The relationship between Chief Operating
Officer equity incentives and Real Earnings
Management
Name author: Tim van Leeuwen Student number: 10016201 Date: 31st May 2014
Paper version: Master Thesis (2.0) Name supervisor: M. Schabus Track: Management Accounting
Table of contents
1 Introduction ... 3
2 Literature review and hypotheses ... 5
2.1 Agency theory ... 5
2.2 Executive equity incentives: two conflicting theories ... 6
2.2.1 Convergence-of-interest theory ... 7
2.2.2 Entrenchment theory ... 9
2.3 Earnings management ... 10
2.3.1 Real earnings management ... 11
2.3.2 Accrual-based earnings management ... 13
2.3.3 Real and accrual-based earnings management as substitutes ... 14
2.4 Chief Operation Officers and their responsibilities ... 15
3 Methodology ... 18
3.1 Sample ... 18
3.2 Variable measurements ... 19
3.2.1 Real earnings management ... 19
3.2.2 COO Equity incentives ... 22
3.2.3 Control variables ... 24 3.3 Empirical model ... 26 4 Results ... 27 4.1 Descriptive statistics ... 27 4.2 Multivariate analysis ... 33 4.3 Additional analyses ... 38 5 Conclusion ... 44 References ... 47 Appendix ... 50 2
1 Introduction
According to the agency theory there are potential conflicts of interest between the shareholders (the principals) and the management of a firm (the agents) (Walker, 2013). Executives will not always act in the best interests of the firm’s shareholders, because interests will differ and each maximizes his own utility (Jensen and Meckling, 1976). Due to uncertainty moral hazard may arise. This means that the shareholders are unable to observe whether the firm’s executives make action choices that are in the general interest of the firm, because they cannot observe those executives’ action choices (Walker, 2013), and total monitoring is prohibitively costly. Therefore shareholders aim to convince the executives to behave as if they are maximizing the welfare of the shareholders, that is, they try to align interests with the transfer of ownership. They can do this, among others, by establishing appropriate incentives (Jensen and Meckling, 1976), like incentives arising from the upward potential of equity holdings. Actually, from the 1990s on, firms dramatically increased the use of equity incentives in the form of stock-based and option-based executive compensation (Bergstresser and Philippon, 2006). The reaction of executives to equity ownership in the firm they serve can differ (Jensen and Meckling, 1976; Fama and Jensen, 1983; Pergola and
Joseph, 2011). The convergence-of-interest theory and the entrenchment theory are two competing theories about this reaction.
According to the entrenchment theory, equity incentives also have some adverse effects. Within a certain managerial equity ownership range, they can incentivize executives to increase the short-term stock price of the firm, which will increase the value of their shares and option holdings (Cheng and Warfield, 2005). One of the ways to do so is by earnings management. Earnings management is “the use of managerial discretion over (within GAAP) accounting choices, earnings reporting choices, and real economic decisions to influence how underlying economic events are reflected in one or more measures of earnings” (Walker, 2013, p. 446), which is referred to as accrual-based earnings management and real earnings management. Prior research shows that managers use those two types of earnings
management as substitutes (Cohen, Dey and Lys, 2008; Cohen and Zarowin, 2010; Zang, 2012; Chi, Lisic and Pevzner, 2011), and firms may have shifted from accrual-based to real earnings management during the last decade (Bartov and Cohen, 2009; Cohen, Dey and Lys, 2008).
Accruals-based earnings management is about exploiting managerial discretion, within the generally accepted accounting principles (GAAP), over the level of accruals to influence
reported earnings (Walker, 2013; Dechow and Dichev, 2002). Firms that engage in real earnings management may change economic decisions, such as pricing and production decisions, to achieve financial targets (Walker, 2013). Roychowdhury defines real earnings management 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” (2006, p. 337). As managers use real and
accrual-based earnings management as substitutes, they trade off those two methods based on their relative costs. When accrual-based earnings management is relatively more costly, firms engage more in real earnings management, and the other way around (Zang, 2012). After the enactment of the Sarbanes-Oxley Act in 2002, which increased CEOs’/CFOs’ personal cost of misreporting the financial performance of the firm (Indjejikian and Matějka, 2009), firms shifted from using accrual-based to real earnings management, possibly because real earnings management may be harder to detect (Cohen, Dey and Lys, 2008).
There are plenty of studies focusing on the relationship between CEO’s/CFO’s equity incentives and earnings management (Cheng and Warfield, 2005; Bergstresser and Philippon, 2006; Burns and Kedia, 2006; Jiang, Petroni and Wang, 2010). They conclude that firms with CEOs/CFOs who have higher equity incentives face higher levels of accruals-based earnings management. Besides a CEO and CFO, some firms also have another executive, the Chief Operating Officer (COO). Nowadays, CEOs face more and more pressure and the
appointment of a COO, who should focus on the company’s short-term goals, should enable the CEO to fully focus on the firm’s longer-term challenges (Bennett and Miles, 2006). There may be two separate roles of COOs: an heir apparent training for the position of CEO and a co-leader delegated with some internal operating authority (Zhang, 2006). Internal operating matters are regulated by the COO, which makes him responsible for a wide range of firms operations, such as the allocation of resources, the handling of disturbances and the
communication and implementation of strategies (Hambrick and Cannella, 2004). Therefore, the COO may have the opportunity to intentionally change economic decisions to achieve financial targets (real earnings management). COO equity incentives can form an incentive to actually engage in real earnings management to maximize their own personal wealth by increasing the short-term stock price of the firm.
Because of the above opportunity of COOs to change economic decisions, I address the following question: what is the relationship between Chief Operating Officer equity incentives and Real Earnings Management?
Even if there are a lot of studies about CEO/CFO equity incentives and earnings management, there is a gap in the existing literature regarding the relationship between COO equity incentives and earnings management. Because COOs control a wide range of firm operations, these executives may have the knowledge and possibility to influence real economic decisions to achieve financial targets. Real earnings management is worth
investigating because of two reasons. The first reason is that real earnings management may have real cash flow consequences and can therefore influence future operating performance (Roychowdhury, 2006; Cohen and Zarowin, 2010; Walker, 2013). Actions taken in the current period to increase earnings can have negative effects on future periods’ cash flows, and therefore can reduce firm value (Roychowdhury, 2006). The second reason is the increasing prevalence of real earnings management in the post-SOX period (Bartov and Cohen, 2009; Cohen, Dey and Lys, 2008).
This paper proceeds in four sections. In the next section, the agency theory, two conflicting theories of equity ownership, two types of earnings management and the duties and tasks of Chief Operating Officer will be explained. Chapter three discusses the empirical approach and the data used in this paper. The empirical results will follow in section four. The last section consists of a conclusion and discussion.
2 Literature review and hypotheses
In this chapter, related research will be discussed, which will be used to develop the hypotheses.
2.1 Agency theory
According to the agency theory there are potential conflicts of interest between the principals and the agents of a firm (Walker, 2013). They have a so called agency relationship: “a contract under which one or more persons (the principal(s)) engage another person (the agent) to perform some service on their behalf which involves delegating some decision making authority to the agent” (Jensen and Meckling, 1976, p. 308). Because the interests of
both will differ and each maximizes his own utility, the agent will not always act in the best interests of the principal(s) (Jensen and Meckling, 1976).
An example of an agency relationship is the relationship between the shareholders and the management of a firm, which is characterised by a separation of ownership and control: the shareholders own the resources of the firm, while the management effectively controls those resources on behalf of the shareholders (Jensen and Meckling, 1976). Firms operate under conditions of uncertainty. Due to this uncertainty, two potential information
asymmetries between the shareholders and the management of the firm may arise: moral hazard and adverse selection, of which the first one is of interest for this study. Moral hazard arises when the shareholders cannot observe the firm’s executives’ action choices, and total monitoring is unpayable. Therefore, the shareholders are unable to observe whether the executives make action choices that are in the general interest of the firm (Walker, 2013).
To solve this problem, the shareholders (the principals) try to align the interests with the transfer of ownership, by aiming to induce the executives (the agents) to behave as if they are maximizing the welfare of the shareholders. Managerial ownership levels can be increased by awarding the executives with stock options and/or shares of stock (Core and Guay, 1999), which are examples of incentives arising from the upward potential of equity holdings, or equity incentives. According to Cheng and Warfield (2005), executives are more likely to act in the interests of the shareholders when they own more shares of the firm.
2.2 Executive equity incentives: two conflicting theories
As discussed in the prior paragraph, incentives arising from the upward potential of equity holdings may be used to align executives’ interests with the interests of the
shareholders. Jensen and Murphy (1990) claim that CEOs were under-incentivized in the period 1974-1986, because the value of the CEOs’ stock and option portfolios increased only with $3 for every $1,000 increase in shareholder wealth in this period. Moreover, a change in the U.S. tax code increased the attractiveness of stock- and option-based compensation. Due to this, firms dramatically increased the use of stock-based and option-based executive compensation from the 1990 on (Bergstresser and Philippon, 2006).
There are five elements of managerial equity incentives: stock ownership, restricted stock grants, current-period option grants and unexercisable and exercisable options (Cheng and Warfield, 2005). In their sample of 9472 firm-years, between 1993 and 2000, with CEOs’ compensation data available from the ExecuComp database, stock ownership is, by far, the biggest part of equity incentives: on average, CEOs own 4.2% of the shares of the firm they serve, while the exercisable options, the second biggest part of the equity incentives package, count on average for 0.64% of the firm’s outstanding shares (Cheng and Warfield, 2005).
According to prior literature equity incentives do not necessarily align interests because executives may react in different ways to stock ownership in the firm they serve (Jensen and Meckling, 1976; Fama and Jensen, 1983; Pergola and Joseph, 2011). Two competing theories about this reaction exist: the convergence-of-interest theory and the entrenchment theory.
2.2.1 Convergence-of-interest theory
According to the convergence-of-interest theory executives are self-oriented when they have no stock ownership. However, they lack the power to overcome controls, which are designed to align the executives’ actions with the interests of the shareholders (Pergola and Joseph, 2011). The executives will gradually align their interest with the interests of the shareholders as equity ownership rises. This will lead to action choices that are in the general interest of the firm (Jensen and Meckling, 1976; Pergola and Joseph, 2011).
Figure 1 illustrates the convergence-of-interest theory, which suggests a uniformly positive relationship between executives’ equity ownership and the quality of decision making by executives (Pergola and Joseph, 2011).
According to the convergence-of-interest theory, the quality of decision making by
executives, and therefore also the market value of the firm, increases with managerial equity ownership (Jensen and Meckling, 1976; Morck, Shleifer and Vishny, 1988; Pergola and Joseph, 2011). Due to this better alignment, the executives will be involved in fewer fraud activities, and would be less inclined to intentionally manipulate earnings to overstate the firm’s performance. At the extreme, the executives own all the shares of the firm. In this case, all actions taken against the firm’s interest hurt the executives themselves, which will prevent them from taking those actions.
According to a study of Warfield, Wild and Wild (1995), managers’ earnings manipulation behaviour is indeed related to the level of managerial ownership. They find a negative relationship between the magnitude of discretionary accruals (which indicate accrual-based earnings management) and managerial ownership: the absolute value of discretionary accruals decreases with managerial ownership. This finding supports the convergence-of-interest theory.
On the other hand, a study of Pergola and Joseph (2011) shows that the convergence-of-interest theory may not accurately reflect reality. They conclude that another theory of equity ownership, the entrenchment theory, is more reflective of how executives react to stock ownership. This theory will be discussed in the next subparagraph.
2.2.2 Entrenchment theory
The second theory about the reaction of executives to stock ownership is the entrenchment theory of equity ownership. For extremely low and extremely high levels of managerial equity ownership, this theory has similar predictions about the quality of
executives’ decisions as the convergence-of-interest theory. The interests of the executives are not well aligned with those of the shareholders at low managerial equity ownership levels. Nevertheless, they cannot undermine governance mechanisms, because they do not possess enough power to do so. At high managerial equity ownership levels, inappropriate actions of the executives would only hurt themselves, as the executives are the shareholders in that case (Pergola and Joseph, 2011).
However, the expectations of the entrenchment theory about the quality of executives’ decisions differ from those of the convergence-of-interest theory in the middle range of the Figure 1. Executives might have enough power to overcome governance mechanisms at relatively large levels of equity ownership and might therefore consume perquisites, which reduce the value of the firm, but outbalance their loss from reduced firm value (Short and Keasey, 1999). They become entrenched, which means that they are able to act in their own self-interests without having to fear for removal or sanctions (Fama and Jensen, 1983; Pergola and Joseph, 2011). This can be seen in Figure 2, where the quality of decision making
decreases when managerial equity ownership increases within the so called entrenchment range.
According to Morck, Shleifer and Vishny (1988), executives can become entrenched at managerial equity ownership levels of somewhere between 5% and 25%1. However, they also state that entrenchment not only depends on equity ownership, but also on the executives’ tenure with the firm and personal characteristics. According to the entrenchment theory of equity ownership, executives may conduct more fraud activities, make poor accrual decisions or intentionally manipulate earnings within this entrenchment range (Pergola and Joseph, 2011). A study of Japanese firms by Teshima and Shuto (2008) support this prediction. They find that within a certain range of managerial ownership (10-35%) the relationship between managerial ownership and discretionary accruals is significantly positive, indicating
managerial opportunistic behaviour in the form of earnings management.
Prior studies find diverse average managerial ownership levels. For example,
Warfield, Wild and Wild (1995) and Klassen (1997) find average percentages of managerial ownership of 17.2% and 15.1% for U.S. firms respectively, which are within the
entrenchment ranges found by Morck, Shleifer and Vishny (1988) and Teshima and Shuto (2008). However, more recent studies show that the average manager of a U.S. firm does not fall within these entrenchment ranges: Cheng and Warfield (2005) find an average ownership percentage of 4.2%, while Custódio, Ferreira and Laureano (2013) show that the mean percentage of shares owned by the managers of U.S. firms is only 1.0%.
Summarized, the entrenchment theory of equity ownership, which accurately reflects reality (Pergola and Joseph, 2011; Morck, Shleifer and Vishny, 1988; Teshima and Shuto, 2008) predicts that within a certain range of equity ownership executives are more likely to opportunistically manage earnings.
2.3 Earnings management
The managers of a firm often have to exercise judgment in financial reporting (Healy and Wahlen, 1999). They use their specific knowledge to make choices about which reporting methods and estimates to apply, and what to voluntarily disclose. This use of judgment in financial reporting can have both advantages and disadvantages. An advantage is that the
1
Pergola and Joseph (2011) find that executives may become entrenched between managerial ownership levels of 30 and 50%.
10
communication of private information to stakeholders can improve. The managers can make reporting choices that best match the firms’ business economics, which will increase the value of the financial reports as a communication tool. However, a major disadvantage of
managerial judgement in financial reporting is that it enables the managers to engage in earnings management. The managers can select reporting methods and estimates that do not accurately reflect the underlying economics of the firm (Healy and Wahlen, 1999).
According to Walker (2013), earnings management is: “the use of managerial discretion over (within GAAP) accounting choices, earnings reporting choices, and real economic decisions to influence how underlying economic events are reflected in one or more measures of earnings” (p. 446), which refers to both accrual-based earnings management and real earnings management. Managers try to alter financial reports to either influence
contractual outcomes that depend on reported accounting numbers or to mislead stakeholders about the underlying economic performance of the firm (Healy and Wahlen, 1999).
Walker (2013) identifies three sets of motives for a firm or executive to engage in earnings management: to achieve earnings-related contractual terms or targets; to influence the information set that external investors and information intermediaries use; and to
influence information that third parties use to judge the firm’s financial strength. The second set of motives, to influence the information set that external investors and information
intermediaries use, is the most relevant one for this study. In order to increase the firm’s stock price, managers overstate earnings assuming that investors do not see through (markets are not fully efficient) and therefore can be misled by earnings management (Walker, 2013).
2.3.1 Real earnings management
One of the two methods firms can use to manage earnings is real earnings management (Walker, 2013). Real earnings management implies that managers change economic decisions to achieve financial targets (Walker, 2013). Roychowdhury (2006) defines real earnings management 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 operations” (p. 337). The ways in which firms can engage in real earnings management are likely to be very industry specific.
However, some real earnings management methods are more common than others. For example, a lot of firms cut their research and development (R&D) expenditures to achieve a certain financial target, as shown by Baber, Fairfield and Haggard (1991).
Roychowdhury (2006) focuses in his study on three methods to engage in real earnings management: sales manipulation, reduction of discretionary expenditures and
overproduction. The first method, sales manipulation, involves giving price discounts or more lenient credit terms to temporarily increasing sales during the year, in the way of accelerating the timing of sales and/or generating additional sales. The second method, reduction of discretionary expenditures, involves reducing discretionary expenditures such as R&D, advertising and maintenance, which are generally immediately expensed when they are incurred. And the last method, overproduction, can be used to lower the reported cost of goods sold (COGS). If a firm produces more goods, the fixed costs will be spread over a larger number of goods, which will lower the fixed costs per product. This means that the total cost per product will also decline, as long as the fixed costs per product reduce more than the marginal cost per product increase. However, overproduction also leads to extra inventory holding costs (Roychowdhury, 2006). Besides these three methods, Gunny (2010) mentions a fourth method to engage in real earnings management: timing of sales of fixed assets.
Managers can manage reported earnings by selling a fixed asset, since a gain on such a sale is reported at the time of the sale.
Real earnings management may destroy firm value (Walker, 2013), as actions taken in the current period to increase current earnings can negatively influence future periods’ cash flows (Roychowdhury, 2006). To illustrate this, an earlier mentioned example of real earnings management can be used: overproduction. Firms can over-produce in the current period to lower the reported cost of goods sold, and therefore to increase current earnings. However, the firm will experience excess inventories because of this overproduction, which will lead to higher inventory holding costs. So, the current period income-increasing action
(overproduction) can have a negative effect on future periods’ cash flows (higher inventory holding costs), and therefore may reduce firm value. Several studies support this statement, as they find that real earnings management leads to a severe decline in future operating
performance (Cohen and Zarowin, 2010; Gunny, 2005).
In contrast to the relationship between executive incentives and accrual-based earnings management, the relationship between executive equity incentives and real earnings
management has not been widely examined. However, Graham, Harvey and Rajgopal (2005) 12
provide evidence that executives prefer engaging in real earnings management activities compared to accrual-based earnings management. 80% of the participants of their survey answers that they would engage in real earnings management, and therefore sacrifice long-term firm value, to meet a short-long-term earnings target. They find that managers want to meet earnings targets to, among others, increase the stock price. As higher equity incentives increase the sensitivity of executives’ wealth to the short-term stock prices of the firm, these incentives may incentivize executives to increase short-term stock prices (Cheng and
Warfield, 2005). They can and, according to Graham, Harvey and Rajgopal (2005), are willing to do this by means of real earnings management.
2.3.2 Accrual-based earnings management
The second method that firms may use to manage earnings is accrual-based earnings management (Walker, 2013). The components of earnings that are not reflected in cash flows are called accruals (Bergstresser and Philippon, 2006). Accruals shift or adjust the recognition of cash flows over time, so that earnings are a better measure of firm performance than cash flows (Dechow and Dichev, 2002). Accrual-based earnings management is the exploitation of managerial discretion over the level of accruals, within GAAP (Walker, 2013). So, managers can manipulate accruals to influence reported earnings (Dechow and Dichev, 2002).
As accruals only shift or adjust the recognition of cash flows over time, accrual-based earnings management does not change the future free cash flows of a firm, and therefore also does not influence the firm’s value, which is calculated by discounting the firm’s expected free cash flows (Walker, 2013).
A lot of prior research focuses on the relationship between executives’ equity
incentives and accrual-based earnings management. Overall, it concludes that companies with CEOs and CFOs with higher equity incentives are associated with higher levels of accrual-based earnings management, measured by discretionary accruals (Bergstresser and Philippon, 2006; Jiang, Petroni and Wang, 2010), the likelihood of meeting or just beating analyst forecasts (Cheng and Warfield, 2005) and the likelihood of misreporting (Burns and Kedia, 2006).
2.3.3 Real and accrual-based earnings management as substitutes
In the last decade, firms shifted from using accrual-based to real earnings management. Managers use real earnings management and accrual-based earnings
management as substitutes (Cohen, Dey and Lys, 2008; Bartov and Cohen, 2009; Cohen and Zarowin, 2010; Zang, 2012; Chi, Lisic and Pevzner, 2011). They trade off those two methods based on their ability to engage in accrual-based earnings management and the relative costliness of both earnings management-methods. Managers engage more in real earnings management when accrual-based earnings management is relatively more costly and when they are less able to engage in accrual-based earnings management, and the other way around. Both methods to manage earnings have their own costs, which depend on the operational and accounting environment of the firm (Zang, 2012).
Scrutiny provided by the accounting practice and potential penalties of detection are the costs of using accrual-based earnings management (Cohen, Dey and Lys, 2008; Cohen and Zarowin, 2010). Zang (2012) states that the scrutiny provided by the accounting practice is likely to be bigger when the firm is audited by one of the Big 8 audit firms (because of more experience, more resources and more reputation risk), when the auditor-client relationship lengthens (longer tenure) and after the enactment of SOX in 2002, which increased managers’ personal cost of misreporting the financial performance of the firm (Indjejikian and Matějka, 2009). The other cost of accrual-based earnings management, the penalties of potential litigation, are likely to be greater for firms in high litigation industries, such as pharmaceuticals/biotechnology, computers and electronics (Cohen and Zarowin, 2010; Zang, 2006; Barton and Simko, 2002).
Moreover, the firm’s ability to manage earnings by means of accrual-based earnings management influences the trade-off between the two earnings management methods. Accrual-based earnings management in previous periods constrains this ability, because GAAP offers little flexibility and accruals need to be reversed in future periods (Cohen and Zarowin, 2010; Zang, 2012; Barton and Simko, 2002). So, the higher the costs of accrual-based earnings management and the lower the ability to use this method, the higher the use of real earnings management will be.
On the other hand, the use of real earnings management will be lower when this method is relatively more costly. Zang (2012) states that the costs of real earnings
management depend on the firm’s operational and accounting environment: real earnings 14
management is more costly for firms that have a less competitive status in the industry, are in a less healthy financial condition, experience higher levels of monitoring from institutional investors and incur greater tax expenses in the current period.
2.4 Chief Operation Officers and their responsibilities
Bennett and Miles (2006) argue that, in recent years, Chief Executive Officers are facing more and more pressure because of many different factors such as globalisation, financial frauds and public concern regarding corporate social responsibility. Because of this increased pressure, many organizations have appointed a Chief Operating Officer, or COO, who is responsible for the oversight of daily organizational operations. The appointment of this extra executive should enable the CEO to fully focus on the company’s longer-term challenges, while the COO should be responsible for the company’s day-to-day, quarter-to-quarter results and should focus on short-term goals (Bennett and Miles, 2006).
Strategic and external activities are among the responsibilities of the CEO, while the internal operating matters are regulated by the COO. This makes a COO responsible for a wide range of firm operations, such as motivating and activating subordinates, transmitting information to the organization’s members, implementing the organizational strategies, the correction of problems within the organization and the allocation of resources in the form of reviewing and approving budgets (Hambrick and Cannella, 2004). Because of these
responsibilities of the Chief Operating Officer, he may have the knowledge and possibility to influence real economic decisions to achieve financial targets and thus may have the
opportunity to engage in real-earnings management.
Zhang (2006) recognizes two separate roles of COOs. The first one is the role of an heir apparent who is training for the position of CEO. In this case, the COO is trained to succeed the current CEO somewhere in the future. As a CEO has a significant influence on the selection of a COO, a CEO is likely to select a COO who is similar to her (Zhang, 2006). And as the CEO has, by definition, more power than the COO (Levinson, 1993), Zhang states that a COO ”tends to partner with rather than compete with the CEO” (2006, p. 285).
The second possible role of a COO is a co-leader delegated with internal operating authority (Zhang, 2006). From this perspective, having a COO creates a different
organizational structure, as this divides the top-level roles that are typically fulfilled by one person between the CEO and the COO (Hambrick and Cannella, 2004). It provides a
separation between strategy formulation (CEO) and strategy implementation (COO) and adds an additional organizational layer. In this role, it is also likely that the COO and the CEO will work together closely in order to work effectively, because the responsibilities of the CEO and the COO are closely related (Zhang, 2006).
In contrast with the above explanations, the CEO-COO relationship can also involve rivalry (Zhang, 2006). In the case of a COO as an heir apparent, the COO may become impatient and wants to become the CEO himself (Levinson, 1993), which can cause rivalry. And when the COO has the role of a co-leader delegated with internal operating authority, he is only one step from the CEO position and can therefore be a power contender to the CEO (Zhang, 2006). So, a COO can either partner with, or compete with, the CEO.
Bennett and Miles (2006) derived seven broad reasons why companies decide to hire a COO from their research. These reasons are:
1. To provide daily leadership in an operationally intensive business; 2. To lead a specific strategic imperative undertaken by management, such as a turnaround (…); 3. To serve as a mentor to a young or inexperienced CEO (often a founder); 4. To
balance or complement the strengths of the CEO; 5. To foster a strong partnership at the top (…); 6. To teach the business to the heir apparent to the current CEO; 7. To retain executive talent that other firms may be pursuing, absent an imperative from the business for creating the position. (p. 7)
Overall, most of these mentioned reasons suggest that the COO will partner with, instead of compete with, the CEO. As Bennett and Miles (2006a) state, the most important success factor of a COO is building a high level of trust between CEO and COO. The CEO’s vision must be truly supported by the COO, while the CEO must grant real decision rights and authority. In accordance with the assumption that COOs will generally partner with the CEO, Rajan and Wulf (2006) provide evidence that COOs truly act as intermediaries between CEOs and the rest of the organization.
Concluding, COOs are responsible for the company’s day-to-day, quarter-to-quarter results and should focus on short-term goals. They may have the opportunity to engage in real earnings management, because of their responsibility over a wide range of firm operations. COOs can have two separate roles: an heir apparent or a co-leader delegated with internal
operating authority. Furthermore, the COO can decide to either partner with the CEO, or compete with the CEO. Given the COO’s responsibilities I expect real earnings management to be positively related to COOs equity incentives. However, if the CEO partners with the COO, and because the CEO is the more influential executive, it is important to control for coexisting CEO incentives.
Based on that, the following hypothesis is formed for this study:
H1: The magnitude of real earnings management is positively associated with Chief Operating Officer equity incentives.
Roychowdhury (2006) and Gunny (2010) examine real earnings management by the abnormal levels of cash flow from operations, discretionary expenses, production costs and gains on asset sales. Firms that manage earnings upwards by means of real earnings
management are characterized by unusually low cash flow from operations, unusually low levels of discretionary expenses, unusually high production costs and/or unusually high gains on asset sales (Cohen and Zarowin, 2010; Gunny, 2010). This leads to the following four sub hypotheses:
H1a: The magnitude of the abnormal level of cash flow from operations is negatively associated with Chief Operating Officer equity incentives. H1b: The magnitude of the abnormal level of discretionary expenses is negatively
associated with Chief Operating Officer equity incentives.
H1c: The magnitude of the abnormal level of production costs is positively associated with Chief Operating Officer equity incentives.
H1d: The magnitude of the abnormal level of gains on asset sales is positively associated with Chief Operating Officer equity incentives.
3 Methodology
3.1 Sample
This study on the relationship between COO equity incentives and real earnings
management focuses on U.S. S&P 1500 firms, as the other studies on the relationship between CEO/CFO incentives and earnings management (Bergstresser and Philippon, 2006; Cheng and Warfield, 2005; Burns and Kedia, 2006; Jiang, Petroni and Wang, 2010), mentioned in the background section, also investigated these firms. This enables a comparison of the results of this study with the outcomes of prior studies on the relationship between executive
incentives and earnings management. Another reason is that, for this research, data about COO equity incentives from the ExecuComp database is needed, which only contains
information about U.S. firms’ executives. COOs will be identified based on executives’ titles in ExecuComp (data item “title”) that include COO, Chief Operating, Chief Operations, President of Operations or President of Operating.
The original sample of 102,788 firm-year observations is collected from the
Compustat database for the period 2004-2013. These years are post-SOX years, in which the levels of real earnings management significantly increased (Cohen, Dey and Lys, 2008). Banks and financial institutions (SIC codes 6000-6500) and firms in regulated industries (SIC codes 4400-5000) are eliminated, which reduces the sample size by 20,792 observations. Each firm-year observation is required to have the data that is necessary to compute the real
earnings management measurements used in this study. Furthermore, for each industry-year group, 15 observations are required, as the normal cash flow from operations, discretionary expenses and production costs are estimated for every industry-year with at least 15
observations, based on the study of Roychowdhury (2006). The sample is reduced by another 42,845 firm-year observations due to these data requirements. Because of missing data to calculate the equity incentives of COOs and CEOs, another 38,031 firm-year observations are dropped. Finally, 46 observations are dropped because they lack the data to calculate
additional control variables. These procedures result in a final sample consisting of 1074 firm-year observations. The sample selection procedures are presented in Table 1. Furthermore, the number of firm-year observations per (group of) two-digit SIC code(s) is shown in Appendix B.
3.2 Variable measurements
3.2.1 Real earnings management
To measure real earnings management, the focus will be on the abnormal levels of cash flow from operations (OCF), discretionary expenses, production costs and gains on asset sales, which Roychowdhury (2006) and Gunny (2010) also examined. Roychowdhury (2006) focused on three manipulation methods: “accelerating the timing of sales or generating additional unsustainable sales through increased price discounts or more lenient credit terms; reduction of discretionary expenditures; and overproduction, or increasing production to report lower COGS” (2006, p. 339), where the sum of R&D, advertising and SG&A expenses are the discretionary expenditures. The normal levels of OCF (OCFi,t), discretionary expenses
(DiscExpi,t) and production costs (Prodi,t) will be calculated using the models of Dechow,
Kothari and Watts (1998):
, where OCF is cash flow from operations (annual Compustat data item OANCF); the sum of research and development expenses (annual Compustat data item XRD), advertising expenses (annual Compustat data item XAD) and selling, general and administrative expenses (annual Compustat data item XSGA) represents DiscExp; and Prod is the sum of the cost of goods sold (annual Compustat data item COGS) and the change in inventories (annual Compustat data item INVT).
To reflect the impact of industry-wide economic conditions during each year on production costs, the equations for normal OCF, production costs and discretionary expenses are estimated cross-sectionally for each industry-year with at least 15 observations
(Roychowdhury, 2006; Zang, 2012).
The abnormal levels of cash flow from operations (AbnCFO), discretionary expenses (AbnDiscExp) and production costs (AbnProd) are calculated by deducting the normal levels of OCF, discretionary expenses and production costs (calculated by using the above
equations) from the actual OCF, discretionary expenses and production costs. As mentioned before, those abnormal levels of OCF, discretionary expenses and production costs are used as proxies for real earnings management: uncommon low cash flow from operations, and/or uncommon low levels of discretionary expenses, and/or uncommon high production costs are characteristics of firms that manage earnings upwards by means of real earnings management (Cohen and Zarowin, 2010).
Gunny (2010) uses a fourth individual measure of real earnings management, namely the gains on asset sales. The normal level of these gains can be estimated in this way:
, where GainAsset is the income from asset sales (annual Compustat data item SPPIV
multiplied by negative one); and Assets represents total assets (annual Compustat data item AT). ln(MV) is the natural logarithm of the firm’s market value, which is calculated by
multiplying the firm’s closing share price (annual Compustat data item PRCC_F) with the number of common shares outstanding (annual Compustat data item CSHO) and which controls for size effects. Furthermore, TQ represents Tobin’s Q (((annual Compustat data item PRCC_F * data item CSHO) + data item PSTK + data item DLTT + data item DLC) / data item AT) and IntF represents internal funds, which is the sum of income before extraordinary items (annual Compustat data item IB), R&D expense (annual Compustat data item XRD) and depreciation and amortization expenses (annual Compustat data item DP). Lastly,
AssetSales and InvestSales represent the long-lived assets sales (annual Compustat data item
SPPE) and the long-lived investment sales (annual Compustat data item SIV) respectively (Gunny, 2010).
The abnormal level of gains on asset sales (AbnGainAsset) is calculated by deducting the normal level of those gains from the actual gains on asset sales. Therefore, the higher the residual from the above model, the higher the asset sales manipulation (Gunny, 2010).
The three individual measures of real earnings management introduced by Roychowdhury (2006) can also be combined into three comprehensive measures of real earnings management (REM_1, REM_2 and REM_Index), as shown by studies of Zang (2006) and Chi, Lisic and Pevzner (2011). REM_1 is calculated by multiplying the abnormal
discretionary expenses by negative one and then add it to abnormal production costs. A high amount for this comprehensive metric indicates a high likelihood that the firm engaged in real earnings management activities (Zang, 2006).
The second comprehensive measure, REM_2, is calculated by multiplying both abnormal discretionary expenses and abnormal cash flow from operations by negative one and add them together (Zang, 2006). The higher this comprehensive measure is, the more likely that the firm manipulates its sales and cuts its discretionary expenses to manage its earnings upwards.
REM_Index, the third comprehensive measure of real earnings management, is
calculated as follows: -standardized AbnCFO + standardized AbnProd – standardized
AbnDiscExp, where AbnCFO, AbnProd and AbnDiscExp represent the abnormal levels of
cash flow from operations, production costs and discretionary expenses respectively. The standardized measures of those variables are calculated by subtracting the variable’s mean from the variable and then divide it by the variable’s standard deviation (Chi, Lisic and
Pevzner, 2011). A high REM_Index indicates a high likelihood that the firm has manipulated its earnings upwards by means of real earnings management.
As using these comprehensive measures may dilute the different implications of the four individual measures of real earnings management (abnormal levels of cash flow from operations, discretionary expenses, production costs and gains on asset sales) for earnings (Cohen and Zarowin, 2010), both the comprehensive as well as the four individual measures of real earnings management will be used to find out if there is a relationship between COO equity incentives and real earnings management.
3.2.2 COO Equity incentives
To measure the power of the equity incentives of Chief Operating Officers, I need to capture the share of the COO’s total compensation that would come from a one percentage point increase in the value of his company’s equity (Bergstresser and Philippon, 2006), using the following formula:
COO_IncRatioi,t = PRCSENi,t / (PRCSENi,t + SALBONi,t),
where PRCSENi,t is the dollar change in the value of the stock and option holdings of a COO
resulting from a 1% increase in the company’s common stock price (Bergstresser and
Philippon, 2006; Brockman, Martin and Unlu, 2010), and SALBONi,t (ExecuComp data item
TOTAL_CURR) represents the COO’s total current compensation, consisting of his salary and bonus.
The sensitivity of the COO’s stock holdings to the firm’s stock price is relatively straightforward, because the delta (Δ) of the stock holdings is one, which means that a dollar increase in the firm’s share price translates one-for-one to the value of a stock holding. However, this is not the case for options, especially those whose exercise price is above the current share price (out-of-the-money options) (Core and Guay, 2002; Bergstresser and Philippon, 2006). Therefore, I follow the “one-year approximation” (OA) method of Core and Guay (2002) to estimate the sensitivity of the COO’s option holdings to the stock price, which explains over 99% of the variation in this sensitivity (Brockman, Martin and Unlu, 2010). This OA method divides the option portfolio into three portions: options granted in the current year, previously granted exercisable options and previously granted unexercisable options. It
then calculates the sensitivity of each group of options to the stock price (Δ) using the Black-Scholes model:
Δ = e-dT
N(Z);
,
where d is the natural logarithm of the expected option’s life time dividend yield, T is the option’s time to maturity in years, S represents the stock price, X stands for the option’s exercise price, r is natural logarithm of the risk-free interest rate and σ is the option’s expected stock-return volatility (Core and Guay, 2002; Brockman, Martin and Unlu, 2010).
For all three groups of options, S, σ, r and d are readily available from the Compustat database (data items PRCC_F, OPTVOL, OPTRFR and OPTDR respectively). For options granted in the current year, T and X can be derived from the Outstanding Equity Awards database within ExecuComp (data items EXDATE and EXPRIC).
However, the exercise prices of previously granted options need to be estimated, by dividing the realizable values of the unexercisable options (ExecuComp data item
OPT_UNEX_UNEXER_EST_VAL) and exercisable options
(OPT_UNEX_EXER_EST_VAL) of the COO by the number of unexercisable options (OPT_UNEX_UNEXER_NUM) and exercisable options (OPT_UNEX_EXER_NUM) and subtracting these amount from the firm’s stock price (Core and Guay, 2002). Before
computing the average exercise price of the unexercisable options, the realizable value and number (OPTION_AWARDS_NUM) of the newly granted options is subtracted from the realizable value and number of unexercisable options to avoid double counting of the current year’s grant. In case the option grant in the current year is bigger than the number of
unexercisable options held by a COO, the excess of the current year’s grant’s realizable value and number over those of the unexercisable options is deducted from the realizable value and number of the exercisable options held by the COO (Core and Guay, 1999; Brockman, Martin and Unlu, 2010).
The times-to-maturity of previously granted options need to be estimated too. These estimates depend on whether the firm has granted new options in the current year or not. If the company granted new options in the current year, the time-to-maturity of the previously granted unexercisable options is estimated to be one year less than the time-to-maturity of the
new options and the time-to-maturity of previously granted exercisable options is set equal to three years less than that of the unexercisable options. If the company did not grant new options in the current year, times-to-maturity of nine and six years are used for the previously granted unexercisable and exercisable options respectively (Core and Guay, 2002).
The delta’s of the new options granted in the current year, previously granted exercisable options and previously granted unexercisable options are used to calculate the dollar change in the value of the COO’s stock and option holdings resulting from a 1% increase in the firm’s stock price (PRCSENi,t). I use the following equation to do this:
PRCSENi,t= S/100 * (ΔNG * NNG+ ΔPGEX * NPGEX + ΔPGUNEX * NPGUNEX + NSTOCK),
where S stands for the firm’s stock price, N denotes the number of options/stocks of the COO, and NG, PGEX, PGUNEX and STOCK represent new option grants, previously granted exercisable options, previously granted unexercisable options and stock holdings respectively (Brockman, Martin and Unlu, 2010).
Although this OA method explains the sensitivity of option holdings to the stock price well in most cases, it may contain some biases in subsamples with large fractions of out-of-the-money stock options (Core and Guay, 2002).
3.2.3 Control variables
Managers engage more in real earnings management when their ability to engage in accrual-based earnings management is lower and when accrual-based (real) earnings management is relatively more (less) costly for their firm, based on its operational and accounting environment (Zang, 2012). Therefore, I include control variables that capture the firm’s ability to engage in accrual-based earnings management and the costs of accrual-based and real earnings management in my study.
Accrual manipulation activities in prior years constrains the ability of a manager to use accrual-based earnings management in the current year, due to the reversal of accruals and the limited flexibility that general accepted accounting principles offer (Zang, 2012). Therefore, I include the firm’s beginning of the year net operating assets divided by lagged sales (NOAi,t-1)
in the regression, which proxies for the prior years’ accrual-based earnings management 24
(Barton and Simko, 2002). Another proxy for the firm’s ability to engage in accrual-based earnings management is the length of the firm’s operating cycles (Cyclei,t-1), computed as:
days receivable + days inventory – days payable at the beginning of the year (Dechow, 1994; Zang, 2012).
To control for the first category of costs of accrual-based earnings management, scrutiny provided by the accounting practice (Cohen, Dey and Lys, 2008; Cohen and Zarowin, 2010), the presence of a Big 4 auditor (Big4i,t) and auditor tenure
(Auditor_Tenurei,t) are added as control variables, as accrual-based earnings management is
more constrained when a firm is audited by a Big 4 audit firm (Francis, Maydew and Sparks, 1999; Chi, Lisic and Pevzner, 2011) and the auditor-client relationship lengthens (Myers, Myers and Omer, 2003). Furthermore, to control for the second category of costs of accrual-based earnings management, the potential penalties of litigation, I include another dummy variable (Litigationi,t), which equals to one if a firm’s industry is characterized by high
litigation risk (computers, electronics and pharmaceuticals/biotechnology industries), and zero otherwise. These industries are captured by the SIC codes 2833-2836, 8731-8734, 7371-7379, 3570-3577 and 3600-3674 (Barton and Simko, 2002; Zang, 2006; Cohen and Zarowin, 2010). Real earnings management is more costly for firms that have a less competitive status in the industry and are in a less healthy financial condition. Therefore, I add control variables to the regression for ratio of the firm’s sales to total its industry’s total sales at the beginning of the year (MarketSharei,t-1) and the firm’s financial health, using a modified version of the
Z-score of Altman (2000) (ZScorei,t-1) (Zang, 2012; Altman, 2000). The lower the values of
MarketSharei,t-1 and ZScorei,t-1, the higher the costs for real earnings management are.
Lastly, following studies of Bergstresser and Philippon (2006) and Jiang, Petroni and Wang (2010), I control for firm size (Sizei,t), firm age (Agei,t), lagged book leverage
(Leveragei,t), market-to-book ratio (MTBRatioi,t), and CEO equity incentives
(CEO_IncRatioi,t-1), to, among others, make sure that the relationship between COO equity
incentives and real earnings management is not driven by the more volatile operating environments of firms that grant a lot of stock and options or by the equity incentives of the firm’s CEO. All continuous variables in the empirical model of this study are winsorized at their respective 1st and 99th percentiles, to rule out the influence of potential outliers and year and industry fixed effects are included in the empirical model. Appendix A contains the definitions (and calculations) of all variables used in this study.
3.3 Empirical model
To test the hypotheses, seven regressions of measures of real earnings management on COO incentives will be conducted, one for each method to engage in real earnings
management as well as one for each of the comprehensive measures of real earnings management. This results in the following empirical model:
REMi,t= βo + β1 × COO_IncRatioi,t-1+ β2 × NOAi,t-1 + β3 × Cyclei,t-1+ β4 × Big4i,t +
β5 × Auditor_Tenurei,t + β6 × Litigationi,t + β7 × MarketSharei,t-1 +β8 ×
ZScorei,t-1+ β9 × Sizei,t+ β10 × Agei,t+ β11 × Leveragei,t-1+ β12 ×
MTBRatioi,t+ β13 × CEO_IncRatioi,t-1 + β14 × YearDummy+ β15 ×
IndustryDummy +εt ,
where:
REMi,t = various real earnings management metrics (abnormal levels of cash flow from
operations, discretionary expenses, production costs, gains on asset sales, and the three comprehensive measures);
COO_IncRatioi,t-1 = the lagged share of a COO’s total compensation that would come from a
one percentage point increase in the value of his company’s equity;
NOAi,t-1 = net operating assets at the beginning of the year;
Cyclei,t-1 = the length of the firm’s operating cycle at the beginning of the year;
Big4i,t = dummy variable that equals one if the firm is audited by a Big 4 auditor, zero
otherwise;
Auditor_Tenurei,t = the length of the firm’s relationship with the current auditor in years;
Litigationi,t = dummy variable that equals one for firms in the computers, electronics and
pharmaceuticals/biotechnology industries, zero otherwise.
MarketSharei,t-1 = ratio of the firm’s sales to total its industry’s total sales at the beginning of
the year;
(Zang, 2012; Altman, 2000);
Sizei,t = natural logarithm of lagged total assets;
Agei,t = dummy variable which is 1 if the firm is listed for more than 20 years on Compustat,
and 0 otherwise;
Leveragei,t-1 = the lagged book leverage of the firm (total liabilities deflated by total assets);
MTBRatioi,t = the market-to-book ratio of the firm;
CEO_IncRatioi,t-1 = the lagged share of a CEO’s total compensation that would come from a
one percentage point increase in the value the CEO’s company’s equity;
YearDummy = this dummy controls for year fixed effects;
IndustryDummy = this dummy controls for industry fixed effects.
4 Results
4.1 Descriptive statistics
To calculate AbnCFO, AbnDiscExp, AbnProd and AbnGainAsset, the normal levels of cash flow from operations, discretionary expenses, production costs and gains from asset sales need to be estimated using the models presented in paragraph 3.2.1. These regression models are carried out per industry-year group with at least 15 observations. Due to this requirement and after excluding banks, financial institutions and firms in regulated industries, 382 such industry-year groups remain in Compustat for the period 2004-2013. All observations in Compustat which meet the data requirements to estimate the four individual measures of real earnings management are included in these regressions. Table 2 presents the descriptive statistics for the distribution of the estimated coefficients and R2s from these models (panel A, B, C and D respectively).
These statistics from this study are similar to those of Roychowdhury (2006), Chi, Lisic and Pevzner (2011) and Zang (2012). Similar to those studies, the regression models have a quite high explanatory power. The average adjusted R2 of the models to estimate the
normal levels of cash flow from operations is 63.1% across industry-years, compared to R2s of 45% and 53.5% reported by Roychowdhury (2006) and Chi, Lisic and Pevzner (2011), respectively. For the models to estimate the normal levels of discretionary expenses, the average adjusted R2 is 55.5% in this study, while this was 38%, 73.9% and 57.55% in the studies of Roychowdhury (2006), Chi, Lisic and Pevzner (2011) and Zang (2012),
respectively. Although the average adjusted R2 of the regressions to estimate the normal levels of production costs is lower than those reported by Roychowdhury (2006) (89%), Chi, Lisic and Pevzner (2011) (93.5%) and Zang (2012) (90.61%), it is still quite high (69.8%).
The average explanatory power of the regressions to estimate the normal levels of gains on asset sales (21.7%) is not as high as those of the other three regressions in this study. However, it is not significantly lower than the average R2 of these regressions reported by Gunny (2010). Overall, the percentages of the total variation in the normal levels of cash flow from operations, discretionary expenses, production costs and gains from asset sales that is explained by the regression models are quite high and similar to those reported by prior studies.
The descriptive statistics of the final sample for all the variables included in my empirical model are presented in Table 3. The mean AbnCFO, AbnDiscExp, AbnProd and
AbnGainAsset are 0.122, -0.068, -0.039 and 0 respectively, which are comparable to those
reported by Cohen, Dey and Lys (2008), Gunny (2010), Chi, Lisic and Pevzner (2011) and Zang (2012). These numbers indicate that, on average, firms only manage their earnings upwards by means of reducing discretionary expenditures (negative AbnDiscExp). Combining the first three individual measures into the three comprehensive measures of real earnings management results in a mean REM_Index, REM_1 and REM_2 of -0.023, 0.032 and -0.037 respectively. Only the mean REM_1 (0.032) suggests that firms, on average, manage their earnings upwards through the use of real earnings management, as positive means for those variables indicate this.
The mean COO_IncRatio is equal to 0.143, which indicates that 14.3% of the average COO’s total compensation would come from a one percentage point increase in the value of his company’s stock price. This ratio is 30.0% for the average CEO, suggesting that CEOs have, on average, higher equity incentives than COOs. This mean of CEOs’ equity incentive ratio is comparable to those reported by Bergstresser and Philippon (2006) (19.2%) and Jiang, Petroni and Wang (2010) (23.6%). Besides the mean CEO’s equity incentives ratio, Jiang, Petroni and Wang (2010) also calculate the equity incentive ratios of CFOs. On average, this ratio is equal to 10.5%. The mean COO_IncRatio of this study is more similar to this number, which could be expected, as the CEO is by definition the most powerful executive within a firm. Therefore, it is more important to align the CEO’s interests with the interests of the shareholders, resulting in a higher equity incentive ratio for CEOs than for somewhat less powerful CFOs/COOs.
The mean operating cycle (Cycle) of the sample firms is equal to 79.541 days and the average net operating assets (NOA) represent 72.0% of the average firms’ sales. With regard to the control variables for the costs of accrual-based earnings management (auditor scrutiny
and litigation penalties): the percentage of the firms in my sample that is audited by a Big 4 audit firm is equal to 90.7 (Big4) and the average length of the firm’s relationship with the current auditor (Tenure) is 12.290 years. Furthermore, 28.3% of the firms in the sample are in an industry characterized by high litigation risk (Litigation). Regarding the costs of real earnings management, the average market share of the firms in the sample is 12.3% (MarketShare) and the mean ZScore is 4.848. These numbers are all comparable to those reported by, among others, Zang (2006), Chi, Lisic and Pevzner (2011) and Zang (2012), except for the average operating cycle. For example, Cohen, Dey and Lys (2008) and Zang (2012) find average operating cycles of 142.56 and 141.77 days, which is about twice the average operating cycle of firms included in this study.
Table 4 provides the Pearson correlations among the variables included in the empirical model. First, in accordance with Roychowdhury (2006) and Cohen and Zarowin (2010), there is a significant and negative correlation between AbnCFO and AbnDiscExp (Pearson correlation of -0.47), which suggests that firms use those types of real earnings management as substitutes. Furthermore, AbnProd is significantly and negatively correlated
with both AbnCFO and AbnDiscExp (Pearson correlations of respectively -0.25 and -0.31), indicating that firms combine overproduction (higher AbnProd) with either sales manipulation (lower AbnCFO) or reduction of discretionary expenditures (lower AbnDiscExp), consistent with Cohen and Zarowin (2010) and Zang (2012). Moreover, the significant and negative correlation between REM_Index and COO_IncRatio (Pearson correlation of -0.06) suggests that, contrary to the hypothesis of this study, real earnings management may be negatively associated with Chief Operating Officer equity incentives. Besides, the equity incentives of CEOs and COOs are also significantly and positively correlated (Pearson correlation of 0.46) and financial healthy firms are likely to award their COOs and CEOs with higher equity incentives (significant correlations between ZScore and COO_IncRatio/CEO_IncRatio of 0.25 and 0.32, respectively).
Other factors that significantly impact the magnitude of real earnings management are the firm’s financial health (ZScore), the presence of a Big 4 auditor and potential penalties of litigation. In accordance with Zang (2012), ZScore is significantly and negatively correlated with REM_Index (Pearson correlation of -0.16), which suggests that more healthy firms are less likely to engage in real earnings management, especially through sales manipulation (Pearson correlation of 0.16 with AbnCFO) and overproduction (Pearson correlation of -0.14 with AbnProd). On the other hand, consistent with Cohen and Zarowin (2010), firms audited by a Big 4 auditor are associated with higher levels of real earnings management (significant and positive Pearson correlation of 0.06 between Big4 and REM_Index), suggesting that scrutiny provided by the accounting practice constrains accrual-based earnings management. Furthermore, potential penalties of litigation are expected to incentivize firms to shift from accrual-based to real earnings management (Zang, 2006; Cohen and Zarowin, 2010), which is supported by a positive and significant Pearson correlation of 0.05 between Litigation and
REM_Index.
Moreover, MarketShare is positively and significantly correlated with both Age and
Size (Pearson correlations of 0.28 and 0.44 respectively), consistent with Zang (2012). Bigger
firms are likely to be financially less healthy and are likely to have a higher leverage (Pearson correlations of -0.32 and 0.41 respectively). Lastly, there is a significant and negative
correlation between Leverage and ZScore (Pearson correlation of -0.56), suggesting that less healthy firms generally have a higher debt/asset-ratio than financially healthy firms.
4.2 Multivariate analysis
To test whether real earnings management is positively associated with equity
incentives of Chief Operating Officers, the regression model of real earnings management on COO incentives, discussed in section 3.3, is performed for each of the seven measures of real earnings management (REM_Index, REM_1, REM_2, AbnCFO, AbnDiscExp, AbnProd and
AbnGainAsset). The results of those regressions are summarized in Table 5. Within Table 5,
there is a distinction between the measures of real earnings management that should be
positive when a firm is engaging in upwards real earnings management (REM_Index, REM_1,
REM_2, AbnProd and AbnGainAsset) and those that should be negative if a firm manages its
earnings upwards by means of real earnings management (AbnCFO and AbnDiscExp), resulting in opposite expected signs on the independent variables included in the models.
As can be seen in Table 5, a positive coefficient on COO_IncRatiot-1 is expected for
REM_Index, REM_1, REM_2, AbnProd and AbnGainAsset, as positive values for those real
earnings management measures indicate that a firm managed its earnings upwards by means of real earnings management. For two of the three comprehensive measures of real earnings management (REM_1 and REM_2) this coefficient is indeed positive (0.345 and 0.112
respectively). In contrast to the predictions, the coefficient on COO_IncRatiot-1 is negative for
both REM_Index, AbnProd and AbnGainAsset. However, none of the mentioned coefficients is significant at the 0.1 level.
On the other hand, the coefficient on COO_IncRatiot-1 is expected to be negative for
AbnCFO and AbnDiscExp, as firms that manage earnings upwards by means of those forms
real earnings management should be characterized by negative values for AbnCFO and
AbnDiscExp (Zang, 2006; Cohen and Zarowin, 2010; Chi, Lisic and Pevzner, 2011). For these
two measures of real earnings management, the findings are mixed as well. In accordance with the expectations, I find a negative (but insignificant) association between the equity incentives of COOs and AbnDiscExp. However, the coefficient on COO_IncRatiot-1 is
positive and significant at the 0.1 level for AbnCFO (0.171), which I expected to be negative. So, the coefficient on COO_IncRatiot-1 is only significant in predicting the abnormal level of
cash flow from operations, and this coefficient implies that firms are less likely to engage in real earnings management by means of sales manipulation if their COO’s potential total compensation is more closely tied to the value of their stock and option holdings. Therefore, none of the four sub hypotheses of this study is supported by the results of the main analysis, due to a lack of significance of most of the coefficients on COO_IncRatiot-1 and a significant,
but positive, association between COOs’ equity incentives and the abnormal level of cash flows from operations (which should be negative to support H1a). While prior studies
conclude that executives with high equity incentives opportunistically manipulate earnings by means of accrual-based earnings management to increase the value of their stock and option portfolios (Cheng and Warfield, 2005; Burns and Kedia, 2006; Bergstresser and Philippon, 2006; Jiang, Petroni and Wang, 2010), I do not find any evidence that higher equity incentives induce COOs to engage in real earnings management. The same holds for the equity
incentives of CEOs. The coefficient on CEO_IncRatiot-1 is insignificant in all the regressions,
suggesting that the equity incentives of CEOs do not induce those executives to engage in real earnings management.
I controlled for the ability to engage in accrual-based earnings management and the costs for both accrual-based and real earnings management, which are used by managers as substitutes, in the regressions. As higher net operating assets in the previous year, the
presence of a Big 4 auditor, a longer relationship between a firm and its auditor and operating in an industry with high litigation risk are all expected to reduce (the ability to engage in) accrual-based earnings management (Myers, Myers and Omer, 2003; Zang, 2006; Cohen, Dey and Lys, 2008; Cohen and Zarowin, 2010; Chi, Lisic and Pevzner, 2011; Zang, 2012), and are therefore expected to increase the use of real earnings management, the coefficients on
NOAt-1, Big4t, Auditor_Tenuret and Litigationt are predicted to be positive (negative) in the
regressions shown in the first (second) part of Table 5. In contrast to Cohen and Zarowin (2010) and Zang (2012), I do not find any evidence that there is a significant association between net operating assets at the beginning of the year and the various methods of real earnings management. For just one of the regressions, the Big4 coefficient is significant with the predicted sign (-0.062 for AbnCFO), indicating that the presence of a Big 4 auditor increases the use of sales manipulation, possibly due to a decreased ability to engage in accrual-based earnings management. The significant coefficients on Litigationt (-0.224,
-0.179 and 0.177 for REM_1, REM_2 and AbnDiscExp, respectively) all indicate that, in contrast to the prediction and the study of Cohen and Zarowin (2010), firms that operate in an industry with high litigation risk engage less in real earnings management than firms that face less litigation risk. Lastly, the coefficient on the Auditor_Tenuret variable is significant with
the predicted sign for AbnCFO (-0.003), suggesting that a longer firm-auditor relationship increases sales manipulation, possibly because it reduces the ability to use accrual-based management (Myers, Myers and Omer, 2003). On the other hand, the significant coefficients on this variable for REM_1 and REM_2 (-0.009 and -0.005, respectively) do not have the expected sign, and therefore suggest exactly the opposite: a longer relationship between a firm and its auditor decreases real earnings management. This could be due to a more friendly relationship between the management of a firm and its auditor, because of a longer auditor tenure, which increases the ability of the firm to manipulate its earnings through accruals (Chi, Lisic and Pevzner, 2011). This is in contrast to prior studies of Cohen and Zarowin (2010), Chi, Lisic and Pevzner (2011) and Zang (2012), which conclude that a longer auditor tenure increases the use of real earnings management.
