The effect of CFO independency on the relation between CEO incentives and earnings management

(1)

Amsterdam Business School

Faculty of Economics and Business

The effect of CFO independency on the

relation between CEO incentives and

earnings management

Student: Enes Karatas

Student number: 10876529

Date: 20-06-2016

Education: MSc Accountancy & Control, specialization Control Supervisor: Mr. A. Sikalidis

(2)

2

Statement of Originality

This document is written by student Enes Karatas who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

3

Abstract

In this study, a database research is conducted to examine the moderating effect of CFO independency on the relation between CEO incentives and earnings management. Specifically, it is hypothesized that CFO independency weakens this relationship. I define independent CFOs as executives with long tenures and work experience in a company who are able to strongly monitor CEOs. To construct CEO incentives, compensation information including short and long term incentives are taken. Using compensation and earnings management data over a period of time between 2006-2015, I find evidence that CFO independency weakens the relationship between CEO incentives and accrual-based earnings management, while there is not sufficient evidence that it weakens the relationship between CEO incentives and real earnings management. This finding is in accordance with the argument that CFOs have the fiduciary responsibility to produce financial statements, and thus are better able to affect accruals. Moreover, I find that CEO incentives are positively related to real earnings management, while it is negatively related to accrual-based earnings management. This finding is consistent with the argument that CEOs prefer real earnings management, because it is harder to detect.

Key words: CFO independency, CEO incentives, accrual-based earnings management, real earnings management, bonus incentives, equity incentives, agency theory. Data Availability: Data used in this study are obtainable from sources that are mentioned in this study.

(4)

4

1. Introduction

Recent corporate accounting scandals in combination with the financial crisis in 2008 have led to broader discussions about financial reporting and executive compensations. A lot of research has been done about CEO incentives in relation with earnings management (Bergstresser & Philippon, 2006). This study will focus on CFO independency. More specifically, this research will use a setting which will look to the extent of earnings management where the CFO is appointed after the CEO compared to a situation where the CFO has a longer tenure than the CEO.

Prior research has shown that accounting misrepresentations can lead to extremely negative capital market effects (Karpoff et al., 2008). Accounting fraud issues have led to investor losses and overall reductions in investor confidence. The authors found that those firms that untruthfully increase their market value with $1, will be punished by the market with one dollar when its misbehavior is revealed plus an additional $3,08. Beasley et al. (2010) found in their study that top executive involvement in accounting fraud is very common. CEOs and CFOs were together involved in 89% of the fraud cases. CEOs were involved in 72% of the fraud cases, while CFOs in 65%.

Some may argue that CFOs will elaborate with CEOs to manipulate financial statements. This derives from the thoughts that CFOs are subordinates from CEOs. Feng et al. (2011) state in their paper that CFOs indeed fail to resist to pressure from CEOs. They fail to resist because CEOs can influence the future career opportunities and compensation of CFOs and this gives CEOs power to exert pressure to manipulate financial reporting. Also Friedman (2014) found that powerful CEOs reduce the independency of CFOs by pressuring the CFO to report bias which leads to implications for incentive compensation, reporting quality, firm value and information rents. However, the main task of the CFO is increasing the financial reporting quality. As a result, if the CFO is caught for manipulating the financials, this will lead to forced CFO turnover (Engel et al., 2015). The CEO bears less risk in such a situation because he has more tasks besides maintaining the financial reporting quality. This means that CFOs are main responsible for the financials and they have failed in their monitoring role if a firm gets accused of accounting manipulations. Matejka (2007) argues that CFO independency could be increased by involving boards or audit committees more in CFO hiring decisions, performance evaluation and in CFO retention decisions.

In the literature the agency theory suggests that managers (agents) behave in their own interest, by preferring leisure activities over productive effort, instead of behaving in the

(6)

6 interest of the owners of the firm (principal). This behavior will lead to an agency problem (Jensen and Meckling, 1976). Management control systems can solve this problem by aligning the interests between the principal and the agent, so that agents carry out organizational objectives and strategies (Merchant and Van der Stede, 2012). These management control systems are often complemented with incentive systems to mitigate the agency problem. This is done for example by combining an incentive with a stock price target. The CEO will get a reward when it succeeds to increase the share price with a certain amount. Remaining agency problems can cause financial losses, reputation damage, and possibly can even lead to organizational failure (Merchant & Van der Stede, 2012). The downside of management control systems is that CEOs can be inclined to manage earnings. This is done to increase personal wealth. There are two ways of earnings management: by using discretionary accruals (accrual-based earnings management) and by giving real economic decisions (real earnings management).

1.1 Research question

RQ; Does CFO independency has an effect on the relation between CEO incentives and earnings management?

The aim of this study is to research whether the appointment of the CFO before the CEO has an impact on the relation between CEO incentives and earnings management. We know that if the CFO is appointed by the CEO it is considerable that the CFO will be inclined to follow the CEO in many cases and will not question some actions that can be classified as misbehavior. Feng et al. (2011) state in their paper that CFOs indeed fail to resist to pressures from CEOs. They fail to resist because CEOs can influence the future career opportunities and compensation of CFOs and this gives CEOs power to exert pressure to manipulate financial reporting. But what if the CFO was already in place before a new CEO came to power. Will the CFO than act more independently? This question remains largely underexplored. In a figure it looks as follows:

Earnings management CEO incentives

(7)

7

1.2 Contribution of the study

This research will have a scientific contribution and a societal contribution. As mentioned before, there has been done a lot of research about earnings management and CEO incentives. Prior research has for example shown that CEO equity incentives will lead to earnings management or that earnings management occurs when a CFO is appointed by the CEO. This study will distinguish from prior research because it looks to an additional aspect, namely whether CFOs with longer tenures than CEOs will have an impact on the relation between CEO incentives and earnings management. This study will therefore give new scientific insights into existing literature, because this setting is not researched earlier.

This study will also provide a societal contribution. The findings of this study could be used for example by regulators, national governments and stakeholders. Conservatism is important in the accounting profession, which can be reached by low levels of earnings management. If for instance the findings will suggest a decrease in earnings management when CFOs have longer tenures than CEOs, the regulators can use these findings to propose in an official statement firms to bring CFOs to power for a minimum number of years to increase the independency of the CFOs, which will lower earnings management and increase conservatism. Further, stakeholders of firms can use these findings to enforce the management to hire CFOs with the intention to work together on the long run. Conversely, if findings suggest that independent CFOs do not decrease the relationship between CEO incentives and earnings management, then regulators, national governments and stakeholders should look to other possibilities to solve this issue.

(8)

8

2. Literature review

Organizations separating management and ownership face difficulties in determining proper compensations for top managers. The literature section will illustrate this issue by providing the key terminology and basic elements of this study in order to better understand the hypothesis and the importance of the research question.

This section starts with defining the key terminology and theories that will be frequently used during this study, in order to give the reader the knowledge that is necessary to understand the research and to avoid misconceptions. First, earnings management will be defined, the motives to use it and the types of earnings management will be explained. Secondly, the agency theory and the problems that can arise will be discussed. Furthermore, the essentials regarding incentives for executives will be defined and specifically equity incentives will be briefly analyzed. Since the executives are the key of this research, the role of CEOs and CFOs in earnings management will be discussed. Lastly, the hypothesis regarding the impact of CFO independency on the relation between CEO incentives and earnings management will be developed by using literature that is presented in this paragraph.

2.1 Earnings management

In this section the term ´earnings management´ will be discussed. The paragraph will start with defining the term earnings management before describing the two main types of it. Furthermore, the reasons and incentives to engage in earnings management will be presented. Lastly, four strategies in earnings management will be analyzed.

2.1.1 Accounting principles and the definition of earnings management

Organizations have the possibility to record their financial transactions in two different ways; using the cash-basis accounting method and the accrual accounting method (Kwon, 1989). The cash method records the effects of an event when it is realized. This means that revenues are recorded in the financial statements only when the cash is received, and expenses are recorded when the invoices are paid. This method has some downsides in maintaining the true economic value of a firm. Especially service providing firms can’t record revenues until they complete a job. This can distort the actual economic value of the firm, since those firms could have provided services that entitled them to a fraction of the earnings. Kwon (1989) suggests that the accrual method is therefore more useful, since this concept records the effects of

(9)

9 events when it occurs, rather than when the event is realized. This means that future income can be recorded partly as current revenues in the financial statements, which will provide a better overview of the true economic value of the firm. However, there are some issues that accrual accounting yields regarding managerial discretion. Accrual accounting is noisier than the cash method due to the estimation errors and/or the possible management manipulations (Kwon, 1989). For instance, managers can use less conservative estimation methods in order to reduce bad debt expenses so that earnings can be managed upwards (Chen et al., 2015).

The question remains whether these forms of managerial discretion can be considered as earnings management. Berstresser and Philippon (2005) state that one component of earnings management is that the use of accruals temporarily boosts or reduces reported income. According to Davidson et al. (2004) earnings management is using flexible accounting principles in order to influence reported earnings that result in a deviation of the true income number. Healy and Wahlen (1999) come up with a more comprehensive definition, but in essence they state the same; ‘Earnings management occurs when managers use judgment in financial reporting and in structuring transactions to alter financial reports to either mislead some stakeholders about the underlying economic performance of the company or to influence contractual outcomes that depend on reported accounting numbers’.

The use of judgment by managers doesn’t directly relate to earnings management. Managers are required to include discretion in their decisions regarding several business activities. Some examples are; depreciation methods, expenditures such as R&D, and lease contracts (Healy and Wahlen, 1999). Managers using their discretion for accounting choices that are acceptable for internal and external parties are not considered as managing earnings. However, some managers may be inclined to use the discretion for personal benefits or other reasons. The use of judgment by managers to meet benchmarks and mislead investors is considered as earnings management (Chen et al. 2015).

2.1.2 Types of earnings management

There are two main types to manage earnings; accrual-based earnings management and real earnings management. According to Zhu et al. (2015) accrual-based earnings management encompasses managerial involvement in the financial reporting process by making accounting choices and estimates within the general accepted accounting principles (GAAP). The authors argue that accrual-based earnings management is more costly in the short term, while on the long run it has less impact on firm performance. Furthermore, misleading accruals that are detected by auditors have to be reversed in future periods.

(10)

10 The academic literature makes a distinction between nondiscretionary accruals also known as normal accruals and discretionary accruals which are abnormal accruals (Geiger and North, 2006). Nondiscretionary accruals and discretionary accruals form together the total accruals. Accruals that are preordained to capture adjustments of normal business activities such as company size, industry and revenue growth are considered as nondiscretionary accruals. Conversely, discretionary accruals contain unexpected mechanisms that are reported by managers, which distort the reporting process due to the application of accounting rules (Geiger and North, 2006). Depreciation choices and inventory write-downs are some examples of discretionary accruals. Much empirical research in accounting has focused on abnormal accruals since this contains more managerial judgment and therefore more earnings management (Geiger and North, 2006).

Another way to manipulate earnings is real earnings management. Zhu et al. (2015) state that this type of earnings management occurs when managers take actions that influence the cash flow in a certain direction by changing the organizations operating, investing or financing transactions. Some examples of real earnings management is reducing R&D and advertising expenses, overproduction of products and boosting sales by giving price discounts. Real earnings manipulations can decrease the overall firm value since the actions that managers take to increase the earnings from the current period, can have a negative impact on future cash flows (Roychowdhury, 2006). For example, by giving price discounts with the intention to boost sales and thereby meet short term targets, can create expectations by customers to get the same discounts also in future periods. This will obviously lower the margins on future sales. Secondly, overproduction of products will lower the cost of goods sold in the current period. However, there are higher levels of inventories needed for the extra products which will cause more inventory holding costs. This will lead to lower earnings in future periods.

Depending on the circumstances managers prefer one strategy above the other. Even though real earnings management has more negative effects on overall firm value, it is not necessary the case that managers therefore will use accruals to manipulate earnings (Roychowdhury, 2006). As discussed before, using real earnings management will cause more costs in future periods. On the other hand using accruals bears more private costs, especially in the short run. Graham et al. (2014) have found that managers are inclined to use real earnings manipulations above accruals. The authors give as reason for this that real earnings manipulations are hard to detect by auditors because it has to do with operational activities, while accrual-based earnings management is easier detectable.

(11)

11

2.1.3 Reasons to engage in earnings management

According to Chen et al. (2015) managers use earnings management for promotion prospects. They found evidence that interim CEOs engage more in earnings management to improve firm performance than non-interim CEOs. They argue that interim CEOs in this way create a positive impression on shareholders who are involved in appointing executives, and thus increase their probability to be promoted to the permanent CEO position. Also Kuang et al. (2014) found that CEOs that are appointed from outside the company are inclined to engage in earnings management, especially in the short run. When they are offered to stay at the job for a longer term, than the earnings management measure decreases to a level which is equal to CEOs promoted from inside. This decrease occurs because the CEOs can face the long-term consequences of their actions.

Another phenomenon is to use earnings management for capital market motivations. Teoh et al. (1998) state that companies going public (initial public offerings) engage in earnings management to get a higher price for their shares and to attract investors to buy the shares. Investors that are unaware of the inflated earnings pay too much for the shares. However, over time after information is revealed about the financials of the company, investors can lose their optimism about the investment when the portrayed earnings are much lower than the actual earnings. In that case, companies using high degrees of earnings management before going public will face a higher share price correction.

Cahan (1992) states that firms manage earnings for regulatory motivations. According to the antitrust laws of the United States it is forbidden for firms to have monopolies in business sectors. Firms accused of creating monopolies can be investigated for monopoly-related violations. Cahan (1992) found that firms which are investigated use income-decreasing abnormal accruals in the investigation years in order to avoid new unfavorable ruling and the costs associated with it.

DeFond and Jiambalvo (1994) argue that firms have contracting motivations to engage in earnings management. Firms that need money from debtholders want this against the most favorable contract terms. Debtholders on the other hand constrain actions and decisions of managers that decrease the value of debtholder claims. The researchers found that firms seeking for debt covenants increase their earnings in the year prior to the collaboration to loosen their debt constraints and to lower interest obligations. Also in the second year they manipulate their earnings upward to have a better position during renegotiations of the debt agreement.

(12)

12 Earnings management is also used in combination with forecast management (Burgstahler and Eames, 2006). Managers of firms prefer zero and small positive earnings above small negative earnings. The authors found that firms’ financial reports contain a high frequency of zero and small positive earnings surprises, and a low frequency of small negative earnings surprises. Managers achieved this by managing earnings upward, and on the other hand managing forecasts downward. Finally, earnings management is also used to increase executive compensation contracts. Bergstresser and Philippon (2006) state that earnings management occurs when a large part of the compensation of CEOs comprises stock and option holdings. These findings will be further discussed in the paragraph about the role of the CEO and CFO in earnings management.

2.1.4 Strategies in earnings management

The earnings management issues discussed in the previous subparagraph are all part of strategies in earnings management. In this subparagraph four strategies in earnings management will be analyzed; income maximizing, income minimizing, big bath accounting and smoothing income (Scott, 2011).

The first strategy, income maximizing, is the most common strategy. Analysts, investors and boards are interested in the firms’ earnings. Managers are therefore inclined to use income increasing accruals to achieve preferred targets. Degeorge et al. (1999) state that managers manipulate earnings upward because they are interested to meet three thresholds. Firstly, managers use income increasing accruals to report positive profits, so above zero. Secondly, to perform at least equal compared to last years’ earnings. Lastly, managers manipulate earnings upward to meet analysts’ expectations.

The second earnings management strategy that managers use is income minimizing. As the term already indicates, managers are inclined to reduce earnings for several purposes. Perry and Williams (1994) investigated the relation between management buyouts and income-reducing accruals. They found that managers are inclined to use earnings manipulation when they plan a buyout transaction. By using income-decreasing accruals in earlier periods, the shareholders are more likely to accept a lower buyout price which makes it less expensive for the management to buy the company.

Another earnings management strategy is big bath accounting. This strategy contains only income-decreasing accruals. Managers that can’t meet their bonus targets, no matter which accounting method they use, are inclined to further reduce current earnings (Healy,

(13)

13 1985). They reduce earnings by accelerating write-offs or deferring revenues to increase the probability to meet their target in the following year.

Smoothing income is the last earnings management strategy. Managers engage in earnings manipulation to ‘smooth’ reported earnings. This means that managers switch income to achieve pre-determined targets. DeFond and Park (1997) found that managers switch earnings for both, to achieve current targets but also future targets. In a situation where current performance is ‘good’ and expected future earnings are ‘poor’, managers are inclined to switch earnings from the current period to the future. Conversely, when the current performance is ‘bad’ and the expected future earnings are ‘good’ than managers will shift future earnings to the current period.

2.2 Agency theory

In this section the ´agency theory´ will be discussed. This theory will lead to a better understanding of earnings management, because earnings management can be a result of agency problems. To achieve this, the costs that arise due to agency problems will be described. Furthermore, notions such as information asymmetry, risk attitudes and optimal contracts will be analyzed.

2.2.1 Principles of agency theory

Prior literature has used the agency theory as a tool to better understand executive compensations. Jensen and Meckling (1976) defined the agency relationship as a contract in which the stakeholders (principals) delegate corporate decision rights to executives (agents). The decision rights are assigned to executives because they have more knowledge about the financial situation of the firm. The stakeholders strive for a relationship in which the executives make decisions that are in the best interest of the company. However, executives will not always work in order to maximize the welfare of the principal. In several cases executives will behave according to their personal interests. The principal can take actions to limit the misalignment between the stakeholders and the agent (Jensen and Meckling, 1976). Principals can for example offer appropriate incentives to executives or they can incur monitoring costs to limit deviant actions of the agent. Another option is to give executives resources and thereby test the agent whether he/she will use this opportunity to spend the amount to maximize the welfare of the firm. These costs are qualified as bonding costs. If costs occur due to imperfect monitoring of the agent or inappropriate use of the bonding

(14)

14 expenditures, than these latter costs are qualified as residual loss. Jensen and Meckling (1976) define agency costs as the sum of; the monitoring costs, bonding costs and residual loss.

Despite the effort of shareholders, managers can engage in earnings manipulations due to information asymmetry (Beyer et al. 2014). This means that executives have private information which leads to informational advantage over shareholders. Information asymmetry contains moral hazard and adverse selection (Eisenhardt, 1989). Moral hazard problems occur when executives have private information about effort choices. In this situation executives can choose for leisure activities over productive effort. Adverse selection problems occur when executives have private information about certain characteristics. The executive will for example claim to have certain skills or abilities during the job application. Secondly, the executive will not reveal negative private information since this can reduce his/her job security. Adverse selection occurs here when the principal is unable to completely verify these abilities during the job application or after the executive has started working.

Another reason for the existence of agency problems is the difference in risk attitudes between the two parties. Eisenhardt (1989) argues that executives are often more risk-averse than shareholders actually want them to be. The economic resources and reputation of the executives are tied to the firms they work for, which makes it harder for them to take large risks since they have much to lose. Shareholders on the other hand, have investments in different companies which makes their holdings diversified. It is therefore easier for shareholders to take larger risks in order to maximize returns, because the collapse of one company will not hurt them financially that much.

The agency literature focuses on determining the optimal contract between the principal and agent to avoid or mitigate agency problems. In a situation in which the behavior of an agent is unobservable due to moral hazard or adverse selection issues, the principal has two options for optimal contracting (Eisenhardt, 1989). The first option is to invest in information systems such as budgeting systems and reporting procedures in order to get a picture of the agents’ behavioral activities. Another option is to implement an outcome-based contract to align the interests of the principals and agents. Outcome-based contracts contain risk because outcomes can only partly be addressed to behavior. Outcomes may also change for example due to competitor actions, government policies, technological developments and more. This contract ensures that the risk of outcome uncertainty is transferred and carried by the agent. Outcome-based contracts are attractive when the outcome uncertainty is low, which means that the costs of shifting the risk to the agent will also be low. Outcome-based contracts

(15)

15 become less attractive when the outcome uncertainty increases, which means that transferring the risk to the agent becomes more expensive.

2.3 Incentives

Organizations provide incentives to align their interests with those of the executives. Incentive systems are key management tools used by companies to influence the behavior of individuals and groups. These contracts tie rewards (and in a lesser extent punishments) to the performance evaluation of executives and can especially be important in maximizing shareholder value when agency problems often occur. There are three functions of providing incentives; effort effect, selection effect and retention effect (Merchant & Van der Stede, 2012). The effort effect informs and reminds executives about rewards that can be obtained when important and competing performance areas are beaten. For example, including ‘increasing stock price’ in the incentive contract as an important performance area, will stimulate the executives to exert extra effort to achieve this.

Another phenomenon is the selection effect. Compensation packages of executives contain both base salaries and performance-dependent rewards. The selection effect of incentives is there for companies that try to hire employees of a specific type. If for instance the company offers compensation contracts with low base salaries and high variable payments, then the company will likely seek for executives that are confident and non-risk averse.

A company can also decide to overpay their executives compared to the competitors in the industry. In such a situation, executives will be inclined to work hard to achieve corporate goals. Further, there will be an intention to stay at the company for a longer term, because dismissal will be costly for the executive. This example is also described as the retention effect.

An important consideration for companies is the way in which they provide incentives (Merchant & Van der Stede, 2012). Strong incentives will for instance stimulate effort towards the performance measures. The performance measures should therefore be in line with the main goals of the company. In most cases the objective of companies is to increase shareholder value, so the performance measures that are chosen should be of high quality, which means that it contains high congruence and low noise.

Although various forms of incentives exist, companies commonly use monetary incentives to reward their employees (Merchant & Van der Stede, 2012). One example of an

(16)

16 incentive that every company provides is salary increases. When this occurs due to compensation of the national inflation it can’t be considered as an incentive. Salary increases are incentives when the employee ‘earns’ it through high performance. Salary increases are valued by executives because it is not just a one-time payment, it applies to future years.

Another form of incentives are short-term incentives. It includes piece-rate payments, bonuses and commissions. These incentives can be earned through high performance measured over a period of one year or less. Firms providing these incentives try to reward executives in accordance with their contributions to the organization. This creates risk-sharing between the agent and principal which is valued by firms. Another feature that firms value is that these incentives are one-time payments instead of annuities. Bonus payments are not always rewarded through financial measures, but can also be provided with nonfinancial measures, such as customer/employee satisfaction. Banker et al. (2000) state that the major reason for firms to use nonfinancial performance measures is that these measures are able to better indicate future financial performance than the more ‘hard’ financial measures, such as earnings per share and return on investment. This is because financial measures only show the effects of the chosen activities on firm value, so it doesn’t comprise the effects that could occur in the future, i.e. it is short term oriented. Nonfinancial measures indicate whether the actions that are taken will result in an increase/decrease of the shareholder value in the future, i.e. it contains long-term information. Further, nonfinancial measures are less susceptible to noise than financial measures. Hay Group (2015) surveyed 147 CEOs and found that firms prefer to use financial measures despite the advantages of nonfinancial measures. There were no reasons given but one could be that nonfinancial measures are ambiguous and susceptible to manipulation (Merchant & Van der Stede, 2012). The interviewer could for instance only approach customers which are positive about the company and avoid interviews with customers who are negative, in order to get a higher score on customer satisfaction.

The last form of incentives are long-term incentives (Merchant & Van der Stede, 2012). The purpose of these incentives is to stimulate executives to maximize firm value on the long run. Secondly, by postponing the bonuses to a far future, executives are obliged to stay at the firm for a long period which means that firms can retain talented executives. Long-term incentives are provided over periods greater than one year, but often it covers a period of 3-4 years. One way to provide such incentives is for example to set a performance target for earnings per share (EPS) that has to be achieved within 4 years, the so-called end-of-period target. Another way to provide long-term incentives is with equity-based plans.

(17)

17

2.3.1 Equity incentives

Equity-based incentives are provided in order to stimulate executives to change their firms share prices upward. These incentives can be provided in three ways; stock option plans, restricted stock plans and performance stock/option plans (Merchant & Van der Stede, 2012). Stock option plans give executives the opportunity to buy shares against an exercise price (Hall & Murphy, 2003). These options cannot be exercised immediately but only after a pre-specified period. This is often after 3-5 years dependent on the terms of the contract that the executive and firm signed together. Stock options expire often within 10 years. The options generate money for the executives if for example the actual stock price after 3 years is higher than the exercise price. Conversely, if the actual stock price after 3 years is below the exercise price then vesting the options will not generate income; the executives will only get their invested money back. The executives will therefore be incentivized to exert extra effort to enhance the stock prices, otherwise they will miss a big amount of bonuses over a period of 3 years. Executives will be less motivated if they get the impression that an increase in the value of the shares is hard to realize, which will lead to retention problems for firms. The advantage for firms to provide stock option plans is that they can reward executives without paying cash (Hall & Murphy, 2003). However, a downside of stock options for firms is that executives are inclined to accept riskier projects because they only benefit from share price increases but they cannot be punished for decreases. Furthermore, since stock options are affected by market-wide events, executives can be rewarded because market-wide improvements can cause an increase in firms’ actual share price while in fact the executives and thus the firm did not perform well.

Restricted stock plans are another form of equity-based incentives that firms frequently use (Carlson & Vogel, 2006). Here, executives don’t need to purchase shares but receive these for free from firms. However, it is not allowed to sell the shares before a pre-specified period. This period is often between 3-5 years. Executives can exercise their shares after the vesting date only if they are still working there. If executives take another job before the stocks vest, they are obliged to give the shares back to the company. Restricted stock options generate in any case money for executives, even if the shares prices fall. The only requirement is continued employment. This gives the idea that the shares are given away to the executives without expecting any performance. Although this form of incentives will decrease retention problems for firms, on the other hand executives will not be inclined as with stock option plans to exert extra effort to increase the share prices.

(18)

18 Lastly, equity-based incentives can also be provided with performance stock or option plans (Merchant & Van der Stede, 2012). Here, executives receive the shares again for free but now they have to achieve stock and non-stock goals within a pre-specified period. The shares can’t be exercised without meeting the performance targets. Like in any target setting process it is important that targets are not easily achievable, because that will create the perception that the shares are given away. On the other hand the targets should not be set too high since this can lead to discouragement, excessive risk taking and frequent executive turnovers.

The optimal long-term incentive plan for executives depends on various factors. Laux (2015) investigated the impact of innovation and career concerns on long-term incentive plans, by using stock options, restricted stock and severance pay. Firstly, he found that firms considering innovation and growth as an important key for future performance and where executives are less concerned about the risk of dismissal, the optimal compensation packages consists of stock options and restricted stock. Stock options are provided to induce effort to discover innovative projects and restricted stock are there to avoid overinvestments in risky projects. Secondly, he found in a situation where firms consider innovation and growth as less important for future performance and where the risk of dismissal is high for executives, the optimal compensation packages consists of stock options and severance pay. Stock options are provided to incentivize innovative projects and severance pay to stimulate executives to be not risk-averse.

2.4 Role of the CEO and CFO in earnings management

This study focuses on two types of executives; CEOs and CFOs. The role of the executives in earnings management differ from each other since the tasks and incentives of the executives are different. Friedman (2014) describes the tasks of the executives in a very broad manner; CFOs have the function to manage the financial information and reporting system within a firm, while CEOs have the function to generate value for shareholders. Engel et al. (2015) state that the role of CFOs within organizations has expanded from basic fiscal management and financial reporting to more complex job responsibilities such as comprehensive financial reporting and analysis, regulatory compliance, risk management responsibilities and strategic business support. This means that with the specific knowledge of the CFO he/she is responsible for improvements in the financial reporting process. The CEO on the other hand has besides maintaining the financials the responsibility to lead the development and

(19)

19 execution of long term strategies in order to create shareholder value. Further, the CEO has to make day-to-day management decisions and implement the long and short term plans of the company. Lastly, the CEO has to communicate on behalf of the company to all involved parties such as the board of directors, shareholders, employees and public and government authorities.

In current corporate settings the separation of ownership and control leads to agency problems. Companies provide incentives to executives to better align the interests of shareholders and executives in order to maximize firm value. However, the provision of incentives can distort the behavior of executives and lead to earnings management. Many have argued that the explanation of this lies in equity incentives because these are tied to the personal wealth of executives. Baker et al. (2003) found that CEOs manipulate income downward before a period where options will be awarded to them. This is done to simplify the task to increase the actual share price above the exercise price in order to have a high option compensation. The findings regarding income decreasing events are even stronger when CEOs have the authority to announce earnings prior to the date that options are awarded. CFOs engage also in earnings management to maximize their personal incentives. Beaudoin et al. (2015) have conducted an experiment to examine the impact of incentive conflict and earnings management ethics on CFOs discretionary expense accruals. They found that when CFOs have personal financial incentives conflicting with corporate incentives, CFOs with ‘low’ earnings management ethics have a tendency to go for their personal incentives by booking higher expense accruals, in order to increase current period expenses and therefore maximize bonus potential over a two-year period. If CFOs have ‘high’ earnings management ethics, they will resist their personal incentives by booking lower expense accruals, in order to minimize expenses and thus achieve corporate targets.

The question remains whether the CEO or CFO has the most influence on earnings management. Jiang et al. (2010) has examined this question. They assume in their research that the equity incentives of CFOs are more vigorous than those of the CEOs in earnings management, because CFOs are mainly in charge for the financial reporting system and have therefore the most expertise over the financials. By using multiple proxies such as beating analyst forecasts, discretionary accruals and total accruals the authors found that the extent of earnings management is more increasing due to CFO equity incentives than to those of the CEO. The authors came to these results even though CFOs have lower equity incentives than CEOs. The findings of the research imply that CFOs have more influence on earnings management than CEOs. However, providing bonuses to CFOs for self-reported financial

(20)

20 performance raises questions about whether this is appropriate. Indjejikian and Matejka (2009) researched the CFO compensation practices of private and public firms. They found in their sample that private firms did not change their compensation structures of CFOs while public firms reduced with about 6 percent the bonuses of CFOs that were tied to financial performance. This implies that firms try to avoid manipulation practices of CFOs by reducing their bonuses that are contingent on financial performance, in order to increase the integrity of the financial reports.

Erickson et al. (2005) investigated the impact of equity incentives on accounting fraud using a sample containing firms that were accused of committing accounting fraud in the past and firms not accused of fraud. In contrast with all researches above they could not find any consistent evidence that providing stock-based compensation to executives was causing accounting manipulations. Beaudoin et al. (2015) states that in a situation where incentive conflicts (between equity incentives of CFOs and corporate incentives) is absent, CFOs with ‘low’ earnings management ethics have a tendency to do everything to reach corporate incentives by booking lower expense accruals, in order to minimize expenses and thus achieve corporate targets. If CFOs have ‘high’ earnings management ethics, they will resist pressures to reach corporate targets and book higher expense accruals to lower company-related earnings management. This will give a fair view of the firm.

2.5 Hypotheses development

The hypothesis which will be presented in this paragraph is based on the theory described above. The literature suggests that CEOs manage earnings to hit bonus-linked targets (Bergstresser & Philippon, 2006). In order to receive those bonuses, CEOs can be inclined to force CFOs to manipulate the financials to be able to reach the targets (Feng et al., 2011). In particular, the authors argue that CFOs collaborate with the CEO due to the pressures of them and the fact that CEOs can influence the future career opportunities and compensations of CFOs. Moreover, Friedman (2014) found that powerful CEOs reduce the independency of CFOs by pressuring the CFOs to report bias which leads to implications for incentive compensation, reporting quality, firm value and information rents. Similarly, Dichev et al. (2016) found in their CFO survey that 90% of the CFOs agreed that pressures from inside the company, is a foremost reason to manipulate earnings through the accounting system. However, recent studies have shown that the implementation of SOX of 2002 which requires that CFOs and CEOs need to certify the financial reports, has resulted that the importance of

(21)

21 CFOs have increased (Bedard et al., 2014). Currently, CFOs have the fiduciary responsibility to produce financial statements that present the firm’s financial performance as accurately as possible (Indjejikian and Matejka, 2009). Further, the authors state that despite the fact that CEOs have also responsibilities regarding the financials of firms, CFOs are main responsible for overseeing the financial reporting process because they have more expertise and capacity to determine what numbers are reported. Ali and Zhang (2015) found that stronger monitoring of the CEOs, results in a decrease of opportunistic behavior of CEOs, which means that especially the extent of earnings management in this situation decreases compared to firms that weaker monitor their CEOs. Another study of Aier et al. (2005) found that the number of years of experience that the CFOs have in the role of chief financial officer is negatively related to earnings restatements. This means that longer tenure as CFO in a certain company is related to less earnings manipulations. The authors also found that restating companies have CFOs with less financial expertise.

Based on the previous findings I predict that CFOs that have worked in a company for a long period, especially longer than CEOs, that these CFOs will be better able to stand against the pressures of CEOs to manipulate earnings and will have less career concerns. Secondly, these CFOs will have a greater understanding of accounting treatments unique to his/her company and will therefore be better able to oversee the financial reporting process. This expertise will make it possible for CFOs to stronger monitor the CEOs, and thus will enhance the independency of CFOs. Hence, it is predicted that if CFOs have a longer tenure than CEOs, then the relationship between CEO incentives and earnings management is weaker. This all leads to the hypothesis.

H1: Relation between CEO incentives and earnings management is weaker when there is CFO independency.

(22)

22

3. Research methodology

This paragraph starts with describing the data that is used, and the sample that is selected. Further, the proxies for earnings management are explained, which is divided in two sections; accrual-based earnings management and real earnings management. Lastly, the regressions that will be used to test the hypothesis and the control variables are provided.

3.1 Data and sample selection

The research question will be answered by conducting a database research. Firstly, for the moderating variable CFO independency, data from the Execucomp database will be collected, which means that the data sample will only consist of U.S. firms. Initially the intention was to use a sample between 2002-2015 in order to avoid the effects of the Sarbanes-Oxley act which was implemented in 2002. However, the data item CFOANN has only data from 2006 up till now. Therefore, the sample will cover the period between 2006-2015.

Also for the independent variable, data from the Execucomp database will be collected. Especially, compensation data of CEOs will be obtained. From 2006-2015 there are a total of 15.965 firm-years with compensation data available for CEOs. After winsorizing 1% of the extreme values from all variables, a total of 2190 cases were left where all variables were filled in. As in Burns and Kedia (2005) the pay-for-performance incentives of the CEO compensation will be measured. The authors separately measure the sensitivity of various pay-for-performance incentives such as bonus, options and restricted stock on earnings manipulations. I will measure the ratio of the total CEO incentives on total CEO compensation and then I will look to what extent the total incentives of CEOs form a reason to manipulate earnings when it is moderated by CFO independency. Therefore, I will first measure the total bonus of CEOs with the following formula:

Total bonus = bonus + non-equity incentive plan

Then I will measure the total incentive compensation ratio of CEOs, that will show the part of total compensation which is tied to incentives. The total incentive compensation ratio of CEOs will be used as main measure and will be computed as follows:

Ratio of CEO incentives = Total bonus + option awards + restricted stock awards ____________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________ Total compensation

(23)

23

3.2 Measurement of earnings management

Prior literature has measured earnings management commonly through discretionary accruals (Bergstresser & Philippon, 2006) and real activities manipulation (Zhu et al., 2015). Similarly to those researches, this study will also use the same proxies to measure earnings management. In the following subparagraphs the measurement of the proxies and the related variables are described.

3.2.1 Discretionary accruals

The first proxy, discretionary accruals, will be measured by using the cross-sectional version of the modified Jones model as it is done in the research of DeFond and Subramanyam (1998). This model is preferred over the normal Jones model, because due to its less restrictive data requirements it suffers less from measurement problems as the normal Jones model does. Secondly, the absolute value of discretionary accruals (ABS_DA) will be used for the main analysis, because earnings management can include both income maximizing and income minimizing events (Degeorge et al., 1999; Perry and Williams, 1994). Specifically the following regression will be estimated:

TAit / Ait-1 = β0 (1/ait-1 ) + β1(ΔREVit – ΔRECit) / Ait-1 + β2 PPEit /Ait-1 + β3IBXIit-1 /Ait-1+ εit ,

Where:

TAit = total accruals for firm i at year t;

Ait-1 = lagged total assets

ΔREVit = the difference in net revenues between year t and t-1;

ΔRECit = the difference in net receivables between year t and t-1;

PPEit = gross property, plant and equipment;

IBXIit-1 = income before extraordinary items at year t-1;

εit = the residual of the model of firm i in year t; and

(24)

24

3.2.2 Real earnings manipulations

According to Zhu et al. (2015) real earnings manipulations occur when executives take actions that influence the cash flow in a certain direction by changing the organizations operating, investing or financing transactions. Commonly, these actions are taken by executives to meet or beat earnings thresholds. The proxies for real earnings manipulations will be developed as in Kim et al. (2012). The following four measures will be used to detect real earnings manipulations: (1) abnormal levels of operating cash flows (AB_CFO), (2) abnormal production costs (AB_PROD), (3) abnormal discretionary expenses (AB_EXP) and (4) a combined measure of real earnings manipulations (COMBINED_REM). The abnormal levels of the first three measures will be measured as the residual from the relevant models that will be estimated by year.

The first measure, abnormal levels of operating cash flows, will be computed by first measuring the normal level of operating cash flows. This is essential because sales manipulations can lead to a decrease in current-period operating cash flows (Roychowdhury, 2006). Therefore the following model will be used:

CFOt /A t-1 = β0 + β1 (1/At-1) + β2(St /A t-1) + β3(ΔSt /A t-1) + εt,

Where:

CFOt = Total cash flow from operation in year t;

A t-1 = Total assets;

St = Net sales; and

ΔSt = the difference in net sales between year t and t-1

Subsequently the abnormal cash flow from operations (AB_CFO) will be measured. This will be done for every firm-year by taking the residual (i.e. εt) from the firm-year’s lagged assets

and sales.

Another way to measure real earnings manipulations is with abnormal production costs. According to Cohen et al. (2008) and Roychowdhury (2006) production costs (PROD) are defined as the sum of cost of goods sold (COGS) and changes in inventory (ΔINV) during the year, where expenses are seen as linear function of contemporaneous sales. The normal production costs will be estimated with the following equation:

PRODt / A t-1 = β0 + β1 (1/A t-1) + β2 (St /A t-1) + β3(ΔSt-1 /A t-1) + β4(ΔSt-1 /A t-1) + εt,

(25)

25 The abnormal discretionary expenses are the third measure of real earnings manipulations. Also here, first the normal level of discretionary expenses will be estimated following the researches of Cohen et al. (2008) and Roychowdhury (2006) using the next equation:

DISEXPt /At-1 = β0 + β1 (1/A t-1) + β2 (St-1 /A t-1) + εt,

Here DISEXPt is the discretionary expenses in year t, which is defined as the sum of

advertising, research & development (R&D) and selling, general & administrative (SG&A) expenses. For every firm-year, the abnormal discretionary expenditure (AB_EXP) can be found by taking the residual from the model.

The last measure of real earnings manipulations is a combined measure that takes the previous three real earnings manipulation proxies, AB_CFO, AB_PROD and AB_EXP, together (Cohen et al. 2008). This combined measure, COMBINED_REM, can be calculated with AB_CFO – AB_PROD + AB_EXP.

3.3 Regression models

The regression analysis is done to test the hypothesis. The first regression model will test accrual-based earnings management and the second will test real earnings manipulations. Both regressions will include the variable CFO_SCORE. The elementary model of the regressions is as follow:

DAti = β0 + β1(CFO_SCORE)t + εt

REMti = β0 + β1(CFO_SCORE)t + εt

Where

DAti and REMti denotes the discretionary accruals and real earnings manipulations for firm i in year t. CFO_SCORE means that if the score is 0 the CEO had a longer tenure than the CFO and with a score of 1 the CFO had a longer tenure than the CEO.

In order to improve the power of the regression models several control variables will be added. Depending on the circumstances firms prefer DA above REM, or vice versa. Roychowdhury (2006) found for example that managers are inclined to use real earnings manipulations above accruals. The author gives as reason for this that real earnings manipulations are hard to detect by auditors because it has to do with operational activities,

(26)

26 while accrual earnings manipulations are easier detectable. On the other hand, real earnings manipulations have more negative effects on overall firm value. Firms often use one manipulation strategy, but can also use a mix of strategies. As in Cohen et al. (2008) the regression model of DAti will include real earnings manipulations (COMBINED_REM) as a control variable to control for the substitutive nature of the manipulation strategies. Conversely the regression model of REMti will include discretionary accruals (DA) as a control variable. Furthermore, for both regression models, SIZE and growth opportunities (MB) are added as control variables. This is done to control for systematic variation in abnormal CFO, discretionary expenses and production costs (Roychowdhury, 2006). Growth opportunities, described in the regression model as MB (market-to-book-ratio), will be measured by taking the ratio of market value of equity to book value of equity. Further, to control for firm performance, ROA and LOSS will be included in the regression models as control variable. According to Dechow, Sloan and Sweeney (1995) a positive relationship between ROA and earnings management can be expected. Also Minutti-Meza (2013) expects higher absolute discretionary accruals for clients with negative net income (losses).

Another control variable that will be added to the regression models is BIG4. According to Becker et al. (1998) the level of earnings management is lower for firms that are audited by big audit firms compared to firms that are audited by non-big audit firms. The authors state that big audit firms are able to provide higher quality audits which results in less possibilities to manage earnings. Also Chi, Lisic and Pevzner (2011) state that firms audited by big audit firms have lower levels of earnings management. Further, they argue that this leads to a switch from discretionary accruals to real earnings manipulations. This means that the control variable BIG4 is expected to be negatively related to discretionary accruals and positively related to real earnings manipulations.

Leverage is also a control variable that will be used in both regression models. Watts and Zimmerman (1990) have tested the debt/equity hypothesis which indicates that firms tend to shift earnings from future periods to current periods when their debt/equity ratio becomes larger. This means that the variable LEV is expected to have a positive relationship with earnings management.

According to Osma and Young (2009) firms adjust R&D expenditures downward to boost short-term earnings. This is especially the case when firms fail to report positive earnings or when they want to achieve earnings growth. Therefore, the R&D intensity variable is expected to have a positive association with earnings management. Another discretionary expenditure that will be included in the regression models as control variable is

(27)

27 advertising expenditures. According to Roychowdhury (2006) advertising expenditures are expensed in the same period as they are incurred. Hence, firms can be inclined to reduce advertising expenditures to increase earnings, in order to meet earnings targets. The variable advertising intensity is expected to have a positive relation with earnings management.

According to Burns and Kedia (2005) stock market performance (STOCK_PER) has an effect on aggressive accounting practices. CEOs for instance, benefit from an increase in the stock price, while they do not or limitedly get penalized when the stock price decreases. Therefore, CEO wealth that is tied to stock performance is positively related to earnings management.

This all combined yields the modified regression models which are shown below:

ABS_DAt = β0 + β1COMBINED_REMt + β2Sizet-1 +β3MBt-1 + β4STOCK_PERFORMANCE + β5ROAt-1 + β6LOSSt + β7BIG4t + β8LEVt-1 + β9RD_INTt + β10AD_INTt +

β11CFO_SCORE*CEO_INCENTIVES_RATIO + β12CFO_SCOREt + β13 CEO_INCENTIVES_RATIOt

REM_PROXYt = β0 + β1ABS_DAt + β2Sizet-1 + β3MBt-1 + β4STOCK_PERFORMANCE + β5ROAt-1 + β6LOSSt + β7BIG4t + β8LEVt-1 + β9RD_INTt + β10AD_INTt +

β11CFO_SCORE*CEO_INCENTIVES_RATIO + β12CFO_SCOREt + β13 CEO_INCENTIVES_RATIOt

Where:

ABS_DA = The value of absolute discretionary accruals, computed using modified Jones model.

REM_PROXY = Sum of the three real earnings management proxies: AB_CFO, AB_PROD and AB_EXP. Also named COMBINED_REM.

AB_CFO = The level of abnormal cash flow from operations

AB_PROD = the level of abnormal production costs, computed by taking the sum of cost of goods sold and the change in inventories

AB_EXP = the level of abnormal discretionary expenses, computed by taking the sum of advertising expenses, R&D expenses and SG&A expenses

(28)

28 CFO_SCORE = measures the net score between the variables CFO_tenure and CEO_tenure. If the score is 0 the CEO had a longer tenure than the CFO, and with a score of 1 the CFO had a longer tenure than the CEO.

Size = logarithm of the market value of equity

MB = market-to-book equity ratio, computed as market value of equity divided by book value of equity (MVE / BVE).

STOCK_PER = stock performance, computed as (stock price end of the year -/- stock price at begin of the year) divided by stock price at begin of the year

ROA = return on asset, computed as income before extraordinary items divided by lagged total assets

LOSS = indicates ‘one’ if net income is negative and ‘0’ otherwise

BIG4 = indicates ‘one’ if the firm is audited by a Big 4 audit firm and ‘0’ otherwise LEV = leverage, computed as long-term debt divided by total assets

RD_INT = R&D intensity, computed as R&D expense divided by net sales

(29)

29

4. Results

This section is divided in two paragraphs. The first paragraph gives the results of the accrual-based earnings management model. The second paragraph gives the outcomes of the real earnings management model. The first paragraph starts with the descriptive statistics of the full sample where the initial results such as mean, median and other metrics of the variables are given. Then, the outcomes of the Pearson correlation matrix will be described for the full sample. Lastly, the main results regarding the impact of CFO independency on the relation between CEO incentives and earnings management will be described which are given in the coefficients tables.

4.1 Accrual-based earnings management model

In table 1 the descriptive statistics are given for the variables that are used in the regression models. The overview shows that on average the absolute discretionary accruals (Abs_DA) of the sample firms are about 4,8% of their total accruals. Further, COMBINED_REM is 12,6% which means that from the $100 dollar earnings on average $12,60 is earned through real earnings management. Furthermore, the average size of the companies is 3,3 which means that their assets are on average 8,4 billion dollars. The mean of the market-to-book ratio indicates the growth opportunities for the sample firms. With an average MB of 3,3 there are substantial possibilities for the sample firms to grow. The stock prices of the sample firms are increasing with an average of 13,6%. Further, the ROA indicates that firms receive on average 5,4% return on their assets. The variable LOSS indicates that 16% of the firms have on

Mean Median Std. Deviation 25th percentile 75th

percentile Min Max N Variables Abs_DA 0,0482 0,0319 ,05370 0,0147 0,0607 0,00 0,41 2910 COMBINED_REM 0,1258 0,1062 ,42498 -0,1434 0,3652 -1,60 2,19 2910 LOGSize 3,3054 3,2332 ,76297 2,7759 3,7975 58,00 5,63 2910 MB 3,2766 2,5242 2,67808 1,6436 3,8880 0,17 20,44 2910 STOCK_PERFORMA 0,1360 0,0940 ,44219 -0,1376 0,3563 -0,76 2,26 2910 ROA 0,0537 0,0634 ,09408 0,0260 0,1013 -0,63 0,34 2910 LOSS 0,1601 0,0000 ,36680 0,0000 0,0000 0,00 1,00 2910 BIG4 0,8883 1,0000 ,31503 1,0000 1,0000 0,00 1,00 2910 LEVERAGE 0,1453 0,1106 ,15272 0,0000 0,2477 0,00 0,77 2910 RD_INT 0,0616 0,0235 ,08154 0,0000 0,1056 0,00 0,85 2910 AD_INT 0,0224 0,0100 ,03214 0,0041 0,0293 0,00 0,26 2910 CEO_Incentives_CENT 0,0000 0,7708 ,23736 0,6054 0,8542 0,00 1,00 2910 CFO_Score_CENTRE 0,0000 0,0000 ,41169 0,0000 0,0000 0,00 1,00 2910 Interaction 0,0051 ,09768 2910

Table 1: descriptive statistics of full sample

(30)

30 average a net loss. BIG4 indicates that about 89% of the firms have a BIG4 auditor. The control variable LEVERAGE indicates that firms finance 14,5% of their assets with debt. Firms spend on average 6,2% of their net sales for research and development. Further, the advertising expenditures are for the sample firms on average 2,2% of their net sales. Lastly, the independent variable and moderator are mean-centered to control for their interaction. This resulted in values equal to 0. Before mean-centering CEO_Incentives_Ratio, it was 0,69 which means that 69% of the total compensation of CEOs comprises incentives. CFO_SCORE was 0,22 which means that in 22% of the cases CFOs have longer tenures than CEOs.

Table 2 gives an overview of the Pearson correlations. The purpose of the correlation matrix is to control for multicollinearity. This latter exist when the correlation between two variables is above 0.90. In order to look to the moderating effect of CFO_SCORE on the relation between CEO_INCENTIVES and earnings management, an interaction effect is implemented in the analyses. Because this interaction effect correlated for 94% with CFO_SCORE, mean-centering was conducted to solve the multicollinearity issue. The correlation between the two variables is now 8,6% (0.086).

The first thing to note in the table is that the variable NET LOSS has the highest impact on accruals (0,241). Further, it is observed that larger firms give their CEOs higher compensations (0,447). However the size of firms is negatively related to accrual-based earnings management (-0,194). Further the table shows that CEO_INCENTIVES have a significant negative correlation with accruals (-0,091), while prior studies found a positive relation between CEO incentives and accrual-based earnings management (Cohen et al., 2008). CFO_SCORE has also a significant negative correlation with accrual-based earnings management (-0,057) which is consistent with the expectations. The

(31)

31 INTERACTION_EFFECT has also a significant negative correlation (-0,062) which was predicted.

Using real earnings management (COMBINED_REM) as dependent variable, we see that R&D intensity has the highest significant correlation with the main variable (0,296). Further, the table shows that CEO_INCENTIVES are significant and positively correlated (0,096) to real earnings management as expected. Another finding is that CFO_SCORE is negatively related (-0,053) to real management which is consistent with the predictions. Lastly, the INTERACTION_EFFECT is negatively related (-0,002) to real earnings management which is consistent with my expectations.

Table 3 gives an overview of the coefficient table. What immediately is noticeable is that the control variable LOSS has again the most effect on accruals. This variable has the highest significant beta (0,228; p = 0,000). This can be justified, because accrual-based earnings management is a good tool to manipulate financial statements. CEOs prefer zero and small positive earnings above small negative earnings (Burgstahler & Eames, 2006). In cases where firms have losses, CEOs are inclined to use income increasing accruals in order to report positive profits (Degeorge et al., 1999). At this point, earnings management is not done for

Standardi zed Coefficie nts B Std.

Error Beta Tolerance VIF

(Constant) ,065 ,005 12,154 ,000 COMBINED_REM ,001 ,003 ,008 ,361 ,718 ,669 1,494 LOGSize -,012 ,002 -,176 -7,559 ,000 ,573 1,744 MB ,002 ,000 ,078 3,793 ,000 ,732 1,366 STOCK_PERFORMANCE ,007 ,002 ,059 3,204 ,001 ,907 1,102 ROA ,024 ,012 ,042 1,949 ,047 ,683 1,464 LOSS ,033 ,003 ,228 10,970 ,000 ,720 1,389 BIG4 ,013 ,003 ,074 3,908 ,000 ,865 1,156 LEVERAGE -,015 ,007 -,042 -2,084 ,087 ,753 1,328 RD_INT ,013 ,013 ,020 1,021 ,307 ,803 1,245 AD_INT ,069 ,031 ,042 2,204 ,028 ,875 1,142 CEO_Incentives_CENTRERED ,000 ,005 -,001 -,026 ,979 ,767 1,303 CFO_Score_CENTRERED -,005 ,002 -,039 -2,151 ,032 ,966 1,035 Interaction -,027 ,010 -,050 -2,794 ,005 ,982 1,018

Table 3: Coefficients table with Abs_DA as dependent variable

Coefficients Variables Unstandardized Coefficients t Sig. Collinearity Statistics

The effect of CFO independency on the relation between CEO incentives and earnings management

Amsterdam Business School

Faculty of Economics and Business