Errors and irregularities restatements : a study on executive and director turnover

(1)

FACULTY OF ECONOMICS AND BUSINESS

Errors and Irregularities Restatements:

A study on Executive and Director Turnover

Master thesis

Name: Vivian Xie

Student number: 10190171

Thesis supervisor: M. Schabus MSc Date: August 15th, 2016

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

(2)

Statement of Originality

This document is written by student Vivian Xie 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)

Abstract

Over the past several years the number of restatements have increased resulting in a significant increase in research on restatements. However, the majority of research conducted on restatements has used databases that contained both error and irregularity restatements. This resulted in several studies that showed no significant evidence in the relationship between restatements and turnover. Hennes, Leone & Miller (2008) state that these lower than expected results could be due to the data samples containing both error and irregularity restatements. This study has followed Hennes et al.’s (2008) footsteps and looked into whether error restatements have a significantly different impact on turnover compared to irregularity restatements. Besides executive turnover this study has also looked at director turnover. The analysis shows a significantly higher turnover rate for directors in the irregularity sample compared to the error sample. This study also demonstrates significantly higher turnover rates for executives compared to directors for both the error and irregularity sample confirming my expectations based on the legitimacy theory.

Keywords: Restatements, irregularities, errors, executive turnover, directors, audit committee members, influences and consequences.

Data availability: The data used in this study is available from public sources identified in the text. Initial sample was extracted from the U.S. database

(4)

1 INTRODUCTION

Since the financial crisis the number of resignations of executives seems to have increased significantly in general compared to the 1990s, resulting in higher executive turnover rates (Favaro, Karlsson & Neilson, 2011). A reason for executive resignation could be a company’s pressure to perform. In other words, a company might feel the need to hire a new executive in the hope to improve performance and keep up with the competition (Karaevli, 2007). However, a more common reason for executive resignation is company failure due to for example poor performance, an accounting scandal or fraud (Hennes, Leone & Miller, 2008; Arthaud-Day, Certo, Dalton & Dalton, 2006; Gilson & Vetsuypens, 1993; Warner, Watts & Wruck, 1988). Farber’s (2005) study about restoring trust after fraud emphasizes that one of the priorities of a company involved in a fraud case is limiting a company’s reputation damage and correspondingly restoring financial reporting credibility, which according to Hennes, et al. (2008) can be acquired through the termination of the members in charge, namely executives. However, Arthaud-Day et al.’s (2006) study shows that directors and audit committee members’ turnover can also be affect by company failure like for example financial restatements. Infamous companies that faced accounting scandals are for example Enron, Tyco, WorldCom and Imtech (Agrawal & Chadha, 2005).

Arthaud-Day et al.’s (2006) study has examined the relationship between financial restatements and executive/director turnover and shows that there is a significant relationship between the two. Their study shows that higher restatements lead to significantly higher executive and director turnover. The chance of Chief Executive Officers (CEOs) and Chief Financial Officers (CFOs) exiting their companies was twice as high for companies that filed material financial restatements compared to their counterparts in a matched sample while the directors and audit committee members’ chances were slightly lower and were 70 percent more likely to lose their jobs.

Even though existing literature (e.g. Kacmar, Andrews, Van Rooy, Steilberg & Cerrone, 2006; Sagie, Birati, & Tziner, 2002; Agrawal, Jaffe & Karpoff, 1999) shows that replacing top management is time consuming and is associated with high costs, studies (e.g. Hennes et al., 2008; Arthaud-Day et al., 2006; Gilson & Vetsuypens, 1993; Warner et al., 1988) still show that there is a significant effect on

(6)

managerial turnover after a restatement. This can be explained by the following studies conducted on restatements.

Dechow, Ge & Schrand’s (2010) paper on earnings quality examines the consequences of restatements and states that besides directors experiencing a significantly higher turnover, the chances of losing firm value by a decline in stock price after a restatement announcement is likely. Additionally, their paper shows there is a significant increase in a company’s cost of capital after a restatement, consistent with Dechow, Sloan & Sweeney’s (1996) findings. Also, Palmrose and Scholz (2004) find that “the likelihood of litigation increases with the impact of restatements on earnings (magnitude) and the fraudulent nature of restatements”, which implies that restatement can result in higher litigation costs (p. 375). Coming back to why companies might choose to replace top management even if the costs of replacement are high. According to Agrawal et al. (1999) a company will increase managerial turnover if the benefits of replacing management outweigh the costs. So in other words, if a firm’s value, cost of capital, litigation risk and/or reputation can be restored substantially and these benefits offset the costs of replacing top management, turnover rates will likely increase.

Additionally, Arthaud-Day et al. (2006) show that an increase in executives, directors or audit committee members’ turnover despite the high costs can also be explained by looking at a non-economic theory, the legitimacy theory. According to Arthaud-Day et al. (2006) restatements can lead to potential loss of legitimacy (i.e. reputation of being reliable/trustworthy) causing a company to lose (their access to) certain resources that can give them a competitive advantage. So they explain that a company will consequently opt for replacing its members in charge and responsible for the restatements in order to restore potential loss of legitimacy. Existing literature states that leaders of companies (i.e. CEO/CFO) are considered to be the face of their companies. Hence, a leader’s reputation affects his or her company’s reputation and vice versa (Sutton & Callahan, 1987; Hambrick & Mason, 1984). This can explain the higher turnover for executives compared to directors or audit committee members after an organizational crisis such as a restatement since executives are hold directly responsible for the actions and operations of the company as compared to the other parties that only have a monitoring role, consistent with the findings in Arthaud-Day et al.’s (2006) paper.

(7)

and restatement are for example Collins, Reitenga & Sanchez-Cuevas (2005), Agrawal et al. (1999) and Beneish (1999). However, these studies find no significant difference in turnover, which is not in accordance with Arthaud-Day et al. (2006) and Desai, Hogan & Wilkins’ (2006) findings. Hennes et al. (2008) state that these lower than expected results are “likely due to the mixing together of restatements due to both errors and irregularities” (p. 1490). They emphasize the importance of this distinction in their study by stating that restatements should be categorized into errors and irregularities because the nature of a restatement can have an impact on the decision whether to appoint a new executive or not.

As mentioned earlier, besides executives restatements could also have a significant effect on director turnover as well since they are the ones in charge of monitoring the daily operations, respectively the oversight of the financial reporting process (e.g., the audit, internal control systems and compliance with legislation) (Arthaud-Day et al, 2006). However, based on the legitimacy theory restatements will have a higher impact on executive turnover than director turnover because executives play a bigger part in obtaining legitimacy since they are the face of a company (Finkelstein & D’aveni, 1994). Nonetheless, Hennes et al. (2008) only look at the impact of financial restatements on CEO and CFO turnover just like many other studies (e.g. Collins, Masli, Reitenga & Sanchez, 2009; Desai et al., 2006) conducted on the consequences of and the relationship between financial restatements and executive turnover and does not look at the impact of restatement on other parties that are also involved in and accountable for the accounting process like directors and audit committee members.

These studies mentioned above show that financial restatements can have a significant impact on a company’s management as well as a company’s value and the ability to attract capital. However, existing literature has not yet looked into the turnover of executives compared to director and/or audit committee members taking into consideration the importance of the distinction between error and irregularity restatements with publically available data like Compustat. Therefore this thesis predicts that by making a distinction between error and irregularity restatements a more significant relation between turnover and restatements will be shown. Also by conducting this study I test whether using publicly available data will result in the same outcome as Hennes et al.’s (2008) study using hand-collected data.

(8)

study and examine what the impact is of errors and irregularities restatements on executive, directors and audit committee members’ turnover, and whether there is a significant difference in impact on turnover between these groups. The research questions are as follows: ‘Do irregularity restatements have a higher impact on executive and director turnover than error restatements?’ and ‘Do error and irregularity restatements have a higher impact on executive turnover than director turnover?’.

The remainder of the thesis is organized as follows: section 2 discusses prior literature related to error and irregularity restatements, the role of executives, directors and audit committee board members and the legitimacy theory. Following this, the hypotheses of this thesis are developed. In section 3 the sample selection is explained and the empirical models and the variables are introduced. Section 4 provides main results of the analysis. Finally, section 5 concludes.

2 LITERATURE REVIEW, THEORY AND HYPOTHESIS

2.1 Error and Irregularity Restatements

Financial restatements are a form of company failure, which are the result of the violation of accounting standard (e.g. GAAP). They are adjustments of financial figures presented in the original financial reports that have passed through the internal control mechanisms of the company, the external auditor and have been processed by the Securities and Exchange Commission (SEC). Once this “inaccuracy” has been detected, all parties are required to report and disclose this. The restatement can either be announced by the company itself or by an external party (e.g. external auditor, SEC). However, research shows that restatements that have been prompted externally are regarded as a sign of either severe lack of oversight or the conscious intent to deceive (Arthaud-Day et al., 2006).

Financial restatements can be the result of creative accounting1 but can also have a more innocent cause like unintentional misinterpretation of accounting standards. The majority of the studies conducted on restatements are mostly interested on the first type of restatement, which are restatements caused by intentional

1

Also known as aggressive accounting or earnings management (Collins et al., 2009; Desai et al., 2006).

(9)

misconduct (Hennes et al., 2008; Arthaud-Day et al., 2006; Gilson & Vetsuypens, 1993; Warner et al., 1988).

Literature shows there can be several reasons for executives to participate in aggressive accounting or even earnings manipulation (i.e. fraud) with one of the reasons being the incentive to maintain positive earnings and the need for external financing. By presenting a company’s earnings more positively they can attract external financing at lower cost due to the lower perceived risk (e.g. chance of bankruptcy). The incentive of executives to maintain positive earnings can also be due to their compensation plans if those are equity based. By portraying the earnings more positively the market value of the company could go up, especially if the company’s performance exceeds stakeholders’ expectations, resulting in an increase of the executives’ stock or options. However, studies show that when the restatement is made public the market value will likely decrease, the cost of capital will increase, their litigation risk will increase and management will get penalized with the worst-case scenario being fired and low chances to find subsequent employment due to the reputational damage caused by the restatement (Collins et al., 2009; Desai, Hogan, Wilkins, 2006). According to Hennes et al. (2008) firing the people that are held responsible for the restatement can help a company limit its reputational damage and restore (financial) trust and credibility, which is one of the top priorities of a company involved in a fraud case (Farber, 2005).

Nevertheless, several studies conducted on the relationship between financial restatements and management turnover find no significant difference in turnover between the restatement firms and their matched sample (Collins, Reitenga & Sanchez-Cuevas, 2005; Agrawal et al., 1999; Beneish, 1999). Possible explanations for why these conducted studies do not show the same results as some of their peer papers2 on turnover could be due to the size of the samples as well as the time windows3 of these studies. Another reason might be that the data they used for their sample contained unintentional as well as intentional misstatements (e.g. data from the GAO database) while the study indicates that it intents to look at restatements

2

Examples of studies conducted on turnover showing a significant increase in turnover after a restatement are Collins et al. (2009), Hennes et al. (2008), Arthaud-Day et al. (2006), Desai et al. (2006) and Srinivasan (2005).

3

The time window of a study might influence the results due to being too short to adequately analyzing turnover rates or by not looking at management turnover prior to restatement announcements, which can also be linked to the restatement.

(10)

caused by severe shortcomings in governance mechanisms, mainly manipulation, like Arthaud-Day et al.’s (2006) study and therefore resulting in less significant outcomes than expected.

Hennes et al. (2008) emphasize the importance of distinguishing restatements by nature in their study by stating that restatements should be categorized into errors and irregularities because the nature of a restatement can have an impact on the decision whether to appoint a new executive or not. Furthermore, their study documents that the amount of restatements due to unintentional errors or applying new accounting standards retrospectively has increased in recent years causing restatements to be a noisy proxy for intentional misstatements. Error restatements, which are unintentional misstatements (e.g. bookkeeping errors or external environmental factors), will have a less significant impact on executive turnover than irregularity restatements, which is intentional misreporting, because error restatements due to for example misinterpretation of an accounting standard are likely to be more easily forgiven by stakeholders as opposed to irregularity misstatements like fraud. Further findings that Hennes et al. (2008) emphasize to illustrate the importance of distinguishing restatements by nature (i.e. intentional or unintentional) is that the market reacts differently to error restatements than to irregularity restatements. Their study shows that in their sample the market reaction after irregularity restatements was -14% compared to -2% after error restatements, which is significantly less negative, in line with Palmrose, Richardson & Scholz’s (2004) findings. Lastly, the sample in their study shows that there are significantly more fraud-related class action lawsuits after irregularity restatements than error restatements, which had only one lawsuit. These findings indicate that irregularity restatements (i.e. restatements involving fraud) are more likely to result in higher cost associated with these consequences and a loss of firm value than error restatements.

This thesis will therefore make a distinction between error and irregularity restatements while looking at the impact of restatements on turnover since the impact of error restatements will most likely differ from irregularity restatements as shown in Hennes et al.’s (2008) study. Following the existing literature and studies mentioned above this study predicts that irregularity restatements will have a higher impact on turnover than error restatements. This is in line with the results from the study conducted by Hennes et al. (2008). Therefore the first hypothesis is formulated as follows:

(11)

Hypothesis 1: Irregularity restatements have a higher impact on turnover than error restatements.

2.2 Executives, Directors and Audit Committee Members

Companies have grown larger and more complex since the late nineteenth and late twentieth century (Williamson, 1996). As a result we see more and more companies growing to the extent that they become too big to be controlled by the founder and only shareholder alone. Therefore often ownership and management are no longer in the hands of one person or a very select group of people anymore. Shareholders appoint agents (i.e. executives) to manage their business for them (Jensen & Murphy, 1990; Baysinger & Butler, 1985; Jensen & Meckling, 1979). In the United States this separation of ownership and management results in the following parties; 1) the executives who are charged with management, 2) the Board of Directors who are charged with governance and have a monitoring task, and 3) the shareholders who have corporate ownership. The next subsections of the thesis will go into further detail on the tasks and responsibilities of executives and the Board of Directors and how these parties’ turnover can be affected by restatements.

2.2.1 Executives and the Board of Directors

Executives are those that are charged with the management of the company. Depending on the size, complexity and branch the company is operating in there can be multiple executive positions with the most common and well known being the CEO, also known as President or Chairman, and the CFO4. The tasks and responsibilities of executives can vary depending on the type of organization (e.g. not-for-profit or not-for-profit). However, the overall duty of an executive is carrying out the strategic plans and goals that have been set by those who are charged with governance, which is the Board of Directors. While the CFO is in charge of managing the financial operations of the company to reach the goals that have been set by the board, the CEO is responsible for managing the overall operations of the company.

4

Other executive positions are for example Chief Operational Officer (COO), Chief Information Officer (CIO) and Chief Sales Officer (CSO).

(12)

The CEO oversees all the operations of a company and makes sure that all tasks are being carried out by lower management in accordance with the directions provided by the Board of Directors. He or she is generally also considered to be the spokesman or woman of the company and stands above the CFO. Regulatory bodies, however, see the CEO and CFO as equally accountable for the financial reporting process and both are required to certify a company’s periodic financial reports with the SEC by law (Sutton & Callahan, 1987; Hambrick & Mason, 1984).

According to the agency theory, hiring executives to manage the company can result in a principle-agent problem since executives have the incentive to act in their own best interest, which is not necessarily the best interest of the shareholders (Jensen & Murphy, 1990; Jensen & Meckling, 1979). In order to mitigate this risk and align the interests of these executives with the shareholders’ interests the executives are often given a compensation that involves a fixed component with a bonus based on a company’s performance and company stocks and options. However, study shows that these compensation plans do not always necessarily lead to the outcomes that are in the shareholders’ best interest. Executives may want to overstate earnings in order to improve performance and maximize their own personal gain (e.g. higher bonus or stock value) with restatements as a consequence (Xie, Davidson & DaDalt, 2003).

Besides compensation plans to align executives’ interests with the shareholders’ a supervisory board can also help to mitigate the risk of the executives acting in their own best interest at the expense of the shareholders. This supervisory board is also known as the Board of Directors. The Board of Directors is not only responsible for the monitoring and evaluating the performance of the executives to mitigate the risk of the conflicting interests between executives and shareholders but they also have the power to hire and fire top management and determine their remuneration (Baysinger & Butler, 1985). Within the Board of Directors there can be multiple subcommittees like for example the audit committee, remuneration committee and nomination committee, with each committee carrying out the tasks they have been assigned to (Xie et al., 2003).

Executives can also be elected to be on the Board of Directors either with or without voting rights, which may feel counterintuitive since this implies that they would be monitoring and reviewing their own performances. Opinions on whether this is in the best interest of a company and its shareholders are divided. Some state that companies with executives on the Board of Directors can result in less

(13)

independence and weaken the governance of the company. They state that there needs to be a clear separation between managerial and monitoring role in order to function optimal. By having the CEO on the board other board members might not feel comfortable expressing certain concerns about the company’s performance or operation as compared to when the CEO would not be on the board and can be regarded as a (executive) staff member whose performances need to be monitored and evaluated. By having the CEO as a board member can create a situation where some board members might be reluctant to express their opinion that are conflicting to the CEO’s view, especially if the CEO is also on the nomination committee, resulting in giving the CEO too much power (Xie, Wallace & Davidson, 2002; Dechow et al., 1996). Dechow et al.’s (1996) study shows that companies with executives dominating the Board of Directors, a CEO that serves as the Chairman of the board and a CEO who is also the founder of the company are more likely to engage in manipulating earnings, which are conducted in their own interest (e.g. higher bonus). Therefore some even opt for a majority of outside directors on the Board of Directors since they are (more) independent and less prone to the CEO’s influence and therefore will be more likely respond to poor performance by dismissing the CEO (Hermalin & Wiesbach, 1991).

However, other studies show that there are positive effects of having inside directors like executives on the board. They state that having inside directors on the board can result in more informed decisions and due to their senior-level management experience they can take over the tasks of the CEO if he or she falls short (Klein, 1998; Baysinger & Butler, 1985). Klein’s (1998) study shows that overall board composition is unrelated to a company’s performance. However, she does find that inside directors can add value to boards like the finance and investment committees due to their inside information on the company resulting in higher returns and consequently better performance. Therefore she argues that the trend of replacing inside directors on the board by outside directors may not be beneficial to the company and it could be in the company’s best interest to reevaluate their board composition.

Another important factor in constraining top management from committing fraud is the financial expertise of the Board of Directors, more specifically the audit committee (Carcello, Hollingsworth & Neal, 2006; DeFond, Hann & Hu, 2005; Xie et al., 2003). This will be discussed in the following subchapter.

(14)

2.2.2 The Audit Committee

The audit committee is a subcommittee of the Board of Directors, which is responsible for the “oversight of corporate governance, financial reporting, internal control structure, and audit functions” (Arthaud-Day et al., 2006, p. 1123). So they are required to monitor the accounting and financial reporting process as well as the audit of the financial statements. While not all companies are required to have an audit committee since the SEC does not mandate it, it is a requirement if the company is listed on the NYSE or NASDAQ stock market. Studies show that having an audit committee reduces the likeliness of companies engaging in earnings management and consequently having financial restatements. However, underlying factors do play a roll in the effectiveness of the audit committee in preventing management from engaging in aggressive accounting. Components that contribute to the effectiveness of the audit committee can be the activity level of the committee (e.g. how often do they have meetings?), the incentive to protect their reputation, their independence from management and the financial expertise of the members on the committee (Krishnan & Visvanathan, 2009; Arthaud-Day et al., 2006; Xie et al., 2003; Daily, 1996; Kesner, 1988). The latter shall be discussed in more detail in the following subsection.

In contrary to the directors on the board, who are usually appointed for a fixed period of time (typically three years) to serve on the board, directors on subcommittees like the audit committee can be dismissed from the board without being removed from the board. Reorganisation of the board or its subcommittees is usually done by the board itself (i.e. nomination committee). However, other stakeholders, of whom the shareholders can exercise the most pressure on the board members, can also indirectly affect the board composition (Jungmann, 2006). Nevertheless, Jungmann (2006) states that pressure to remove certain directors on the board will only be exercised if there is serious misconduct (e.g. intentional misreporting) or extreme underperformance (due to failing to effectively carry out their monitoring task).

Srinivasan (2005) looks at the consequences of financial reporting failure and finds that after a restatement, outside directors on the audit committee do not necessarily get penalized by lawsuits and there are only limited SEC actions but do receive significant labour market penalties that increase with the severity of the

(15)

restatement. These labour market penalties can be reputational but also financial, which are due to the loss of board positions following the restatement. Srinivasan (2005) states that by having board positions can result in multiple benefits like prestige, a great network and learning opportunities, which are even considered more important than just the financial benefits like their paycheck. Therefore directors on the audit committee will likely have the incentive to efficiently monitor management since poor monitoring and consequently poor performance or worse, restatements will likely lead to the loss of board positions and its corresponding benefits.

Unlike Srinivasan (2005) who argues that restatements are a sign of lack of supervision of management some think that “boards generally fail in their responsibility to monitor management and guide their companies, and have called for regulations requiring that boards be composed of a majority of outsiders” (Hermalin & Weisback, 1991, p. 101). However, others argue that the benefits of effective monitoring will provide the members with sufficient incentive to fulfill their monitoring task (Fama & Jensen, 1983; Fama, 1980).

2.2.2.1 Financial Experts

As mentioned in the prior subchapter the Board of Directors can have multiple subcommittees like the audit committee. Among these directors and/or committee members there usually is one or few directors that have financial expertise. Several studies conducted on board composition and earnings management show that the financial expertise of the audit committee members impacts the likeliness of companies engaging in aggressive accounting (e.g. earnings management) (Krishnan & Visvanathan, 2009; Carcello et al., 2006; Bédard, Chtourou & Courteau, 2004; DeFond et al., 2005). The Sarbanes-Oxley Act (SOX) of 2002 mandates the disclosure of whether the audit committee has a financial expert5 and if not the reason why there isn’t a financial expert should be disclosed. The SEC argues that having financial experts on the audit committee will likely improve the quality of information provided to investors, which is supported by studies showing significant lower levels of earnings management, fewer restatements and lower cost of debt (Krishnan & Visvanathan, 2009; Carcello et al., 2006; Abbott, Parker & Peters, 2004; Bédard et al.,

5

According to the Sarbanes-Oxley Act (SOX) of 2002 there should be at least one member with financial expertise on the audit committee.

(16)

2004; DeFond et al., 2003). However, opinions are divided about what kind of expertise an individual should poses in order to be considered a financial expert.

Initially the SEC defined a financial expert as someone who has prior accounting-related experience with SEC financial reporting like auditors, controllers and Certified Public Accountants (CPA). Nevertheless, this definition has been broadened to also include nonaccounting financial experts (e.g. CEO) after there was some controversy about whether the initial definition of a financial expert might be too narrow (Krishnan & Visvanathan, 2009; DeFond et al., 2005). Krishnan & Visvanathan’s (2009) study looks at whether auditors value audit committees with accounting financial expertise more compared to audit committees with nonaccounting expertise and they find that audit pricing and earnings management is negatively related to accounting financial expertise but not significantly related to nonaccounting financial expertise on the audit committee. This shows that auditors only value accounting financial expertise in their assessment of the control risk and overall audit risk, which affect the audit fee6. This is in accordance with best practices that suggest that accounting-related financial expertise is important since audit committee members’ tasks require a relatively high degree of accounting expertise. So their study suggests that only accounting financial experts will add value to audit committees by contributing to more effective monitoring due to their experience with accounting-related financial information resulting in lower risk of governance failure which is in line with DeFond et al.’s findings (2005).

The benefits of accounting-related financial expertise on the audit committee have been proven by several studies (e.g. Krishnan & Visvanathan, 2009; Carcello et al., 2006; Bédard et al., 2004; DeFond et al., 2003). Nevertheless, Krishnan & Lee (2009) find a substantial amount of companies that do not have accounting financial expertise on their audit committees. They also find that not every accounting expert is appointed as audit committee financial expert and not every financial expert has accounting expertise. According to them another reason for companies not having a financial expert on the audit committee might be due to the unwillingness of potential candidates to serve as an accounting financial expert on the audit committee because of potential litigation risk.

6

Higher assessment of the control risk and consequently the overall audit risk result in higher audit fees due to more substantive testing to mitigate the higher perceived risk.

(17)

Since studies show that accounting financial experts on the audit committee can contribute to better monitoring of management and consequently result in less restatements I assume it is not unlikely that restatements can impact the turnover of financial experts on the audit committee since restatements are an indication that the financial experts have been failing to effectively monitor the company’s management regarding the accounting process. Unlike the other members on the audit committee that do not have accounting-related financial expertise accounting financial experts cannot hide behind the fact that they do not have enough accounting expertise to detect and prevent the accounting inaccuracy. However, many public available databases do not always disclose whether directors are assigned as financial expert on the board, if there are any, making it difficult to obtain a substantial sample for possible research.

2.3 Legitimacy Theory

Executive directors and non-executive directors like financial experts and audit committee members bear responsibility for the financial reporting process. Based on the legitimacy theory restatements will have an impact on executive turnover and director and audit committee members’ turnover since they play a part in obtaining legitimacy7 for a company (Finkelstein & D’aveni, 1994).

The legitimacy theory suggest that in order for a company to properly operate one must obtain legitimacy, which is defined by Suchman (1995) as follows:

“Legitimacy is a generalized perception or assumption that the actions of an entity are desirable, proper, or appropriate within some socially constructed system of norms, values, beliefs, and definitions” (p. 574).

7

There are two types of legitimacy; regulatory & normative legitimacy. Irregularity restatements threaten a company’s regulatory legitimacy because it indicates violations of the legal standards imposed by the SEC. Irregularity restatements also decreases normative legitimacy because they reveal a company’s violation of societal expectations of proper conduct (Arthaud-Day et al, 2006). This thesis will not make a distinction between the two types of legitimacy, as this will not contribute to the quality of conduct of this study.

(18)

So in other words, a company can obtain and maintain legitimacy by giving its stakeholders (i.e., employees, clients, shareholders, suppliers, etc.) the impression (i.e., signalling) that the company is operating in a socially accepted and responsible way. This implies that in order for a company to gain legitimacy, their performance doesn’t necessarily has to be outstanding or even good since being legitimate is about operating in a socially acceptable and responsible way. Therefore, legitimacy isn’t performance related but rather reputation/image related. So we could state that a company’s legitimacy is dependent of its image (Arthaud-Day et al., 2006). Suchman’s (1995) paper shows that there can be many reasons for companies to seek legitimacy. However the most important one; companies that are considered legitimate are also considered (more) thrust worthy. Consequently this trust can be used to maintain and/or improve access to multiple resources (e.g., intellectual, material, social or financial capital), which the company can benefit (i.e., competitive advantage) from. Nevertheless, losing or the threat of losing legitimacy, due to for example a material restatement, will likely result in losing access to these multiple resources. Even to the extent of stakeholders dissociating themselves from the company that has lost or is threatening to lose their legitimacy (Arthaud-Day et al, 2006; Suchman, 1995). Since restatements can endanger a company’s legitimacy it is likely that a company will opt for replacing its (top) management in order to restore potential loss of legitimacy as shown in Hennes, et al. (2008).

In order to remain legitimate a company should meet the expectations of stakeholders and society in general (e.g., costumers, environmentalists, suppliers, etc.). So legitimacy can be obtained and maintained by a company by choosing to do “the right thing”. However, if a company is endangering its legitimacy due to a form of company failure a company can take decisive action to limit the loss of legitimacy and secure their access to key resources by using one or a combination of the following two methods.

The first strategy a company can use is disassociation, which is the restructuring of a company by symbolically distancing the company from bad influences. By distancing itself from the executives that have been stigmatised by association with the restatement a company can show its willingness to accede to external demands and mitigate the stigma attached to the company, which may be enough to relieve pressure on a company. The second strategy is creating or revamping monitors and watchdogs, which means changing up the oversight

(19)

functions to show a commitment to prevent the company failure that damaged or threatened the company’s legitimacy from reoccurring (Arthaud-Day et al., 2006; Suchman, 1995).

According to the agency theory directors (i.e. non-executive directors) on the Board of Directors, are responsible for effective oversight of operating managers (i.e. executive directors). The agency theory suggests that when ownership and management are separated, the agents (i.e. executive directors) will have the incentive to act in their own interest at the expense of the principles (i.e. shareholders) (Jensen & Meckling, 1976). The non-executive directors on the board with their monitoring role should mitigate this tendency but according to Arthur Levitt, former SEC chairman, they have been failing to do so. Financial restatements can therefore be regarded as an indication that the non-executive directors (e.g. audit committee members) have been failing to effectively monitor the company’s management regarding the accounting process (Arthaud-Day et al., 2006; Suchman, 1995; Jensen & Meckling, 1976).

According to Suchman, (1995) neither one of these strategies can directly repair legitimacy damage but they can signal to stakeholders that they can safely continue their relationship with the troubled company in question. So by applying one or a combination of both methods the company facing restatements and risking the loss of legitimacy can guard itself against any further damage and restore trust of stakeholders, which is one of the priorities of a company involved in a fraud case according to Farber (2005).

Since executives are the face of a company and are naturally held responsible for the decisions and actions of a company, they are the ones who are most involved in the legitimacy process (Finkelstein & D’aveni, 1994). Executives serve as symbols for a company’s successes and failures, which implies that these executives will likely serve as scapegoats when there is company failure like for example a financial restatement. Executive succession can be used to signal to stakeholders the company’s willingness and intention to improve and change (Day et al., 2006). Arthaud-Day et al. (2006) state that since executives are held directly responsible for restatements, removing these executives will help the company’s recovery process. While non-executive directors like the members of the audit committee also play a role since they are responsible for the monitoring of management this thesis predicts that the executives will likely face higher turnover rates since they are the internal

(20)

strategic leaders and primarily responsible for a companies financial health. Therefore the second hypothesis of this study, following the results of Arthaud-Day el al. ‘s (2006) study, is as follows:

Hypothesis 2: Error and irregularity restatements have a higher impact on executive turnover than director turnover.

3 METHODOLOGY

3.1 Sample

The data used for the test is retrieved from publically available databases, which are as follows: 1) Compustat Fundamentals Annual database (Compustat), 2) Compustat Executive Compensation - Annual Compensation database (Execucomp), 3) Audit Analytics - Non-Reliance Restatements database (AUDIT), and 4) Audit Analytics - Director & Officer Changes database (D&O). The restatement data is retrieved from Audit Analytics, which covers all SEC registrants who have disclosed a financial statement restatement in electronic filings since 1 January 2001. The data has been extracted principally from the following form types: 8-K, 8-K/A, K, 10-Q, 10-Q/A, 10-K/A, 10KSB, 10KSB/A, 20-F, 20-F/A, 40-F and 40-F/A’s.

Due to the SEC’s requirement for companies to file an 8-K with the title 4.02 “Non-Reliance of Previously Issued Financial Statements” when previous financial statements should no longer be relied upon, it is likely that more restatements will be identified after 2004. Even though Hennes et al. (2008) do not believe the SEC requirement per August, 2004 will bias their test, to out rule any possible effects on the test this thesis will use a sample after 2004. Therefore this study has selected a sample from January 1st, 2007 until December 31st, 2015 from the AUDIT database for its test.

In order to be included in the test sample the restatement has to be identified as an Accounting Rule (GAAP/FASB) Application Failure and classified as fraud, irregularities or errors. The initial sample from AUDIT consisted of 317 error restatements and 96 irregularity restatements, which is a total of 413 observations. Subsequently I excluded companies that are in the financial services industry (SIC

(21)

codes8 6000-6999), following Hennes et al.’s (2008) study, which leaves the sample with 257 error observations and 88 irregularity observations. The sample contained several companies with multiple restatements between January 1st, 2007 and December 31st, 2015, however I only selected one restatement per company for the sample. Therefore in case of multiple restatements of a company during this time period the restatement to be included in the sample was selected based on the following three criteria: 1) restatement due to irregularity, 2) restatement has a negative effect on the financials, and 3) restatement with earlier filing date. Lastly, I eliminated companies that have no data available on either one of the other three databases (i.e. Compustat, Execucomp & D&O) used for company, executive and director data. This leaves me with 170 error restatement observations and 67 irregularity restatement observations, which totals 237 observations. Appendix A gives a summary of the restatement sample selection.

At first sight this sample seems relatively small. However, since fraud cases are a rare and uncommon practice this sample is sufficient for the purpose of this study, which is to analyze the differences between error and irregularity restatements and their impact on executive, director and audit committee turnover. Panels A-C of Appendix B show the distribution of the sample by year and industry.

TABLE 1

Litigation and Error/Irregularity Classification

Errors % Irregularity % Frequency

Securities Class Action Litigation

14 8.24 21* 31.34 54

No litigation 156 91.76 46 68.66 183

Total 170 100.00 67 100.00 237

* Indicates significance of difference across error and irregularity group at p < 0.01.

According to previous studies like Hennes et al. (2008) and Palmrose et al. (2004) the chances of risking litigations will likely be higher for companies disclosing

8

North American companies in Xpressfeed are assigned a 4-digit SIC code that identifies the line of business of the company as a whole.

(22)

irregularity restatements than companies disclosing error restatements due to the nature of the restatement. For comparison reasons this study also takes an initial look at the number of litigations following the restatement disclosures. The sample shows that 31.34% of the irregularity sample faced some type of litigation following a restatement as compared to only 8.24% of the error sample which is significantly lower, which supports previous studies’ findings. Table 1 gives an overview of the number of litigations per category of the sample (i.e. error/irregularity restatement).

3.2 The Turnover Window

Several studies conducted on executive and director turnover have shown both significant (e.g. Arthaud-Day et al., 2006; Gilson & Vetsuypens, 1993; Warner et al., 1988) and insignificant (e.g. Collins, Reitenga & Sanchez-Cuevas, 2005; Agrawal et al., 1999; Beneish, 1999) increases in turnover after restatements. However, most studies only consider turnover after the announcement date of a restatement. This study will consider CEO, CFO and director turnover before as well as after the restatement disclosure date following Hennes et al.’s (2008) approach. The reason it is important to also take turnover before the actual restatement announcement into consideration is because it is possible that the misstatement that has been discovered during the new management’s tenure but was caused by prior management. So in this case it would be unlikely that current management will be penalized by dismissal for previous management’s wrong doings. Another reason it is important to look at turnover prior to the restatement announcement according to Hennes et al. (2008) is that it might occur that executives are terminated as a consequence of the deliberate misreporting while an investigation is still going on resulting in termination prior to the actual announcement date of the restatement. Therefore I agree with Hennes et al.’s (2008) approach, who use a thirteen-month window around the restatement announcement (six months before and six months after), and also take turnover prior to restatement announcements into account.

In Table 2 I have summarized the turnover over time relative to the restatement announcement date for the three groups (e.g. CEO, CFO and Director & Committee Members) differentiated into error and irregularity restatements. The table shows that for the error sample CEO, CFO and Director & Committee Member’s turnover occur respectively for 71 percent, 75 percent and 68 percent between six

(23)

months prior and twelve months after the restatement announcement. This also holds for the irregularity sample which also shows that the majority of the CEO, CFO and Director & Committee Member’s turnover occur between that window (63 percent for all three groups). Both the error and the irregularity sample show that turnover is mostly concentrated between zero and twelve months after the restatement announcement. According to Hennes et al. (2008) and Parrino (1997) turnovers that occur close to the restatement announcement are more likely to be related to the restatement. So this analysis suggests that the majority of the turnover related to restatements in the sample will likely be captured when using a window of eighteen months, six months prior and twelve months after the restatement announcement.

TABLE 2

Turnover over Time Relative to Restatement Announcement Date Variables …-7 months prior 6-0 months prior 0-12 months after 13-24 months after Total CEOs Error 6.35 31.75 39.68 22.22 100.00 Irregularity 26.31 26.31 36.84 10.54 100.00 CFOs Error 5.88 19.12 55.88 19.12 100.00 Irregularity 20.93 25.58 37.21 16.28 100.00 Directors & Committee Members Error 10.52 26.32 42.11 21.05 100.00 Irregularity 18.75 41.67 20.83 18.75 100.00 Error Total 7.58 25.73 45.89 20.80 100.00 Irregularity Total 21.99 31.19 31.63 15.19 100.00 Total 14.79 28.46 38.76 17.99 100.00

(24)

3.3 Variable Measurement and Control Variables

The dependent variables used in this study to test the two hypotheses in the previous sections are CEO, CFO and director turnover. Using event history data from Audit Analytics’ Director and Officer database, which indicates the date of the employment termination of the directors and officers retrieved from 8-K filings, the turnover was determined for the sample. Individuals retaining their position during the turnover window of six months prior to twelve months after the restatement announcement were assigned a value of 0. Those who’s employment was terminated (either resigned, dismissed or ceased) during the turnover window were assigned a value of 1.

The first independent variable is the dummy variable IRREGULARITY which is assigned a value of 1 if the restatement is an irregularity, and a value of 0 otherwise (i.e. error). Prior restatement studies (e.g. Hennes et al., 2008; Land, 2006; Palmrose and Scholz, 2004) state that restatements with a larger magnitude will likely have a larger impact on stakeholders (e.g. shareholders and potential investors) than similar restatements with a smaller magnitude and introduce a severity variable in their model which is a scaled measure of the cumulative effect of the restatement. Hence, this thesis introduces the variable MAGNITUDE which is calculated as follows:

MAGNITUDE =Cumulative Amount of Net Income Overstated Total Assets

The cumulative amount of net income overstated is calculated as the sum of changes in the net income for all the periods affected by the restatement. The total assets correspond to the total assets per year end prior to the restatement announcement which can be found as a balance sheet item.

Since prior research shows that revenue recognition restatements are more often associated with stronger market effects the model controls for these revenue recognition related restatements by assigning a value of 1 to the dummy variable REVREC if the restatement is related to revenue recognition, and 0 otherwise. Also added is the dummy variable ANNUAL to differentiate between annual restatements (e.g. 10-K filings) and unaudited interim restatements (e.g. 10-Q filings) because prior studies (e.g. Land, 2006; Palmrose and Scholz, 2004) suggest that unaudited interim restatements will likely be regarded as less severe as compared to audited annual restatements. The dummy variable ANNUAL is assigned a value of 1 if the company

(25)

restated an annual report, and assigned a value of 0 otherwise.

To control for CEO power that has been argued to (negatively) affect turnover the model introduces the variable CEO_EQUITY which is the percentage of equity ownership of the CEO in the year prior to the restatement.9 The next variable LEVERAGE is added to indicate if the restating company is financially distressed since previous studies (e.g. Hennes et al., 2008; Gilson, 1989) show that financially distressed companies will more likely experience turnover. The variable LEVERAGE is computed as follows:

LEVERAGE = Total Debt Total Assets

The total debt is calculated as the sum of total long-term debt (#9 Compustat) and total debt in current liabilities (#34 Compustat) in the year prior to the restatement announcement. The total assets (#6 Compustat) correspond to the total assets per year end prior to the restatement announcement. Besides the variable LEVERAGE this study also added the organizational control ROA, which is a commonly used indicator for a company’s performance. The variable ROA is calculated as follows:

ROA =Operating Income After Depreciation Total Assets

The operating income after depreciation is in accordance with data item number 178 of Compustat. The total assets component used to calculate the ROA is similar to the total assets (#6 Compustat) in the denominator of the equation of the variable LEVERAGE. Lastly, the model contains a size indicator variable (SIZE) which is based on the total assets (#6 Compustat) of the restating company and is computed by taking the natural logarithm of the value.

The director turnover model also contains the dummy variable AUDIT to control for directors that are also on the audit committee board. So the variable AUDIT is assigned a value of 1 if the director had a seat on the audit committee board, and 0 if this was not the case. The CEO, CFO and director turnover models developed to test the two hypotheses of this study are shown in the following section.

9

CFO and director ownership is not included in their respective models because this is not mandated by law to be disclosed (Hennes et al., 2008).

(26)

3.4 Empirical Model

The models used for CEO, CFO and Director turnover are as follows: The CEO turnover model

CEO_TURNOVER = 0 + 1IRREGULARITY + 2MAGNITUDE

+ 3REVREC + 4ANNUAL + 5CEO_EQUITY

+ 6LEVERAGE + 7ROA + 8Firm_Size + 

The CFO turnover model

CFO_TURNOVER = 0 + 1IRREGULARITY + 2MAGNITUDE

+ 6LEVERAGE + 7ROA + 8Firm_Size + 

The Director turnover model

DIR_TURNOVER = 0 + 1IRREGULARITY + 2MAGNITUDE

+ 6AUDIT + 7LEVERAGE + 8ROA + 9Firm_Size + 

where:

TURNOVER = 1 if the CEO (or CFO/Director respectively) leaves the firm within the

six months before or the twelve months after the restatement announcement, and 0 otherwise;

IRREGULARITY = 1 if the restatement is classified as an irregularity, and 0

otherwise;

MAGNITUDE = the cumulative amount of the net income overstatement scaled by

total assets in the year prior to the restatement announcement;10

REVREC = 1 if any part of the misstatement relates to revenue recognition, and 0

otherwise;

10

The MAGNITUDE variable is consistent with the measure used in Palmrose et al. (2004), Srinivasan (2005) and Hennes et al. (2008).

(27)

ANNUAL = 1 if the firm restated an annual report, and 0 if the firm restated only an

interim report;

AUDIT = 1 if the director also has a seat on the Audit Committee Board, and 0

otherwise;

CEO_EQUITY = the percentage of equity ownership of the CEO in the year prior to

the restatement;

LEVERAGE = total debt scaled by total assets (Compustat (#9 + #34) / #6);

ROA = operating income after depreciation scaled by total assets (Compustat #178/

#6);

SIZE = indicator variable for company size based on total assets .

4 RESULTS

4.1 Descriptive Statistics

Table 3 shows the descriptive statistics of the error and irregularity sample from January 1st, 2007 till December 31st, 2015. To mitigate the possible effect of outliers I have calculated the z-scores of the variables to determine observations that are not within the normal distribution. Subsequently I have winsorized the data of the sample at the bottom and top which had a z-score above 2.68 or below -2.68 and replaced these values the highest or lowest observation in the sample.

Comparing the descriptive statistics of the error and irregularity sample it shows that the average size of the companies in the error sample, based on total assets, (mean = 5.09, median = 5.38) does not significantly differ from the average size of the companies in the irregularity sample. Surprisingly the companies in the irregularity sample do not show a significantly lower performance compared to the error sample. The ROA for the error sample (mean = -1.88 percent, median = 0.03 percent), which is calculated as the operating income after depreciation divided by total assets, is lower than the ROA of the irregularity sample (mean = -0.12 percent, median = 0.05 percent) showing that the error sample has relatively more negative return than the irregularity sample. This is in line with the higher leverage that is shown in the error sample (mean = 0.70, median = 0.15) compared to the irregularity sample (mean = 0.35, median = 0.20). Companies that perform more poorly will

(28)

likely need more finance from external parties which results in higher levels of debt. However, differences in performance and leverage across the error and irregularity group are not significant.

As anticipated, the irregularity sample shows a significantly (p < 0.05) more negative MAGNITUDE (mean = -0.10 percent, median = -0.01 percent) compared to the error sample (mean = -0.01 percent, median = 0.00 percent) indicating that the irregularity restatements has more severe restatements on average. This might sound counterintuitive but since the severity variable, MAGNITUDE, is computed by the cumulative amount of the change in net income (with a more negative change indicating a larger restatement) divided by total assets a more negative MAGNITUDE indicates a more severe restatement.

The next descriptive statistic is REVREC, which is a dummy variable that indicates whether or not the misstatement is related to revenue recognition with 1 indicating a relation to revenue recognition. The REVREC of the irregularity sample (mean = 0.42, median = 0) shows a significantly (p < 0.01) higher value compared to the REVREC of the error sample (mean = 0.15, median = 0) which indicates that the irregularity restatements are more often related to revenue recognition than the error restatements in our sample. This is similar to Hennes et al. (2008) and Dechow et al.’s (2007) studies on restatements and turnover that shows a 59 percentage and 54 percentage respectively in the irregularity sample compared to only 16 percent for the error sample in Hennes et al.’s (2008) study. A reason irregularity restatements are more likely related to revenue recognition is because revenue is easier to manipulate compared to other income statements while errors could occur in any part of the financials explaining the lower percentage in the error restatement related to revenue.

Lastly, the irregularity sample shows a slightly higher mean for the variable CEO_EQUITY (mean = 3.13 percent, median = 1.11 percent) compared to the error sample (mean = 2.76 percent, median = 1.44 percent) which means that CEOs in the irregularity sample have a higher percentage of equity ownership on average than the CEOs in the error sample. However, the difference in CEO equity ownership between the two groups is not significant.

(29)

TABLE 3 Descriptive Statistics

Error Irregularity

n Mean Median Std. n Mean Median Std.

SIZE (= ln (total assets)) 170 5.09 5.38 2.31 67 5.39 5.60 2.27

ROA 170 -1,88% 0.03% 12.22% 67 -0.12% 0.05% 0.62%

LEVERAGE 170 0.70 0.15 2.40 67 0.35 0.20 0.69

MAGNITUDE 127 -0.01% 0.00% 0.04% 39 -0.10%* -0.01%## 0.22%

REVREC 170 0.15 0 NA 67 0.42** 0# NA

CEO_EQUITY 170 2.76% 1.44% 4.69% 67 3.13% 1.11% 6.83%

**, * Indicate t-tests of difference across groups significant at p < 0.01 and 0.05, respectively. ##, # Indicate Wilcoxon signed-rank test across groups significant at p < 0.01 and 0.05, respectively.

4.2 Empirical Results

In order to answer the first hypothesis I will first analyse the differences between the error and irregularity sample on CEO and CFO turnover. The first hypothesis is as follows:

Hypothesis 1: Irregularities restatements have a higher impact on turnover than error restatements.

Panel A of table 4 shows that in the error sample 25.88 percent of the CEOs experience turnover while CEOs in the irregularity sample experience a higher turnover rate of 35.82 percent, which is a relative increase of 38.41 percent. The CFO turnover rate also increases from a 30.00 percent turnover rate in the error sample to a 40.30 percent turnover rate in the irregularity sample, which is a relative increase of 34.33 percent.

Similarly I compare the turnover rates of CEOs and CFOs in the two groups related to annual restatements to those of quarterly restatements. Restatements that were made for annual reports, which are filed through for example 10-K forms, are categorized under annual while restatements that were made for interim reports, which are filed through for example 10-Q forms, are categorized under quarterly. I expect the annual restatements to show higher turnover rates compared to quarterly restatements since quarterly restatements are most often unaudited and therefore

(30)

considered less severe.11 The results show that for the annual restatements there is an increase of 62.24 percent in turnover rate from 26.80 percent for CEOs in the error group to 43.48 percent for the CEOs in the irregularity group. A similar increase is witnessed for CFOs where there is an increase of 40.58 percent in turnover rate from 30.93 percent in the error group to 43.48 percent in the irregularities group. However, looking at the turnover rates of the quarterly restatements the results show that while the CFO turnover rate slightly increases with 15.85 percent from 28.77 percent in the error group to 33.33 percent in the irregularity group the CEO turnover rates

decreases with 22.75 percent from 24.66 percent in the error group to only 19.05 percent in the irregularity group. By looking further into the quarterly sample of CEO turnover rates I find that the average equity ownership of the CEOs in the error group is 0.80 percent while the CEOs in the irregularity group have an average ownership of 1.67 percent, which is more than twice the equity CEOs in the error group hold on average. This may explain the lower than expected CEO turnover rate of the quarterly sample because as shown by previous studies (e.g. Hennes et al., 2008; Arthaud-Day et al., 2006) CEO power is likely to negatively affect turnover rates.

I also ran an additional overall analysis for the total sample to see whether CEOs that do not experience turnover have significantly more power, measure by the equity ownership the CEO, than CEOs that do experience turnover. The analysis showed that this to be the case and supported the assumption with a significance of p < 0.05 that CEOs with more equity ownership have more power resulting in a smaller chance of being dismissed after a restatement for the overall sample (both error and irregularity).

Going back to the annual versus quarterly samples the results show that while the CEOs (CFOs) turnover rates in the error sample experiences a slight increase of 8.68 percent (7.51 percent) from 24.66 percent (28.77 percent) to 26.80 percent (30.93 percent) the irregularity sample experiences a greater increase in turnover rates for the annual restatements compared to the quarterly restatements. Comparing the annual restatements to the quarterly restatements in the irregularity group I observe an increase of 128.24 percent (30.45 percent) in CEO (CFO) turnover rates from 19.05

11

Interim reports of publically traded companies are not subject to audits mandated by law as opposed to annual reports. However, it could be possible that a company chooses to audit their interim reports voluntarily. I expect the majority of the interim reports to be unaudited since this is not a common practice.

(31)

percent (33.33 percent) to 43.48 percent (43.48 percent).

The above findings are consistent with three of my expectations. Firstly, executives are more likely to be penalized by dismissal if the nature of the restatement is related to fraudulent practices (i.e. irregularity restatements). Secondly, higher levels of CEO power, measured by CEO equity ownership, will likely negatively affect turnover rates because the CEO will more likely be able to exercise more control. Lastly, I expect that annual restatements will more likely have a higher turnover rate compared to quarterly restatements due to prior research like Hennes et al. (2008) arguing that annual restatements are considered more severe than quarterly restatements. However, the differences in turnover rates for the groups all fail to be statistically significant. Therefore based on the results stated above and shown in panel A of table 4 hypothesis 1 is not accepted for the CEO and CFO sample due to lack of statistical evidence on the significance of the results.

TABLE 4

Panel A: CEO & CFO Turnover Frequency Error Turnover % Irregularity Turnover % Total Turnover % n CEO CFO n CEO CFO n CEO CFO

Total 170 25.88 30.00 67 35.82 40.30 237 28.69 32.91

Annual vs. Quarterly

Annual 97 26.80 30.93 46 43.48 43.48 143 32.17 34.97

Quarterly 73 24.66 28.77 21 19.05 33.33 94 23.40 29.79

Panel B: Director Turnover Frequency Error Turnover % Irregularity Turnover % Total Turnover %

n Director _CommitteeOn Audit n Director _CommitteeOn Audit n Director _CommitteeOn Audit

Total 170 29.41 7.65 67 44.78a 16.42 237 33.76 10.13

Annual vs. Quarterly

Annual 97 32.99 5.15b 46 50.00 17.39 143 38.46c 9.09

Quarterly 73 24.66 10.96 21 33.33 14.29 94 26.60 11.70

a _{Significance of difference comparing turnover rates in the error and irregularity samples at p < 0.05.} b _{Significance of difference comparing turnover rates in annual versus quarterly restatements at p < 0.05.} c

Errors and irregularities restatements : a study on executive and director turnover