The influence of audit committee characteristics on
the trade-off between accrual-based earnings
management and real earnings management
Nathalie Denekamp 10216480
MSc Accountancy and Control, specialization Accountancy Amsterdam Business School
Faculty of Economics and Business, University of Amsterdam Supervisor: Dr. W.H.P. Janssen
Word count: 14.655 Date: 22-06-2015 Version: Final
2
Statement of Originality
This document is written by student Nathalie Denekamp 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
This study examines to what extent an entirely independent audit committee and an audit committee with at least one financial expert has an influence on the trade-off between accrual-based earnings management and real earnings management. The results show a negative influence of an entirely independent audit committee on accrual-based earnings management. A significant negative relation between audit committee independence and three real earnings management measures (abnormal cash flow from operations, abnormal discretionary expenses, and abnormal production costs) is found. So, even if managers decide to switch from accrual-based earnings management to real activities manipulation, an entirely independent audit committee has the ability to lower real earnings management. Furthermore, I find no significant complementary relation between the earnings management forms. I conclude that an entirely independent audit committee is able to lower both earnings management forms, thereby having a significant influence on the earnings management trade-off.
Concerning audit committee financial expertise, no significant relation was found with accrual-based earnings management. Additionally, there is no significant association between audit committee financial expertise and real earnings management. There is a significant complementary relationship between real earnings management and accrual-based earnings management in firms with audit committees with at least one financial expert. This is probably because the audit committee fails to detect both earnings management forms. The result that in none of the cases having a financial expert on the audit committee proves to be beneficial should be a useful insight for regulators such as the SEC.
Keywords: corporate governance, audit committee, audit committee independence, audit committee financial expertise, earnings management, accrual-based earnings management, real earnings management, trade-off
4 Table of Contents 1 Introduction ... 6 1.1 Background ... 6 1.2 Contribution ... 8 1.3 Structure... 9 2 Literature review ... 10 2.1 Earnings management ... 10 2.1.1 Definition ... 10 2.1.2 Agency theory ... 11
2.1.3 Reasons for managing earnings ... 11
2.1.4 The trade-off ... 12
2.2 The audit committee ... 14
2.2.1 Definition ... 14
2.2.2 Responsibilities of the audit committee ... 14
2.2.2.1 Responsibilities with regard to earnings management ... 15
3 Hypotheses development ... 16
3.1 Audit committee independence ... 16
3.2 Audit committee financial expertise ... 18
4 Research design ... 22
4.1 Measurement of accrual-based earnings management ... 22
4.2 Measurement of real earnings management ... 23
4.3 Measurement of audit committee characteristics ... 25
4.4 Empirical models ... 26
4.4.1 Control variables ... 27
4.5 Sample selection ... 28
5
5.1 Descriptive statistics ... 31
5.2 Univariate analysis ... 34
5.2.1 Pearson correlation matrix ... 36
5.3 Multivariate analysis... 39
5.3.1 Accrual-based earnings management ... 39
5.3.2 Real earnings management ... 40
5.3.3 Control variables ... 43
5.4 Additional analysis ... 44
5.4.1 REM summary measure ... 45
5.4.2 Complementary or substitutive relation ... 45
6 Conclusion ... 48
7 References ... 50
6
1
Introduction
1.1 Background
Recently, the Chairman of the U.S. Securities and Exchange Commission Mary Jo White has emphasized in a speech the extraordinarily important role that the audit committee, as part of the corporate governance, plays in creating a culture of compliance through their oversight of financial reporting (2014). Jay D. Hanson, board member of the Public Companies Accounting Oversight Board, also said that the importance of the audit committee cannot be overstated, as they are in the center of the financial reporting process. Thereby he stressed the need for financial expertise of audit committee members (2015).
The audit committee concept was first endorsed by the New York Stock Exchange (Klein, 2002). As a consequence of various scandals in the beginning of the 21st century
however, the Sarbanes-Oxley Act (SOX) was enacted and included a series of measures designed to improve corporate governance at public companies, including new audit committee requirements (PCAOB, 2015). With the passage of SOX, two audit committee characteristics became essential. First, SOX amended section 10A of the Securities Exchange Act of 19341, requiring all audit committees to be independent of firm management. Second, it required firms to report if their audit committee contains at least one financial expert, and if not why. However, having a financial expert on the audit committee is not obligated (SOX, 2002).
Many studies have researched the effectiveness of several audit committee characteristics, including independence and financial expertise, on audit and earnings quality. For example, Ghosh, Marra & Moon (2010) investigate the effect of audit committee characteristics on accrual-based earnings management. They find proof that audit committee size, activity and tenure are associated with earnings management, however financial expertise is not. Bedard, Chtourou & Courteau (2004) find a significant reduction in the likelihood of aggressive earnings management when 100% of the audit committee is independent and a significant negative relation between financial expertise and accrual-based earnings management. Klein (2002) concludes that audit committees composed of less than a majority of independent directors are more likely to have larger adjusted abnormal accruals, whereas there is a negative relationship between adjusted abnormal accruals and the percentage of independent audit committee members.
1 With this Act, Congress created the Securities and Exchange Commission. The Act empowers the SEC with
7 Many attention in past research has been paid to accrual-based earnings management, which involves managing earnings by exercising discretion over accruals. Less research however has focused on real earnings management. Real earnings management involves manipulating real activities. Roychowdhury (2006, p. 335) defines real earnings management as “management actions that deviate from normal business practices, undertaken with the primary objective of meeting certain earnings thresholds”. This kind of earnings management affects cash flows and eventually firm value. Managing real activities includes for example giving price discounts to meet sales targets, overproducing to decrease the cost of goods sold or decreasing investments (Roychowdhury, 2006). Sun, Lan & Liu (2014) have examined the effect of five audit committee characteristics on real earnings management. They find that audit committee members’ additional directorships are significantly positively related with real earnings management. However they find no significant association between the other four characteristics (accounting financial expertise, board tenure, block shareholdings, audit committee size) and real earnings management.
As both earnings management forms exist, it is possible that when managers decide to manage earnings, they trade-off the two types of earnings management. This decision might be influenced by several factors. In recent research, some studies have examined what factors and how these factors might influence this trade-off made by managers. For example, Zang (2012) finds that managers use real earnings management and accrual-based earnings management as substitutes based on their relative costs. Cohen, Dey & Lys (2008) conclude that firms switched from accrual-based earnings management to real earnings management after the introduction of SOX. Achleitner, Günther, Kaserer & Siciliano (2014) document that family firms engage less in real earnings management and more in earnings-decreasing accrual-based earnings management than non-family firms. Wongsunwai (2013) finds that companies conducting an initial public offering, backed by higher-quality venture capitalists, show less aggressive financial reporting, which is reflected in lower accruals and lower real activities manipulation. The effect that corporate governance, and more specifically the audit committee, has on the earnings management trade-off, if any, has however not yet been examined. This study addresses this gap in literature by examining the following research question:
“To what extent do audit committee characteristics influence the trade-off between accrual-based earnings management and real earnings management?”
8 The audit committee characteristics that are the focus of this study are the two that SOX requires to report on, audit committee independence and audit committee financial expertise. These variables have also been measured as how they are prescribed in SOX, thereby enabling the outcomes to be meaningful to regulators such as the SEC.
The results show a negative relation between a fully independent audit committee and accrual-based earnings management. I also find a significant negative relation between audit committee independence and all three real earnings management measures. Even if managers decide to switch from accrual-based earnings management to real earnings management, the audit committee is able to lower real earnings management. Furthermore, both earnings management forms are used as complements, however this relation is not significant. I conclude that an entirely independent audit committee is able to reduce both earnings management forms, thereby having a significant influence on the earnings management trade-off.
Concerning audit committee financial expertise, I find no significant relation with accrual-based earnings management. Additionally, there is no significant association between audit committee financial expertise and real earnings management. There is a significant complementary relation between the two earnings management forms in firms with audit committees with at least one financial expert. This is probably because audit committees with a financial expert fail to detect both earnings management forms, thereby questioning if having at least one financial expert on the audit committee is beneficial.
1.2 Contribution
This study makes important contributions to the literature. First, the effect of audit committee characteristics on accrual-based earnings management and the effect of audit committee characteristics on real earnings management have only been examined in isolation, i.e. studies only examine the relation between audit committee characteristics and accrual-based earnings management or the relationship with real earnings management. To date there is no study that examines the earnings management trade-off that managers make in relation to the audit committee. Therefore, this study contributes to the literature by bringing more perspective on how these two earnings management types co-exist and the influence of audit committee independence and audit committee financial expertise on this. Second, the relation between audit committee independence and real earnings management has not yet been examined. Thus, this study adds to the limited research available on real earnings management.
This study also has important implications to regulators such as the SEC whose goal it is to protect investors. It is assumed that the two characteristics, audit committee independence
9 and audit committee financial expertise, enable the audit committee to reduce earnings management through executing their function of monitoring the financial reporting process. However, as appears from the results, audit committee financial expertise is in none of the cases associated with both accrual-based and real earnings management. Apparently, having at least one financial expert on the audit committee is not enough to reduce earnings management. This should be a useful insight to regulators.
1.3 Structure
The remainder of this study is structured as follows. In section 2 the concepts accrual-based earnings management, real earnings management and audit committee will be further elaborated on. In section 3, hypotheses for the relation between audit committee independence and audit committee financial expertise and both earnings management forms will be developed. Subsequently, in section 4 the research methodology will be discussed. In section 5 the results of the regression analysis will be discussed and in the final section, the conclusion will be drawn.
10
2
Literature review
In section 2.1 the concept earnings management is discussed and subsequently in section 2.2 the audit committee.
2.1 Earnings management
In this section first a definition of earnings management is provided. Then, the agency theory will be described. Subsequently, incentives for managing earnings are described. Finally, some previous literature on the decision between the two types of earnings management is discussed.
2.1.1 Definition
Healy and Wahlen (1999, p. 368) describe earnings management as follows: “Earnings management occurs when managers use judgment in financial reporting and in structuring transactions to alter financial reports to either mislead some stakeholders about the underlying economic performance of the company, or to influence contractual outcomes that depend on reported accounting numbers”. Several ways in which managers can use this judgment are mentioned by Healy and Wahlen (1999). Judgment can be used in estimating future economic events, such as expected lives and residual values of long-term assets, obligations for pension benefits, bad debt expenses and asset impairments. Managers can also use judgment in choosing between different accounting methods to report economic events. For example, inventory can be valued using the Last-In First-Out, First In-First Out or weighted average method. So, managers can exercise judgment in when to recognize certain economic events, i.e. earnings and losses. If managers use this judgment to alter financial statements or mislead stakeholders, they manage earnings. This is what is called accrual-based earnings management.
Roychowdhury (2006, p. 337) defines real earnings management or real activities manipulation as “departures from normal operational practices, motivated by managers’ desire to mislead at least some stakeholders into believing certain financial reporting goals have been met in the normal course of operations”. Roychowdhury (2006) mentions three ways in which real activities manipulation can be achieved. The first is sales manipulation. This means that managers accelerate the timing of sales or generate additional unsustainable sales through increased price discounts or more lenient credit terms. This will increase the revenue from sales for the current year, but might harm future profits. Second they can reduce discretionary expenses. Discretionary expenses consist of (1) advertising expenses, (2) research and development (R&D) expenses and (3) selling, general and administrative (SG&A) expenses.
11 Reducing these expenses increases earnings in the short run, but might harm firm value in the long run. For example, cutting R&D expenses might have as a consequence that a firm falls behind its competitors, thereby reducing profits. The last way is overproduction, which means that managers increase production so that the fixed costs can be spread over more units resulting in lower cost of goods sold, higher earnings but also a higher inventory on the balance sheet.
2.1.2 Agency theory
In this section, the agency theory will be discussed. This theory explains why managers might mislead stakeholders and explains the role of an audit committee hereby.
Jensen & Meckling (1976, p. 308) define an agency relationship as a contract under which one person (the principal) engages another person (the agent) to perform a service on their behalf, which involves delegating some decision making authority to the agent. If both the principal and the agent want to maximize their utility, it is likely that the agent will not always act in the best interest of the principal. Eisenhardt (1989) describes two problems with which the agency theory is concerned. The first problem is the agency problem. This problem means that it is hard for the principal to verify if the agent has behaved appropriately. The second problem is the risk sharing problem. Here, the principal and the agent may prefer different actions because they have different attitudes towards risk.
The principal-agent relationship can be applied to several relationships, such as the relation between managers (agents) and shareholders (principals). As managers have no share in their company, they have an incentive to not always act in the best interest of the company (and thus the shareholders), but instead act in their own best interest. This could imply that managers might manage earnings to meet a certain target for one year, in order to receive their bonus or they manage earnings to meet analyst forecasts. The task of the audit committee is to monitor the financial performance and processes on behalf of the shareholders, which implicitly means that they monitor management and their behavior and make sure that they do not engage in opportunistic behavior.
2.1.3 Reasons for managing earnings
There are several reasons why managers might manage earnings. As mentioned in the previous section, managers might manage earnings in order to meet a target to get their bonus in a certain year. Healy (1985) describes patterns of earnings management used in reaching a bonus target: income minimizing and income maximizing. With income minimizing, managers decide to use
12 income decreasing accruals, for example if they have already reached their bonus target. All the earnings above the target do not count for the bonus. Thus, through using income decreasing accruals, they store these earnings in order to use them to reach their target in the subsequent year. With income maximizing, income increasing accruals are used to reach a bonus target.
Other patterns of earnings management are income smoothing and big bath accounting (Stolowy, Lebas & Ding, 2010). Stolowy et al. (2010) describe income smoothing as the objective of creating a steadily growing (predictable) profit stream. This is done by shifting the date of recognition of some costs or revenues. Graham, Harvey & Rajgopal (2005) find in response to their survey that 78% of CFOs admit to smooth earnings, thereby sacrificing long-term value. Furthermore, CFOs find it important that their earnings are predictable, because for less predictable earnings a risk premium is being asked in the market. Taking a big bath often takes place when a new CEO is appointed. The new CEO might take a big bath to easily obtain an improved income figure, whereby assets are reduced in order to have lower expenses in the future (Stolowy et al., 2010).
Dichev, Graham, Harvey & Rajgopal (2013) held a survey under CFOs in order to obtain insight in what drives CFOs to manage earnings. The desire to influence the stock price, outside pressure to hit earnings benchmarks and inside pressure to hit earnings benchmarks are the most given answers to why CFOs misrepresent economic performance. Also, 88,6% state earnings are managed to influence executive compensation. 80,4% believe that earnings are misrepresented to avoid adverse career consequences if poor performance is reported. Other reasons to manage earnings that stem from the survey of Dichev et al. (2013) are to avoid violating debt covenants, pressure to report smooth earnings and the feeling of executives that managers believe that managing earnings will not be detected.
2.1.4 The trade-off
In this section, the trade-off between real earnings management and accrual-based earnings management will be discussed. Most of the previous research has focused on accrual-based earnings management. However, managers can also manage their earnings through real activities. The survey of Graham et al. (2005) also indicates they do this. In their survey executives are being asked what they would do, within the boundaries of GAAP, if they might not reach the desired earnings target near the end of the quarter. 79,9% of the respondents indicate that they would decrease discretionary expenses like advertising, R&D or maintenance costs. So, since both types of earnings management exist, it is likely that managers make a trade-off between both types based on certain circumstances/factors.
13 Several studies have investigated what factors might influence managers to decide between the forms of earnings management. First of all, Zang (2012) studies if managers use real earnings management and accrual-based earnings management as substitutes. Her findings indicate that managers decide between the two earnings management types based on their relative costs. Managers use more accrual-based earnings management if real earnings management is costly to them. This can be due to having a less competitive status in the industry, being in a less healthy financial condition, experiencing higher levels of monitoring from institutional investors or incurring greater tax expenses. If accrual-based earnings management is more costly for managers, due to being under a higher level of scrutiny of accounting practices after the implementation of SOX, limited accounting flexibility because of accrual manipulation in prior years or shorter operating cycles, managers use more real earnings management (Zang, 2012).
Cohen et al. (2008) examine the use of the types of earnings management around the application of SOX. They find evidence that after the application of SOX, real earnings management increased. They conclude that firms switched from accrual-based earnings management towards real earnings management with the passage of SOX. This finding is consistent with Zang (2012), who argues that after implementation of SOX accrual-based earnings management is more costly for managers and therefore managers switch to real earnings management.
Another factor that has been researched by Cohen & Zarowin (2010) is real and accrual-based earnings management around seasoned equity offerings (a new equity issue by a company that is already publicly trading). They find evidence that firms use both types of earnings management around SEOs. The choice between the two types varies as a function of the firm’s ability to use accrual-based earnings management and the costs of using it. Also they find that the decline in operating performance after the SEO attributable to real activities management is more severe than the decline that is attributable to the management of accruals, indicating that the operating performance decline is not just driven by the reversal of accruals.
Furthermore, Achleitner et al. (2014) find evidence that family firms engage less in real earnings management and more in downwards accrual-based earnings management than non-family firms do. Wongsunwai (2013) examines the effect of the quality of venture capitalists on earnings management in companies that conduct initial public offerings (i.e. the first time a private company sells stock on the public market). He finds that companies conducting an IPO backed by higher-quality venture capitalists show less aggressive financial reporting, which is reflected in lower accruals and lower real earnings management. Badertscher (2011)
14 investigated the use of earnings management in overvalued companies. The findings show that the amount of total earnings management increases with the time that the firm is overvalued. He also finds that managers engage in accruals management in the early stages of overvaluation and then move towards real earnings management, in order to sustain the overvaluation.
As shown there are several factors that can influence the earnings management trade-off. In the next section the factor that is subject of this study will be discussed.
2.2 The audit committee
2.2.1 Definition
Merchant and van der Stede (2007, p. 553) define corporate governance as “the sets of mechanisms and processes that help ensure that companies are directed and managed to create value for their owners while fulfilling the responsibilities to other stakeholders”. Part of the corporate governance is the board of directors. The board of directors' primary activity is to monitor the actions of the management. The board of directors has a fiduciary duty to promote the long-term success for the benefit of the shareholders (Merchant and van der Stede, 2007). The board of directors typically consists of several committees. One of these specialized committees is the audit committee. SOX (2002, section 2) defines an audit committee as: “a committee (or equivalent body) established by and amongst the board of directors of an issuer for the purpose of overseeing the accounting and financial reporting processes of the issuer and audits of the financial statements of the issuer”.
2.2.2 Responsibilities of the audit committee
In this section, the responsibilities of the audit committee are discussed and subsequently the specific responsibilities with regard to earnings management.
Following the definition of an audit committee by SOX (2002), the primary responsibility of an audit committee is to monitor the accounting and financial reporting process of a firm. Another main responsibility which is apparent from the SOX definition is that the audit committee is directly responsible for the appointment, compensation and oversight of the work of any registered public accounting firm hired by a firm for the purpose of preparing or issuing an audit report or related work. This includes the resolution of disagreements between management and the auditor regarding financial reporting. Also, the hired registered public
15 accounting firm needs to report directly to the audit committee (SOX, 2002, section 301, section 204 and section 202).
Another responsibility of the audit committee is that it has to establish procedures for complaints. First of all for “the receipt, retention, and treatment of complaints received by the issuer regarding accounting, internal accounting controls, or auditing matters” (SOX, 2002, section 301). Second, they need to set up procedures for the confidential, anonymous submission by employees of a firm of concerns regarding questionable accounting or auditing matters (SOX, 2002, section 301). The audit committee also has the right to engage independent counsel and other advisers, if the audit committee deems this necessary to carry out its duties (SOX, 2002, section 301). Finally, the audit committee also has the responsibility to approve the provision of non-audit services by a registered public accounting firm that are not specifically prohibited by SOX (2002, section 201).
2.2.2.1 Responsibilities with regard to earnings management
As already pointed out, the primary responsibility of an audit committee is to monitor the accounting and financial reporting process of a firm. In executing their monitoring function, one would expect that the audit committee would notice dissimilarities in the financial reporting. For example, an unusual sales growth or extraordinary high inventory on the balance sheet should come to the attention of the audit committee. Sun et al. (2014) also note this. When audit committee members come across abnormal changes in items on the financial statements, they should discuss these with the management. This has been reinforced by SOX which requires audit committee members to expand their understanding of the company and its activities beyond the core competencies of accounting. Sun et al. (2014) further note that since earnings management distorts financial reporting and an important duty of the audit committee is to ensure the integrity of the financial statements, the audit committee has a responsibility for constraining real earnings management.
Furthermore, Bedard et al. (2004, p. 16) quote a speech held by Herdman (staff of the SEC) in 2002, in which he says that “the audit committee may reduce opportunistic earnings management by "evaluating the competence and independence of the external auditors," by engaging in proactive discussions with company management and outside auditors regarding key accounting judgments, and by probing "to find out the nature and extent of issues that management and the auditors gave considerable attention to" as well as "the outcome of these discussions" ”. So, the audit committee also has a responsibility for accrual-based earnings management.
16
3
Hypotheses development
In this chapter the hypotheses that are going to be tested are developed. For both characteristics, one hypothesis with regard to accrual-based earnings management is developed and one hypothesis with regard to real earnings management.
3.1 Audit committee independence
In this section, the hypotheses with regard to the independence of the audit committee are developed. First, the definition of an independent audit committee member will be discussed. For this, the definition given by SOX will be used. SOX (2002, section 301) prescribes that each member of the audit committee is also a member of the board of directors and shall otherwise be independent. SOX describes independence as follows: “In order to be considered to be independent for purposes of this paragraph, a member of an audit committee of an issuer may not, other than in his or her capacity as a member of the audit committee, the board of directors, or any other board committee— ‘‘(i) accept any consulting, advisory, or other compensatory fee from the issuer; or ‘‘(ii) be an affiliated person of the issuer or any subsidiary thereof.” (2002, section 301). However, the SEC can decide to exempt from these requirements if the SEC determines this appropriate in the light of circumstances (SOX, 2002, section 301). So, although I expect a high percentage of independent audit committee members, not all members should by definition of SOX have to be independent.
The reasoning behind the introduction of a totally independent audit committee is that, according to the SEC (2003b) “an audit committee comprised of independent directors is better situated to assess objectively the quality of the issuer's financial disclosure and the adequacy of internal controls than a committee that is affiliated with management”. As described by the agency theory, managers might have incentives to look just at the short term. The SEC (2003b) argues that “an independent audit committee with adequate resources helps to overcome this problem and to align corporate interests with those of shareholders”. As explained in section 2.2.2.1 the audit committee is responsible for earnings management. So, I would expect that an audit committee comprised of independent members should be better able to assess objectively the occurrence of earnings management, than members affiliated with management.
Previous research has investigated the relationship between an independent audit committee and accrual-based earnings management. Klein (2002) has examined whether audit committee characteristics are related to earnings management. Klein finds that audit committees composed of less than a majority of independent directors are more likely to have
17 larger adjusted abnormal accruals. Furthermore, there is a negative relationship between adjusted abnormal accruals and the percentage of independent audit committee members. Klein (2002) argues that one interpretation of the results could be that a negative relation exists between audit committee independence and earnings management for all large traded U.S. firms. However, another interpretation could be that firms with large accruals build-in in their earnings structure are less inclined to have independent audit committees.
Bedard et al. (2004) conclude that their results regarding independence generally support the requirement of SOX that all the members of the audit committee should be independent for efficient monitoring. Contradicting to Klein (2002), they argue that a majority of independent audit committee members is not sufficient to lower earnings management. Bedard et al. (2004) find a significant reduction in the likelihood of aggressive earnings management when 100% of the audit committee is independent. They also find that the exercise of stock options in the short term by independent audit committee members is associated with a higher likelihood of aggressive earnings management.
Furthermore Xie, Davidson & DaDalt (2003) find that the percentage of outside directors is negatively related with accrual-based earnings management. They argue that having outside corporate members on the audit committee is associated with the committee’s ability to monitor the financial performance and reporting. Earnings management is less likely to occur and occurs less often in firms with an independent audit committee.
Finally, Sharma and Kuang (2014) also find evidence that audit committees consisting of independent directors reduce the likelihood of earnings management. Greater stock ownership by non-executive and executive directors serving on the audit committee increases the risk of aggressive earnings management. However, stock ownership by independent directors reduces this risk. This is inconsistent with the finding of Bedard et al. (2004) that stock options held by independent audit committee members increases the likelihood of earnings management.
Overall, previous research seems to indicate that audit committee independence leads to lower accrual-based earnings management. Therefore, I state that an independent audit committee is effective in monitoring accrual-based earnings management:
H1: Audit committee independence is negatively associated with accrual-based earnings management.
18 It might be possible that, because an independent audit committee is effective in monitoring accrual-based earnings management, managers might not use accruals to manage earnings anymore, but switch to managing real activities instead.
If this is true, then I would expect a positive association between real earnings management and audit committee independence. This means that the presence of an independent audit committee does not keep managers from using real earnings management, for example because managers think that it will not come to the attention of the independent audit committee. This could for example be because they might have less knowledge of a company’s activities due to the fact that they have no relations with a company.
However, it could also be that an independent audit committee actually is effective in monitoring real earnings management, which would be reflected in a negative association between real earnings management and audit committee independence. So even if managers decide to switch to real earnings management, in this case the audit committee is able to detect and lower real earnings management. This effectiveness could be the result of the fact that they do not have any relations with the company and thus are better able to look critical at the financial performance of a company. Because both ways could be possible, I pose the following non-directional hypothesis:
H2: Audit committee independence is associated with real earnings management.
3.2 Audit committee financial expertise
The hypotheses with regard to audit committee financial expertise will be developed in this section. First of all, it is necessary to know what the term financial expert means. SOX (2002, section 407) prescribes that firms have to disclose whether or not their audit committee contains at least one financial expert, and if not, why. Note that it is not obligated to have a financial expert on the audit committee. SOX describes that, to be qualified as a financial expert, the person has, “through education and experience as a public accountant or auditor or a principal financial officer, comptroller, or principal accounting officer of an issuer, or from a position involving the performance of similar functions— (1) an understanding of generally accepted accounting principles and financial statements; (2) experience in— (A) the preparation or auditing of financial statements of generally comparable issuers; and (B) the application of such principles in connection with the accounting for estimates, accruals, and reserves; (3) experience with internal accounting controls; and (4) an understanding of audit committee functions.” (SOX, section 407).
19 However, the above mentioned proposed definition of the term "financial expert" proved to be the most controversial part of the proposals of the SEC (2003a). Most of the commenters thought that the proposed definition was too restrictive. In the final rules issued by the SEC, a financial expert needs experience preparing, auditing, analyzing or evaluating financial statements. So this is broader than the original requirement of experience with preparing or auditing financial statements. Also, it is not required to have experience in applying generally accepted accounting principles in connection with the accounting for estimates, accruals, and reserves. The ability to assess the general application of such principles in connection with the accounting for estimates, accruals and reserves is enough. In the final rules, these attributes cannot only be acquired by experience as a public accountant or auditor or a principal financial officer, comptroller, or principal accounting officer, but also by actively supervising these or other relevant experience (SEC, 2003a).
The SEC (2002) believes that the primary benefit of having a financial expert on the audit committee is “that the person, with his or her enhanced level of financial sophistication or expertise, can serve as a resource for the audit committee as a whole in carrying out its functions”. Furthermore I believe that a member with financial expertise would be better at executing their monitoring job, because of their financial expertise.
The relationship between audit committee financial expertise and earnings management has previously been subject of research. Ghosh et al. (2010) state that if financially knowledgeable audit committee members are better at monitoring financial reporting, they expect that accrual-based earnings management will be lower. Their findings indicate that firms that have at least one financial expert on the audit committee appear to have lower discretionary accruals than firms that do not have a financial expert on the audit committee. This is however not a significant relation.
Contradicting to Ghosh et al., Bedard et al. (2004) find a significant negative relation between financial expertise and accrual-based earnings management. This relationship applies to both positive and negative earnings management, i.e. to managing earnings upwards and downwards. They conclude that the presence of at least one financial expert on the audit committee reduces the likelihood of aggressive earnings management.
Xie et al. (2003) as well find that having a financial expert on the audit committee reduces the likelihood of accrual-based earnings management. Audit committee members with a legal, financial or banking background do not have a significant influence on the amount of discretionary accruals, but members with an investment banking background are negatively associated with accrual-based earnings management.
20 Finally, Sharma and Kuang (2014) argue that audit committee members with financial expertise are effective at identifying financial reporting problems, because they better understand the internal and external audit program and internal controls which makes it easier to prevent and detect material misstatements. They find that financial expertise is associated with lower accrual-based earnings management, but only if the independent directors hold this financial expertise.
As it seems from previous literature, audit committee members that have financial expertise have a negative impact on accrual-based earnings management. I therefore predict that:
H3: Audit committee financial expertise is negatively associated with accrual-based earnings management.
As financially knowledgeable audit committee members seem to be effective in monitoring accrual-based earnings management, managers might switch to using real activities manipulation.
If this is the case, then I would expect a positive association between real earnings management and audit committee financial expertise. This means that the presence of financial experts on the audit committee does not keep managers from using real earnings management, for example because managers think that financial experts on the audit committee will not notice the real earnings management. A positive association also means that an audit committee with a financial expert is actually bad in monitoring real earnings management.
However, a negative association between real earnings management and audit committee financial expertise is also possible, reflecting that an audit committee with a financial expert is effective in monitoring and constraining real earnings management. Even if managers decide to make a trade-off, an audit committee with a financial expert would effectively reduce real earnings management. This could be because financial experts have a better idea of the possibilities for real earnings management. For example, audit committee members with financial knowledge could or in theory should earlier detect an unusually high inventory on the balance sheet or an extraordinary sales growth while executing their monitoring function than someone who does not possess this financial expertise.
Previous research of Sun et al. (2014) shows no association between financial expertise and real earnings management, however they only use a sample of 100 firm-years, because they focus on earnings management in order to avoid reporting an annual loss. As has been pointed
21 out in section 2.1.3, there are other reasons for managers to manage earnings, for example manipulating sales in order to reach a sales target and receive a bonus, which is a form of real earnings management. Carcello, Hollingsworth, Klein & Neal (2006) find a positive association between financial expertise and abnormal discretionary expenses but fail to find an association between abnormal production costs and financial expertise. They did not examine the relation between audit committee financial expertise and abnormal cash flows. My sample will contain more observations and I examine all three earnings management forms. Because both a negative and a positive association could be possible, I pose the following non-directional hypothesis:
22
4
Research design
In this chapter, the research design is explained. The research method that is used is database research. First of all the measure for accrual-based earnings management is discussed, then the measure for real earnings management and finally the way that the two audit committee characteristics are operationalized. Subsequently, the empirical models will be established and finally the sample selection process is shown.
4.1 Measurement of accrual-based earnings management
Accrual-based earnings management can be measured by the amount of discretionary accruals. Estimating discretionary accruals starts with total accruals. Total accruals are defined as the difference between income before extraordinary items and cash flow from operations (Klein, 2002).
Total Accruals = Income before extraordinary items – Cash flow from operations (1)
Total accruals can be decomposed into a discretionary and a non-discretionary part.
Total Accruals = DA + NDA (2)
Non-discretionary accruals are the part of the accruals that managers cannot influence, but instead are the result of normal business activities. However, managers can exercise discretion over part of the total accruals, which are called discretionary accruals. There are several models to estimate discretionary accruals. The starting point of these models is estimating total accruals. Then these models try to estimate the non-discretionary part of the total accruals and the residual is the discretionary part of the accruals (Dechow, Sloan & Sweeney, 1995).
I will use the Modified Jones Model to estimate discretionary accruals. Dechow et al. (1995) evaluate alternative models for detecting earnings management. They conclude that their Modified Jones Model exhibits the most power in detecting earnings management. In the original Jones Model (1991), non-discretionary accruals are measured as a function of the change in revenues and the level of property, plant & equipment. The proportion of accruals that then is not explained by total accruals, are the discretionary accruals (Dechow et al., 1995). So, the original Jones Model (1991) assumes that no discretion is being exercised over the revenues in the event period. However, Dechow et al. (1995) state that managers can use
23 discretion over the recognition of credit sales. Managers can accrue revenues at the end of the year, when cash is not yet received and it is questionable if these revenues have been earned. The Modified Jones Model assumes that all changes in credit sales are the result of earnings management (Dechow et al., 1995). So, the non-discretionary part of accruals following the Modified Jones Model is the sum of changes in cash revenues (total revenues – credit sale revenues) and the level of property, plant & equipment. The proportion of accruals that is subsequently not explained by total accruals, are the discretionary accruals. This results in the following model (Modified Jones Model):
TACt / TAt-1= α0 + α1 (1/TAt-1) + α2 ((ΔREVt – ΔRECt)/TAt-1) + α3 (PPEt / TAt-1) + εt (3)
Where:
TACt = total accruals
TAt-1 = total assets at t-1 (beginning of the year)
ΔREVt = revenue in year t – revenues at t-1
ΔRECt = receivables in year t – receivables at t-1
PPEt = gross property plant & equipment in year t
εt = residuals of the regression in period t
α1, α2, α3 = firm-specific parameters
In the model above, the discretionary accruals are the residuals of the regression (εt). Those will
be used in regression model R1 (see section 4.4) as the dependent variable ABEM.
4.2 Measurement of real earnings management
As already explained in section 2.1.1, Roychowdhury (2006) mentions three ways in which managers can exercise real earnings management: sales manipulation, reduction of discretionary expenses and overproduction. Roychowdhury (2006) has developed measures to detect these three kinds of real earnings management. Those measures will be explained in this section. For this explanation, Roychowdhury (2006), Sun et al. (2014) and Chi, Lisic & Pevzner (2011) have been used.
The first way to manipulate real activities was sales manipulation. Sales manipulation can be measured by the abnormal cash flow from operations (ABN_CFO):
24 Where:
CFOt = cash flow from operations in period t
TAt-1 = assets at t-1 (beginning of the year)
SALESt = sales at t
ΔSALESt = change in sales (SALESt - SALESt-1)
εt = residuals of the regression in period t
ABN_CFO is measured as the residuals of the regression. Since offering price discounts and more lenient credit terms lead to lower cash flows in period t, a low value of abnormal cash flows (ABN_CFO) indicates high real earnings management.
Reducing discretionary expenses was the second way to manipulate real activities. By cutting these expenses, earnings go up because these expenses are usually expensed as they are incurred. Following Roychowdhury (2006), to measure if managers have reduced discretionary expenses (DEXP) to manipulate real activities, we use discretionary expenses as the sum of (1) advertising expenses, (2) research and development (R&D) expenses and (3) SG&A expenses.
DEXPt/TAt-1 = α0 + α1 (1/TAt-1) + α2 (SALESt -1/TAt-1) + εt (5)
Where:
DEXPt = discretionary expenses in period t
SALESt -1 = sales at t-1
The same reasoning as with sales manipulation applies here. Abnormal discretionary expenses (ABN_DEXP) is measured by the residual of the above regression. If managers reduce discretionary expenses to boost earnings, this will result in lower abnormal discretionary expenses. So, a low value of ABN_DEXP means high real earnings management.
The final form of real earnings management was overproduction. When firms produce more goods than needed, they can spread the fixed costs over more units, thereby decreasing cost of goods sold and increasing earnings. If managers engage in such real activities manipulation, it is expected that the level of production costs is abnormally high. Like Roychowdhury (2006),
25 production costs is defined as the sum of cost of goods sold and change in inventory. To estimate abnormal production costs (ABN_PROD) we use the following model:
PRODt/TAt-1 = α0 + α1 (1/TAt-1) + α2 (SALESt /TAt-1) + α3 (ΔSALESt /TAt-1)
+ α4 (ΔSALESt -1/TAt-1)+ εt (6)
Where:
PRODt = production costs in period t (sum of cost of goods sold and Δ inventory)
ΔSALESt -1 = change in sales (SALESt-1 - SALESt-2)
Abnormal production costs (ABN_PROD) is measured by the residual of the regression. A high value of ABN_PROD indicates high real earnings management, because overproduction leads to higher production costs.
4.3 Measurement of audit committee characteristics
Data on the two audit committee characteristics that are subject of this study is collected from Institutional Shareholder Services (ISS, formerly known as RiskMetrics). This database examines proxy statements for firms listed in the Standard and Poor’s (S&P) 500, the MidCap 400, and the SmallCap 600 (Ghosh et al., 2010). In these proxy statements firms report on the independence and financial expertise of directors as defined by SOX.
In ISS (2010), directors (including audit committee members) are classified as having one of the following board affiliations: Insiders / Employees (E), Affiliated Outsiders / Linked (L) or Independent Outsiders (I). A director is classified as Independent Outsider if he or she has no material connection to the company other than a board seat. Looking at the detailed guidance in ISS (2010), which can also be found in Appendix A, I conclude that if a director, other than in his or her capacity as a member of the audit committee, receives any consulting, advisory or other compensatory fee or is otherwise an affiliated person of the issuer or any subsidiary thereof, he or she is not classified as Independent Outsider. Thus in classifying audit committee members, RiskMetrics takes into account the SOX requirements for an independent audit committee. In this study, the variable audit committee independence (ACIND) will be measured as a dummy variable which is ‘1’ if all audit committee members per company and per year are classified as Independent Outsiders, and ‘0’ otherwise. Since audit committee independence is measured as required by SOX, the outcomes might be meaningful to regulators
26 such as the SEC. This measurement method using a dummy has previously been adopted by Bedard et al. (2004), Sharma and Kuang (2014) and Klein (2002).
In proxy statements companies also report who on the audit committee is identified as financial expert as required by SOX. In ISS, the variable “Financial Expertise” flags the directors that companies identified as financial experts (see Appendix A). In this study, audit committee financial expertise (ACFIN) will be measured using a dummy. The dummy has the value of ‘1’ if, per company and per year, the audit committee contains at least one financial expert (indicated by the word “yes” in the field “Financial Expertise”) and ‘0’ otherwise. SOX requires companies to report if their audit committee contains at least one financial expert, so this measurement method captures if it is beneficial to have at least one financial expert on the audit committee. Bedard et al. (2004) and Ghosh et al. (2010) also measured audit committee financial expertise in this way.
4.4 Empirical models
To test the hypotheses, two empirical models will be estimated. With the first empirical model, hypotheses H1 and H3 will be tested:
ABEM = β0 + β1 ACIND + β2 ACFIN + β3 SIZE + β4 ROA + β5 DEBT + β6 LOSS + β7 CRISIS
+ β8 BIGN + ε (R1)
Where:
Dependent variable
ABEM = accrual-based earnings management, measured as the residual of equation 3
Independent variables
ACIND = audit committee independence ACFIN = audit committee financial expertise Control variables
SIZE = firm size, measured as the natural log of total assets ROA = ratio: net income / total assets
DEBT = ratio: total debt / total assets
LOSS = dummy: ‘1’ if firm reports a loss, ‘0’ otherwise CRISIS = dummy: ‘1’ if year = 2008 or 2009, ‘0’ otherwise
27 With the second empirical model, H2 and H4 with regard to real earnings management will be tested:
REM = β0 + β1 ACIND + β2 ACFIN + β3 SIZE + β4 ROA + β5 DEBT + β6 LOSS + β7 CRISIS +
β8 BIGN + ε (R2)
Where:
Dependent variable
REM = real earnings management, measured as either ABN_CFO (4) or ABN_DEXP (5) or ABN_PROD (6)
4.4.1 Control variables
The first control variable that will be used is firm size (SIZE), which is measured as the natural log of the total assets. Managers of large firms are more likely to exploit the latitudes in accounting to reduce political costs (Warfield, Wild & Wild, 1995). However, large firms are also under more scrutiny from the outside world (Watts & Zimmerman, 1990). Therefore, firm size is expected to have an association with earnings management.
The second control variable is Return on Assets (ROA), which is measured as net income divided by total assets. ROA is included to control for changes in firm performance (Dechow et al., 1995). As third control variable DEBT is included. DEBT is measured as the ratio of total long-term debt to total assets. As stated in section 2.1.3, an incentive for managing earnings is to avoid violating debt covenants. So firms with more debt are more likely to manage earnings (DeFond & Jiambalvo, 1994). Therefore, I expect a positive association between the amount of debt a firm has and earnings management.
LOSS is included as control variable because managers of firms in financial distress might have more incentives to exercise discretion in accruals. DeAngelo, DeAngelo & Skinner (1994) also found that firms in financial distress have large negative accruals. So, a positive association is expected between financially distressed firms and earnings management.
Another control variable is CRISIS, a dummy which takes the value ‘1’ for the years 2008 and 2009, and ‘0’ otherwise. In the crisis there could be an incentive to manage earnings upwards because there is more pressure to report a small profit instead of a loss. There could also be an incentive to take a big bath because the current year is a lost cause anyway, thereby
28 managing earnings downwards. So, I expect the crisis to be positively associated with earnings management.
The last control variable is BIGN. This is a dummy that takes the value ‘1’ if a firm was audited by one of the Big Four audit firms (Deloitte, EY, PwC or KPMG) and ‘0’ otherwise. Prior research shows that large audit firms are better in detecting accrual-based earnings management (Becker, DeFond, Jiambalvo & Subramanyam, 1998; Chi et al., 2011). So, I expect a negative association between BIGN and accrual-based earnings management. However, Chi et al. (2011) found that, when accrual-based earnings management is constrained by higher quality auditors, the use of real earnings management increases. Hence, a positive association between BIGN and real earnings management is expected.
4.5 Sample selection
For testing the hypotheses I will use a sample of U.S. firms, because that is where SOX applies and has to be reported on audit committee independence and financial expertise. The financial data has been extracted from Compustat and data on audit committees has been extracted from ISS (formerly RiskMetrics) as explained in section 4.3.
This study examines the years 2008 until 2013. This is because I want to use the most recent data to test the hypotheses. The reason to start at 2008 is that in 2007 RiskMetrics changed the methodology to collect data to follow ISS specifications2. Data from 1996 till 2006
is available through the old database and could be merged by using CUSIP. However, in the old database there is no data available on the financial expertise of audit committee members and also for the year 2007 this data is not available. Thus I decided to take a sample from 2008 until 2013.
I started with the collection of data on the audit committee characteristics. After creating the dummy variables (see section 4.3) needed for the regression analysis and dropping duplicate observations, 8.895 observations remain. I used the CUSIP of these observations to extract the accompanying financial data from Compustat. To avoid duplicates, I dropped 2.704 observations with industry format “FS” which resulted in 15.316 remaining observations from Compustat. I then merged these two datasets based on CUSIP.
After merging, 8.370 observations are matched and 7.480 observations are deleted. This is quite high given that I used the output from ISS as input for Compustat, but this is because in Compustat I chose the years 2005-2013 as I needed the years 2005 and 2006 to calculate
29 certain parts of both earnings management formulas for subsequent years. Furthermore, whereas in ISS output data for a company sometimes is only available for a few years, in Compustat data for all years is available. These are the reasons why 7.480 observations are not matched.
To test the hypotheses, I run regression models R1 and R2 as described in section 4.4. Because in this study a trade-off is being examined, I want to do both regressions using the same data to increase the reliability of the results. As a consequence the sample is smaller than when I would have taken a separate sample for each earnings management form. First, observations with missing data on accrual-based earnings management and real earnings management are deleted. Furthermore, I deleted all observations in the finance, insurance and real estate sector, because discretionary accruals are difficult to define for financial services firms (Klein, 2002). Following Roychowdhury (2006) I also deleted all observations in regulated industries. Subsequently, I dropped observations with missing control variables. During the early stages of descriptive analysis it appeared that one holding firm in the sample merged with a financial company in 2012, though the SIC code was unchanged (since it concerns a holding company their business activities vary across different industries). Because of the merger the financial instruments this company held increased fifteen-fold. This firm came forward in the analysis because after winsorization, there was an outlier for one of the real earnings management metrics (a residual above 1). Therefore I decided to delete all data relating to this firm. Finally, since residuals are being calculated for each 3-digit, 2-digit and 1-digit SIC industry membership, I dropped all observations that are in a 1-digit SIC industry-year group that contains less than 10 observations. This reduces the final sample to 5.681 observations. In table 1 this sample selection procedure is displayed.
30
Table 1
Sample selection procedure
After merging two datasets 8.370 observations
- missing data on ABEM and REM - 1.952 observations - finance, insurance, real estate (SIC codes 6000-6799) - 296 observations - regulated industry (SIC codes 4400-4999) - 297 observations - missing data on control variables - 32 observations
- merged firm - 5 observations
- # observations < 10 in 1-digit SIC industry-year group - 107 observations
31
5
Empirical results
In this chapter I will first discuss the descriptive statistics of the sample. Then I will continue with a univariate analysis. Subsequently, I will continue with the multivariate analysis in which the regression models developed in section 4.4 will be tested. Finally some additional tests are run to provide more insight into the trade-off.
5.1 Descriptive statistics
As described in section 4.5, the final sample contains 5.681 firm-year observations. 1.210 unique firms are identified in the sample. In table 2 the distribution of ACIND per year is shown. As explained, with the introduction of SOX it was required that all audit committees should be totally independent, unless the SEC decides to exempt from these requirements if the SEC determines this appropriate in the light of circumstances. I therefore expected a high percentage of independent audit committees though not 100%, which is also reflected in the sample. Table 2 shows that on average 97,04% of the audit committees consists of entirely independent members. 2,96% of the audit committees is not fully independent and thus does not comply with the SOX requirement that the audit committee should be fully independent. Furthermore it can be seen that the amount of fully independent vs. not fully independent audit committees is evenly distributed over the years.
Table 2
Distribution of ACIND per year
Year
ACIND 2008 2009 2010 2011 2012 2013 Total %
0 26 35 28 26 30 23 168 2,96
1 914 921 938 957 960 823 5.513 97,04
Total 940 956 966 983 990 846 5.681 100
In table 3 the distribution of ACFIN per year is shown. SOX prescribes that firms have to disclose if their audit committee contains at least one financial expert, and if not, why. So it is not obligated by SOX to have a financial expert on the audit committee. However, as shown in table 3, on average 98,89% has at least one financial expert on the audit committee. Only 1,11% of the observations does not have a financial expert on the audit committee at all. Also from table 3 it seems that ACFIN is evenly distributed over the years.
32
Table 3
Distribution of ACFIN per year
Year
ACFIN 2008 2009 2010 2011 2012 2013 Total %
0 12 8 8 9 11 15 63 1,11
1 928 948 958 974 979 831 5.618 98,89
Total 940 956 966 983 990 846 5.681 100
In table 4 the descriptive statistics for all variables are shown. All continuous variables that are used to calculate accrual-based earnings management and real earnings management and all continuous control variables are winsorized to the 1st and 99th percentiles of their distributions. In table 4 all amounts are rounded to three decimal places. The mean of accrual-based earnings management and all three measures of real earnings management is close to 0. This means that on average there is equally as much upwards earnings management as downwards earnings management. Furthermore table 4 shows that on average 97% of the audit committees is fully independent, what was already shown in greater detail in table 2. On average 98,9% audit committees have at least one financial expert, what already appeared from table 3. Concerning the control variables table 4 shows that the mean size of the firm-year observations, measured as the natural logarithm of total assets, is 7,659 with a minimum of 4,83 and a maximum of 11,638. The average return on assets of the sample firms is 5%. The average debt ratio is 17,2% with a minimum of 0% and a maximum leverage of 63,7%. 15,6% of the observations has a negative net income and 33,4% of the observations are in the time of the crisis. This is consistent with what I expected since two of the six sample years are defined as crisis years. Finally table 4 shows that 92,9% of the sample observations are audited by one of the Big Four audit firms (Deloitte, EY, PwC or KPMG).
33
Table 4
Descriptive statistics
Observations Mean St. deviation Min Max
ABEM 5.681 + 0,000 0,054 - 0,308 0,218 ABN_CFO 5.681 - 0,000 0,059 - 0,250 0,272 ABN_DEXP 5.681 + 0,000 0,145 - 0,575 0,794 ABN_PROD 5.681 - 0,000 0,132 - 0,864 0,710 ACIND 5.681 0,970 0,169 0 1 ACFIN 5.681 0,989 0,105 0 1 SIZE 5.681 7,659 1,495 4,830 11,638 ROA 5.681 0,050 0,091 - 0,379 0,261 DEBT 5.681 0,172 0,154 0 0,637 LOSS 5.681 0,156 0,363 0 1 CRISIS 5.681 0,334 0,472 0 1 BIGN 5.681 0,929 0,258 0 1
To indicate the amount of earnings management, I look at the average explanatory power of the regressions (adjusted R-squared). Residuals are estimated for each 3-digit, 2-digit and 1-digit SIC industry membership with 10 yearly observations or more. Table 5 provides the mean adjusted R-squared of the regressions. For accrual-based earnings management the mean adjusted R-squared is 12,6%. This means that on average 12,6% of the total accruals are explained by normal business activities of the sample firms. The rest of the accruals (87,4%) are discretionary accruals. For abnormal cash flow from operations the mean adjusted R-squared is 23,5%. This means that on average 76,5% of the total cash flow from operations are abnormal cash flows. The adjusted R-squared of abnormal discretionary expenses is 21,6%, which indicates that on average 78,4% of the total discretionary expenses are abnormal discretionary expenses. Finally table 5 shows that the adjusted R-squared of total production costs is 81,6%. So on average 18,6% of total production costs are abnormal production costs.
The adjusted R-squared of the earnings management measures found in this study correspond to those found by Roychowdhury (2006). He finds an adjusted R-squared of 45% for cash flows, 38% for discretionary expenses, 89% for production costs and for accruals an adjusted R-squared of 28%. Just like in my sample, the average adjusted R-squared is smallest for accruals, then discretionary expenses, cash flows and biggest for production costs.
34
Table 5
Adjusted R2 for earnings management regressions
Observations Mean St. deviation Min Max
ABEM 5.681 0,126 0,219 - 0,483 0,911
ABN_CFO 5.681 0,235 0,250 - 0,463 0,907
ABN_DEXP 5.681 0,216 0,227 - 0,285 0,949
ABN_PROD 5.681 0,816 0,171 - 0,043 0,998
5.2 Univariate analysis
In the next table, the descriptive statistics from table 4 are split up for when an audit committee is totally independent (ACIND=1) and when they are not (ACIND=0). This allows us to do a first univariate analysis to test the hypotheses. This is done by performing a t-test, which tests whether or not the difference in the means is significantly different from 0. In the last column of table 6 the p-value of the alternative hypothesis, which states that the difference in the mean is not equal to 0, is reflected.
First, table 6 shows us that when the audit committee is not independent, the mean for accrual-based earnings management is positive whereas the mean is negative when the audit committee is independent. However the difference in the means is not significant, meaning that in firms with an independent audit committee there no difference in accrual-based earnings management than when the audit committee is not independent. So hypothesis 1 cannot be supported based on this analysis. Concerning real earnings management, table 6 shows that an independent audit committee has a positive mean of abnormal cash flow from operations and an audit committee which is not independent has a negative mean. The difference in the means is significantly different from 0 at a 1% significance level (p = 0.0018). This confirms hypothesis 2 that audit committee independence is associated with real earnings management for the first measure of real earnings management. For real earnings management through abnormal discretionary expenses, table 6 indicates that an independent audit committee has a positive mean of abnormal discretionary expenses and a not independent audit committee has a negative abnormal discretionary expenses mean. The difference in the means is significantly different from 0 at a 10% significance level (p = 0.0707). This reconfirms hypothesis 2 that audit committee independence is associated with real earnings management. Finally table 6 shows that an independent audit committee has a negative mean of abnormal production costs whereas observations that do not have an independent audit committee have a positive mean. The difference in the means is also significantly different from 0 at a 1% significance level (p
