Discussion and Limitations
In this section the results found in the previous chapter will be discussed. Furthermore limitations are discussed as well as suggestions for future research.
Financial restatements and gender differences has become increasingly prevalent phenomenon in corporate America and has attracted considerable attention from regulators, the financial press and investors (Catalyst, 2004; Francis et al. 2014). Extant studies identify a wide range of determinants of financial restatements such as various firm attributes, compensation incentives, and managerial characteristics (e.g. Aier et al. 2005; Collins et al., 2009; Abbott et al., 2012; Francis et al., 2014). In this thesis, I have linked the risk aversion of female CFO’s to firms financial restatements, because prior literature has proven that females are more risk averse than their male counterpart in different aspects of accounting choices (e.g. Barua et al. 2010; Peni and Vähämaa, 2010; Habib and Houssain, 2013; Huang and Kisgen, 2013; Francis et al. 2014).
Consistent with the methodology of Francis et al. 2014, but with some modifications, I constructed a sample with male to female CFO transitions and then examined whether there is a significant negative relationship between financial restatements and male to female CFO
transition. The first hypothesis which suggested a negative relationship between CFO gender and financial restatement was rejected, because the results of the chosen model has not provide significant evidence to support this prediction. An additional test that compared female CFO with male CFO’s in a sample containing both transition and non-transition firms and while comparing female CFO’s with male CFO’s has also not been able to provide statistical evidence supporting a negative relationship between female CFO’s and financial restatements.
The relationship between CFO discretion and financial restatements has been tested in a setting containing only transition firms. While testing the second hypothesis, which suggested that CFO discretions strengthens the relationship between CFO gender and financial restatement, based on Ge et al. (2011). CFO discretion was measured based on CFO tenure, which is a
different approach from Ge et al. (2011). The results has not provided the same statistical results as was suggested by Ng and Feldmann, 2010; Kuang et al., 2015) which indicated that higher CFO tenure can provide accumulated firm-sepcific and job-related knowledge and is less likely to get a financial restatement (Aier et al., 2005; Kuang et al, 2015). The reason for this might by
39 because of the small sample size and the way the sample has been constructed, which has been discussed in chapter 3.
The third hypothesis which assumes that financial restatements increases the likelihood of a male-to-female CFO transition is also rejected. Prior literature has found that, from a legitimacy perspective, a firm distance itself from the management team associated with the restatement. By distancing themselves, they send the message that the firm has taken steps to regain its legitimacy. This thesis has not found significant evidence in line with this theory. For this reason, and for the reason that this thesis is to the best of my knowledge the first of his kind, I encourage future research to not only approach this subject from a gender perspective but also incorporate personal characteristics which may help to further understand gender differences at top management level.
To the best of my knowledge this is the first paper that researches the possible effect of CFO gender and financial restatement through CFO transitions. Furthermore, this thesis answers Habib and Houssain (2013) calls for more research on managerial characteristics influences on reporting quality. However, to compliment this line of research it is important to acknowledge the limitations of this thesis. Limitations regarding the data and model used should be
considered. The set of control variables was not extensive, but ideally more financial control variables should have been included in the model. Hence, the size and scope of the dataset is not as far reaching as one might expect. Therefore, to further improve this model hand collected data regarding CFO’s gender might improve the model used in order to identify more transitions. In addition, ideally the use of one sample would provide a solid base for making comparisons.
Gender differences is interesting both from a financially and psychological perspective. I share the same view as Francis et al. (2014) that not only researches, but also government and firms would benefit from research based on gender differences with the increasing amount of females in top management team to fully understand potential benefits and costs of having female CFO’s.
