The$effect$of$audit$quality$on$the$relationship$between$equity6based$
executive$compensation$and$earnings$management
Student Name: Marco Wallenburg Student number: 10003653
Date of final version: 19-06-2015 Word count: 15,254
MSc Accountancy & Control, variant Accountancy Amsterdam Business School
Faculty of Economics and Business, University of Amsterdam Supervisor: dr P. Kroos
Statement of Originality
This document is written by student Marco Wallenburg 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
ABSTRACT
The purpose of this thesis is to examine the relationship between equity-based executive compensation and accrual-based earnings management, and whether audit quality has a negative effect on this relationship. I will contribute to existing theory by specifically looking at the potential moderating effect of audit quality. I also contribute by adopting a
comprehensive perspective on audit quality by looking at different proxies. Moreover, I incorporate the commonality of these different proxies for audit quality by means of a factor analysis. Based on a sample of 9,758 firm-year observations I am able to determine that there is a positive relationship between equity-based compensation and earnings management. Furthermore, using three different proxies for audit quality I find that audit quality has no negative effect on this positive relationship. A supplemental analysis has been performed on all regression analyses to validate the obtained results. Based on the performance-matched model of Kothari et al. (2005) I document that high audit quality in the form of hiring a big4 audit firm does have a moderating effect on the relationship between equity-based
compensation and earnings management. In addition to the main results I performed a factor analysis to incorporate the commonality of the different proxies for audit quality. This factor analysis documents the same results as the main analyses.
TABLE OF CONTENTS
1. INTRODUCTION... 6
1.1 Background ... 6
1.2 Research question... 7
1.3 Motivation and contribution... 8
1.4 Structure ... 8
2. LITERATURE REVIEW AND HYPOTHESES ... 9
2.1 Agency theory ... 9
2.2 Earnings management ... 11
2.2.1 Definition...11
2.2.2 Types of earnings management...12
2.2.3 Incentives for earnings management...13
2.3 Executive compensation... 14
2.3.1 Components of an executive compensation plan...14
2.3.2 Relationship between equity-based compensation and earnings management...17
2.4 Audit quality... 18
2.4.1 Introduction...18
2.4.2 Determinants of audit quality...19
2.4.3 Relationship between audit quality and earnings management...22
3. RESEARCH METHODOLOGY... 25 3.1 Sample selection... 25 3.2 Regression models... 26 3.2.1 Main model...26 3.2.2 Auxiliary model...27 3.2.3 Control variables...28 4. EMPIRICAL RESULTS ... 30 4.1 Descriptive statistics... 30 4.2 Main results ... 34 4.3 Supplemental analyses ... 38
4.3.1 Performance-Matched discretionary accruals...38
5. CONCLUSION ... 42 REFERENCES... 44 APPENDICES... 49 !
1. INTRODUCTION
1.1 Background
$Following the accounting manipulation scandals of the late 1990s, regulators and investors started raising their concerns that equity-based compensation could lead to the manipulation of the firms’ reported earnings, reducing the usefulness of financial reporting (Cheng and Warfield, 2005). It is argued that tying management compensation to the stock price increases the incentives for managers to use earnings management to affect the firms’ stock price, and in turn, managers’ wealth (Cheng & Warfield, 2005; Bergstresser & Philippon, 2006). Despite these developments, equity-based compensation was initially seen as a possible solution of the agency problem caused by the separation of ownership and control. Due to the separation of ownership and control, managers’ wealth is not affected by the value of the company. This may provide incentives for managers to act in ways that, while privately beneficial, reduce the value of the firm as a whole (Bergstresser & Philippon, 2006). To align the incentives of the managers with the interest of the shareholders, firms began to design compensation packages that provide incentives for managers to act appropriately by selecting those actions that are in the best interest of the shareholders. Firms accomplished this, for example by granting stock options to managers. In this way the compensation of managers is tied to the performance of the firm. Stock options become more valuable with higher share prices, and therefore encourage managers to take actions that increase share price, thus making shareholders better off (Meek, Rao & Skousen, 2007). However, these changes caused unintended consequences. It has been suggested that tying management compensation to the stock price increases the incentives for managers to manipulate their firms’ reported earnings to inflate the stock price (Cheng & Warfield, 2005; Bergstresser & Philippon, 2006).
Various articles have found that there is a positive relationship between equity-based compensation and earnings management.1 Gao and Shrieves (2002) examined how the components of compensation influence earnings management behavior. Their empirical analysis shows, amongst others, that the amounts of stock options are positively related to earnings management, meaning that compensation contract design does influence earnings management (Gao & Shrieves, 2002). Moreover, Cheng and Warfield (2005) find that !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
1!Earnings management is defined by Scott (2011) as the choice that managers make of accounting policies or actions that have an effect on earnings, whereby managers have the aim of achieving some specific reported objective (p. 403). !
managers with high equity incentives are more likely to sell shares in the future. To increase the value of these shares managers engage in income-increasing earnings management (p. 441). They find that managers with high equity incentives are more likely to report earnings that meet or just beat analysts’ forecasts (Cheng and Warfield (2005). Overall, prior literature states that the use of earnings management is more evident at firms where the executives have been awarded greater equity-based incentive compensation (Aboody & Kasznik, 2000; Gao & Shrieves, 2002; Cheng & Warfield, 2005; Bergstresser & Philippon, 2006).
However, there are factors that have a moderating effect on the positive relationship between equity based compensation and earnings management (e.g., Dechow, Sloan & Sweeney, 1996; Meek, Rao & Skousen, 2007). One factor that is being mentioned by the literature as a possible moderating factor is audit quality. A number of studies found that the presence of an external auditor functions as a constraint for management to engage in earnings management, and that the effectiveness of this constraint depends on the quality of the audit (Becker et al., 1998; Brown and Pinello, 2007; Prawitt, Smith and Wood, 2009). For instance, Becker et al. (1998) examined the effect of audit quality on earnings management through discretionary accruals. Their findings demonstrate a direct relationship between audit quality and earnings management. They describe that high quality audits ensure that
misreporting is quicker detected and revealed, in comparison to low quality audits. As a result, audits of high quality operate as an effective restraint to earnings management. In this thesis I will examine whether higher audit quality will function as a constraint and will result in fewer opportunities for executives to engage in opportunistic earnings management to increase their equity-based compensation.
1.2 Research Question
The purpose of this study is to examine the relationship between equity based executive compensation and accrual-based earnings management, and whether audit quality will have a negative effect on this positive relationship. Based on the information that is provided in the background section the following research question can be formulated:
“Does audit quality have a negative effect on the relationship between equity-based executive compensation and earnings management?”
1.3 Motivation and contribution
Many studies examined the positive relationship between equity-based incentives and
earnings management (Aboody & Kasznik, 2000; Gao & Shrieves, 2002; Cheng & Warfield, 2005; Bergstresser & Philippon, 2006). Relatively few studies have looked into potential moderators of this positive relationship (e.g., Dechow et al., 1996; Meek, Rao & Skousen, 2007). Meek et al. (2007) found firm size to be a significant mitigating factor. The authors describe that large firms can be associated with less information asymmetry, stronger governance structures and, most importantly, stronger external monitoring. Their line of argument is that large firms are more often audited by large audit firms (big4), and are therefore associated with higher audit quality. They suggest that the stock option incentive effect to manage earnings is lower in large firms partially because of this higher audit quality of big4 audit firms (DeAngelo, 1981; Becker, Defond, Jiambalvo and Subramanyam, 1998; Meek et al., 2007).
The scientific contribution of this paper is that I specifically look at audit quality. I also contribute by adopting a comprehensive perspective on audit quality by looking at different proxies. To the extent that different proxies focus on different dimensions of audit quality and/or contain some measurement error, I also incorporate the commonality of these different proxies for audit quality by means of a factor analysis.
Lastly, as a consequence of the inaccurate financial information created by managers due to opportunistic earnings management, shareholders and other stakeholders may make non–optimal financial and operational decisions. The implication of this study is that if audit quality increases, managers have fewer opportunities to engage in earnings management. Moreover, a better understanding of the potential role of audit quality in addressing the accrual-based earnings management induced by managers’ equity incentives may also have policy implications as prior literature documented that managers that manage earnings have a higher likelihood to move towards the slippery slope to fraud (Schrand and Zechman, 2012).
1.4 Structure
The structure of this paper is as follows. In section two the literature review and hypotheses development will be discussed. The research methodology is described in section three. This section will consist of a description of the sample selection, empirical models, variable
measurements, and control variables. The empirical results are presented in the fourth section. Lastly, section five will present the conclusion and some research limitations.
2. LITERATURE REVIEW AND HYPOTHESES
2.1 Agency theory
The agency theory offers an explanation of why managers do not always act in the best interests of stakeholders. The theory describes the structure of the economic exchange between two parties: the principal and the agent. In this relationship one party (the principal) delegates the decision-making authority to the other party (the agent). The principal will compensate the agent, who performs the task. Jensen and Meckling (1976) were one of the first who provided a definition of the agency relationship:
“We define an agency relationship as a contract under which one or more persons (the principal) engage another person (the agent) to perform some service on their behalf, which involves delegating some decision-making authority to the agent. If both parties to the relationship are utility maximizers there is good reason to believe that the agent will not always act in the best interests of the principal” (p. 308).
The agency theory is concerned with resolving two problems than can occur in the agency relationship: the agency problem and the problem of risk sharing (Eisenhardt, 1985; Gottschalk, 2010).Information asymmetry is often held as the cause of many agency problems. Asymmetric information implies that the agent generally has access to more information than the principal does, which enables the agent to function in their own self-interest rather than in the best self-interests of the principal. For example, managers know better than shareholders whether they are capable of meeting the objectives of the shareholders. There are two possible forms of asymmetric information, depending on the information that is known at the moment of decision-making respectively, the moral hazard problem and the problem of adverse selection (Jensen & Meckling, 1976; Eisenhardt, 1988; Scott, 2011).2 !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
2!There is also the problem of risk sharing that arises when the principal and agent have different opinions towards risk (Eisenhardt, 1988; Scott, 2011). The agent is typically more risk-averse than the principal. In an arrangement between the principal and the agent, the former bears much of the risk. Moreover, if the principal is risk-neutral, he or she bears all of the risk (Aumann & Hart, 1994). For example, the executives’ equity
compensation can be linked to a single firm. In this case the agent (manager) is not well diversified, so he or she will take less risk. The principal (investor), however, is able to invest in one specific firm or can diversify by investing a small amount in several firms. The principal will choose the latter, and is therefore willing to take more risk.
Moral hazard results from the inability of the principal to monitor an agent’s actions. It is often difficult or expensive for the principal to verify what the agent is actually doing. Therefore, the principal is usually not able to completely monitor every action of the agent and determine whether the agent has behaved appropriately (Picard, 1987; Eisenhard, 1988; Scott, 2011). In this case, agents may act in their own best interest at the expense of the interests of the principals. The agents take opportunistic actions reflecting the managers’ objectives instead of the objectives of the principles (Hölmstrom, 1979). Adverse selection exists because the principal has imperfect knowledge of the agent’s quality and skills, the efficiency of his actions (Zou, 1992). In this instance, the agent has an information advantage, which he or she can exploit at the expense of the principal.
To resolve the agency problem companies started to develop a compensation structure that provides incentives for managers (the agents) to act appropriately by selecting those actions that are in the best interest of the shareholders (the principals). One way to encourage the alignment of interests of the managers and shareholders is by giving managers stock options, a form of equity-based compensation. By granting managers stock and/or equity the compensation of managers is tied to the performance of the firm. Stock (options) becomes more valuable with higher share prices, and therefore encourage managers to take actions that increase share price, thus making shareholders better off (Meek, Rao & Skousen, 2007, p. 305). Equity based compensation can therefore be seen as a solution to the agency problem (Eisenhard, 1988; Meek, Rao & Skousen, 2007).3
However, the incentives provided in the compensation plan may have undesirable effects. Cheng and Warfield (2005) argue that equity incentives can lead managers to focus on short-term stock prices, thereby leading to incentives for earnings management (p. 442). According to Beck (2003) stock options provide management with incentives to manipulate earnings to inflate the stock price. For example, managers who want to boost earnings have incentives to reduce R&D (research and development) spending. Managers may expect that R&D spending negatively impact the short-term stock prices, due to the decrease in current earnings (Dechow and Sloan, 1991; Baber, Fairfield and Haggard, 1991; Cheng, 2004;
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! !
3!The stock price development is also used as a measure to assess whether the abilities of the agent still match the requirements of the firm. A lower stock price is found to be associated with a higher likelihood of dismissal as the agent has less incentive to perform well.!!
Atchinson II, 2008).4 In conclusion, instead of creating wealth for the firm, stock options may encourage executives to manage for the short term. They engage in earnings management to increase the stock price that is, their potential total compensation. As a result, equity-based compensation is an imperfect solution of the agency problem because of earnings
management (Beck, 2003; Cheng and Warfield, 2005; Meek, Rao & Skousen, 2007).
2.2 Earnings management
2.2.1 DefinitionIn this section the theory of earnings management will be explained. The academic literature provides various, and somewhat different, definitions for earnings management. I will discuss three definitions that are often used to define earnings management.One of the first papers that defines earnings management is “Commentary on Earnings Management” written by Katherine Schipper (1989). She uses the following definition:
By "earnings management" I really mean "disclosure management" in the sense of a purposeful intervention in the external financial reporting process, with the intent of obtaining some private gain (Schipper, 1989, p. 92).
Healy and Wahlen (1999) suggest that earnings management takes place when managers use judgment in financial reporting and in structuring transactions to modify financial reports. According to Healy and Wahlen managers do this either to mislead stakeholders about the true economic value of the firm or to influence contractual outcomes that depend on reported accounting numbers. For example, certain compensation structures partially depend on the reported earnings or share price (stock options). This provides incentives for managers to alter the reported accounting numbers to maximize their total compensation. Lastly, Merchant and Van der Stede (2007) define earnings management as “any action that changes reported earnings while providing no real economic advantage to the organization and, sometimes, actually causing harm” (p.185). This is equivalent to what Xie et al. (2003) state, that earnings management practices can influence the firm value in a negative way.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
4!Investments in R&D have a negative impact on the current earnings as R&D spending is instantly expensed (Cheng, 2004). !
In conclusion, these various definitions indicate that managers purposely engage in earnings management by the intentional manipulation of accounting information. Moreover, the primary objective of managers to engage in earnings management is to achieve some specific reported objective or to increase managers’ wealth. However, the authors argue that this kind of behavior misleads stakeholders, which can negatively affect the value of the firm (Merchant & Van der Stede, 2012).
2.2.2 Types of earnings management
There are two forms of earnings management: accrual-based earnings management and real-based earnings management (Richardson; 2000).
Real earnings management can be described as the attempt of managers to alter reported earnings in a particular direction by making suboptimal decisions on the timing and structure of the underlying business activities (Healy and Wahlen, 1999; Roychowdhury, 2006; Taylor and Xu, 2010; Zang, 2012). According to Roychowdhury (2006) managers engage in real-based earnings management to mislead stakeholders into believing certain financial reporting goals have been met in the normal course of operations (p. 337). Managers achieve this by reducing spending on R&D, advertising and maintenance, or by delaying investments (Graham, Harvey, and Rajgopal, 2005). These decisions increase short-term earnings, but will result in a negative net present value in the long run (Scott, 2011). Real-based earnings management differs from accrual-Real-based earnings management as it has direct cash flows effects and it involves real economic decisions. Executives prefer to manage earnings through real activities instead of accruals, because real activities are easier to camouflage as normal activities. Real-based earnings management is therefore less likely to draw auditor or regulatory scrutiny (Cohen and Zarowin, 2008).
Accrual-based earnings management is achieved by changing the accounting methods or estimates used when presenting a given transaction in the financial statements (Zang. 2011, p. 676). Managers are able to opportunistically manage earnings because various estimations and judgments go into the process of preparing financial statements. Companies can increase or decrease current earnings by creating accruals. These accruals can be distinguished by non-discretionary and non-discretionary accruals. Non-non-discretionary accruals result from transactions in the current period, which are regular for the firm. It concerns factors such as the
performance level, industry conventions and macro-economic events. Discretionary accruals arise from transactions or accounting treatments executed with the purpose of manipulating earnings (Cohen et al., 2008; Ising, 2013). According to Scott (2011) examples of altering
earnings by using discretionary accruals are: increasing amortization charges, record generous provisions for doubtful accounts, record excessive liabilities for product guarantees and obsolescence of inventories (p. 292). Manipulating reported accounting numbers with accrual-based earnings management is in particular of interest for this thesis. Therefore, when this thesis mentions earnings management it refers to accrual-based earnings management. 2.2.3 Incentives for earnings management
The literature provides a variety of reasons for why managers engage in earnings
management. Graham, Harvey and Rajgopal (2005) have conducted a survey and interview among 400 executives, and have determined several factors that drive earnings management. According to the authors, meeting or exceeding benchmarks is one of the main reasons for managers to engage in earnings management. They state that managers are even willing to make sacrifices in the (long-term) economic value of the firm in order to meet the earnings expectations of analysts and investors. Hitting an earnings target is important for several reasons, for example the effect on the stock price (Graham et al., 2005; Dichev, Graham, Harvey and Rajgopal, 2013). Graham et al. (2005) argue that managers believe that hitting earnings benchmarks builds credibility with the market and helps to maintain or increase the firm’s stock price (p. 5). When managers don't meet an earnings benchmark, the markets’ perception of the firms’ future value will be negatively influenced, which decreases the firm’s stock price. Other reasons for meeting or exceeding benchmarks include, higher managers’ bonus compensation, enhancing external reputation and to avoid adverse career consequences (Graham et al., 2005; Dichev et al., 2013).
Another motive for management to exercise their accounting discretion is to reduce the probability of violating a debt covenant (Watts and Zimmerman, 1990; Graham et al., 2005). More specifically, when high leverage firms are close to the violation of debt
covenants, managers are more likely to shift income from future periods to the current period to avoid those violations, and therefore mitigate potential constraints on their behavior (Press and Weintrop, 1990; Sweeney, 1994; Warfield et al., 1995; Scott, 2011). Furthermore, research conducted by Dichev et al. (2013) shows that executives manage earnings due to their preference for smooth earnings. This preference can be explained by various reasons. For example, smoother earnings make it easier for analysts and investors to predict future earnings, which increase the firms’ stock price (Graham et al., 2005, p. 66). Moreover,
investors perceive smooth earnings as less risky, and it also reassures suppliers and customers that the business is stable (Graham et al., 2005; Dichev et al., 2013).
Lastly, prior literature argues that management uses their reporting discretion to signal private information regarding the firm’s future growth prospects through accruals (Louis & Robinson, 2005). To explain in more detail, a manager can be unable to meet analysts’ forecast when an order gets canceled. However, if the manager is aware that various orders are in process, the manager may decide to manage earnings to signal this information to the market. If the manager does not engage in earnings management, the market will only have knowledge regarding the forecast that is not achieved, and will think that the firm is
underperforming. Prior literature shows that earnings management is usually viewed as opportunistic, because managers are often assumed to use their discretion to mislead investors or to obtain some private gain (Schipper, 1989; Healy & Wahlen, 1999). However, this example concerning signaling private information suggests that earnings management can also be used in a positive way: providing the market with information regarding the firm’s future growth prospects that is not yet recorded.
2.3 Executive compensation
2.3.1 Components of an executive compensation plan
Despite substantial heterogeneity in executive compensation structures across firms and industries, most executive compensation packages contain the following basic components: a base salary, an annual bonus tied to accounting performance, long-term incentive plans, stock options and restricted stock. Besides the previous components, executives often additionally receive contributions to defined,benefit pension plans, perquisites (perks) and severance payments (Murphy, 1999; Milkovich and Newman, 2004; Frydman and Jenter, 2010). Below follows a description of the most common components of an executive compensation plan:
Base Salary
Base salary is the standard wage paid to an executive for the core role and responsibilities of the day-to-day running of the organization. Only about 20 percent of an executive’s total compensation consists of base salary; the remainder is made up of incentives based on the firms’ corporate performance (Ellig, 2002; Valenti, 2013). The reason is that if the firm is performing well and the shareholders are making money, the executives should share in that success. Moreover, a base salary can be considered to be a (free) risk premium as managers receive this with more certainty than, for example, a bonus that is dependent on managers’
performance. A downside to base salaries is that executives get rewarded when the company performs well, but still receive the same reward when the company performs poorly. In other words, the level of salary does not fluctuate with performance. Therefore, base salaries, on their own, offer little incentives for executives to work hard and efficient (Murphy, 1998). Short-term (annual) incentives
Annual bonuses often play a major role in executive compensation and are primarily designed to motivate better performance and reward executives for meeting annual performance
objectives. Executive bonus plans can be categorized in terms of three components:
performance measures, performance standards and the pay-performance relation. The figure below illustrates these components. Under a typical bonus plan, no bonus is paid until a lower performance threshold is achieved, and a “hurdle bonus” is paid at this lower performance threshold. Target bonuses are paid for achieving the performance standard, and the bonus is usually capped at an upper performance threshold. The range between the thresholds and cap is the “incentive zone”, indicating the range of performance realization where an increase in performance is associated with an increase in the bonus (Murphy, 1999; Murphy and Jensen, 2011).
Long-term incentive plans
Long-term incentives are structured to reward executives for achievement of the company’s strategic objectives. Moreover, it functions as a reward system for executives for maximizing shareholder value and for staying with the company.5 It may be provided in the form of stock-based compensation, such as stock options, restricted stock, performance shares, or cash. Generally, long-term incentives are a combination of types of equity and may include a cash component (Lambert and Larcker, 2004).
Stock options
Stock options provide managers with the right to buy or sell shares in the future at a pre-specified exercise (or strike) price for a pre-pre-specified term. After this given date, the option ceases to exist. The price at which options are exercised is usually the price of the firms’ stock on the date the options are granted. When a firm performs well and the stock price increases above the exercise price, the stock options contain value and therefore will reward the executive for his/her part in the company’s success (Murphy, 1998). Besides incentives to increase performance, stock options are also used to provide risk-taking incentives to CEO’s. That is, managers can benefit from value-increasing projects, but do suffer less when projects ultimately turn out to be only value-destroying. Therefore, stock options provide managers with incentives to adopt more risky projects characterized by more extreme positive and negative outcomes (Cohen, Hall and Viceira, 2000; Rajgopal and Shevlin, 2002; Sanders and Hambrick, 2007).
Restricted Stock
An executive who receives a restricted stock grant typically pays nothing for the shares. The shares however are subject to restrictions on transferability and to risk of forfeiture under certain conditions, for example failure to meet performance goals or termination of
employment. This means that executives are prohibited from selling the shares for a specified period of time (vesting period), often a period of three or four years (Cheng and Warfield, 2005). So, when the executive decides to leave the firm or is dismissed during the vesting period the executive will forfeit his shares. This makes it costly for the executive to leave the !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
5!Rewarding the executives with long-term incentives will motivate them to stay at the company. If the firm keeps their well performing executives satisfied the firm will save money in the long run as recruiting and training new workers is costly (Chambers, Foulon, Handfield-Jones, Hankin & Michaels, 1998).
firm, as he/she will lose the value of his/her shares, which provides an incentive for the executive to stay at the firm. When the conditions are met, the shares are no longer restricted, and become transferable (Foulkes, 1991; Lipman and Hall, 2008). Moreover, from the
previous discussed compensation components, the stock options and restricted stock are directly tied to the share price. When this thesis mentions equity-based compensation, these two components are referred to.
2.3.2 Relationship between equity based compensation and earnings management
Prior literature has extensively examined the conflicts between managers and shareholders. It is argued that a strong relation between compensation and firm performance is important to resolve the agency problems created by the separation of ownership and control (e.g., Jensen and Murphy, 1990; Peng and Röell, 2008). Jensen and Meckling (1976) argue that to reduce the agency costs the compensation of the manager should be linked to shareholder wealth by rewarding managers with options or shares of stock. Theoretically, as managers own more shares, they are more likely to act in the interests of shareholders (Core and Guay 1999). Therefore, according to Burns and Kedia (2006) options can be viewed as an influential means to align the interests of managers and shareholders. Consistent with this perspective, prior research argues that granting options to managers is associated with firm value
maximization (Hanlon, Rajgopal and Shevlin, 2003).
However, the assumption that options are used only to align the interests of
management with those of shareholders has come under scrutiny (Burns and Kedia 2006). Due to the increased use of equity-based compensation, the compensation of managers has become more dependent on performance incentives (Kedia, 2003). This is because, the value of stock options are related to the share price of the company. As a result, the wealth of managers is increasingly sensitive to the performance of the firm. Managers therefore have the incentive to focus their attention on the share price. This focus on share price has the intended effect of making managers choose those actions that maximize firm and shareholder value (Kedia, 2003, p. 2).
However, Cheng and Warfield (2005) state that it may also have the unintended effect of inducing managers to focus on short-term stock prices, thereby leading to incentives for earnings management. If managing earnings affect stock prices, then managers with equity-based compensation might have an incentive to maximize their wealth through accounting choices (Burns and Kedia, 2006). Jensen (2005) supports this finding by stating that “equity-based compensation can encourage managers to increase short-term stock prices in order to
benefit from subsequently selling shares of their own firms' stock” (p. 442). According to Bange and Bondt (1998) many investors give too much weight to reported accounting
earnings. This provides incentives for managers to inflate the current earnings, as the value of their stock (options) may depend on it. For example, managers may be inclined to reduce R&D spending to meet earnings goals and to signal firm value (Bushee, 1998). Thus, managers with high equity incentives could benefit from earnings management with the objective of keeping stock prices high and increasing the value of the shares in order to increase their total compensation (Cheng and Warfield, 2005).
Several studies have investigated this possible relationship between equity-based compensation and earnings management. Gao and Shrieves (2002) have examined how the components of compensation influence earnings management behavior. Their empirical analysis shows that the amounts of stock options and bonuses are positively related to earnings management (Gao & Shrieves, 2002). Moreover, Cheng and Warfield (2005) find similar results. Their findings indicate that managers with high equity incentives are more likely to engage in earnings management to increase the value of the shares (p. 441). Lastly, Bergstresser and Philippon (2006) provide evidence that the use of discretionary accruals to manipulate reported earnings is more evident at firms where the executives have been
awarded greater equity-based incentive compensation. In addition, they state that during years of high accruals, executives exercise unusually large amounts of options and sell large
quantities of shares (Bergstresser and Philippon, 2006, p. 511). Overall, prior literature has found that there is a positive relationship between equity-based compensation and earnings management (e.g., Gao & Shrieves, 2002; Cheng & Warfield, 2005; Bergstresser &
Philippon, 2006; Meek, Rao & Skousen, 2007).
Consequently, I formulate the following hypothesis regarding the relation between equity based compensation and earnings management:
2.4 Audit quality
2.4.1 IntroductionAn audit is the examination of the financial report of an organization by someone independent of that organization. The purpose of an audit is to form a view on whether the information presented in the financial report reflects the true financial position of that organization, and whether the financial statements are free from error. Audit services are demanded as monitoring devices because of the potential conflicts of interest between owners and managers (Watts, 1977; Watts and Zimmerman, 1981; Gray and Manson, 2007).
The quality of audit services is defined to be "the market-assessed joint probability that a given auditor will both (a) discover a breach in the client's accounting system and (b) report the breach” (DeAngelo, 1981, p. 186). This definition underlines two aspects of audit quality that are of importance. First, the competence of the auditor that determines how likely it is that a misstatement will be detected. Second, the independence of the auditor that
determines what the auditor is likely to do about a detected misstatement (DeAngelo, 1981; Watts and Zimmerman 1981; Knechel, 2009). Thus, auditor independence depends not only on the ability of the auditor to identify problems, but also whether the auditor appropriately reports those problems.
Auditor independence is usually defined as “the ability to act with integrity and objectivity” (Watts and Zimmerman, 1981). Audit quality is believed to depend on factors that could affect the competence and independence of the auditor. Prior literature has found a variety of factors that could have an impact on the auditor’s competence or independence. These factors include the following: economic dependence on a client, magnitude of audit fees, provision of non-audit services, the length of auditor-client tenure, litigation, external regulation, audit firm size and industry specialization (Carson, Fargher, Geiger, Lennox, Raghunandan and Willekens, 2012).
2.4.2 Determinants of audit quality
Researchers have used various proxies in an attempt to assess the quality of audits and, in turn, determine whether a differential in audit quality exists. The literature provides several different indicators, for example: audit firm size, non-audit service fees, and auditor tenure. I will now discuss these three measures to determine audit quality in order to give more depth regarding the subject matter.
Auditor size
Numerous studies show that audit firm size and audit quality are positively associated (e.g., DeAngelo, 1981; Becker et al., 1998; Francis, 2004; Lawrence, Minutti-Meza and Zhang, 2011). Audits conducted by large audit firms (big4) are generally viewed as high quality, because they can be associated with higher levels of available resources and greater degree of personnel training and expertise (Bing, Huang, Li and Zhu, 2014). Large audit firms can invest more in training courses and other resources necessary to ensure that their auditors are competent, able to audit to a high standard, and are less likely to be compromised by actions of clients (Sawan & Alsaqqa 2012, p. 213). In addition, large firm size has found to be associated with a higher probability of error detection. O’Keefe and Westort (1992) and Lawrence et al. (2011) state that this can be explained due to the fact that personnel in larger firms are involved in more professional training programs. Hence, auditors in large audit firms are expected to have greater in-house expertise and technological knowledge than those in small firms, and to be more skilled in detecting material problems in the financial
statements (Sawan & Alsaqqa, 2012; Francis and Yu 2009).
According to DeAngelo (1981) large audit firms are (financially) less dependent on a specific client because they have a wide client base. As a result, the auditor has less incentive to behave opportunistically, which increases the perceived quality of the audit (p. 197). Moreover, large audit firms will less likely give in to the pressure of the client to issue an audit opinion that’s in their favor because they have a greater reputation to protect (Choi, Kim, Zang and Kim, 2010; Malihi, Mahdavikhou and Khotanlou, 2012). Furthermore, various studies document that large audit firms receive higher audit fees relative to small audit firms (Simunic, 1980). Francis (2004) argues that these higher audit fees can be associated with higher audit quality because more hours are put in the audit and/or the auditor has greater industry expertise.6 Lastly, prior literature has found that clients audited by larger audit firms have lower abnormal accruals and a lower likelihood of meeting benchmark earnings targets, which implies less aggressive earnings management behavior and therefore higher audit quality (Becker et al., 1998; Francis and Krishnan, 1999, Nelson et al., 2002; Francis, 2004; Francis and Yu 2009). In conclusion, prior literature demonstrates that there are numerous reasons why large audit firms can be associated with higher audit quality.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
6!Solomon et al. (1999) argue that auditors with industry expertise have a deeper knowledge than non-experts due to greaterexperience in that specific industry, which enables them to make more accurate audit judgments (Francis, 2004, p. 354). !
Auditor tenure
Multiple studies suggest that there is a positive relationship between auditor tenure and audit quality (Johnson, Khurana and Reynolds, 2002; Geiger and Raghunandan 2002; Myers, Myers and Omer, 2003; Gosh and Moon, 2005).7 According to Ghosh and Moon (2005) the independence of auditors and audit quality increase with longer tenure, because of improved auditor expertise from client-specific knowledge. Client-specific knowledge gained through years of on-the-job experience helps them to understand the client’s business processes and risks, and will make the auditors more competent (Lim & Tan, 2010). This allows the auditors to rely less on management estimates, which increases the likelihood that they will detect financial misstatements (Libby and Frederick, 1990; Johnson, Khurana and Reynolds, 2002; Myers et al., 2003).
Opponents express the belief that lengthy auditor tenure could be negatively associated with audit quality as the independence of auditors may be compromised. For example, Carey & Simnett (2006) state that long audit tenure will impair the independence of the auditor due to the development of personal relationships with the client, which negatively influences the auditors’ critical judgment. Moreover, it is argued that the auditor can become over familiar with the client and develop blind spots (Lim & Tan, 2010, p. 926). Furthermore, auditors may have economic incentives to become less independent. They may agree to the demands of the client in order to preserve the audit fees in the future (Lim & Tan, 2010). However, several studies provide evidence that shorter audit firm tenures tend to be associated with higher levels of earnings management and lower audit quality (Manry, Mock and Turner 2008). They argue that longer auditor tenure results in auditors placing greater constraints on extreme management decisions in the reporting of financial performance (Myers et al., 2003; Gosh and Moon, 2005). In fact, Geiger and Raghunandan (2002) show that there are significantly more audit reporting failures in the earlier years of the auditor-client relationship due to the lack of sufficient knowledge of new auditors regarding firm-specific risks (Gosh and Moon, 2005; Manry, Mock and Turner, 2008). This argument is consistent with research indicating that most of the audit failures by audit firms take place at newly acquired clients (Myers et al., 2003, p. 782). In conclusion, most studies show evidence consistent with that a longer auditor-client relationship is associated with higher audit quality.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
Non-audit services (fees)
External parties and regulators have expressed their belief that auditor independence can be impaired when non-audit services are provided (Parkash and Venable 1993; Gore, Pope and Singh, 2001; Gul et al., 2007). This concern is based on the premises that the provision of non-audit services increases the total amount of fees derived from one client, thereby further increasing the economic bond between the audit firm and the client, and hence will decrease the objectivity of the audit firm (Hoitash, Markelevich and Barragato, 2007; Svanstrom, 2012; DeAngelo 1981; Simunic 1984). However, some studies suggest that auditors have incentives to act independent. For example, Watts and Zimmerman (1983) conjecture that auditors have concerns with regard to the loss of reputation, and therefore will remain independent.
Prior literature has found mixed evidence regarding the effect of non-audit services on audit quality. Various studies have found a positive relation between non-audit fees and the level of earnings management (Gore et al., 2001; Frankel, Johnson and Nelson, 2002). For instance, Gore et al. (2001) found evidence that earnings management behavior is positively associated with the ratio of non-audit fees to total fees for the non-big4 audit firms. However, Ashbaugh, LaFond and Mayhew (2003) state that firms purchasing non-audit services don’t manage earnings to a greater extent than other firms. Moreover, Chung & Kallapur (2003) provide evidence that auditor independence is not impaired when non-audit services are provided.
A potential explanation for this mixed evidence could be that there is a non-linear relationship between non-audit services and audit quality. When the non-audit fees represent a small portion of the total fees auditor independence will not be impaired immediately.
However, as the non-audit fees increase and form a larger part of the total fees auditor independence can be impaired, which reduces the quality of the audit. So, the non-audit fees have to be sufficiently high to initiate a lack of independence. This potential explanation is consistent with the findings of Wines (1994) who investigated whether higher levels of non-audit services impair the independence of an non-auditor, and found a negative relationship between non-audit fees and audit quality. In conclusion, empirical research and literature is somewhat inconclusive on whether non-audit services actually impair audit quality. However, most studies show evidence consistent with auditor independence impairment when the amount of non-audit fees is large.
2.4.3 Relationship between audit quality and earnings management
Prior literature found that higher audit quality functions as a constraint for management to engage in earnings management (Becker et al., 1998; Prawitt, Smith and Wood, 2009). For example, Becker et al. (1998) examined the effect of audit quality on earnings management through discretionary accruals. The authors describe that high audit quality ensures that misreporting is quicker detected and revealed. As a result, audits of high quality operate as an effective restraint to earnings management (Becker et al., 1998).
Furthermore, prior research shows that various determinants of audit quality influence the level of earnings management. Firstly, several studies provide evidence that large audit firms (big 4) are associated with higher audit quality (e.g., DeAngelo, 1981; Francis, 2004; Lawrence et al., 2011). Large audit firms are associated with higher levels of available resources and a greater degree of personnel training and industry expertise (Bing, Huang, Li and Zhu, 2014). They are also (financially) less dependent on a specific client because they have a wide client base (DeAngelo, 1981). More importantly, prior literature has found that clients audited by larger audit firms have lower abnormal accruals and a lower likelihood of meeting benchmark earnings targets, which implies less aggressive earnings management behavior and therefore higher audit quality (e.g. Nelson et al., 2002; Francis & Yu 2009). Secondly, multiple studies show that there is a positive relationship between auditor tenure and audit quality (Johnson et al., 2002; Myers et al., 2003; Gosh & Moon, 2005). The independence of auditors and audit quality increase with longer tenure, because of improved auditor expertise from client-specific knowledge (Lim & Tan, 2010). Furthermore, short audit tenure tends to be associated with higher levels of earnings management and lower audit quality due to the lack of sufficient knowledge of new auditors regarding firm-specific risks (Gosh & Moon, 2005; Manry, Mock and Turner 2008). Lastly, prior research shows that non-audit fees are negatively related with non-audit quality as non-non-audit services increase the
economic bond between the audit firm and the client, and hence will decrease the independence of the auditor (DeAngelo 1981; Hoitash et al., 2007; Svanstrom, 2012).
Moreover, various studies have found a positive relation between non-audit fees and the level of earnings management (Gore et al., 2001; Frankel, 2002). For instance, Frankel et al. (2002) show that non-audit fees are positively associated with small earnings surprises and the magnitude of discretionary accruals.
In conclusion, previous studies have found a direct relationship between audit quality and earnings management. High audit quality due to large audit firms and/or long auditor tenure has found to be associated with less earnings management behavior. In addition,
studies demonstrate that higher levels of non-audit services are negatively related with audit quality, and incentivize earnings management behavior. Meek et al. (2007) have described, among other things, that the relationship between equity-based compensation and earnings management is weaker in larger firms due to the associated higher audit quality. In addition, several studies demonstrate that high audit quality ensures that earnings management is quicker detected and revealed, and therefore functions as a constraint for management to engage in earnings management. Altogether, I expect that higher audit quality will result in fewer opportunities for executives to engage in opportunistic earnings management to increase their equity-based compensation. The above discussion leads to the following hypothesis:
H2: Audit quality has a negative effect on the positive relationship between equity based executive compensation and earnings management.
3. RESEARCH METHODOLOGY
3.1 Sample selection
I select all firms covered by the Execucomp database from 1992 to 2006, which provides me with an initial sample of 19,470 firm-year observations. The Execucomp database roughly covers the S&P 1500 firms. From the Execucomp database, I select variables on executive compensation. From Fundamentals Annual, I select data on balance sheet and income statement items. The AuditAnalytics database will be used to retrieve data regarding non-audit services.8I exclude financial intuitions with SIC codes between 6000 and 6999 since banks and financial institutions are subject to different accounting regulations. This leads to the exclusion of 3,441 firm-year observations. Moreover, firms in the utilities industry with SIC codes between 4400 and 4999 are as well eliminated because executives might have different motivations to manage earnings (Becker et al., 1998; Burgstahler and Eames, 2003; Cheng & Warfield, 2005). This results to the exclusion of 2,160 firm-year observations. Furthermore, dropping duplicates (1,300) and missing data on executive compensation (2,567) results in a sample of 10,002 firm-year observations. After merging Execucomp data with Compustat data the sample size is reduced to 9,974 firm-year observations. Lastly, missing data on the absolute value of the discretionary accruals (101) and control variables (115) leads to the exclusion of 216 observations. This yields a final sample of 9,758 firm-year observations derived from 1,665 firms. I winsorized all variables at the 1st and 99th
percentiles of their distribution to mitigate the potential impact of outliers.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
8!The AuditAnalytics database will be used to retrieve data regarding the non-audit fees. For the associated regression analysis a customized sample will be applied, this will be further explained in the empirical results section. !
3.2 Regression models
3.2.1 Main model
The following regression model will be used to test for the first hypothesis:
DA = β0 + β1Equity + Controls + ε
I will use unsigned discretionary accruals (DA) as a proxy to measure the amount of accrual-based earnings management estimated by mean of a modified Jones model (1991). “Equity” represents the amount of stock options and restricted stock (equity-based compensation) the executives receive. The fraction of equity-based compensation as a percentage of total compensation will function as a proxy to measure the degree of equity incentives. β1
describes the association between equity-based compensation and earnings management. On the basis of my first hypothesis, I expect that β1>0.
The following regression model will be used to test for the second hypothesis:
DA = β0 + β1Equity + β2AQ + β3Equity*AQ + Controls+ ε
Three different proxies are employed to define audit quality: auditor size (BIGN), auditor tenure (AUDTEN) and non-audit services (NAS). Prior research has found audit firm size to proxy for audit quality. The dummy variable is equal to one if the auditor is a big4 auditor, and zero if otherwise (DeAngelo, 1981; Becker et al., 1998; Francis, 2004). Furthermore, a number of studies have found that a shorter audit tenure year is a viable proxy to measure low audit quality (Johnson et al., 2002; Geiger and Raghunandan, 2002; Myers et al., 2003). Prior research by Johnson et al. (2002) found that a dummy variable that is equal to one if the audit tenure is lower than four years, zero otherwise, can be associated with significantly lower audit quality. Lastly, the effect of non-audit services on audit quality could be measured by calculating the ratio of non-audit fees to the total audit fees (e.g., Gore et al., 2001; Frankel, Johnson and Nelson, 2002; Ferdinand, Gul, Bikki, Jaggi and Krishnan, 2007).
The interaction variable has been added to the first empirical model to test for the second hypothesis. β3 represents how audit quality moderates the relationship between the equity incentives of executives and earnings management. Based on the second hypothesis I expect that β3<0.
3.2.2 Auxiliary model
Following prior literature, (e.g., Dechow, Sloan and Sweeney, 1995; Becker et al., 1998; Shuto, 2007; Cohen, Dey and Lys, 2008) I use unsigned discretionary accruals as a proxy to measure the amount of accrual-based earnings management within a firm. Dechow et al. (1995) evaluated the ability of alternative models to detect earnings management. They found that a modified model of Jones (1991) provided the most powerful tests of earnings
management. I will estimate the discretionary accruals by applying this model cross-sectionally for every two-digit SIC classification, with a minimum of 20 observations. The use of two-digit SIC levels controls for industry-wide differences in economic conditions that affect the total accruals.
As many prior studies, this study calculates discretionary accruals using the Modified Jones Model. The original model developed by Jones makes the faulty assumption that
managers have no discretion with regard to the revenues. For example, management could use their discretion to recognize revenues at the end of the year when the cash has not yet been received and it is highly questionable whether the revenues have been earned. Because management recognizes revenues in advance through discretion, this will automatically result in an increase in accounts receivable, and therefore an increase in discretionary accruals (Dechow et al., 1995). Therefore, Dechow et al. (1995) have developed a modified version of the Jones Model that resolves this inaccuracy. This modified model incorporates that changes in revenues are adjusted for the changes in receivables in the event period. This modification results in a more accurate estimation of earnings management (Dechow et al., 1995, p. 199). In this model the discretionary accruals (DA) are defined by the absolute value of the residual in the equation below, or put differently, the difference between total accruals (TA) and the non-discretionary accruals (Dechow et al., 1995; Cohen et al., 2008). The total accruals are defined as: TA = EBXIit – CFO. I calculated this in Compustat by taking the income before extraordinary items and deducting the operating activities net cash. After calculating the total accruals for each firm each year, estimation can be made of α1, α2, and α3 by using the
Ordinary Least Squares (OLS) regression. When the OLS regression has determined the estimations for α1, α2 and α3,the calculation can be made for the non-discretionary accruals,
Where,
i = Stands for company i
TAit = The total accruals for the year t
Ai,t-1 = The total assets in the past year
ΔREVit = Revenue in year t less revenues in year t-1
ΔRECit = Net receivables in year t less net receivables in year t-1 PPEit = The gross property
εit = The error term in year t
α1, α2 and α3 = coefficients in the linear model
3.2.3 Control variables
I added several control variables to my empirical model to control other factors that may affect the level of earnings management. First, the natural logarithm of total assets (SIZE) will be included as a proxy for the size of a company. Prior literature states that firm size and the level of earnings management are positively associated (Rangan, 1998; Barton and Simko, 2002; Nelson et al., 2002). For instance, Barton and Simko (2002) argue that large firms face more pressures to meet or beat the analysts' expectations. Moreover, Nelson et al. (2002) document that auditors are more likely to ignore earnings management attempts by large clients. Further, the theory suggests that managers of large firms are more likely to look at the possibilities of legal creative accounting to reduce potential political costs (Zmijewski and Hagerman, 1981; Warfield, Wild and Wild, 1995). Therefore, large firms may experience greater levels of earnings management.
The market-to-book ratio will be added to control for growth (GROWTH).
The market-to-book ratio has been used extensively in previous research to proxy for future growth opportunities (e.g. Smith and Watts, 1992; Cheng and Warfield, 2005; Meek et al., 2007). It is argued that firms with growth opportunities are less transparent, which creates more opportunities for managers to engage in earnings management (Meek et al., 2007). Therefore, growth may be positively related with the level of earnings management.
The return on assets (ROA) will be included to control for differences in performance. The return on assets is measured as net income divided by total assets (Cohen et al., 2008). Further, the leverage of the firm (LEV) will also be incorporated in the empirical model as a control variable (Cohen et al., 2008). Leverage is measured as total debt divided by total assets (Cheng and Warfield, 2005). The literature argues that there is a positive relationship between the leverage of a firm and the manipulation of earnings. For example, when high
leverage firms are close to the violation of debt covenants, managers are more likely to shift income from future periods to the current period to avoid those violations, and therefore mitigate potential constraints on their behavior (Press and Weintrop, 1990; Sweeney, 1994; Warfield et al., 1995; Scott, 2011).
Lastly, the loss of the firm (LOSS) will be included as a control variable. Various authors suggest that firms that reported a loss experience greater levels of earnings
management (DeFond and Subramanyam, 1998; Scott, 2011). It is argued that during periods of financial distress when a firm has to report a loss, managers can opt to report a large loss by using income decreasing discretionary accruals and take a bath and in this way shift income towards the future (DeFond and Subramanyam, 1998). This will increase the probability of future reporting profits (Scott, 2011). Moreover, managers can also decide to use income increasing discretionary accruals in order to avoid reporting a loss (Burgstahler and Dichev, 1997). Hence, I would expect that in times of poor financial health managers engage more in earnings management. This control variable will be measured by creating a dummy variable, which is equal to one when the firm reports a loss in a particular year. To conclude, I will incorporate industry and year dummy variables to control for year and industry effects.
4. EMPIRICAL RESULTS
4.1 Descriptive statistics
Table 1 illustrates the number of observations for each year. The distribution over the sample years shows that a majority of the observations can be attributed to the years 1997 till 2005. Moreover, the year 2006 has less observations compared to the other years.9Overall, the sample is reasonably well distributed over the years.
Table 1: Distribution of observations per year
In appendix B the sample distribution is illustrated by the two-digit SIC code industry (see appendix). The sample contains firms from 47 different industries. The most heavily represented industry is the two-digit SIC code 73 ‘Business Services’ with 1,162 firm-year !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
9 The independent variable “equity” is defined by the stock options plus the restricted stock, and dividing this amount by the total executive compensation. There were substantially more missing values regarding the restricted stock and stock options of executives in the year 2006. This explains why 2006 has less observations compared to the other years.
observations, representing 11.91 percent of the sample. In general, the sample composition seems to be relatively evenly distributed across industries.
Table 2 reports the descriptive statistics regarding the variables used in the empirical models. The mean of the dependent variable “discretionary accruals” (DA) is comparable with prior studies (Teoh, Welch & Wong, 1998; Reynolds, Deis Jr & Francis, 2004; Meek, Rao & Skousen, 2007). Regarding the independent variables the table indicates that the total compensation of executives consists for 41.1% out of equity compensation (Equity).
Furthermore, the table describes that of the total sample 84.5% of the observations are being audited by big4 audit firms (BIGN), and that 15.3% of the observations have a client-auditor relationship of less than four years (AUDTEN). Moreover, on average the total fees charged by the audit firm consist for 36.3% out of non-audit fees (NAS). With regard to the control variables the table shows that the market-to-book ratio (GROWTH) has a mean of 3.2, which indicates that the market value is significantly higher than the book value. Furthermore, leverage (LEV) has a mean of 0.18, meaning that 18% of total assets is financed by debt. Lastly, the mean of the control variable “loss” (LOSS) is 0.20, which implies that 20% of the sample shows a loss as seen over the total sample period.
Table 2: Descriptive statistics of variables
a
A customized sample will be applied for the regression analyses of the determinant “non audit services”. From the AuditAnalytics database, I select data from 2000 to 2006, which provides me with an initial sample of 86,360 firm-year observations. Dropping duplicates (3,790) and missing data on non-audit services (2,403) results in a sample of 80,167 firm-year observations. After merging AuditAnalytics data with Compustat data and Execucomp data the sample size is reduced to a final sample of 4,764 firm-year observations.
Table 3 presents the Pearson correlations matrix. To test the relationships between the independent variables a test for multicollinearity is performed.10 Multicollinearity can cause strange outcomes when studying the contribution of independent variables in the
understanding of the dependent variable (Matignon, 2005). The Pearson correlations matrix shows the correlations between the independent variables.11 The correlation coefficients illustrate how strong the relationship is between two variables. When the independent
variables are highly correlated the chance of multicollinearity is significant (Kennedy, 2003). Prior literature states that coefficients higher than 0.7 or lower than -0.7 will influence the reliability of the model (Dancey and Reidy, 2004). All the coefficients are within the range of -0.7 and 0.7.
As expected, equity-based compensation (Equity) is significantly positively correlated with earnings management (DA). The correlation coefficient between BIGN and DA is -0.017 and is significantly correlated at the 0.1 level, which indicates that there is a negative
association between the two variables. This is equivalent to prior literature stating that firms audited by big4 audit firms experience lower levels of discretionary accruals (e.g., Becker et al., 1998; Nelson et al., 2002; Francis, 2004). Furthermore, we can observe a significant positive correlation between auditor tenure (AUDTEN) and earnings management (DA), suggesting that a short auditor-client relationship leads to more earnings management (note that short audit tenure proxies for lower audit quality). This is consistent with the belief that audit quality increases with longer tenure due to improved auditor expertise from client-specific knowledge (Gosh and Moon, 2005; Lim & Tan, 2010). Moreover, the non-audit services (NAS) are not significantly correlated with the discretionary accruals. When examining the table further, SIZE is significantly positively correlated with both BIGN and NAS, and significantly negatively correlated with both DA and AUDTEN. This means that larger firms are more likely to be audited by a big4 audit firm, and that audit firms provide more non-audit services for large firms. In addition, large firms can be associated with lower levels of earnings management, and they less often have short auditor tenure. Lastly, Table 3 shows that LOSS is significantly positively correlated with DA.
!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
10!Multicollinearity is the extent to which a variable can be explained by the other variables in the analysis (Brooks, 2014).!!
11!The value of a correlation can vary between -1 and 1.!If the coefficient is zero there is no correlation between the two variables. !
Table 3: Pearson Correlation Matrix
***, **, * Corresponds with 1%, 5%, and 10% significance level (two-tailed).
!
DA Equity BIGN AUDTEN NAS SIZE GROWTH ROA LEV LOSS
DA 1.000 Equity 0.041*** 1.000 (0.000) BIGN -0.017* 0.067*** 1.000 (0.098) (0.000) AUDTEN 0.069*** 0.033*** -0.079*** 1.000 NAS (0.000) 0.014 (0.329) (0.001) 0.087*** (0.000) (0.000) -0.058*** (0.000) -0.033** (0.024) 1.000 SIZE -0.113*** (0.000) 0.143*** (0.000) 0.118*** (0.000) -0.093*** (0.000) 0.041*** (0.004) 1.000 GROWTH 0.078*** 0.152*** 0.012 0.010 0.000 -0.041*** 1.000 (0.000) (0.000) (0.247) (0.344) (0.978) (0.000) ROA -0.221*** -0.032*** 0.023** -0.103*** -0.005 0.160*** 0.148*** 1.000 (0.000) (0.002) (0.023) (0.000) (0.726) (0.000) (0.000) LEV -0.011 -0.050*** -0.023** -0.012 0.039*** 0.265*** -0.138*** -0.150*** 1.000 (0.264) (0.000) (0.021) (0.235) (0.007) (0.000) (0.000) (0.000) LOSS 0.153*** 0.057*** -0.011 0.084*** 0.055*** -0.148*** -0.091*** -0.695*** 0.118*** 1.000 (0.000) (0.000) (0.259) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) ! ! ! ! ! ! ! ! ! ! !
4.2 Main analyses
The following tables report regression estimates for the empirical models described by equations (1) and (2). Instead of the OLS regression the robust regression is performed to test for the hypotheses. Robust regression is an alternative to OLS regression when data may include outliers or influential observations. Robust regression involves allocating a weight to each data point. For each data point an equal weight is allocated and the coefficients are determined using OLS. After that, weights are recalculated so that data points further away from model predictions (outliers) are given lower weight. The coefficients are then
recomputed using weighted least squares (Lawrence and Arthur, 1990). Hence, the robust regression is a method that is more robust and resistant to outliers.
There are two hypotheses than can be examined by the following tables. Table 4 reports regression estimates for the empirical model described by equation (1) where controls are included for year-effects as well as year- and industry-effects, respectively. It examines the first hypothesis that predicts that equity-based executive compensation has a positive effect on earnings management. Model 1 shows that the coefficient between equity-based compensation and earnings management is positive and significant (coefficient: 0.007, p<0.01). When I control for both year and industry effects, the results are similar (Model 2). Consistent with my predictions, I find that equity-based executive compensation is positively associated with earnings management (coefficient: 0.005, p<0.05).
With respect to the control variables, firm size, accounting performance, and leverage are negatively correlated with earnings management, while growth options and incurring losses are positively and significantly associated with earnings management. Both models are significant (p<0.01).
Table 4: Regression results of the empirical model related to hypothesis 1 DA (1) (2) Equity 0.007*** 0.005** (0.002) (0.021) SIZE -0.003*** -0.003*** (0.000) (0.000) GROWTH 0.002*** 0.002*** (0.000) (0.000) ROA -0.075*** -0.071*** (0.000) (0.000) LEV -0.014*** -0.007 (0.001) (0.130) LOSS 0.008*** 0.007*** (0.001) (0.005) N 9758 9758 F-value 29.73*** 15.80***
Year dummies Yes Yes
Industry dummies No Yes
***, **, * Corresponds with 1%, 5%, and 10% significance level (two-tailed).
Table 5 reports the regression estimates for the empirical model described by equation (2). The left (middle) {right} two columns report the findings for Big 4 (auditor tenure)
{provision of non-audit services} as a proxy for audit quality. It examines the second hypothesis that predicts that audit quality has a negative effect on the positive relationship between equity-based executive compensation and earnings management. In general, the findings on the second hypothesis are not significant. For example, the coefficient on the interaction of equity and BIGN is insignificant for both models where I control for year effects, and both year and industry effects respectively. Likewise, the coefficients on the interaction of equity with the other two proxies for audit quality (i.e., auditor tenure and non-audit services) are also insignificant. In conclusion, I find that all three proxies for non-audit quality are not significant. As a result the second hypothesis can be rejected, meaning that audit quality has no negative effect on the positive relationship between equity-based executive compensation and earnings management.
Regarding the control variables the regression analyses demonstrate that SIZE, GROWTH and ROA are significant in all cases. Meaning, smaller firms, firms with more growth opportunities, and less profitable firms are associated with more earnings !
