The Effect of CEO Compensation on Real Earnings Management

(1)

Degree Project, 30 hp

International Business programme, 240 hp Spring term 2020

Supervisor: Jesper Haga

The Effect of CEO

Compensation on Real Earnings Management

Douglas Grambo

(2)

1

Abstract

Keywords: Agency theory, real earnings management, CEO compensation

Real earnings management has been a subject of increasing debate ever since the passing of the Sarbanes-Oxley act in the united states. As research has pointed towards real earnings management increasing this has sparked discussions on whether real earnings management is damaging to companies, or if it is benefiting them, or if it lies somewhere in between. For this paper we wanted to examine how the financial incentives of a CEO would affect the usage of real earnings management. Are CEO’s being poorly motivated, and as a result harming their companies?

To guide the paper, we decide to formulate our research question thusly:

How do different forms of CEO compensation affect real earnings management?

In this paper we attempt to find correlations between indicators of real earnings management and three different forms of CEO compensation. For our indicators we follow to a paper by Roychowdhury, titled “Earnings Management Through Real Activities Manipulation” and calculate abnormal cash flow from operations, and abnormal production. These indicate usage of overproduction, reduction of discretionary expenses, and moving sales across periods (Roychowdhury, 2006).

For forms of CEO compensation, we measure them as a ratio of total compensation. We track salary, bonuses, and stock ownership. In our results we can see that all three of these are significantly correlated to both of our real earnings management indicators. Bonuses have a positive correlation to abnormal production, and a negative correlation to abnormal cash flow from operations. Salary is positively correlated to both our indicators, and ownership is negatively correlated to both our indicators. Our final conclusion is that yes, the makeup of a CEO’s compensation has a significant effect on the usage of real earnings management within the company.

Acknowledgements

I would like to thank my supervisor, Jesper Haga for his invaluable guidance in the writing of this paper, and for answering my many confused questions with endless patience. I would also like to extend my thanks to my parents, and girlfriend for putting up with my frustrated ramblings when the writing process got the better of me. Finally, I would like to thank Umeå University for all the assistance offered during the writing process, even as the spread of COVID-19 is hampering many efforts and making many ordinary methods unfeasible.

(3)

2

Clarifications of Terms

SOX - Sarbanes-Oxley Act. The Sarbanes-Oxley Act was passed by U.S congress to increase oversight over auditors, and to ensure an auditor’s independence from their clients.

ACFO - Abnormal Cash Flow from Operations

CFO - Cash Flow from Operations

APROD - Abnormal Production

COGS - Cost of Goods Sold

R&D - Research and Development

NPV - Net Present Value

SG&A - Selling, General, and Administrative Expenses

SIC - Standard Industrial Classification, a code assigned to a company to classify them by company activity

The board/A board, etc. - Board of directors

(4)

3

1. Introduction

In this chapter we give a basic explanation as to the background knowledge required to understand the subject, as well as explaining why we find the subject to be interesting, and why the subject warrants study.

1.1 Research Background

Real earnings management is defined as manipulating earnings by changing the firm’s operations with the intention of meeting certain thresholds (Roychowdhury, 2006). A common method of doing real earnings manipulation is for a firm to reduce their expenditures on certain accounts, usually accounts like research and development, commonly referred to as R&D (Roychowdhury, 2006). This is done to reduce costs for a period, resulting in higher profits. This is often done to meet or exceed certain previously set benchmarks for profit, in the hopes of improved stock performance (Gunny, 2010).

Real earnings manipulation can be an interesting subject for the owners of any company as managers manipulate their departments performance, risking decreasing the long-term performance of a company, usually for a perceived personal gain (Gunny, 2010). In certain situations, however, there has been evidence that both forms of earnings manipulation can in fact increase company value in the long term, by creating short term gains that can be tapped to increase long term gains by more than it has been reduced through the manipulation (Gunny, 2010). Managers have different reasons for committing to real earnings management as there are many incentives for it. A manager might want to look better for one quarter, even at the cost of the next for instance. If a manager has met their target for a bonus one year, they might attempt to move further business forward in time until the next year in order to more easily meet the target again in the next period (Roychowdhury, 2006). In the same vein, a CEO could manage earnings to improve short term gains, at the cost of long term performance if they find that that will net them something, usually bonuses, or other financial incentives (Bergtresser and Philippon, 2006).

At its core earnings management is a problem best explained by, and examined through the lens of, agency theory. The agents, in this case managers and CEOs, have incentives that go directly against the incentives of their principal. In this case the principal could be either their own upper management, or the board of directors and the investors, this is referred to as the “agency problem” (Eisenhardt, 1989). It then stands to reason that altering the managers’ or CEO’s compensation will alter their incentives and either reduce or increase the level of real earnings management.

Real earnings management has been found to potentially significantly negatively impact company performance (Gunny, 2010). However, using real earnings management exclusively to meet the benchmarks not only removes this negative impact, but actually provides short term gains that can be utilized to increase future performance (Gunny, 2010). It has been found that real earnings management is associated with significant positive future performance when benchmarks were met, but negatively associated when they were not, further supporting the conclusions made by Gunny earlier (Al-Shattarat et al., 2018).

(7)

6

This divides real earnings management into two forms, opportunistic, and non- opportunistic.

Opportunistic

This form of real earnings management refers to when an agent is using real earnings management not to meet external market expectations or benchmarks, but to increase metrics in the short term that will instead provide value to the CEO instead of the company. In this case the agent is causing damage to it’s principal in the pursuit of real or perceived personal gains. Companies will want to minimize this to decrease the agency cost of the company, but do so effectively, as to not spend more on reducing it than they would save. (Gunny, 2010)

Non-opportunistic

Non opportunistic real earnings management refers to when a manager is using real earnings management to meet external benchmarks or expectations that will increase company value in the long term. In this case, according to several papers made several years apart, (Gunny, 2010) (Al-Shattarat et al., 2018) value is being created for the principal as short term gains created from real earnings management are reinvested to create more long term value than what was lost from the real earnings management. Value being created means that the decision is not being made opportunistically by a manager, but rather is the correct course of action for the company as a whole. Companies want to maximise this as it is correct behaviour in a functioning agent-principal relationship, again however, it must be done effectively, to avoid spending more than would be gained.

Note that there is some dispute and debate within the scientific community as to whether or not meeting benchmarks with real earnings management is actually creating value, and whether this is always the case, or if it is ever the case. Some argue that evidence for any of these theories are inconclusive. But for this study we will operate with our assumptions based on the findings of these papers (Gunny, 2010) (Al-Shattarat et al., 2018). While we have found that the evidence presented in these papers is conclusive enough for the purposes of our paper, we felt that it would be prudent to notify the reader of the existence of this debate so that they might form their own opinion on the subject.

This paper will focus on real earnings management used by CEOs because of their role as final decision maker, and an interesting source of the agency problem between investors and the CEO. A common way CEOs manage earnings can be by manipulating investing activities to improve financial statements (Gunny, 2010). Of course, the markets are aware of the occurrence of firms using real earnings management to attempt to improve the appearance of their numbers, and firms that engage in more of it have been found to have a higher cost of capital, as lenders want a premium for the risk they are taking on (Kim and Sohn, 2013). Furthermore it was found that in the presence of sophisticated investors, real earnings management was less effective as investors were able to detect that the earnings had been managed to improve financial statement performance, and would make investment decisions accordingly (Roychowdhury, 2006).

There has been an overall increase of real earnings management since the beginning of the twenty-first century (Cohen et al., 2008). While real earnings management requires

(8)

7

more effort from the actor, by forcing a company to alter its operations rather than just its accounting practices, companies are clearly still of the opinion that the gains of earnings management are worth even this higher cost. Research has shown that the passing of Sarbanes-Oxley act in 2002, usually referred to as SOX, in the U.S.A after the bankruptcy of Enron Corporation reduced the usage of accruals earnings management (Cohen et al., 2008). Accruals earnings management is a form of earnings management that involves only accounting, and no changes to a company’s operations. The act sought to increase oversight over auditors and limit an auditor’s ability to provide non-auditing services to clients (SOX, 2002). It would seem that as accrual earnings management got harder, and in many cases illegal, companies have phased out accrual earnings management in favour of real earnings management.

But why might companies engage in real earnings management, at cost to their long-term value? One answer could be the phenomenon of short-termism. Short-termism is defined as when companies make “Decisions in which firms pursue short-term gains (for example, seeking to maximise quarterly profits) at the expense of long term strategies (for example, investing in basic research or laying the groundwork for new core competencies.)”

(Laverty, 2004. p. 1).

One reason for why companies tend to favour short term gains so much more than long term gains is risk aversion. Managers are aware that they might not stay at a company long enough to enjoy the benefits of any long term decision making, and therefore decide to maximise gains in the short term, when they are more likely to still be at the company and reap the rewards of the increased performance (Palley, 1997). This combined with high pressure from other managers and the markets as a whole, to meet and exceed benchmarks and projections put high pressure on managers to meet short term goals, but little focus is put on any long term gains (Dichev et al., 2013) (Palley, 1997). This research implies that these managers are not being compensated appropriately. While the company as an entity wants the largest gains, regardless of timeframe, they are failing to properly reward strong long term gains, causing decision makers to go for the safest option, taking quick short term gains, to secure themselves in case they lose their position in the company. Clearly, companies must make extra effort to motivate managers to make good long-term decisions or they are incentivised to risk the company’s future as a going concern for personal short term gains.

Highly connected to agency theory is information asymmetry, when two entities involved in something have different levels of knowledge about said thing (Johnson and So, 2018).

It has become a natural side effect of the way companies are structured. The head of a specific department will almost always have greater insights into the everyday workings of their department than the person overseeing this head of the department. This information asymmetry opens up the opportunity for real earnings management where the head of the department can make these self serving decisions without the upper management being able to distinguish whether or not that action is justified, since they lack the insights into the climate the department operates in. The same can be said for CEOs where they often can have greater insights than the board, and much greater insights than investors. This is the reason we have decided to focus on compensation as a tool of impacting the occurrence of real earnings management. In theory, if an employees compensation is structured to reward only decisions that benefit the company

(9)

8

as a whole, opportunistic real earnings management would disappear, as the agent is now more incentivised to benefit the company, than to go against it (Dalton et al., 2007).

Little research is made on the effects of compensation on real earnings management. One study has found that “clawbacks”, the ability for the board to reclaim given bonuses upon discovering financial misstatements, would reduce the occurrence of accrual earnings management, but this also had the side effect of increasing real earnings management (Chan et al., 2015). This further strengthens the idea that real earnings management and accruals earnings management are connected. They fulfill the same need, but accruals is simply easier and cheaper to utilize than real earnings management, but when accrual earnings management is taken off the table, real earnings management quickly takes its place as a well functioning substitution.

Interestingly, research has however found that the higher executives beneath the CEO’s compensation is relative to the CEO’s compensation, the less likely real earnings management is to be used by these lower executives, and this effect was stronger when CEO’s are less powerful, and other executives have more influence in the company (Cheng et al., 2006). This seems to indicate that when managers below the CEO level are given enough compensation they either find the risk of backlash from opportunistic real earnings management is too high, either through an increase in risk of detection, or increase in the scale of consequences. Or the increased influence within the company is altering the manager’s perceptions on the benefits of real earnings management. It is also possible that when lower managers receive higher compensation, they are not just seeing their compensation upscaled, but perhaps also changed to include or increase methods of compensation that will somehow incentivise not pursuing real earnings management.

This paper is limited to only real earnings management and will not put any focus on accrual earnings management. This decision was made since there are easier ways than compensation to affect the occurrence of accrual earnings management. The passing of SOX made many older methods of accrual earnings management not legally usable, and larger companies are continuously audited to ensure accrual earnings management is kept to a minimum. This would mean that the effects of CEO compensation are likely insignificant when compared to these factors. Similar efforts are however not relevant to real earnings management, making it a much more interesting subject of further study.

The increase of the occurrence of real earnings management (Cohen et al., 2008) also indicates that it is a trend that is here to stay, and that large-scale accrual earnings management is a thing of the past. We are therefore interested to dig into the motivations that incentivise real earnings management and how they might be affected by changes in compensation.

Furthermore, this paper will also forego researching managers and executives below the CEO. One reason for this is the different nature of the agent relationship between a top manager and their lower department heads, and the agent relation between a CEO and their board of directors and investors. This relationship is very different, and it might prove extremely difficult, or even impossible, to explain both situations in one uniform paper, with a uniform theoretical framework. The main reason we chose CEO’s over other executives is that the CEO sems a more interesting fit as the decision maker who has final say of what goes on in the company. Finally, previous research on CEO behaviour and

(10)

9

compensation will serve as an excellent ground to stand on, that is not available for lower managers (Bergtresser and Philippon, 2006) (Demers and Wang, 2009) (Jiang et al., 2013). We do however still hope that some of the revelations made throughout this paper will still be applicable or useful outside of real earnings management by CEO’s, but that they could also be used to explain real earnings management by lower executives, at least to some degree.

The study will be limited to U.S equity markets, this decision was made due to ease of access to a large sample working under similar regulations. The U.S is also one of the largest markets in the world containing multitudes of companies and a very high CEO- to-worker pay ratio of 287 to 1 in the S&P 500 (AFL-CIO, 2018) meaning the effects of CEO compensation are likely to be magnified compared to areas where CEO’s hold a lower relative compensation since they stand to gain or lose more with changes in compensation. The sample is spanning a time period from 2006 - 2019. We do hold that these 14 years of observations should prove more than adequate for our study.

1.2 Research question

The background information provided, and surface level research made digging into the subject has led us to formulate a research question to serve as a cornerstone, and general guide in our paper.

How do different forms of CEO compensation affect real earnings management?

1.3 Purpose

This paper is attempting to answer not only whether or not CEO compensation affects real earnings management, but also which kinds of compensation affect what kinds of real earnings management, and how. It is our hope that this paper could shed light on reasons for real earnings management, and how CEO’s reason when they utilize them.

This could shed further light on the discussion as to whether or not earnings management is pursued opportunistically by CEO’s for their own gain, and whether or not CEO’s are of the opinion that earnings management has the capability to improve the metrics that their financial compensation is measured against.

1.4 Theoretical and practical contributions.

Theoretical contributions include a greater understanding of how compensation impacts the efforts and goals of decision makers in various business industries by observing the effects of different forms of compensation. It will serve to support previous research into the effects of CEO incentives on real earnings management by Bergtresser and Philippon (2006) as they focused on all forms of earnings management, and we narrow our research down to just real earnings management. Likewise, it will support research that has found that managers are more likely to utilize accruals earnings management if they have larger bonuses (Guidry et al., 1999). Furthermore, it will also bring a greater understanding of what motivates CEO’s to utilize real earnings management by observing and explaining what compensation methods are most likely to lead to increased or decreased real earnings management. The paper can also be used in the debate of defining opportunistic versus non-opportunistic real earnings management to observe how CEO’s perceive and

(11)

10

utilize real earnings management, and in doing so complement the research of Gunny (2010). It will do so by observing if there are any changes when financial incentives attempt to align the goals of the CEO’s with the goals of the company. Finally, our paper will unite agency theory, one of the broadest theories in business research, with our current understanding of real earnings management to attempt to see if they can support each other and function in one system. It will support Meek et al. (2007) as they researched the relationship between stock options and earnings management. Our research is broader in that we include more forms of CEO compensation, but also narrower in that we only observe real earnings management.

Practical contributions will be an empirical examination of how companies can affect usage of real earnings management in order to attempt to increase their efficiency and unity. It will help companies maximize any non-opportunistic real earnings management, while simultaneously minimizing costly opportunistic real earnings management. It will attempt to further explain whether real earnings management is always damaging to the company or if there are times when real earnings management could be a net zero, or even a net positive endeavour. Furthermore, it will serve as a basis to educate board members, investors and other agents in this principal- agent relationship as to what real earnings management is and how it impacts them. Finally, it can hopefully decrease some of the information asymmetry between CEO’s and their agents and thus create a healthier business environment for all actors.

(12)

11

2. Theoretical Framework

This chapter focuses on constructing a theoretical framework that will be used to guide our approach to the study. It will mention various previous theories and argue for how they relate to our specific case, and what expectations we can have going into this study.

2.1 Agency Theory

This paper is very tightly connected to agency theory. This is because the shareholders in a firm are owners of the company but have little oversight as to what is going on inside the company that they are owners of. Thus, the CEO in a firm could go against the interest of the people whose money they are supposed to utilize for the investor’s benefit and instead benefit themself. In agency theory this is called an “Agency Problem” (Dalton et al., 2007). Agency theory is entirely focused on motivation and incentives. There is only an agency problem if there is a real or perceived incentive for an agent to go against their principal’s interests (Eisenhardt, 1989). These incentives can be minimized, and systems can be established to incentivise agents to benefit their principal. These safeguards minimize the agency problem with a sort of stick and carrot approach, where an agent is rewarded for making the decisions that are perceived to benefit the principal, and punished for doing the opposite (Dalton et al., 2007).

Eisenhardt (1989) summarises previous research on the subject of agency theory, diving into its origins and the two main research fields, positivism and principal-agent.

Throughout the paper Eisenhardt remains cautiously optimistic in regard to the applicability of the theory and closes the paper by describing agency theory as “providing a unique, realistic, and empirically testable perspective of problems of cooperative effort.”

(Eisenhardt, 1989

p. 72). In the paper Eisenhardt notes several studies of positivist agency theory in which results were consistent with predictions from agency theory¹. Eisenhardt uses these results to argue that there exists an agency problem between top executives and investors, as described by agency theory, in various fields where executive’s and investor’s interests were theorized to diverge, such as acquisitions, takeover attempts, and debt versus equity financing, to name a few of the examples. The paper is important and fills the research gap of whether or not agency theory actually holds any water. It succinctly gives a list of several occasions in different fields in which the predictions of agency theory have come to pass and provides strong evidence that agency theory is a worthwhile field of study.

Agency theory has two separate viewpoints, positivist and principal-agent (Eisenhardt, 1989). Principal-agent research has been broader and focused on a general theory of any principal- agent relationship. In this paper, however, we will take the perspective of, and

put additional focus on, the positivist research into agency theory. Positivist agency theory is focused on finding situations where a principal and their agent have conflicting goals and attempting to find solutions to limit and reduce an agent’s incentives and

1 Walking and Long (1984) Amihud and Lev (1981) Wolfson (1985) Kosnik (1987)

(13)

12

abilities to go against the wishes of their principal. Furthermore, positivist agency theory has placed a large focus on the relationship between the owners and managers of large publicly traded companies, making it a perfect fit for our study. (Eisenhardt, 1989)

In the same paper from there is also a discussion on two main propositions of positivist agency theory. (Eisenhardt, 1989)

Proposition 1

When the contract between the principal and agent is outcome based, the agent is more likely to behave in the interests of the principal.

Proposition 2

When the principal has information to verify agent behaviour, the agent is more likely to behave in the interests of the principal.

While the two propositions are entirely complementary, and both very relevant to our paper we will put additional focus onto the first. The second proposition states that an agent will behave according to a principal’s interests when the principal can better monitor agent behaviour. The agent’s incentives are altered because of greater risk of negative consequences. The first proposition, however, focuses on aligning the interests of the agent with that of the principal through compensation formulated to reward better results, or outcomes, for the company. Since our paper is focused around CEO compensation, this first proposition is of particular note. (Eisenhardt, 1989)

2.2 Choice of management theory

In this paper we focus on the perspective of agency theory. In our opinion it is a perfect fit for out paper since we focus on the effects of CEO compensation’s ability to alter a CEO’s incentives. Agency theory is a great foundation for this type of research as it focuses entirely on the incentives of an agent.

Another theory that could have been used is stewardship theory. Stewardship theory argues differently from agency theory. Stewardship theory argues, unlike agency theory, that managers seek to maximise value for their company, even if it does not maximise their own utility (Davis et al., 1997). We decided against this since we have evidence that CEO’s will go against their company, if they feel they could receive benefits from doing so; research has found that managers are willing to manipulate earnings, at cost to the company if it would help them meet targets (Graham et al., 2005). There is also evidence that real earnings management is more common in public companies, where agency theory’s agent problem is held to be bigger (Haga et al., 2018).

Agency theory perfectly encapsulates the issue the paper is focused around and does so with no regard for any inherent wish of an agent to do good from a sense of morality (Eisenhardt, 1989). This is great for our paper as we also focus entirely on direct incentives and have no focus on a CEO’s motivations stemming from any other source than financial incentives.

(14)

13

2.3 Information Asymmetry

Information asymmetry is when two entities do not have the same information about something that is of interest to both of them (Johnson and So, 2018). In our case it is the difference in access to information between a board of directors, or an investor, and the CEO of a company. They have different levels of understanding of the company, despite both of them standing to gain significantly from attaining knowledge of it. This can give CEO’s an advantage when the interests of the CEO and that of the board, or investors clash.

When one mentions agency theory, information asymmetry is always implicit in the conversation. An agent that knows more than it’s principal has additional opportunities to go against the interests of their principal as the information asymmetry can allow them to do so not only without repercussions, but without the principal ever knowing their interests have been hampered (Johnson and So, 2018) (Dalton et al., 2007). The main method for attempting to combat this information asymmetry specifically between companies and their would-be investors is auditing (Adams, 1994). A neutral third party is to come in, be allowed insights into a company and then decide if, in their opinion, the company has wrongfully stated anything in their public financial statements. This has proven effective at reducing the usage of real earnings management in companies (Alhadab and Clacher, 2018) and does so at no direct cost to investors as the company foots the bill.

This paper, however, approaches the issue from another independent angle. To what extent can companies circumvent this information asymmetry by incentivizing the CEO to always work for the best interest of the firm. In this ideal scenario, while there is still the same level of information asymmetry, it is less relevant as the CEO with more information is now using that information to the betterment of their principal, and in turn themselves, through compensation.

2.4 Signalling Theory

Signalling theory states that when two parties are in a state of information asymmetry, any decision taken by the more informed party will signal to the less informed party, but signals can also take the typical form of public statements or other explicit information.

(Connelly et. al, 2010). A good example of implicit signals is an all-in in poker. When a player goes all-in, they are signalling that they are confident in their current hand, and this gives other players information as to what their hand could be, even without ever seeing the hand. Even if a player is bluffing the effect can be seen through other players reassessing the situation, and their assumptions about the other players hand, without ever seeing said hand.

The signaller is an insider who possesses information that would typically be valuable for an outsider, and the receiver is anyone who can benefit from said information (Conelly et al., 2010). In our circumstance the sender of the signal is the CEO and the receiver are the investors and would be investors. When a CEO utilizes earnings management to improve financial statements, they are signalling good performance. Signalling can also be negative and unintentional, a company being caught in a scandal will send negative signals to investors (Connelly et al., 2010).

(15)

14

2.5 Earnings

Earnings serve as an estimate of a firm’s income over a period, but because of the complexities of modern commerce and accounting it is always an estimate. Earnings are calculated as cash flows plus accruals (Hribar and Collins. 2002). The cash flows in turn are composed of every cash inflow from a period, minus the cash outflows during the same period. Either inflows or outflows can be manipulated by changing operations to alter the earnings published in the final financial statement.

2.6 Earnings Management

Earnings management is defined as: ‘‘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 practices.” (Healy and Wahlen 1999 p. 6)

Research has been done that attempts to contextualize earnings management in regard to accounting regulation. It defines three groups of motivations for earnings management.

Capital market, contracting, and regulatory. And attempts to explain the three groups.

(Healy and Wahlen, 1999)

Incentives to manipulate earnings can be placed into different groups (Healy and Wahlen, 1999):

Capital market motivations.

These are incentives based on influencing stock price in the short term.

Contracting motivations.

These are incentives based on contracts, such as bonuses for managers.

Regulatory motivations.

These are incentives that arise because of regulation, usually to circumvent negative effects of said regulation.

There have been attempts to explain and define the concept of earnings management (Schipper, 1989). Schipper argues that earnings management is not about altering one number, but rather is the manipulation of what she refers to as “information content”, the altering of the information the number will convey to the receiver. She further mentions that groups with high proficiency in finance, who have gains at stake with no contractual frictions serving as receivers are likely to undo earnings management, as it’s purpose will no longer be fulfilled since the receivers of the information can see through the misdirection this is supported by similar findings regarded sophisticated investors (Roychowdhury, 2006). It is worth mentioning that Schipper argues that at the time of writing there were no conclusive evidence as to whether or not earnings were being

(16)

15

managed. Since this paper was published in 1989 there has arisen plenty of evidence of earnings management (Roychowdhury, 2006) (Gunny, 2010).

The root causes of earnings management seem to be that people are heavily incentivised by market forces and upper management to meet certain expectations in the short term, along with incentives to raise stock prices (Dichev et al., 2013). These metrics are then more easily increased not by doing better at what they are supposed to measure, but instead by finding workarounds and loopholes. It stands to reason the usage of these methods then increase metrics, and cause markets to have even higher expectations for these metrics in the future, this can cause a vicious cycle in which people feel forced to resort to earnings management in order to be able to keep up with market expectations.

Another perspective of earnings management is that of short-termism, in which companies value short term gains over long term gains. While the time value of money states that money today is worth more than money tomorrow companies have been found to pursue short term performance to an inordinate degree (Laverty, 2004). This is partially explained by employee turnover often being quite high among executives, leading to these decision makers undervaluing long term projects since they might not still be working for the company by the time these long term projects start bearing fruit (Palley, 1997). This short-termism is another cause of earnings management, while the company as a whole wants strong long-term performance, all the decision makers want short term profits. This is a large cause of an agency problem as the incentives can be large. You could join a company, sink long term gains in order to boost numbers for a handful of years, leave while everything looks great for a new company that is impressed by your performance, rinse and repeat.

Earnings management can be done in one of two ways referred to as “Accruals earnings management” and “Real earnings management” (Healy and Wahlen, 1999). These two methods vary in their approach as earnings are made up of two parts, the first part being operating cash flows, and the second part being accruals (Healy and Wahlen, 1999).

Companies can choose two different methods to achieve similar results, when attempting to make their earnings look better. Manipulate the accruals or manipulate the operating cash flows. Of the two methods accruals earnings management is generally easier since it is made entirely in the accounting books, as opposed to the real earnings management approach requiring the firm to alter their business operations by increasing production, offering discounts or other methods (Roychowdhury, 2006).

2.7 Accrual earnings management

Accrual earnings management is earnings management utilizing only manipulation of accruals. Firms for instance change classifications of certain entries to hide them from financial statements or to change certain figures and or ratios. The operations of the actual company are unchanged, but the methods in which they report these operations are altered to accomplish certain external goals, like a higher stock price and increased investment.

(Healy and Wahlen 1999).

(17)

16

2.8 Real Earnings Management

Real earnings management is a form of earnings management that involves a company altering their operations to influence various entries on their financial statements through changing operating cash flows (Healy and Wahlen, 1999).

Earnings management is often used simply to attempt to reach certain benchmarks (Dichev et al., 2013). It then seems fair to assume that CEO’s, when struggling to meet short term benchmarks and targets to receive their bonuses would turn to earnings management in attempt to reach these benchmarks and targets in the short term to maximise their gains.

This could be worth even the risk of real earnings management hurting company value, like has been theorized (Roychowdhury, 2006). This is because it was discovered that employees favour the short term because of employee turnover rates, they might not be around to feel any long-term negative impacts of their decisions (Palley, 1997). There is also the possibility that next period will be strong enough to outweigh any losses in the long term incurred by real earnings management. In this situation the CEO would still gain their full bonus in this new period, making the utilization of real earnings management a compelling choice for CEO’s.

Throughout a study exploring the occurrence and nature of real earnings management Roychowdhury finds evidence that managers have utilized overproduction, price discounts and reduction of discretionary expenditures in an attempt to increase earnings for a period. (Roychowdhury, 2006)

Roychowdhury, like predicted by Schipper (1989), finds that this manipulation is less prevalent among sophisticated investors. Roychowdhury (2006) argues that this suggests that real earnings management fails to increase long term firm value, and also theorizes that real earnings management could have a negative effect on long term firm value as earnings are moved across periods at cost. Roychowdhury (2006) also argues that this fear of real earnings management negatively impacting long term performance leads to real earnings management being a wasted effort in the presence of sophisticated investors, which is why it is utilized less. Roychowdhury (2006) claims that managers utilize real earnings management rather than accrual earnings management because of an incentive split. While real earnings management can have greater costs for the company, accruals earnings management has a high private cost in the form auditing and other regulatory attention. In other words, accrual earnings management brings larger risk to the manager, whereas real earnings manipulation brings larger risk to the company (Roychowdhury, 2006). Roychowdhury (2006) creates methods to empirically indicate the usage of real earnings manipulation across larger samples, which we will use.

Another paper surveys financial executives on the main drivers of their behaviour regarding performance measurement and voluntary disclosure (Graham et al., 2005).

They find evidence that these executives would avoid positive net present value projects if it would mean falling short of quarterly goals. A large section of the executives would also trade away economic value for smoother earnings. The paper finds that the view of

(18)

17

the surveyed executives matches with those predicted by capital market motivations² from earnings management. The paper however only finds what the authors call modest evidence that debt, political cost and bonus plans has any effect on earnings management.

While writing a study Gunny (2010), initially had found that all forms of real earnings management has a significant negative impact on subsequent performance. The activities she classified as real earnings management were overproduction, reduction of discretionary expenses, altering SG&A spending, and altering timing of revenue recognition. In the final paper from 2010 however, Gunny finds that this association does not hold true if benchmarks are only barely met with the help of real earnings management (Gunny, 2010). Gunny theorizes that in this case real earnings management is not opportunistic, but rather managers utilizing real earnings management are attaining short term gains through their meeting of benchmarks pleasing investors that can be utilized to achieve long term gains that outweigh any cost of the real earnings management, and it is therefore in the investors’ best interest.

Gunny’s paper is interesting because she found that using real earnings management to meet benchmarks actually improves performance, but using it for other purposes negatively

impacts performance (Gunny, 2010). This second result could be explained by signalling theory, where a firm signals strong results, and can then utilize the gains of this improved perception of the company to offset any losses incurred by the earnings management.

This is supported by findings that that earnings management was positively associated with real earnings management only when benchmarks were met, and was negatively associated when they were not (Al-Shattarat, 2018). It seems that either the markets are expecting firms that engage in real earnings management to meet benchmarks, and get more hesitant when they do not, relative to firms who engage in less real earnings management. Alternatively failure to meet benchmarks with real earnings management is not hurting the company more by itself, but real earnings management has a cost to a company, and when not meeting benchmarks there are no gains to offset these costs, resulting in a larger net loss for the company.

There are three main ways to manipulate real earnings, defined by Roychowdhury (Roychowdhury, 2006):

Sales manipulation.

Sales manipulation can take several forms. Roychowdhury divides sales manipulation into two groups, price discounts, and lenient credit terms, while discounts and lenient credit terms are not always real earnings management, they can be classified as such when the lenience, or the discount, is considered excessive. Price discounts are applied at the end of a period to attempt to move sales from the coming period to the current one by offering unreasonably high limited time discounts. While this method increases earnings for the period, it will harm long term earnings as the sales are now less profitable than

2 In the paper they refer to it as “Stock market Motivations” but it holds the same meaning as the term “Capital market motivations” used in this paper.

(19)

18

they would have been without the discount. Lenient credit terms are similar, firms will offer unreasonably low interest rates at the end of periods to attempt to move sales from the coming period to the current, with similar results to the price discounts.

(Roychowdhury, 2006)

Reduction of discretionary expenditures.

The main discretionary expenses mentioned by Roychowdhury are R&D, advertising, and maintenance. As they are usually expensed at the time they are incurred, firms can reduce these to improve earnings for the period. Most commonly firms will focus first on expenditures that do not result in immediate gains, such as R&D, to offset any loss of earnings into future periods, as any lost earnings from foregone advancements will come much later, often several years into the future. (Roychowdhury, 2006)

Overproduction.

Earnings can paradoxically be increased by increasing spending. By buying unneeded amounts of inventory or material each unit sold will have a lower cost associated with it because of the overhead costs being spread over more units. Since purchased materials and inventory is an asset, same as cash, the purchase does not result in a recorded loss, but will result in a reduced COGS for the period. This means that when a sale is recorded, the cost that is recorded to produce the unit is reduced, resulting in a larger income on paper. And costs associated with the excessive purchase, such as storing the inventory, and any eventual discounts to get rid of inventory, are spread out over future periods, resulting in a net gain for the current period. (Roychowdhury, 2006)

Roychowdhury (2006) also states that excessive discounts, and overproduction will result in production costs that are abnormally high when compared to the dollar sales. And that a reduction of discretionary expenses will result in abnormally low discretionary expenses when compared to sales.

Roychowdhury (2006) identifies several other factors that the study claims affect the presence of real earnings management such as industry membership, stock of inventory and receivables, the presence of debt and finally growth opportunities.

Real earnings management has been theorized to possibly reduce firm value as earnings are moved across periods at cost (Roychowdhury, 2006). This has later been corroborated, with some reservations for real earnings management as a tool to meet benchmarks.

(Gunny, 2010).

Real earnings management among companies was a somewhat unpopular practice in the pre-SOX era, but rapidly increased after SOX when accruals earnings management was now harder to do (Cohen et al., 2008). The same trend held true when clawbacks were used to reduce accruals earnings management (Chan et al., 2015). When accruals earnings management went down, it caused real earnings management to go up. It seems that attempting to curtail accruals earnings management is only a band-aid solution.

A study has found that managers were willing to manipulate real activities to meet earnings targets, even at cost to the company. The same paper also found that financial executives were more willing to commit to real earnings management than accrual

(20)

19

earnings management, despite fears that real earnings management can harm a company’s value in ways that accruals earnings management will not. (Graham et al., 2005)

Research has found that cost of capital increases with real earnings management (Kim and Sohn, 2013). This confirms what might seem somewhat obvious: Real earnings management is not a secret to the markets as a whole, and they will change their interactions with firms that are engaging heavily in real earnings management to match this increased information asymmetry, and thus their risk as lenders or investors. This relationship is again, supported by findings that the presence of what was called

“sophisticated investors” reduced the occurrence of real earnings management (Roychowdhury, 2006). It seems that when firms know investors are aware of real earnings management occurring and its effects on financial statements, the perceived value of it is diminished. The company sees no long term value being created through real earnings management, instead it is simply being used as smoke and mirrors to obfuscate the truth to investors, and when it becomes clear investors can see through the mirage, companies stop.

2.9 Short-Termism

Companies should first and foremost focus on long term gains but utilize short term gains to ensure that those long-term gains ever bear fruit. However, companies seem to put an inordinate amount of effort into short term gains, to the point of forgoing long-term gains (Laverty, 2004). A paper also found that managers are influenced towards short-termism because of capital markets and performance measurement (Marginson and McAulay, 2008). And another has found that managers leaned towards short-termism because of a combination of employee turnover and risk-aversion (Palley, 1997). These are forces outside of the company itself that are pushing CEO’s to forego what is actually in the company’s best interest in order to maximise their personal utility. This phenomenon of companies seemingly tunnel-visioning on short term gains is referred to as “Short- termism” (Laverty, 2004).

These incentives to act for short term gains are heavily connected to earnings management as managers feel pressured to perform, and most clear performance metrics mainly measure the short term (Laverty, 2004), as numbers are much clearer as one can look over a timeframe of one quarter, one year, etc. to get a decent picture of performance, as opposed to attempting to measure future gains that might not be fulfilled for another 10- 15 years. Combined with the pressure of meeting certain benchmarks, CEOs are incentivised not only to focus on short term gains, but sometimes also to hamper long term gains in order to cut costs in the short term and meet benchmarks, commonly by cutting spending on research and development and or advertising (Laverty, 2004).

2.10 CEO Compensation

There are several different common methods of compensating CEOs, and commonly firms will apply a combination of methods in their compensation plan. Here we discuss some of the most common methods.

Cash salary.

(21)

20

The normal monthly paid cash salary, it is a previously decided monthly pay given out as cash. A cash salary is completely independent of performance outside of being let go from the company, seeing your future pay being reduced, or the hope of receiving a higher salary in future. Pay cuts are quite rare and thus a cash salary does very little to incentivise hard work and improved performance, other than working to keep your position. (Larcker and Tayan, 2012)

Bonuses.

Bonuses are applied to attempt to motivate the CEO in a way that a strict cash salary cannot do. A bonus is essentially an additional yearly lump sum, decided in advance that is given out if performance is deemed adequate. This performance can be measured empirically by looking at for instance profitability ratios or stock performance, or have the decision be made arbitrarily by the board (Larcker and Tayan, 2012). This method of compensation ideally has the benefit of motivating a worker to perform better in order to secure their additional compensation, but can also cause the CEO to attempt to secure the compensation through methods that do not increase company value, but does improve the ratios that the bonus is hinging on. It could also incentivise a CEO to favour short term gains to an excessive degree and forego long term gains for the company, since they are less likely to still be at the company when long term projects start breaking even, and until then the projects are just reducing their chances of securing a bonus. (Larcker and Tayan, 2012)

Stock options.

This is a method where CEOs are given the opportunity to buy stocks at a set price, usually at market price at the time the option is offered, and the opportunity to buy is spread over time. This should in theory reward long term growth, as it will take several years to be able to buy the stocks, and you want the stock value to be high at that time. In reality however, managers stand to gain from high risk decisions to increase stock price, and lose next to nothing if those risks don’t pay off as they can simply avoid buying the shares if prices plummet. (Larcker and Tayan, 2012)

Stock ownership.

The final method of CEO compensation we will discuss is stock ownership. Stock ownership differs from stock options in that the CEO is not given the option of buying stock at a set price but is instead just given stock in the company. This gives the CEO motivation to increase value over the long term, it helps align the CEO’s goals with the investors, as the CEO is now an investor themself. It is not a perfect solution, however.

The CEO can still attempt to push for short term gains, and sell when value peaks, before investors can realise this upswing in stock value is likely only very temporary. Or they can sell off stock before a coming decrease in value that the CEO is aware of because of their insider position. This gives them an unfair advantage compared to the average investor because of the information asymmetry between them. (Larcker and Tayan, 2012)

CEO compensation has been found to affect earnings management (Bergtresser and Philippon, 2005. A study has found that usage of accrual earnings management was more widespread in firms where the CEO was more heavily compensated by company stock and options. While these compensation offers were added to attempt to align the

(22)

21

incentives of CEO’s and the board they seem to also appear to have had the effect of allowing CEO’s to “game the system” where instead of simply trying to perform better, they would focus on meeting the key targets to increase company stock value and focus less on other aspects necessary for a company. This will always be a problem with any compensation plan because of the agency problem, there is no realistic way to measure performance well enough that it will only measure the performance the board wants, so the CEO could now be motivated to also do things the board does not want, because it’s in the measurement or vice versa, as the CEO avoids doing things that are beneficial for the company, but will negatively impact the measurements.

2.11 Hypothesis Development

The theories discussed in this chapter so far suggest that there are several different ways that are commonly used to compensate CEO’s (Larcker and Tayan, 2012), and according to agency theory these different forms of compensation are going to give the CEO different incentives as they attempt to maximise personal utility (Dalton, 2007).

Furthermore, CEOs are generally willing to utilize real earnings management if they think they stand to gain something from it, even if it comes at the cost of long-term firm value (Roychowdhury, 2006) (Graham, 2005). This leads us to formulate our first hypothesis.

Hypothesis 1:

CEO compensation affects the occurrence of real earnings management.

We also know that different ways of compensating CEOs should affect their incentives as the measurements that determine the requirements to maximise their compensation have now been altered. This change should result in the CEO changing their behaviour, accordingly, changing their behaviour and decision making to that perceived by the CEO to maximise personal gain for this new compensation plan. This leads us into hypothesis 2.

Hypothesis 2:

Different forms of CEO compensation will have varying effects on the occurrence of real earnings management.

These two hypotheses will be empirically tested. For hypothesis 1, if any of our identified methods of CEO compensation are related to the occurrence of real earnings management, we will accept the hypothesis as true. For hypothesis 2, as long as we can detect significant differences in the effects of our different forms of compensation, we will accept the hypothesis as true.

(23)

22

3. Scientific Method

3.1 Ontology

Ontology in the context of a scientific study determines the papers view on the nature of reality (Collis and Hussey, 2009. p. 59). There are two categories of ontology, objectivism and subjectivism, these two hold completely opposite assumptions in regard to the nature of reality. Note that we will only discuss ontology within the context of social studies, as that is the focus of our paper.

Objectivism

Objectivism in social studies states that a social entity exists externally to social actors.

For instance, one could argue that the role of a leader exists separately to person fulfilling that role, this would be an objectivist view. In essence the role has been created formally, and any interactions with other social actors around the role are irrelevant to the role’s existence and nature. This view is commonly used when the role is more important to the study than the decisions of the person in the role. For instance, a manager can be seen as simply a function of a company without much regard for the individual performing this function. The emphasis often lies on the structure of a firm, or other organization.

(Saunders et al., 2009 p. 110)

Subjectivism

Subjectivism takes the opposite approach and argues that any social role is constructed by social actors, and that these roles continue to be affected by the perceptions, actions of, and interactions with, the social actors who surround this role. If one would argue that a manager exists not because of formal reasons, such as the structure of a company, but rather because the people surrounding this manager has created the role of this manager through constant social interactions, this would be a subjectivist argument. (Saunders et al., 2009 p. 111)

For this paper we will utilize the objectivist approach. Our paper is centred around the effects of compensation on CEO’s, our main focuses, real earnings management and compensation are completely separate from social actors. Furthermore, our paper pays no heed to any social actors’ effect on the occurrence of real earnings management through interaction with the CEO. We are essentially assuming that every CEO holds perfect information of the effects of real earnings management and are basing their decision making solely of financial gain. Because of this, every CEO will therefore always make the same decisions. Because of these reasons we find that the objectivist approach is more suitable for the study.

3.2 Qualitative or Quantitative Study

Quantitative data is data that is very raw and unaffected by opinion. Metrics like weight and distance are quantitative measures. They represent a physical reality, even if the thing they measure is not necessarily physical. Qualitative data on the other hand is vaguer and more based on emotion or non quantifiable behaviours. Rating happiness on a scale from 1 to 10 is qualitative data. Most researcher will early on choose to make either a quantitative or qualitative study, although combinations are possible. (Saunders et al., 2009 p. 415)

(24)

23

For our paper we will be using solely quantitative data. This decision was made because quantitative data is easier to collect in larger batches. We can retrieve financial information on several thousand companies with relative ease, as opposed to attempting to interview CEO’s or other relevant people to get an equally large sample size.

Furthermore, most previous research we base or study on was made using quantitative data³ meaning that making ours use quantitative data will make results more easily comparable. Real earnings management is also a tool to influence numbers, that is to say, quantitative data, so the choice felt natural. However, this does mean that there are insights we might be missing regarding CEO behaviour that qualitative data could represent, however we feel the need of such data is outshined by the need of the qualitative data.

3.3 Epistemology

Epistemology regards what would constitute knowledge in context of your study. It is essentially used to determine where you draw the line as holding something to be true, and where you decide that your evidence is inconclusive. There are two main viewpoints in epistemology, a positivist view, and an interpretivist view. (Saunders et al., 2009 p.

112)

Positivism

Positivism is focused on observable reality. Less interested in subjective opinions and feelings, a positivist researcher will favour pure data that can be easily construed to create a theory, and dislikes having to draw it out from vague things like open ended questions.

The positivist approach is generally, but not always, better suited for quantitative studies as it puts a stronger emphasis on the raw data that quantitative studies provide. (Saunders et al., 2009 p. 113-114)

Interpretivism

Interpretivism is focused on more subjective opinions, and less hard facts. people of a interpretivist mindset would argue that human behaviour is too complex to be accurately represented by raw data. While set figures and numbers can be a good fit for natural sciences, it simply isn’t deep enough for the social sciences. Interpretivist researchers prefer to have open interviews and be allowed to “dig deeper” into the thoughts of a person being observed, favouring smaller sample sizes to be allowed to spend more time on each observation. Where positivism favours one correct answer, interpretivism favours personal interpretation of data. Interpretivism is often, but not necessarily, better suited to qualitative

studies, as they are centred around less raw data, and more interaction. (Saunders et al., 2009 p. 114)

For this study we have decided that a positivist approach would be the better choice.

While our study is focused on behaviour, we are focused on the assumption that financial

3 Roychowdhury, 2006. Gunny, 2010. Al-Shattarat et al., 2018 etc.

(25)

24

incentives do affect people's behaviour, and is thing affecting that behaviour. This assumption let’s us assume that we can measure human behaviour in numbers by observing the relation between various variables. Because of this we found that the more raw statistical approach lended itself to positivism more so than interpretivism, as interpretivism would place a larger focus on the underlying feelings and behaviours of individual CEO’s.

3.4 Deductive or Inductive Reasoning

There are two common types of analytical procedures, deductively based, and inductively based (Saunders et al., 2009 p.500). They take generally opposite approaches to building a study. Deductive tends to make a claim based on previous theory, and then attempt to discover if said claim holds true. Inductive approaches tend to first collect data, and then attempt to draw conclusions from the data. (Saunders et al., 2009 p.500)

Deductive

Deductive reasoning is based around pattern matching. At first, you create a framework based on previous theories and use those to predict a pattern of results. You then run tests and compare the actual outcomes to your predictions, and if they match you hold your claims to be true. (Saunders et al., 2009 p.500)

There are two common ways to accomplish this, one is by creating a set of dependant and independent variables. You then use your theoretical framework to predict how these variables are going to interact with each other. This is the approach we have decided to use for this paper. (Saunders et al., 2009 p.500)

This version is well suited towards our research as we are looking for one main conclusion.

This approach benefits us as it allows us to focus on if our claim holds true, and if it does not, only then do we start looking for alternative explanations (Saunders et al., 2009 p.500). This could however have the downside of giving our paper some tunnel-vision by not considering alternatives very heavily in the earlier stages of writing.

Another version of pattern matching is to have only independent variables. For this approach you will create several patterns that are mutually exclusive, if one is true, all others are false. You then run the tests, and discover if any of your patterns matched reality, and accept that as true. (Saunders et al., 2009 p.500)

The reason we decided against using this approach was because it felt less suited to our broader purpose. We are less concerned with how exactly what form of CEO compensation affects what forms of real earnings management, and more inclined to answer if CEO compensation affects real earnings management at all. This approach would have been better suited if we put a higher emphasis on predicting specific effects and might be better suited for further research based on this study. This approach could have improved our paper by offering more nuanced answers, but we felt that this nuance is less relevant until we can prove there is any correlation at all.

Inductive